United Nations
        Environment Programme
            United States
            Environmental Protection
            Agency
October 1986
&EPA
EFFECTS OF CHANGES IN STRATOSPHERIC
OZONE AND GLOBAL CLIMATE
Volume 3: Climate Change

-------
Library of Congress Cataloging - in - Publication Data
Effects of changes in stratospheric ozone and global climate.
Proceedings of a conference convened by the United Nations Environment
Programme and the U.S.  Environmental Protection Agency.
Contents: v. 1. Overview — v. 2. Stratospheric ozone — v. 3.  Climate
change. — v. 4. Sea level rise
1. Atmospheric ozone—Reduction—Congresses.  2. Stratosphere—Con-
gresses.   3. Global  temperature  changes—Congresses.   4.  Climatic
changes—Congresses.  5.  Sea level—Congresses.  6.  Greenhouse effect,
Atmospheric—Congresses. 7. Ultraviolet  radiation—Congresses.
I. Titus,  James G.   II. United States Environmental  Protection Agency.
III. United Nations Environment  Programme.

-------
EFFECTS OF CHANGES IN STRATOSPHERIC
         OZONE  AND GLOBAL CLIMATE
             Volume 3:  Climate Change
                          Edited by
                        James G. Titus
              U.S. Environmental Protection Agency
        This report represents the proceedings of the INTERNATIONAL CON-
        FERENCE ON HEALTH AND ENVIRONMENTAL EFFECTS OF
        OZONE MODIFICATION AND CLIMATE CHANGE sponsored by the
        United Nations Environment Programme and the  U.S. Environmental
        Protection Agency. The purpose of the conference was to make available
        the widest possible set of views. Accordingly, the views expressed herein
        are solely those of the authors and do not represent official positions of
        either agency.  Mention of trade names or commercial products does not
        constitute endorsement or recommendation for use.

-------
PREFACE
     This document  is  part  of a  four  volume  report  that examines the possible
consequences  of projected  changes  in  stratospheric  ozone  and global climate
resulting from emissions of chlorofluorocarbons, carbon dioxide,methane, and
other gases released by  human activities.   In June 1986,  the United Nations
Environment Programme  and the U.S. Environmental Protection Agency sponsored
an International Conference on the Health and Environmental Effects of Ozone
Modification   and  Climate  Change,  which was  attended  by   scientists  and
officials, representing  twenty-one countries from all areas  of  the world.

     This volume examines  the effects  of  the change  in  climate  that might
result  from  a  global  warming.     Volume  1   of the  proceedings  provides  an
overview of the issues as well as the introductory  remarks  and reactions from
top  officials of the  United  Nations  and the United States.   Volumes 2 and 4
focus on the  effects of  ozone depletion and sea level rise.

     This  report  does  not  present  the official  views  of either  the U.S.
Environmental Protection Agency or  the United Nations Environment Programme.

-------
TABLE  OF  CONTENTS
INTRODUCTION

Overview of the Effects of Changing the Atmosphere
     James G. Titus and Stephen R. Seidel	    3


FORESTRY, AGRICULTURE AND ENDANGERED SPECIES

Atmospheric Carbon Dioxide Change:  Agent of Future Forest
  Growth or Decline?
     Allen M. Solomon and Darrell C. West  	   23

Historical Changes in Forest Response  to Climatic Variations
  and Other Factors Deduced From Tree  Rings
     Harold C. Fritts 	»	   39

How Changed Weather Might Change American Agriculture
     Paul E. Waggoner	   59

Drought Policy Implications of C02-Induced Climatic Change  in the
  United States and Australia
     Donald A. Wilhite		   73

An Assessment of  the Potential Economic Impacts of
  Climate Change  in Oklahoma
     Ellen J. Cooter 	   89

Climatic Change — Implications for the Prairies
     R. B. Stewart 	  103

Potential Effects of Greenhouse Warming on Natural Communities
     Robert  L. Peters and Joan D. S. Darling  	  137

-------
WATER RESOURCES
The Effects of Climate Change on the Great Lakes
     Stewart J. Cohen	 163

Climatic Evolution and Variability in Dryland Regions:
  Applications of History to Future Climatic Change
     Sharon Nicholson 	 185

Response of Lake Levels to Climatic Change — Past, Present, and Future
     F. A. Street-Perrott, M. A. J. Guzkowska, I. M. Mason, and
     C. G. Rapley 	 211

Regional Water Resources and Global Climatic Change
     Peter H. Gleick 	 217

Hydrologic Consequences of Increases in Trace Gases and
  COp in the Atmosphere
     John R. Mather and Johannes Feddema 	 251
HEALTH
The Impact of Human-Induced Climatic Warming Upon Human Mortality:
  A New York City Case Study
     Laurence S. Kalkstein, Robert E. Davis, Jon A. Skindlov, and
     Kathleen M. Valimont 	 275
                                     vi

-------
INTRODUCTION

-------
Overview of the Effects of Changing the Atmosphere
James G. Titus and Stephen R. Seidel
Environmental Protection Agency
Washington, DC USA
INTRODUCTION

     Society  is  conducting  a global  experiment  on the earth's  atmosphere.
Human activities are  increasing the  worldwide  atmospheric  concentrations of
chlorofluorocarbons,  carbon  dioxide,  methane,  and several other  gases.    A
growing body  of  scientific evidence suggests that  if  these trends.continue,
stratospheric ozone may decline and global temperature  may  rise.   Because  the
ozone layer shields  the  earth's surface  from damaging ultraviolet  radiation
(UV) future depletion could increase  the  incidence of  skin cancer  and  other
diseases,  reduce crop yields, damage materials,  and  place additional stress on
aquatic plants   and animals.   A global warming from the  "greenhouse  effect"
could also threaten  human health,  crop yields,  property,  fish, and  wildlife.
Precipitation  and storm  patterns  could change,  and the level of  the  oceans
could eventually rise.

     To  improve the  world's understanding  of  these and  other  potential
implications  of  global atmospheric  changes,  the United Nations  Environment
Programme (UNEP) and  the  U.S.  Environmental Protection  Agency  (EPA)  sponsored
an  International Conference  on the Health and Environmental Effects of  Ozone
Modification  and Climate  Change during  the week of June  16-20,  1986.    The
conference brought  together over three hundred researchers  and policy  makers
from  approximately  twenty nations.   This  four-volume  report  presents  the
seventy-three  papers  that  were  delivered at  the  conference  by  over  eighty
speakers,  including  two U.S.  Senators, top officials from  UNEP and EPA, some
of  the  leading scientists  investigating  the  implications  of  atmospheric
change, and representatives  from industry and environmental  groups.  Volume  1
presents a series of overview papers describing  each  of the major areas of
research on the  effects of atmospheric change,  as well as policy  assessments
of  these   issues  by  well-known  leaders   in  government,   industry,  and  the
environmental  community.   Volumes  2,  3,  and 4 present the more  specialized
papers  on  the  impacts of ozone modification,  climate change, and  sea  level
rise, respectively,  and  provide some  of  the  latest research  in  these areas.
This paper summarizes the entire four-volume report.

-------
OZONE MODIFICATION

Atmospheric Processes

     The ozone  in  the  upper part  of the   atmosphere—known as  the strato-
sphere—is  created  by  ultraviolet radiation.   Oxygen  (02)  is  continuously
converted  to  ozone (Oo)  and back  to Op by numerous  photochemical reactions
that  take  place  in  the  stratosphere,   as  Stordal  and  Isaksen   (Volume  1)
describe.   Chlorofluorocarbons  and other gases released  by  human activities
could  alter  the   current  balance of  creative  and  destructive  processes.
Because  CFCs  are  very  stable  compounds, they  do  not break up in the lower
atmosphere  (known as  the troposphere).   Instead,  they  slowly migrate to the
stratosphere,   where  ultraviolet   radiation  breaks  them   down,  releasing
chlorine.

     Chlorine acts as a  catalyst to destroy  ozone;  it promotes reactions that
destroy ozone without being  consumed.  A chlorine  (Cl) atom reacts with ozone
(Oo) to  form  CIO  and 02-   The CIO later reacts with another 0,  to form two
molecules  of  02,  which  releases  the   chlorine  atom.   Thus,  two molecules of
ozone are converted to three molecules of ordinary oxygen,  and the chlorine is
once again free to start the process.   A  single  chlorine  atom  can destroy
thousands  of  ozone molecules.   Eventually,  it returns to  the  troposphere,
where it is rained out as hydrochloric acid.

     Stordal and Isaksen  point out  that  CFCs are not the only gas released by
human   activities   that   might   alter   the   ozone   balance.     Increasing
concentrations of methane in the  troposphere increase the  water  vapor in the
stratosphere,  which helps create  ozone.   Carbon dioxide and other greenhouse
gases  (discussed  below)  warm  the   earth's   surface  but  cool   the  upper
atmosphere; cooler stratospheric temperatures slow the process of ozone deple-
tion.  Nitrous oxide (NpO) reacts with both chlorine and ozone.

     Stordal  and  Isakson  present   results  of  possible  ozone depletion  over
time, using  their  two-dimensional  atmospheric-chemistry model.    Unlike  one-
dimensional models which  provide changes  in  ozone  in  the global  average,  this
model calculates  changes for specific latitutdes  and  seasons.    The results
show that if concentrations of the relevant trace gases grow at recent levels,
global  average  ozone  depletion  by  2030 would  be  6.5  percent.    However,
countries in the higher latitudes (60°N)  would experience 16 percent depletion
during spring.   Even  in  the  case of constant CFC  emissions,  where global
average depletion  would  be  2  percent by 2030,  average  depletion  would  be 8
percent in the high northern latitudes.

     Watson (Volume 1)  presents evidence  that ozone has been changing recently
more than atmospheric models had predicted.  As Plate 1  shows, the ozone over
Antarctica  during  the  month  of  October appears   to  have  declined  over  40
percent in  the  last six  to  eight  years.  Watson also discusses  observations
from ozone monitors that suggest a 2 to 3 percent worldwide reduction in ozone
in the upper portion of the stratosphere   (thirty to forty kilometers above the
surface),  which is  consistent with model predictions.  Finally,  he presents
preliminary data  showing a  small  decrease  since  1978  in the total (column)
ozone worldwide.  However,  he strongly emphasizes that  the  data  have not yet
been fully reviewed and  that it  is   not possible  to conclusively attribute
observed ozone  depletion to  the  gases  released by human activities.   While

-------
there  are  several  hypotheses  to  explain  why  ozone  concentrations  have
declined,  none  have  been  adequately   established;   nor   did  any  of  the
atmospheric models predict the measured loss of ozone over Antarctica.

Ultraviolet Radiation

     Many  of  the  chemical  reactions  investigated  by  atmospheric  scientists
take  place only  in  the  stratosphere  because they  are caused  by  types  of
radiation  only  found in  the  upper  atmosphere.    As  Frederick  (Volume  1)
explains, the sun  emits radiation  over a broad range of wavelengths, to which
the human eye responds in the  region from approximately 400 to 700 nanometers
(nm).  Wavelengths  from 320  to MOO nm are known as UV-A; wavelengths from 280
to 320 nm are called UV-B,  and wavelengths  from  200 to 280  nm  are known as
UV-C.

     Frederick explains why  attention has primarily  focused on the UV-B part
of the spectrum.   The atmosphere  absorbs  virtually  all UV-C,  and is expected
to continue to do  so  under  all foreseeable circumstances.  On  the other hand,
UV-A  is  not  absorbed  by ozone.1.   By contrast,  UV-B is partially absorbed by
ozone, and future depletion would  reduce the effectiveness of  this shield.

     We  now  examine  the  potential  implications  of such  changes  on human
health, plants, aquatic organisms, materials, and air pollution.

Effects on Human Health

     The  evidence  suggests  that   solar  ultraviolet  radiation  induces skin
cancer,  cataracts,  suppression of  the  human  immune  response  system,  and
(indirectly  through  immunosuppression)  the  development  of   some  cutaneous
infections, such as herpes.   Emmett (Volume 1) discusses the  absorption of UV
radiation  by  human tissue and the mechanisms by which damage and repair may
occur.

     Emmett also  examines  UV  radiation as  the cause of aging  of the skin and
both  basal  and squamous skin  cancers.   In reviewing the role of UV  radiation
in  melanoma  (the  most  frequently  fatal  skin cancer),  he states  that some
evidence  suggests  this  link,  but that currently there  is no acceptable animal
model that can be  used to explore  or validate  this relationship.  He  concludes
that  future  studies  must  focus  on three  major  factors—exposure  to solar
radiation,  individual susceptibility, and  personal  behavior.   Waxier  (Volume
1) presents evidence  of a link between UV-B  exposure  and  cataracts.

      Volume  2 presents  specific research  results and provides more  detail on
many  of  the aspects  covered in this volume.  Scotto presents  epidemiological
evidence  linking  solar  radiation with skin cancers,  other than melanoma.  His
analysis  suggests  that Caucasians  in the  United States  have a  12  to  30  percent
chance of developing  these  cancers within their lifetimes, even without ozone
depletion.    Armstrong  examines the role of  UV-B  exposure to  melanoma in  a
study  of  511  matched  melanoma  patients   and  control  subjects  in  Western
Australia.    He  shows that  "intermittent  exposure"  to  sunlight  was  closely
associated with  this  type of cancer.
   However,  $2  anc*  ^2  reflect  some  UV-A  back  to  space.

-------
     In a paper examining nonmelanoma skin cancer in Kuwait, Kollias and Baqer
(Volume  2)  show  that  despite  the  presence of  protective pigmentation,  75
percent of cancers occur on the 10 percent of the skin exposed to sunlight.  A
second paper on skin cancer presents experimental evidence suggesting that the
mechanism  by  which  skin  cancer  could   occur   involves  disruption  of  the
cytoskeleton  from  exposure  to  UV-A  and  UV-B  light  (Zamansky  and  Chow,
Volume 2).

     The  pathways  by  which  suppression  of  the  immune   response  might  be
triggered are  explored  in  papers by  DeFabo and Noonan,  Daynes et  al.,  and
Elmets  et  al (all  Volume 2),   Davies and  Forbes  (Volume 2) show  that mice
exposed to UV-B radiation had a decrease  in lifetime that was proportional to
the quantity of radiation and  not  directly related to the  incidence of skin
cancer.

     Possible   implications   of  immune   suppression  of   diseases   and  the
mechanisms by which it occurs are still uncertain.  However, several papers in
Volume 2 suggest that in addition to skin cancer and contact hypersensitivity,
diseases influenced  by  UV-B induced  immune  suppression  include  leishmaniasis
and herpes  infections.   Fisher  et.  al   (Volume 2}  show  that  at  least  one
sunscreen effectively  protects  mice  exposed to UV-B radiation  from sunburn;
but it  does  not stop the immune suppression from  interfering with  a contact
hypersensitivity (allergic)  reaction.

Effects on Plants

     The effects of increased exposure  to  UV-B  radiation  on plants has been a
primary area  of research for nearly a decade.   Teramura  (Volume  1)  reports
that of the two hundred plants tested for their sensitivity to UV-B radiation,
over two-thirds  reacted adversely;  peas,  beans,  squash,  melons,  and cabbage
appear  to  be the  most  sensitive.    Given the complexities  in  this area  of
research, he warns  that  these  results  may be misleading.   For  example, most
experiments have used growth chambers.   Studies  of plants  in the  field have
shown them to be less sensitive  to UV-B.  Moreover,  different cultivars of the
same  plant  have  shown  very  different   degrees   of  sensitivity   to  UV-B
radiation.    Finally,  Teramura  suggests that potential effects  from multiple
stresses (e.g., UV-B  exposure plus  water  stress or  mineral  deficiency)  could
substantially alter a plant's response to changes in UV-B  alone.

     In Volume  2,  Teramura  draws extensively  from the  results of  his five
years of field  tests  on soybeans.  His data show that a  25 percent  depletion
in ozone could  result in a  20  to 25  percent  reduction  in  soybean  yield  and
adverse impacts on the quality  of that  yield.   Because  soybeans  are the fifth
largest  crop  in  the  world,   a  reduction  in  yields   could  have  serious
consequences for world food supplies.   However, some soybean cultivars  appear
to be less susceptible  to UV-B  radiation, which  suggests  that selective crop
breeding might reduce future losses,  if it does not increase vulnerability to
other environmental stresses.

     BJorn (Volume  2)  examines  the mechanisms  by which plant damage occurs.
His research relates  specific wavelengths with those aspects  of plant  growth
that might be susceptible,  including the destruction of chloroplast,  DMA,  or
enzymes necessary for photosynthesis.

-------
Aquatic Organisms

     Aquatic plants would also  be  adversely affected by increased ultraviolet
radiation.  Worrest (Volume 1) points out that most of these plants, which are
drifters  (phytoplankton),  spend much  of their  time  near the surface  of the
water   (the   euphotic  zone)   and  are  therefore   exposed   to  ultraviolet
radiation.   A reduction  in  their  productivities  would be  important  because
these  plants  directly and  indirectly  provide  the  food for almost  all fish.
Furthermore,  the  larvae  of  many  fish found  in the  euphotic zone would  be
directly affected, including crabs, shrimp, and anchovies.  Worrest points out
that fish account for  18  percent  of the animal protein that people around the
world consume, and 40 percent of the protein consumed in Asia.

     An  important  question  is  the extent to which current  UV-B levels are a
constraint on aquatic  organisms.   Calkins  and  Blakefield  (Volume 2) conclude
that some  species are already  exposed to as much UV-B as  they  can tolerate.
Thomson  (Volume  2)  shows that  a   10 percent decrease in  ozone could increase
the number of abnormal larvae as much as  18 percent.  In a study of anchovies,
a 20 percent  increase in UV-B  radiation  over  a 15-day period caused the loss
of all the larvae within a 10-meter mixed layer  in April and August,

     Many other factors could affect  the magnitude of the impacts on specific
species,  ecosystems,  and  the   food  chain.   An important mechanism by which
species could adapt to higher UV-B incidence would be to reduce  their exposure
by moving  further away from  the  water's surface during  certain times of  the
day or year  when exposure is greatest.   Haeder (Volume 2) suggests*,- however,
that for certain species such avoidance may be  impaired by UV-B  radiation.

     Even  for those  organisms  that  could move  to  avoid  exposure, unwanted
consequences may result.  Calkins  and  Blakefield present model results  showing
that movement by phytoplankton  away from  sunlight  to reduce exposure  to a  10
percent  increase  in  UV-B  would   result  in a  2.5  to  5  percent  decrease  in
exposure  to  the  photosynthetically  active radiation  on which their growth
depends.   Increased  movement  requires  additional energy  consumption, while
changes  in location may affect  the availability  of food for  zooplankton, which
could  cause other changes in shifts in  the  aquatic food chain.

     To  a certain extent, losses  within  a particular species of plankton  may
be compensated by gains in other species.   Although  it  is possible  that no  net
change in productivity will occur, questions  arise concerning  the  ecological
impacts  on  species diversity  and  community  composition  (Kelly,   Volume  1).
Reductions  in diversity may make  populations more  susceptible  to changes  in
water  temperatures, nutrient availability,  diseases,  or pollution.   Changes in
community  composition  could  alter  the protein content, dry weight,  or  overall
food value of the  initial stages of the aquatic  food  chain.

Polymer  Degradation and Urban Smog

     Current  sunlight can cause paints to fade, transparent window  glazing to
yellow,  and  polymer  automobile  roofs to  become  chalky.   These changes  are
likely to occur more  in  places closer to the equator where UV-B radiation is
greater.    They  are   all  examples of degradation  that  could  accelerate  if
depletion of the ozone layer occurs.   Andrady and Horst (Volume 2) present a
case  study of  the potential magnitude of loss due  to increased exposure  to

-------
UV-B radiation on polyvinyl chloride  (PVC).   This chemical is used in outdoor
applications  where  exposure   to  solar   radiation  occurs  over  a  prolonged
period.   It is also  used  in  the  construction  industry in siding  and window
frames and as a roofing membrane.

     To analyze  the potential  economic  impact of  future ozone  depletion on
PVC,  the  authors assumed  that  the future  service  life of polymers  would be
maintained by increasing the quantity of light  stabilizers {titanuim dioxide)
used  in  the product.   As  a result,  the costs associated  with  increased UV-B
radiation  would  be  roughly  equal  to  the  costs  of  increased  stablizers.
Preliminary  results  show  that for  a  26  percent  depletion  by  2075,  the
undiscounted costs would be $4.7 billion (1984 dollars).

     Increased penetration  of UV-B radiation to the earth's surface could play
an  important  role  in  the  formation  of  ground level  oxidants  (smog).   UV-B
affects  smog  formation  through the  photolysis  of formaldehyde,  from which
radicals  are  the  main  source  for   deriving  chain  reactions  that  generate
photochemical smog.   Whitten  and  Gery  (Volume  2)   analyze  the  relationship
between UV-B, smog, and warmer  temperatures.   The results of this preliminary
study of Nashville,  Philadelphia,  and Los Angeles  show that  large depletions
in stratospheric ozone and increases  in  temperature would increase smog by as
much as 50  percent.   In addition, because  oxidants would form  earlier in the
day  and closer  to population  centers  (where  emissions  occur),  risks  from
exposure could  increase  by an  even  higher percentage  increase.   Whitten and
Gery also report a  sensitive relationship  between UV-B and hydrogen peroxide,
an oxidant and precursor to acid rain.

CLIMATE CHANGE

The Greenhouse Effect

     Concern about a possible global warming focuses largely on the same gases
that may  modify the  stratospheric ozone:  carbon dioxide, methane,  CFCs,  and
nitrous oxide.  The report of  a recent conference convened by UNEP, the World
Meteorological  Organization,   and   the  International   Council  of  Scientific
Unions  concluded  that  if  current trends  in  the  emissions  of  these gases
continue,  the  earth  could  warm a few  degrees  (C)  in the next  fifty years
(Villach  1985).   In the next century,  the planet could warm as  much as five
degrees (NAS  1983),  which  would leave  the planet warmer than at any time in
the last two million years.

     A planet's temperature is  determined  primarily by the amount of sunlight
it receives, the  amount  of sunlight  it  reflects, and  the extent  to which the
atmosphere  retains heat.   When  sunlight  strikes the earth,  it warms  the
surface, which then reradiates the heat as infrared radiation.  However, water
vapor, C02, and other gases in the atmosphere absorb some of the energy rather
than allowing it  to pass undeterred  through the atmosphere to space.  Because
the atmosphere  traps  heat  and  warms  the  earth  in a manner somewhat analogous
to the glass panels of a greenhouse,  this phenomenon  is commonly known as the
"greenhouse effect."  Without  the  greenhouse effect of the gases  that occur
naturally in the atmosphere,  the earth would be approximately 33°C colder than
it is currently (Hansen et al.  1984).

-------
     In  recent  decades,  the  concentrations  of greenhouse  gases  have  been
increasing.  Since the  beginning  of the industrial  revolution, the combustion
of  fossil fuels,  deforestation,   and  a  few  other  activities have released
enough C02  to  raise atmospheric  concentrations  by  20 percent; concentrations
have risen  8 percent  since 1958  (Keeling,  Bacastow,  and Whorf  1982).   More
recently, Ramanathan  et al. (1985)  examined the greenhouse  gases  other  than
C02 (such as methane, CFCs, and nitrous oxide), and concluded that these other
gases are likely  to  double the  warming  caused by  C02  alone.    Using  these
results,  the Villach Conference estimated  that an "effective doubling" of C0?
is likely by 2030.2

     Hansen et  al.  (Volume 1) and Manabe & Wetherald  (Volume  1)  present the
results  that their  climate models  predict for an  effective  doubling  of  COp.
Both models  consider a  number of  "climatic feedbacks" that  could  alter the
warming  that would directly result from C02 and other gases released by human
activities.  Warmer temperatures  would allow  the  atmosphere to  retain  more
water vapor, which  is  also a  greenhouse gas,  thereby resulting in additional
warming.   Ice  and  snow  cover would  retreat, causing  sunlight  that  is now
reflected by these bright surfaces to  be absorbed instead, causing additional
warming.  Finally, a change in cloud cover might result, which could increase
or decrease  the projected  warming.   Although the  two models differ  in  many
ways, both conclude that  an effective  doubling of greenhouse gases would warm
the earth's surface between two and four degrees (C).

     Hansen et al. project  the doubling to occur between 2020 and ^2060.  They
also  provide  estimates  of  the   implications  of  temperature  changes  for
Washington, D.C.,  and  seven  other  U.S.  cities  for  the  middle  of  the  next
century.   For  example, Washington  would have  12 and 85 days per  year above
38°C (100°F) and 32°C  (90°F),  respectively,  compared  with 1 and 35 days above
those levels today.  While evenings in which the thermometer fails to go below
27°C (800F)  occur  less than once per year  today  in that  city,  they project
that such evenings  would occur  19 times per  year.   (See Plates  2  and 3 for
worldwide maps  of historical and  projected temperature changes.)

Water Resources

     Manabe  and Wetherald  (Volume 1)  focus  on  the  potential  changes  in
precipitation patterns  that might  result  from the  greenhouse warming.   They
project  substantial increases  in  summer dryness at the middle  latitudes  that
currently support most of the  world's  agriculture.   Their model  also projects
increased rainfall for late winter.

     Beran (Volume  1)  reviews the literature on  the hydrological  and  water
resource  impacts  of climate  change.   He expresses  some surprise  that  only
twenty-one papers could  be  found  that address future water  resource impacts.
One  of  the problems,  he notes,   is  that  there   is  a  better  scientific
understanding of  how global average  temperatures and rainfall might  change,
than for  the changes  that specific regions  may  experience.   Nevertheless,  he
   Studies on the greenhouse effect generally  discuss  the  impacts of a carbon
   dioxide doubling.   By  "effective  doubling" we refer  to  any combination of
   increases in  concentrations of  the various  gases that  causes a  warming
   equal to the warming of a doubling of carbon dioxide alone.

-------
demonstrates  that  useful  information  can  be  extracted  by  studying  the
implications of particular scenarios.

     Nicholson (Volume  3)  shows how historical changes  in water availability
have caused  problems  for society  in the past.   The best  lesson  of climatic
history,  she writes,  "is  that agricultural  and  economic  systems must  be
flexible enough to adapt to changing conditions and,  in the face of potential
water  scarcity,   systems  must  be  designed  that  require  minimum  use  of
resources."  Wilhite (Volume 3) examines drought policies in Australia and the
United  States,  concluding  that  the  lack  of  national  drought plans  could
substantially impair the ability of these two nations to successfully adapt to
hydrologic changes resulting from the greenhouse warming.

     Cohen  (Volume 3)  examines  the  potential  implications  of  the  global
warming for  water levels  in  the  Great Lakes that  separate Canada  from the
United States.  Using results  from the models of both Hansen et al. and Manabe
& Wetherald,  he  concludes that  lake levels could drop  10 to 30 centimeters.
This drop would significantly  reduce the capacity of ocean-going vessels that
enter the Great Lakes.   On the  other  hand,  such  a drop  might  be  viewed as a
benefit by the owners of critically eroding property whose homes are currently
threatened by historically high lake levels.  Street-Perrott et al. (Volume 3)
discuss the  historic  impacts  of changes  in climate  on  the levels  of lakes in
North America, South America,  Australia, and Africa.

     Gleick  (Volume  3) uses  scenarios  from  the  Hansen  et al. and  Manabe &
Wetherald models  (as well  as a  third developed  by  the  National  Center for
Atmospheric  Research) to  drive a water-balance model of the Sacramento Basin
in California.  He finds  that reductions in runoff could occur even in months
where precipitation increases substantially, because of the increased rates of
evaporation  that  take  place at higher  temperatures.  He also points out that
the models  predict that changes  in  monthly runoff patterns  will  be far more
dramatic  than  changes  in annual averages.   For  seven  of ten scenarios, soil
moisture would be  reduced every  month of the  year; for  the other three cases,
slight increases  in moisture  are projected  for winter months.  Mather (Volume
3)  conducts  a  detailed  analysis  for  southern  Texas   and  northern  Mexico;
examines  in  less  detail twelve regions  around the world; and projects shifts
in global vegetation zones.

Agriculture and Forestry

     The  greenhouse  warming  could  affect  agriculture   by  altering  water
availability,  length of  growing  season,  and the  number  of  extremely  hot
days.    Increased C02  concentrations   could  also  have  two  direct  impacts
unrelated to climate change:  At least  the laboratory, plants grow faster (the
C02 fertilization effect) and retain moisture more efficiently.  The extent to
which  these  beneficial  effects offset  the  impacts of climate  change will
depend  on the extent to which  global warming is  caused  by C02 as opposed to
other greenhouse gases, which do not have these positive  impacts.

     Parry (Volume 1) provides an overview of  the  potential  impacts of climate
change  on agriculture  and  forestry.   He points  out that commercial farmers
plan according to the average year,  while family and subsistence farmers must
ensure  that even  in  the  worst years  they can  make ends  meet.    Thus,   the
commercial farmer would be concerned about the impact of  future  climate  change


                                       10

-------
on average conditions and average yields, while farmers at the margin would be
most  concerned  with  changes  in  the probability  of  (for  example) a  severe
drought that causes  complete crop failure.  Parry  notes  that the probability
of  two or  more  anomalous  years  in a  row  could  create  disproportionately
greater problems  for agriculture.   For  example,  a  persistent drought  in the
U.S. Great Plains from  1932  to 1937  contributed to about two hundred thousand
farm bankruptcies.

     Parry discusses  a  number of  historical  changes  in  climate.   The  Little
Ice  Age  in  western  Europe  (1500-1800  A.D.)  resulted in the abandonment of
about half the  farms in Norway, an  end  to cultivation of cereals in Iceland,
and some farmland in Scotland being permanently covered with snow.  Concerning
the late medieval cooling {1250-1500) he writes:  "The failure to adapt to the
changing circumstances  is believed to explain  much of the Norse decline.  The
Norse continued to emphasize stock-raising in  the  face of reduced capacity of
the already  limited  pastures.  The option of  exploiting the rich seas around
them, as the Inuit  (Eskimos) successfully  did, was not taken up  ... This is
an  extreme example  of  how  governments  can   fail  to identify  and implement
appropriate policies of response."   It  also suggests  that effective responses
can reduce damages from climate change.

     The  paper  reviews  a  number of studies   that  project  impacts of climate
change on  agriculture.   "Warming appears  to be detrimental to cereals in the
core  wheat-growing  areas of North America  and Europe."  If no  precipitation
changes take place,  a one-degree warming would decrease yields 1  to 9 percent
while  a  two-degree  (C)  warming would decrease yields 3 to  17 percent.   Parry
also discusses how particular  crop zones might shift.  A  doubling of C02  would
substantially expand the wheat-growing  area  in Canada-due  to  higher winter
temperatures and  increased rainfall.  In Mexico, however,  temperature stresses
would  increase, thereby  reducing yields.

     A  number  of studies  have been conducted using  the  models  of Hansen et
al.,  Manabe  &  Wetherald,  and others.   Although  these  projections cannot be
viewed as  reliable forecasts,  they do provide  consistent  scenarios  that can be
useful  for examining vulnerability  to  climate change.   Parry indicates  that
investigations  of Canada,  Finland,  and  the northern  USSR using the model by
Hansen et  al. show  reduced yields of spring-sown crops such as wheat, barley,
and oats,  due  to   increased  moisture   stress early  in  the  growing period.
However,  switching   to  winter wheat or  winter rye  might reduce this stress.
Parry  goes on  to outline numerous  measures  by which farmers  might adapt to
projected  climate change.

     Waggoner  (Volume 3) points out that  the  global warming would not affect
plants  uniformly.   Some  are more  drought-resistant  than  others,  and  some
respond  to higher COp concentrations more vigorously  than others.  Co plants,
such  as wheat, respond to  increased COg  more than Cjj plants  such as maize.
Thus,  the  CO-,  fertilization effect would not  help  the farmer growing C^  crops
accompanied  By  Co weeds,  Waggoner also examines the  impact of  future climate
change on average  crop yields  and  pests, and the probability  of successive
drought  years.   He  concludes  that  although projections of  future  changes are
useful,  historical  evidence suggests that surprises may  be  in store, and that
"agricultural  scientists will be expected to  aid  rather than watch mankind's
adaptation to an  inexorable  increase in  COp and its greenhouse effect."
                                       11

-------
     The  impact  of  future  climate  change  on  yields  for  spring wheat  in
Saskatchewan,  Canada,  is  the subject  of  the  paper  by  Stewart  (Volume  3).
Using the output  from the  Hansen  et  al.  model (Volume 1), which projects that
the  effective  doubling   of  carbon   dioxide  would   increase  average  annual
temperatures  in  that region  by  4.7°C, he  estimates that  the  growing  season
would start  two  or  three  weeks  earlier  and  end  three  or  four  weeks  later.
Although average  precipitation during the  growing season  would  increase,  he
also finds  that the area would become more prone to drought.  The impact of
climatic change would be to reduce yields 16 to 26 percent.  Stewart estimates
that the fertilization effect of  a C02 doubling would reduce the losses to 6
to  15 percent.   Cooter  (Volume  3) examines the economic  impact  of projected
climate change  on the economy of Oklahoma,  concluding  that the  Gross State
Product would decline 75 to 300 million  dollars.   (The  state's gross product
in  1985 was approximately 50 billion dollars.)

     Fritts  (Volume 3)  examines   tree  rings to  assess  how past  changes  in
climate have  affected forests, and  concludes  that tree  rings  are useful  for
estimating past changes  in climate.   Solomon and  West (Volume  3) discuss  the
results of their  efforts to model  the future  impacts.  Considering the impact
of climate change caused by doubled C02 without the fertilization effect, they
find that  "biomass  (for  boreal forests)  declines for 50-75 years as warming
kills off large boreal forest  species, before new northern hardwoods can grow
into the plot."

     "Warming at  the transition  site  causes  an  almost immediate  response  in
declining biomass from dieback of mature  trees,  and in decline of tree mass as
large trees  die  and are  temporarily replaced  by  small  young  trees,"  they
write.   "The deciduous  forest site  .  .  . results in permanent loss of dense
forest.    One  might expect  the  eventual appearance  of  subtropical  forests
similar to  those  in  Florida  today,  but  the  real difficulty is  the moisture
balance (which  is)   more  similar   to  those  of treeless  Texas today, than  to
those  of  southern   Florida,"    Solomon  and  West  go  on  to  show how  the
fertilization effect from increased  concentrations  of C02  could  offset part
but not all of the drop in forest productivity.

Sea Level Rise

     One of the most widely recognized consequences  of a global warming would
be  a rise  in sea level.   As  Titus (Volume 1) notes,  global temperatures  and
sea level  have fluctuated  over  periods   of one  hundred  thousand  years,  with
temperatures during  ice  ages  being  three to five degrees  (C) lower and  sea
level over  one  hundred  meters   lower  than  today.    By  contrast, the  last
interglacial period  (one hundred   thousand  years  ago) was  one  or two degrees
warmer than today, and sea level was five to seven meters higher.

     The projected global warming could raise sea level by heating and thereby
expanding  ocean  water,   melting   mountain  glaciers,  and  by  causing  polar
glaciers  in  Greenland  and Antarctica  to  melt  and  possibly  slide  into  the
oceans.     Thomas  (Volume  4) (presents   new  calculations  of  the  possible
contribution of Antarctica  and 'combines  them with previous estimates  for  the
other sources,  projecting that a worldwide  rise  in sea  level of  90  to  170
centimeters  by  the  year 2100  with 110 centimeters most  likely.   However,  he
also estimates  that  if  the global warming  is substantially delayed, the rise
in  sea  level could  be  cut in half.   Such a  delay might  result  either from


                                      12

-------
actions to  curtail  emissions or  from  the thermal lag  induced  by  the oceans'
ability to absorb heat.

     On the other  hand,  Thomas  also  estimates  that  if a  warming of  four
degrees results  from a  C02  doubling  (which the  model  of  Hansen et  al.
projects)  and concentrations continue to grow after 2050, the rise could be as
great as 2.3 meters.   He also notes that  an irreversible deglaciation  of the
West Antarctic  Ice  Sheet might begin  in the next century, which  would raise
sea level  another six meters in the following centuries.

     Titus  (Volume  1) notes  that  these  projections  imply that sea level could
rise 30 centimeters by 2025,  in addition to local subsidence  trends that have
been important  in Taipei, Taiwan;  Venice,  Italy; the  Mile Delta,  Egypt; and
most of the Atlantic and Gulf Coasts of the United States.  The projected rise
in  sea level  would  inundate  low-lying  areas,   destroy  coastal marshes and
swamps,  erode  shorelines,   exacerbate   coastal   flooding,  and increase  the
salinity of rivers, bays, and aquifers.

     Bruun  (Volume  4) argues that  with a combination  of coastal  engineering
and  sound   planning,  society can meet  the challenge  of a  rising sea.   He
discusses  a number  of   engineering  options,  including  dikes (levees)  and
seawalls,  and  adding sand  to recreational  beaches  that are  eroding,  with a
section on  the battle that  the  Dutch  have  fought  with  the  sea  for over one
thousand years.   Goemans (Volume  U)  describes  the  current approach  of the
Dutch  for  defending  the  shoreline, and estimates  that the cost  of raising
their  dikes for a one meter  rise in sea level would  be 10 billioi* guilders,
which  is less  than  0.05 percent  of  their  Gross  National Product for a single
year.

     Goemans concludes that there is no need to anticipate such a rise because
they could  keep up with  it.   However,  he  is more  concerned  by the two-meter
scenario: "Almost  immediately after  detection, actions would  be required.  It
is  not at  all  certain that decision-makers  act that  fast.  .  .  .  The present
flood  protection  strategy came about only  after  the  tragic disaster of  1953.
When nobody can  remember a specific disaster,   it is  extremely difficult to
obtain  consensus  on countermeasures."   For  his own  country,  Goemans sees one
positive  impact:    Referring to  the unique  experience  pf Dutch  engineering
firms  in the battle with  the sea, he suggests that "a rising sea may provide a
new global market for this expertise."  But  he predicts  that "the question of
compensation payments may come up," for the poorer countries who did not cause
climate change but must face its  consequences.

     Broadus et  al.   (Volume 4)  examine  two such countries in detail:   Egypt
and Bangladesh.  The  inhabited areas of both countries are river deltas, where
low-lying land has been created by the sediment washing  down major rivers.  In
the  case  of Egypt, the damming of  the  Nile has   interrupted the sediment, and
as  the delta sinks,  land is lost  to  the Mediterranean  Sea.   Broadus  et al.
estimate  that  a  50-centimeter  rise in  global sea level,  when combined with
subsidence  and the loss  of  sediment, would result in  the  loss of 0.3  to 0.4
percent  of  the nation's  land  area;  a 200-centimeter  rise  would  flood 0.7
percent.  However, because Egypt's population is  concentrated in the low-lying
areas,  16  and 21  percent of the nation's  population  currently reside  in the
areas  that  would  be lost  in  the two scenarios.
                                       13

-------
     The situation would be even more severe in Bangladesh.  As Plate 4 shows,
this nation, which is already  overcrowded,  would  lose 12 to 28 percent of its
total  area,  which  currently   houses  9  to  27  percent  of  its  population.
Moreover, floods could penetrate  farther  inland,  which could  leave the nation
more vulnerable  to the type of tropical storm that  killed 300,000  people in
the early 1970s, especially if the frequency of tropical storms doubled due to
warmer water temperatures, which  deSylva  (Volume  4)  projects.   Broadus et al.
conclude  that  the vulnerability  of Bangladesh  to a  rise in sea level  will
depend in large measure on whether future water projects disrupt land-creating
sediment washing down the Ganges.

     Bird (Volume  4) examines  the implications  of  sea level rise  for other
African and  Asian  nations,  as  well as Australia.  While holding  back the sea
may be viable  in Australia, he shows  areas in New Guinea where people live in
small cottages on  the  water's  edge on a  barrier  island that  almost  certainly
would be  unable to  justify  construction  of  a dike.   He also points  to the
Philippines, where many people  have literally  "taken to the water,"  living in
small boats and maintaining fishing nets  in their own plots of bay instead of
land.  Current wetlands,  he suggests,  may convert to these shallow bays,  with
people converting to a more water-based economy.

     Leatherman  (Volume  U)  examines  the implications  of  sea level  rise for
South  America.   He  notes that  such  popular resorts  as Copacabana  Beach,
Brazil;  Punta  del  Este,  Uruguay; and Mar  del Plata,  Argentina,  are already
suffering  serious  erosion.     He  concludes   that  because  of  the  economic
importance  of  resorts,   governments  will  allocate  the  necessary   funds  to
maintain  their  viability.  However,  he  predicts  that  "coastal wetlands  will
receive benign neglect" and be lost.

     Park et al. (Volume 4) focus on the expected drowning of coastal wetlands
in  the United  States.   Using  a computer  model of over 50 .sites,  they project
that 40-75  percent of existing  U.S.  coastal wetlands  could be lost  by 2100.
Although  these  losses  could  be reduced to  20-55  percent if new wetlands  form
inland  as  sea level  rises,   the necessary  wetland  creation  would  require
existing  developed  areas  to  be  vacated  as  sea level  rises,  even  though
property  owners would  frequently  prefer to  construct bulkheads to protect
their property.  Because  coastal  wetlands are  important for many commerically
important seafood  species,  as well as birds and  furbearing  animals,  Park et
al. conclude that  even a  one-meter rise  in sea level would have major impacts
on  the coastal environment.

     DeSylva  (Volume 4)   also  examines the environmental  implications  of sea
level  rise, noting  that  in  addition  to wetlands  being  flooded,  estuarine
salinity would increase.  Because 66 to 90 percent of U.S. fisheries depend on
estuaries, he  writes that these impacts  could  be important.  He also suggests
that coral  reefs could become  vulnerable because of sea level rise,  increased
temperatures, and the decrease in the pH  (increased acidity) of the ocean.

     Kuo  (Volume 4)  examines  the implications of sea level rise  for flooding
in  Taipei,   Taiwan,  and  coastal  drainage in  general.   Although Taipei  is
upstream  from  the sea, Kuo concludes that projected sea  level rise would cause
serious  problems,  especially  because  Taiwan  is also sinking.  He recommends
that engineers around the world take "future sea  level  rise into consideration
... to  avoid designing a system that may become  prematurely obsolete."


                                       14

-------
     Gibbs (Volume 4)  estimates  that sea level rise  could  result  in economic
damages in Charleston,  South  Carolina,  equal to as much as 25  percent  of the
annual product of the community.   Anticipatory measures,  however,  could  reduce
these impacts by half.  Gibbs finds that in some areas actions should be taken
today, in spite of the current uncertainty regarding future rates  of sea level
rise, while for other  areas  it would be more prudent to wait until uncertain-
ties are resolved.

     Ken Smith, a realtor from coastal  New  Jersey,  reacts  to  the  other  papers
presented in  Volume  4.  He argues that  the  issue of  sea level  rise should be
taken seriously today,  but  laments the  fact  that many of  his fellow realtors
make comments such as  "What  do you care?  You won't be around to  see it!" and
the  scientific  community  is  "a bunch of eggheads who don't want  us (to build
on  the  coast) anyway." Smith suggests  that part of  the resistance to  taking
the  issue  seriously  is  that there are  a  number of  "naturalists"  who  oppose
building near the shore, and "most  of  the discussion seems  to come from the
'naturalist1  camp."   Nevertheless,  Smith argues that "the  solutions—if there
are  any—should be  contemplated  now as part  of  a  concerted  global effort.
This is a beautiful world, and we are its stewards."

Human Health  and Ecological Impacts

     Climate  and  weather have  important impacts on  human health.   A global
warming would increase the  stresses due to  heat,  decrease those  due to cold,
and  possibly  enable  some  disease  that require warm year-round temperatures to
survive  at   higher  latitudes.    Kalkstein  et  al.  (Volume  3) vpresent  a
preliminary  statistical assessment of  the relationship  of mortality rates to
fluctuations  in temperature in New  York City.   They  find  that a  two to  four
degree  (C)  warming would substantially increase mortality rates  in New  York
City,  if nothing  else changed.   However, they caution that if New  Yorkers are
able to acclimatize to  temperatures  as well as people  who currently live in
U.S.  cities  to the  south,  fewer  deaths would occur.   Kalkstein  et al. write
that knowledgeable observers  disagree   about  whether and  how  rapidly people
adapt  to higher temperatures; some people undoubtedly adjust  more readily  than
others.

      Although people may  be able  to adapt to  changes  in climate, other species
on  the planet would also be  affected and may not  be as able to control their
habitats.   Peters and Darling (Volume 3) examine the  possibility that changes
in  climate  would place  multiple  stresses on  some  species which would become
extinct,  resulting in a  significant decline in biodiversity.   (Mass extinc-
tions  appear  to have  accompanied  rapid  changes  in temperatures  in the past.)

      Throughout the  world reserves have been set aside where targeted species
can remain  relatively free  of human intrusion.  Peters and Darling ask:   Will
these reserves  continue to serve  the same function  if the  climate changes? In
some cases,  it  will  depend on whether the reserve's boundaries  encompass areas
to   which  plants and  animals could migrate.   Some  species may  be able to
migrate "up the mountain" to find cooler temperatures; coastal wetlands  could
migrate inland.    A  northerly   migration  of  terrestrial  species  would be
possible  in  the undeveloped  arctic  regions  of Alaska, Canada, and the Soviet
Union; but human development would  block  migration of larger animals  in  many
areas.
                                       15

-------
POLICY RESPONSES

     Papers by UNEP  Deputy  Director Genady Golubev and  EPA  Administrator Lee
Thomas (both in Volume  1) provide  official views  on the nature of the effects
from projected changes in the atmosphere and the role of their institutions in
addressing those  changes.    Golubev notes  that  while "the global  issues are
complex, uncertainty  exceeds understanding, and patience  is  prudence," there
is an other  side  to the story:   "Our  legacy to the  future  is an environment
less benign than that inherited from our forbearers.  The risks are sufficient
to generate  a collective  concern  that  forebodes  too  much  to  wait out  the
quantifications of scientific  research.   Advocating patience  is an invitation
to be a spectator to our own destruction."

     Golubev also points out that UNEP has worked  for  the achievement  of the
Vienna Convention for the Protection of the Ozone Layer, in which many nations
have agreed to act  in concert  to  address  an environmental issue whose impacts
have not yet  been detected.   Yet he notes that  the agreement is for coopera-
tion in research and does not yet bind nations to observe limits in production
and emissions of gases that could deplete stratospheric ozone.

     Thomas  points  out  that both  the  potential depletion  of ozone and  the
global  warming  from the  greenhouse  effect  are  examples  of  environmental
problems that involve the "global commons."  Because all nations contribute to
the problem  and  experience  the consequences, only  an international agreement
is likely  to be effective.   He  urges scientists around  the  world to discuss
this issue with their colleagues and key officials.

     Richard  Benedick,  Deputy Assistant  Secretary  in the U.S.  Department of
State (Volume  1), describes  the  emerging  international  process addressing the
ozone issue.   Although  the  process for addressing  climate change has not yet
proceeded  as  far, he writes,  "from my perspective as  a  career  diplomat, it
appears  that  the greenhouse  effect has all  the  markings of  becoming  a high
visibility foreign policy issue.  .  .  .  How we address this issue internation-
ally depends  to  a great extent on  our success or failure in  dealing with the
ozone depletion issue."

     J.P.  Bruce  (Volume 1)   of  Environment  Canada presents  the   issue  of
atmospheric change in the context of air pollution in general.  He writes that
ozone modification and  climate change  are "urgent issues," especially because
important  long-term decisions are  being  made today  whose outcomes  could be
strongly affected by changes in climate and the ozone layer.  Bruce recommends
that emissions of CFCs  be  reduced, and concludes that  "a new approach, a new
ethic towards  discharging  wastes and  chemical  materials  into  the air  we all
breathe must soon be adopted on a international scale."

     Two U.S.  Senators  also provide their  reactions.   John  Chafee from Rhode
Island  (Volume 1) describes  hearings  that his Subcommittee  on Environmental
Pollution  held June  10-11,  1986.   "Why  are policy makers  demanding  action
before  the scientists have  resolved all of the  questions and uncertainties?"
he asks.    "We are  doing  so  because  there is a  very  real  possibility that
society—through  ignorance  or  indifference, or both—is irreversibly altering
the ability of our  atmosphere  to  perform basic life support functions for the
planet."   Albert Gore, Jr.  from Tennessee, who has chaired three congressional
hearings on  the greenhouse  effect,  explains why he  has  introduced  a bill in


                                      16

-------
the U.S. Senate  to  establish an International Year  of  the Greenhouse Effect.
"The  legislations  would  coordinate  and  promote  domestic and  international
research efforts on  both the scientific  and policy aspects of  this  problem,
identify strategies to reduce the  increase of carbon dioxide  and trace gases,
investigate  ways  to minimize   the   impact   of  the  greenhouse  effect,  and
establish long-term  research plans." Senator Gore closes  by  quoting  Sherwood
Rowland (discussed below): "What's the use of having developed a science well
enough to make  predictions,  if  in  the  end  all  we're  willing to do  is stand
around and  wait for  them to come  true?"   Both Senators call  for  immediate
action to reduce global  use of CFCs.

     John S. Hoffman  (Volume 1) emphasizes  the  inertia  of the atmosphere and
oceans.   Because  there  are time  lags  between changes  in   emission  rates,
atmospheric  concentrations,   and   changes   in   ozone  and   global   warming
temperatures, the types of management strategies must  be different  from those
that  are  appropriate  for controlling,   for  example,  particulate  pollution,
where  the problem  goes  away as  soon as  emissions are halted.   CFG emissions
would  have  to  be   cut  80  percent  simply  to  keep  concentrations  from
increasing.    Although constant  concentrations would prevent  ozone depletion
from worsening,  Hoffman points  out that even if  we hold the concentrations of
greenhouse  gases  constant once  the  earth has  warmed one degree,  the planet
would warm another degree as  the oceans come into equilibrium.  Thus it might
be  impossible  to  prevent a  substantial  warming  if  we  wait until  a small
warming has taken place.*

     The final  section  of this volume presents  the  papers from fcfee final day
of the conference.    Peter Usher of  UNEP  recounts the evolution of the ozone
issue.   Following  Rowland  and  Molina's  hypothesis  that  chlorofluorocarbons
could cause a depletion of stratospheric ozone in 1974, UNEP held a conference
in 1977 that led to a world  plan  of action   to  assess the issue and quantify
risks.  Since that  time,  UNEP  has  held  numerous coordinating meetings leading
up the the  Vienna Convention.   However, Usher suggests that motivating inter-
national effort on the greenhouse effect will be more difficult:  "Prohibition
of  nonessential  emissions   of   relatively   small  amounts  (to  control  ozone
depletion)  is one  thing,  limiting  emissions  of  carbon  dioxide from coal- and
oil-burning  is quite another."

     Dudek  and  Oppenheimer of  the Environmental Defense  Fund (U.S.) analyze
some  of  the costs  and  benefits  of  controlling  emissions   of CFCs.   they
estimate that by holding emissions constant,   1.65 million cases of nonmelanoma
skin cancers could be prevented worldwide, and that the cost of  these controls
would  be 196 to 455 million dollars, depending on the availability of alterna-
tive chemicals.

     Two  former high-ranking  environmental  officials   in  the  United States
argue  that  we should be  doing  more to address  these problems.   John Topping
recommends  that CFCs  in aerosol  spray cans,  egg  cartons, fast-food containers,
and  other nonessential  uses be phased  out,  and  that people recognize that
   Titus  (Volume  1) and Thomas  (Volume 4) also  explore inertia,  noting that
   even  if temperatures remained  constant after warming  somewhat,  sea level
   would rise at an accelerated rate as the oceans, mountain glaciers, and ice
   sheets came into equilibrium with the new temperature.
                                       17

-------
along with  energy  conservation,  nuclear power  is  the most likely alternative
to  fossil  fuels over  the next  generation  or  two.   He  also  recommends that
society  take  steps to  minimize  the  impacts  of climate  change  and  sea level
rise, for  example, by  requiring environmental impact  statements to consider
the likely impacts.

     Gus  Speth,  president  of  the  World  Resources  Institute,  recommends
international  efforts  to  stop  tropical deforestation;  a production  cap for
chlorofluorocarbons; increased energy conservation;  advanced technologies for
producing  electricity   from  natural  gas;   and tighter  regulations  to limit
carbon monoxide from automobiles,  which  would indirectly limit increases in
atmospheric  methane.     He  agrees  with  Topping  that   environmental  impact
statements  for  projects  that  could contribute to or be  affected  by climate
change or ozone modification should consider these impacts.

     Doniger  and  Wirth,  from  the  Natural  Resources Defense  Council  (U.S.),
argue that  the  current  uncertainties  are  no longer  a  reason to  wait for
additional information:  "With  the  stakes so high,  uncertainty  is an even more
powerful argument for taking early action."  These authors conclude that sharp
reductions  in CFCs are  necessary,  pointing  out  that even with a production
cap,  atmospheric  concentrations  of  these  gases  will  continue  to  grow.
Therefore,  Doniger and Wirth propose  an 80  percent cut in production over the
next  five  years  for  CFCs  11  and  12,  the  halons,  and  perhaps some other
compounds,  with a complete phaseout in the next decade.

     Richard  Barnett  of  the  Alliance  for  a  Responsible  CFC  Policy  (which
represents CFC  using  industries) agrees that we  should  not delay  all action
until the  effects of  ozone depletion  and  climate  change are  felt;  but  he
"would hardly characterize the activities over the last twelve years as 'wait
and see" .  .  .  The science,  as  we currently understand it,  however, tells us
that there  is additional time in  which to  solidify  international  consensus.
This must be  done  through discussion and negotiation,  not through unilateral
regulation."

     Barnett  adds  that  industry  should  "take  precautionary  measures while
research and  negotiations  continue  at the  international  level.    We  will
continue  to  examine  and   adopt   such  prudent   precautionary  measures  as
recapturing,  recycling,   and  recovery  techniques  to  control  CFC  emissions;
transition  to  existing   alternative  CFCs  that  are  considered to  be  more
environmentally  acceptable;  practices  to  replace  existing  systems  at  the
expiration of their  useful lives  to  equipment using other  CFC  formulations;
practices in  the field  to prevent emissions where possible; encouragement of
CFC users to look"for~processes or substances that are as efficient,  safe, and
productive—or better—than what is presently available."

     Barnett  concludes  that "these  environmental concerns  are  serious,  but
their successful  resolution  will  require  greater global  cooperation  in  con-
ducting the necessary research and  monitoring,  and in developing coordinated,
effective,  and equitable policy decisions for all nations."

     We hope that this  paper has provided the reader with a "road map" through
the papers of this four-volume report on the potential effects  of changing the
atmosphere.    But  we have  barely scratched the surface  of each, just  as the
existing  research  has  barely  scratched  the  surface  in  discovering  and


                                      18

-------
demonstrating the possible risks of ozone modification and  climate  change.   A
continual evolution of our  understanding will be necessary  for  our knowledge
to stay ahead of the global experiment that society is conducting.

REFERENCES

Hansen, J,E., A. Lacis,  D. Rind, and G. Russell. 1984.  Climate sensitivity to
   increasing greenhouse  gases.   In Greenhouse effect and  sea level rise:   A
   challenge for this generation,  eds. M.C. Barth and J.G.  Titus.   New York:
   Van Nostrand Reinhold.

Keeling,  C.D.,  R.B. Bacastow,  and T.P.  Whorf.    1982.   Measurements  of the
   concentration of carbon  dioxide  at  Mauna  Loa,  Hawaii.   Carbon  Dioxide
   Review 1982. 377-382,  ed. by W. Clark.  New York:  Oxford University Press,
   Unpublished data from NOAA after 1981.

WAS.   1983.  ChanginR Climate.  Washington, D.C.:  National Academy Press.

Nordhaus, W.D.,  and G.W. Yohe.   1983.  Future  carbon  dioxide emissions from
   fossil fuels.   In Changing  Climate.   Washington, D.C.:   National Academy
   Press.

Villach.  1985.  International assessment of the role of carbon dioxide and of
   other  greenhouse  gases  in  climate  variations  and  associated  impacts.
   Conference Statement.  Geneva:  United Nations Environment  Program.
                                       19

-------
FORESTRY, AGRICULTURE, AND
ENDANGERED SPECIES

-------
Atmospheric Carbon  Dioxide Change:
Agent of Future Forest Growth or Decline?
Allen M. Solomon and Darrell C. West
Environmental Sciences Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee USA
ABSTRACT

     Increasing concentrations  of atmospheric C02  potentially could generate
multiple  and  even  opposing  effects on  forests.   Greenhouse experiments have
shown that enhanced  C02 positively  affects  woody  seedling  growth,  and that
these effects  may  also  occur in  saplings and mature  trees under elevated C02
concentrations.  Yet, today's close geographic correspondence between certain
climate  variables  and  forest  distributions suggests  that  climate changes
resulting  from future  C02  increases  could  destroy  many currently existing
forests.   The potential response of  forests  to  these  conflicting forces was
examined  using a computer model  of tree growth  and forest stand development.
The model  can  incorporate simultaneous  changes In CO, and climate, as well as
the known responses  of  trees to  these variables.   The  model was run with the
annual modeled climate  and CQ2 changes, suggested  by  current energy use pro-
jections,  in  three different ecosystems for  several  hundred simulated years.
The  results of  these  simulation experiments imply  that  the  initial  forest
responses  to changes  in the  environmental variables associated with increasing
C02 may be minor because of tree longevity.   In the long term, however, nega-
tive effects of climate change  on forest growth may be strong enough to over-
whelm the positive benefits  derived  from enhanced C02.  Direct CO, benefits
could, nevertheless,  change the magnitude and the time required by rorests to
respond to climate change,

INTRODUCTION

     The trace gas composition  of the global  atmosphere continues to  change in
response to natural causes,  such  as volcanic  activity, and anthropogenic ones,
such as fossil fuel use.  The radiatively active gases (carbon dioxide, ozone,
methane,  water vapor,  etc.)  are  of  particular concern.  For example, carbon
dioxide is transparent  to  short  wavelengths  that  compose the sunlight  inter-
cepted by clouds or the earth but not to the  much longer wavelengths  of  infra-
                                    23

-------
red subsequently radiated back  to  the  sky.   A "greenhouse" effect occurs when
some of  this infrared radiation  is  absorbed by C02 and  reradiated  to earth,
rather than to space.

     The  resulting  COp-induced climate  change  should  generate  responses  in
forests.  Many (but not all) trees and forest communities are now, or soon may
be, subjected to a different and therefore more stressful climate than that to
which  they  were adapted  as germinating  seeds  (Solomon  and  West 1985).   In
addition, the  change in CCU may  directly affect plant growth  and  forests in
that enhanced atmospheric C02 concentrations have increased the growth of tree
seedlings in greenhouse and growth chamber experiments (Lemon 1983;  Oechel and
Strain 1985).

     Lemon (1983) and Strain and Cure (1985) discussed the effects of enhanced
C02  on  photosynthesis,  respiration,  growth, and  development  of  plants  in
greenhouse experiments.   As yet,  no research data  indicate  that mature trees
growing  in   forests   will  be  capable of taking  advantage  of  the  measured
increases in  dry matter production  and  in drought tolerance  found  in green-
house herbs and woody seedlings.

     Indeed,  the opposite  response  (growth  loss)  could  actually occur  with
enhanced  atmospheric  C02.   Plants acclimate  (cease to respond)  to  increased
C02  concentrations  after  several days  or  months  (Kramer  and  Sionit  1986;
Oechel and Strain 1985), casting doubt on the long-term implications of short-
term  experiments with  high  C02  concentrations.    Field  studies  measuring
changes  in  tree growth  in  response to  acidic  precipitation and gaseous  air
pollutants  revealed   that  annual  tree  growth  has declined  (Johnson  1983;
McLaughlin,   West,  and  Biasing 1983;  Plochmann 1984),  despite  increases  in
global C02  of 25% to 3055  since about 1850  (Solomon  et al.  1985).   Even  the
growth increases at very high  altitudes  (LaMarche  et  al.  1981), parallel with
C02  increases,  are  ambiguous  at  best.   For  example,  the timing of  enhanced
tree growth  at these temperature-limited growth sites  coincides more closely
with the warming of the past century than with the  C02 increases.

     Forests will respond to changes in climate and C02 "fertilization," if at
all, as a function of changing competitive advantages among species.   Competi-
tion negates the simplistic  view that  all  forest  trees would  benefit  from
increased C02.  Instead, growth advantages conferred on one species must incur
growth  losses in  less competitive  species  in  a  complex,   but predictable,
way.   The  discussion  below  represents  our  inital attempt  to  evaluate  the
importance of future conflicting and compensating forces which will operate on
forests through the mechanism of interspecies competition.

APPROACHES TO ESTIMATING FOREST COMMUNITY RESPONSE TO ENVIRONMENTAL CHANGES

     The most reliable approach to projecting the states of future atmospheric
C02  and  climate would be to wait  for  them to occur.   Otherwise, a  technique
for  projecting future changes  is required.   These projections  will involve
data from past and present  environments,  coupled  with conceptual and mathe-
matical  models  of  the  essential environmental,  biological,  and  ecological
processes  involved   in  future  events  (see  Reichle,   Trabalka, and  Solomon
1985).   The models used  to study the effects  of  climate and  C02  on forests
must consist  of  as much relevant  knowledge as possible.  The results of model
trials  will  identify issues that require scientific research,  by  projecting


                                      24

-------
the system behavior that would occur if our models  faithfully  represented the
natural systems.    There is  no comprehensive,   all-purpose model;   there  is,
rather, a  series of models  and model approaches,  each having  strengths and
weaknesses relative to the problems to  which they are applied.

     Two general model  approaches  are  available for  projecting  the responses
of  forest  trees  to climate  change alone,  each based  on different  assump-
tions.   The first  is  a pragmatic  approach  in  which correlations  among data
sets,  describing potential causes  and  effects,  are  substituted  for knowledge
of  cause  and  effect  processes.    The  present geographic distributions  of
climate  variables   are  correlated  with  geographic  distributions  of  biotic
variables, such  as  the  geography  of ecosystems, communities,  or densities of
tree species.  The  expected  future  values of  these  climate variables can then
be  replaced  directly  with biotic variables (Emanuel, Shugart, and Stevenson,
1985; Solomon et al. 1984).

     Unfortunately, empirical  approaches  are inherently  incapable  of charac-
terizing  the transient  response patterns  of  forest ecosystems.   The maximum
life expectancy  of  trees  of most species  is of  the same order of magnitude as
the  expected  appearance  of  doubled  C02 concentrations (Trabalka  et al.
1985).   Thus, the  static forest  ecosystems  projected from  empirical models
probably will not  be  formed until  many years  after a climatic steady state is
reached.   The number  of years involved and the  nature of the  transient  forest
ecosystems  simply  cannot be  estimated using  empirical  projection  techniques.
Yet,  the  short-term  {i.e.,  100-200 years)  transient  responses are  of most
interest  in any analysis of  the  anthropocentric  climate  impacts  %n forests
(Solomon and West  1985).

      Perhaps an  even more  telling  deficiency in the empirical  approaches is
their  inability  to deal  with carbon  fertilization  effects.    The knowledge
gained from greenhouse  experiments  on  seedlings cannot be simply projected to
populations  or  communities composed of adult trees,  each competing for  light
and nutrients   with  other  trees  in   the forest  stand.    To  remedy   these
deficiencies, a  very different approach is required.

      This  second  approach  uses  simulation  models  of  the   cause-and-effect
 relationships  (along with empirical  relationships when  pause and effect are
 unknown).    The  objective is  to apply basic  data  and principles (i.e.,  tree-
 species  natural  histories,  ecological  and  physiological  processes,  and
 environmental  variables)  to projections  of  the response  of  interacting and
 nonlinear ecosystems  to climate change.   Like the empirical approaches,  simu-
 lations require  estimates of future climate to  drive the  model experiments.

      These models  combine  features of the mechanistic  leaf models  (scaled  in
 minutes and millimeters) favored  by  physiologists  (Oechel and Strain  1985)
 with the empirical spatial models  (scaled in years and kilometers)  favored by
 plant geographers  (e.g.t  see  Bartlein,  Prentice,  and Webb 1986).   The  models
 simulate  tree  responses   that  represent  the  summation of   physiological
 processes,  rather  than dealing with the  actual  physiological  processes  under-
 lying tree  response  to variables  such as temperature,  age, or  moisture.  The
 forest stand  simulation  approach  (Shugart  1984)  is particularly  appropriate
 for examining potential carbon fertilization  effects on forest stands, because
 the few experimental greenhouse data now available can  be used to develop
                                       25

-------
individual  tree  response  functions (e.g.,  see Regehr,  Bazzaz, and  Boggess
1975, Figure 3).

     Forest stand simulation models have been under development  for  the past
fifteen years  (JABOWA,  Botkin, Janak,  and Wallis  1972;  FORET,  Shugart,  and
West  1977,  1980).    Adapting  the  models  to  assessing forest  issues  in  the
global COp  problem has  required  a long  development  period (Solomon  et  al.
1980, 1984; Solomon, West,  and Solomon 1981; Solomon and Shugart 1984; Solomon
and Tharp  1985;  Solomon  and Webb  1985; Solomon and  West  1986;  Solomon 1986).
The concepts and biology incorporated  into  the latest versions  of  the model
are described  in detail by Shugart (1984)  and Solomon  et  al.  (1984).   The
mathematical expressions are provided  by Shugart and West (1977)  and Solomon
and Shugart (1984).   Regional variants of the FORET models have been tested on
vegetation  at  several locations  in the United  States,  Canada,  and  overseas
(see  Shugart  1984;  Dale,  Hemstrom, and  Franklin,  1986;  Solomon 1986).   To
supplement  these studies  of  model  validity in  space,   FORET  has also  been
tested using long temporal  sequences of 10,000 to 20,000 years for which there
are fossil  pollen records  of actual forest composition  (see Solomon  and Webb
1985).

     A diagram of  the   model  structure is  presented in Figure  1.    In  the
idealized forest, growth of each  tree species at  each  age (response  function
of  diameter to  time, center  right) would occur  at  the greatest  rates  ever
measured among forest-grown trees.   However,  such  growth  rarely occurs in the
model  because  annual growth  is  reduced  by  extrinsic  (warmth  and  moisture
response  functions, upper  right)  and  intrinsic  (stand  density and  shading
response  functions,  lower   right)   limits to growth.   New trees  are  added to
simulated plots (establishment, lower left), and established  trees are removed
through increased probability  of  death due to slow  growth (suppressed trees,
center left) or  by  increased probability  of death with  age  (mortality, upper
left).

     The  foregoing  describes  the  stand  simulator  used  recently  in  climate
effects studies  that have not  included  COp fertilization (Solomon and Shugart
1984;  Solomon  et al.  1984; Solomon and Tharp  1985;  Solomon and  West 1985;
Solomon  1986).   The results  from some of these  studies suggest the  need to
include  direct C02  effects in  simulation models  (Solomon  and West  1985),
Modifications of the model to  include fertilization effects of atmospheric C02
would  involve  changes  that might  resemble  those shown  as dotted  lines in
Figure 1.   The optimum growth  with  added CX^ (center right)  would increase at
all ages,  coincidentally enhancing  the maximum age  each  species could attain
and thereby reducing  rates  of  mortality (center left).   The  complex of growth
effects attributable  to  enhanced  C02 is also expected  to increase the photo-
synthetic  optimum  temperature and the  maximum  temperature  at which  each
species  can grow  (warmth,  upper   right).   In  addition,   tree  species should
become more tolerant to drought   (moisture,  upper  right).   These  shifts in
response  functions, combined with increased photosynthesis in shade (shading,
lower right), should enhance biomass per unit area (density,  lower right).

     Such  a realistic  simulation  of  carbon fertilization  is  not  currently
possible.   Quantitative estimates  of  the  direct  COp response  functions  are
unavailable for  any major  species groups.  Indeed,  few  of the functions have
been  measured  in any tree  species.   What little  is  known about COp fertili-
zation effects was simulated to assess the implications  of the  effects.  The


                                      26

-------
                                                                                EXTRINSIC LIMITS
                                                                                           MOISTURE
                              MORTALITY
10
-vl
SUPPRESSED TREES
1. INCREASED MORTALITY
 WITH DECREASED
 GROWTH
                         ESTABLISHMENT

                         1. CLIMATE
                         2. SITE CONDITIONS
                         3. LIGHT LEVELS
                                                                              BIOMASS
                                                                                     PROPOSTION OF
                                                                                     FULL SUNLIGHT
                                   MEASURED M FORESTS

                                   HYPOTHESIZED UNDER ENHANCED COj
           Figure  1.   Diagrammatic  representation of  important processes  in forest gap  dynamics simulated by the
     most recent  version of  the model FORENA  (Solomon  1986).   Solid  curve represents  response functions already
     part of  the  model.  Dashed curves represent possible wa|fs  by which  response functions could be  changed if
     data were available to  characterize the  interrelated  effects of increased carbon  dioxide concentrations on
     tree growth.   Extrinsic stochastic  variables and intrinsic deterministic variables  control growth (right)
     differently,  depending on tree species and tree age on the plot (center).  Trees are removed by mortality as
     they age or  stop  growing (left  center)  and are  replaced  by  stochastic  seed  sources,  sorted  by  site
     conditions (left bottom).

-------
model modifications  included only  the  growth function  (center  right,  Figure
1).   Note that  although  only the  growth  curve is involved,  other processes
change  indirectly.    For  example,  suboptimal  temperature and  precipitation
values become less stressful because of enhanced growth.  Maximum ages of each
species are  unchanged,  but the additional  growth from  CC^  reduces age-inde-
pendent (stress-related)  mortality, increasing  the average  age at death for
trees of any species.

     Simulations contained  the  assumption that  CC^ effects  increase linearly
up to a COp doubling, and from doubling to a CCU quadrupling.  The simulations
began with a 400-year period of  tree growth  from  bare plots,  followed  by a
100-year period during which  CO^ doubled and  climate  changed to that expected
from  a  doubling of  COp.    During  the  following 200  years,  C02  quadrupling
occurred  in  the  climate   effects  [but not  in  carbon  fertilization  effects
(Sionit et al.  1985)].  A final  300-year period followed, during which climate
and  C02  were  stable at  the quadrupled  COp  levels.    The parallel  climate
changes were based  on climate model  results  of Mitchell  (1983) and Mitchell
and Lupton (1984).  Solomon (1986)  provides details of model implementation.

     The climate and atmospheric C02 shifts between doubled and quadrupled C02
were simulated to examine forest response under continuously changing climatic
conditions.   The stable,  quadrupled  C02  climate of  the final  300 years was
simulated to investigate lags in forest response to the imposition of environ-
mental stability.  The  reader is cautioned that neither the specific climate
values used,  nor the  presence  of  specific (X^ concentrations (i.e., a COo
quadrupling), is a condition predicted to occur.  The  reader is also cautionea
to accept the simulation results for what they represent.  They are the impli-
cations of our current, inadequate  knowledge  of the processes  that will domi-
nate  future  forest  growth  in  the face  of  change,  rather  than  being any
realistic projection of future forest dynamics.

SIMULATIONS OF FUTURE ENVIRONMENTS

     The examination of forest  response to concurrent changes  in  climate and
carbon fertilization begins with a simulation of climate change  in the absence
of fertilization effects.   Then the analysis  is broadened to  include the few
carbon  fertilization  effects  measured  in  greenhouses,  and  fertilization
effects greater than those measured.   Forest  dynamics  were simulated in three
places: a boreal forest in west  central Ontario, a coniferous-deciduous trans-
ition forest  in northwest  Michigan,  and a  deciduous  forest  in east central
Tennessee.   These were  the  sites at which  Solomon and West  (1986) assessed
potential  reactions  to C02-induced climate changes  by the forest industry.
The sites were also among twenty-one at which climate  effects of C02 increases
were simulated (Solomon 1986).

Simulated Response to Climate Changes Induced by CO^

     The initial simulations  assume that  climate begins to  change after  year
400  as  atmospheric  COg  increases  (Figures  2  through  4),  without enhanced
growth because  of  carbon  fertilization.  During  the first 100 years of simu-
lated warming (years 400  to  500),  summer  and  winter temperatures rise respec-
tively 2.5°  and 5.0°C at  the boreal site, 2.5° and  3-5°C at  the transition
site, and  3.0°  and 2.0°C  at  the  deciduous forest site, based  on  the climate
simulations of Mitchell (1983)  and Mitchell  and Lupton (1984).   At the same


                                      28

-------
                                                          ONNL-DWO M-11111

                  SIMULATED DYNAMICS AT SITES IN BOREAL, TRANSITION,
                               AND DECIDUOUS FORESTS
             FIGURE 2
            WESTCENTRAL ONT
     FIGURES
    NORTHWEST MICH
     FIGURE 4
   EASTCENTRAL TENN
        0  200 400 600 800
               YEARS
0  200 400  600 800
       YEARS
0  200 400 600 BOO 1000
       YEARS
                       BOREAL FORESTS          ESS3 OAK-HICKORY-PINE FORESTS

                       NORTHERN HARDWOOD FORESTS  EZ3 SOUTHERN MIXED FORESTS
                       MESIC DECIDUOUS FORESTS
     Figure  2.   Simulated  stand dynamics  at  the  boreal  forest site  in west
central  Ontario under  four experimental conditions  of  climate effects.    (a)
Stand biomass  by ecosystem type  in megagrams per hectare. (Mg/ha).  (b) Stems by
ecosystem type in stems per hectare,  (c)  Mass of the average  tree  in megagrams
per stem.   Simulation  conditions during  years 0-400  (1 x COP)  include modern
climate  and   climate   variance;  during  years  400-500  (2   x  COp),  climate
gradually changing to  that determined  by  doubled C02 at year  500;  during years
500-700  (4 x COO, climate gradually changing to that determined by quadrupled
C02 at  year 70S; and  years  700-1000 (4  x COp),  stable  climate  and  climate
variance determined by quadrupled COp.

     Figure 3.  Simulated stand dynamics at  the coniferous-deciduous transition
forest  site  in  northwest  Michigan   under  four   experimental  conditions  of
climate effects.   Same as in Figure 2.

     Figure 4.  Simulated  stand dynamics  at the deciduous forest site  in east
central Tennessee under four experimental conditions of climate  effects.  Same
do xii r x^u"6 b t
                                       29

-------
time,  annual  precipitation  is  not  changed.    During the  next 200  years of
warming  (years  500  to 700), summer and winter  temperatures rise respectively
2.5° and  4.0°C  at the boreal and  transition  sites,  and  2.0°  and 3.0?C at the
deciduous  forest site.  Simulated annual precipitation of the 200-year period
declines  about  25%  at  the transition  and  deciduous forest  sites,  but is
unchanged at the boreal forest site.

     Simulated stand biomass was unaffected at the boreal location until about
year 500, when C02 doubling is reached (Figure 2a).  In contrast, stem numbers
began to decline almost as  soon  as temperature  began to increase (Figure 2b),
and average tree  size on the stand  increased slightly (Figure  2c).   In model
simulations, as in reality,  stress initially  led  to increased mortality among
the  youngest  (and  most  plentiful)  trees,   producing  a  plot  having  fewer,
primarily mature  trees.    This  shift had  little  effect  on simulated stand
biomass.   After  year 500,  biomass declined for  50  to  75 years as warming
killed off  the  large boreal forest species and before new northern  hardwoods
could grow into the  plot.   Once  the hardwoods began to enhance stand biomass,
they continued  to increase in biomass and  numbers to the  end  of  the simula-
tion, although climate change ended some 300 years earlier.

     Stems of deciduous and oak-hickory-pine forest  species were  relatively
common  after  year 600  (Figure  2b)  although  they  rarely  survived  to  a  size
large enough to affect  stand biomass (Figure 2a).   Average tree size (Figure
2c) increased directly  with expansion in nonconifer  populations  (Figure  2a),
primarily because the conifers formed forests of  smaller  stature than did the
deciduous trees.

     Warming at  the  transition  site (Figure 3)   caused  an almost  immediate
response  in declining biomass from dieback  of mature trees (Figure  3a),  in
enhanced stem numbers from increased small young stems as the  canopy  of the
simulated forest opened  (Figure  3b),  and in  decline  of  tree  mass  (Figure 3c)
as large trees died and were temporarily replaced  by small, young trees.  This
immediate response  to  the  warming  is  logical, considering  that almost  all
species  growing  in  transition  communities   belong  further  north  (boreal
species) or  further  south  (temperate deciduous  species).   Thus,  they  were
initially under stress, and a  change in the  climatic  status  quo enhanced the
stress for the  dominant northern species.   As  warming continued during years
500 to 700,  biomass (Figure 3a)  and tree mass  (Figure 3c)  first recovered with
the growth of  northern hardwoods, then declined  again as the continued warming
stressed  the  recent  northern  hardwood  immigrants, forcing  their  demise  in
favor  of more  warmth-adapted  mesic  deciduous and  oak-hickory-pine  forest
types.

     Climate changes  at the  deciduous forest  site  (Figure  4) generated  no
discernible shift  in any  of the  forest  variables until about  year  500,  when
stand biomass (Figure 4a)  and  the mass of  individual  trees (Figure  4c) began
to decline.   Dieback took  the  following  200 years.   Unlike diebacks  at  the
other two sites, this one resulted  in permanent loss of dense forest.  The 60
Mt/ha of stand biomass resembles that in open  oak  woodland and savanna (Olson,
Watts,  and  Allison 1983).   One might expect subtropical forests similar to
those in Florida today eventually to appear, but the eventual moisture balance
excludes subtropical trees.  By  model year  700,  soil moisture values  were more
similar  to  those  of  treeless, central Texas today than to those of southern
                                      30

-------
Florida.  Obviously,  such a change  in biomass could  have  important implica-
tions for the global carbon cycle.

Simulated Responses to Climate and  Tree Growth Changes Induced by CC^

     The  stand  simulations  of   COp-induced   climate  changes  resulted  in
temporally ordered sequences of forest  stand destruction and regrowth.   These
sequences are  likely to  differ  if carbon  fertilization also  affects  forest
growth, particularly  because  carbon fertilization should postpone mortality.
This  idea  was tested  at  the  three sites  where  forest response  to  climate
change was simulated.

     The model was modified to allow  as  much as a 20%  increase in deciduous
tree growth  and  an \\% increase in coniferous  tree growth,  with as much as a
doubling of  CO-  [from  350 to  650 uL  L-1  (Sionit et al.   1985)].   Moisture
effects on growth  were unchanged  in  coniferous trees  (P.J.  Kramer, personal
communication, 1984)  and  were  decreased by as  much as 18$ in deciduous trees
(Sionit et al. 1985), with as much as  a doubling of C02.  Figures 5 through 7
illustrate stand  biomass  response  in  the  following ways:   to climate change
alone (from Figures 2a, 3a, and 4a) and to climate change combined with carbon
fertilization  effects;  as measured  by Sionit  et  al. (1985);  at twice those
measured (40$  increase, in deciduous tree  growth,  2255  increase in  coniferous
tree  growth,  36$  decline in  deciduous  tree  water use); and  at three times
those  measured  (60%  increase  in  deciduous  tree growth,   3355  increase  in
coniferous tree growth, 54$ decline in deciduous tree water use).

     Continuously  increasing  carbon fertilization effects  decreased the time
required after a dieback to repopulate the plot with new trees at  the boreal
site (Figure 5).  Although the dieback began at about  the same  time, both with
and without  measured fertilization, recovery  began  40  years  earlier  and was
completed about  100 years later in the absence of carbon fertilization.  The
simulated dieback  feature  of  forest  response to  climate  change  was  almost
absent  when  the  carbon fertilization  effect  was doubled,  and it disappeared
entirely  under a  carbon   fertilization  effect three  times  that  measured in
growth  chamber experiments (Figure 5).   After climate stabilized at year 700,
total  stand  biomass  reached slightly greater  values with carbon  fertilization
than without.

     In contrast to results at  the boreal site,  the  simulated forest at the
coniferous-deciduous  transition  site  (Figure   6)  required  only  the measured
carbon  fertilization effect to balance, and thus eliminate, the dieback before
a  doubling  of atmospheric C02  occurred  at   year  500.   The  deciduous tree
communities  that  eventually  controlled  the  transition  forests  began  to
dominate much earlier with than without  simulated carbon  fertilization.   In
addition, increasing  fertilization increased stand biomass.  Indeed, the final
rank  order  of stand  biomass  at  year 1000 was  first  established less than 50
years  after  C02-induced  climate and  growth effects  began  (about year 440 in
Figure  6).   The  sensitivity of these simulated  forests to changes  in mortality
rates  was apparently great enough to  generate an  almost instant response to
these  subtle environmental changes.

     The simulated  deciduous forest most clearly responded to  the successively
greater effects  of carbon fertilization (Figure 7).   No response was  evident
before  C02  doubled.  Then, each   succeeding increase  in C02 treatment  reduced


                                       31

-------
                  CO
                  co
                  <

                  O
                  I
200

180

160

140

120

100

 80

 60

 40

 20

  0
                                	COZ EFFECT TRIPLED

                                —— COj EFFECT DOUBLED

                                — —-• CO2 EFFECT AS MEASURED

                                -•-•• NO COa EFFECT     I
                                I    I   III
                           100 200 300 400 500 600 700  800  900  1000
                                         YEARS
     Figure  5.   Simulated  stand biomass  of  boreal forest  ecosystem  in  west
central  Ontario, with  varying climatic  and C02 effects.   Climate  changes are
as  in  Figures  2 through 4,  with successively greater effects of C02  on  tree
species  growth.   See Figure 2 legend and text for simulation conditions during
years 0-400  (1  x C02),  years 400-500  (2 x C02), years  500-700 (4  x  COP), and
years 700-1000 (4 x C02).
                    CO
                    CO

                    O
                    m


                    I
                                     CO, EFECT TRIPLED

                                  — CO, EFFECT DOUBLED

                                  •-• CO, EFFECT AS MEASURED

                                  -•• NO CO. EFFECT
                            100 200 300  400 500 600  700 800 900 1000
                                          YEARS
     Figure  6.  Simulated stand  biomass of the coniferous-deciduous  transition
forest  ecosystem  in  northwestern  Michigan,  with  varying  climatic  and  C02
effects.   Climate  changes are  as  in  Figures 2  through 4,  with successively
greater effects of COn on  tree species  growth.   See Figure  2 legend and  text
for  simulation  conditions  during  years  0-400  (1  x C02),  years  400-500  2 x
C02), years  500-700  (4 x  C02),  and years 700-1000 (4 x  C02).
                                       32

-------
                 
-------
central Canada  (Hare and  Thomas  1979).   In contrast,  the  parallel fertili-
zation  effect  of  a C02  doubling,  as  measured  in  long-term  COp fumigation
experiments, represents only  a  20%  increase in growth  of deciduous -trees and
an  ^^%  increase  in  the growth  of coniferous  trees.   If future  direct C02
effects are  to  rival or  exceed climatic  effects,  then  other,  unmodeled C02
fertilization features will have to be extremely important.

     The simulation  results  also  indicated other  forest responses  that may
occur in the future.  For  example,  significant  changes in boreal or deciduous
forest biomass may not be  detectable  during the first few decades of environ-
mental  changes  that lead  to  doubled C02  and  concomitant  increased tempera-
tures.   However,   the oncoming  shifts  in  forest biomass may be  presaged by
early losses in seedlings  and saplings  relative to  mature trees.  Also, early
detection of forest responses may be  ordered geographically.   The sensitivity
of  simulated   transition   forests   to  environmental   change  implies  that
coniferous deciduous  transition forests  and other  forests near  tree  growth
limits may be the first to respond to  changing C02  and climate.  This impli-
cation  is consistent with  the suggestion by LaMarche  et al.  (1984) that they
measured C02-derived increases in tree growth at high altitude range edges.

     These ideas are worthwhile working hypotheses only as long as the forcing
and response functions  simulated  in the model are  similar to the forcing and
response functions  that  affect  forests in  the  future.    As a  tool,  the model
must  accommodate  both  new  and  revised knowledge.   Accurate  simulations are
currently restricted by the lack of the specific  growth chamber data required
to  characterize  the  alternative  (dashed)  lines  in  each of  the  response
functions illustrated  in  Figure 1.    In  addition,  the scenarios  of  future
climate change are  also  subject to large errors, particularly in the effects
of feedbacks among components of the climate system.  For example, in the most
recent  projections of  C02-induced  climate change  by  Manabe  and  Wetherald
(1986), temperature increases are twice as  great  as  those used in the simula-
tions discussed here, and  soil moisture is  40?  less than that used here.  The
larger  climate  effects  occur because  of  moisture  feedbacks  that  were  not
considered in earlier climate model experiments.

     The stand simulator could  also be greatly improved,  even with available
data.   The  model  we used  (FORENA;  Solomon 1986)  does not  consider certain
features that may  be  important  under climate changes,  such as the incorpora-
tion  of localized soil  nutrients and  turnover,  which  are  available in other
models  (Pastor  and  Post  1985).   Excluding nutrient  cycling from  the  C02-
climate simulations  should generate greater simulated  community productivity
than would be the  case on  present and future landscapes where nutrients limit
and will limit tree and  forest growth.

     Another  feature not   modeled  is  the interacting effects  of  chronic
diseases and  atmospheric  pollutants.   Insects,  disease,  and  their vectors
(e.g., other insects, fungi,  and bacteria)  have their own, often complicated,
life cycles which depend on weather and climatic  events in a manner different
from  that of  the host  trees.   No model has yet been  applied to the complex
ramifications of pathogen,  insect,  and tree-life-cycle interactions under C02-
induced climate change  and other environmental perturbations.   A large-scale
regional research program  is  under way  at  several cooperating institutions to
determine the  chronic  effects  of  acidic precipitation on forests,  based on
field studies and  the  forest-stand  model (for example,  see McLaughlin et al.


                                      34

-------
1983).  This effort might be extended to include insects, pathogens, and other
air pollutants, as well as climate change.

     Finally, the  model  is inherently  limited  by the presence  of mountains,
oceans, and other nonclimatic restrictions upon  the  geographic ranges of tree
species.   Within  the present  form  of the  model,  such  boundaries  must  be
assumed to coincide  with climatic barriers,  although this  is  clearly not the
case for some species.

     The present model experiments on effects of carbon fertilization indicate
that  the  primary  impacts  could  involve accelerated growth,  increased aging,
and  reduced  impacts  of  the  climate-related  environmental  changes simulated
without C02 fertilization.  Even with unrealistically high growth enhancement,
hypothetical  tree growth  and forest community  productivity  did  not exceed
current known  values for  those  communities.   More  data from  many species on
the  responses  of mature  (as well as  seedling)  trees to increased atmospheric
COp  concentrations are  required  to  characterize  potential  COp fertilization
and  increased  water-use  efficiency.    Indeed,   there  is a  critical  need for
evidence  that  any  tree  life  stage  besides  seedlings  will benefit  from C02
fertilization.   At present, we can expect such benefits  only in  plants growing
in  noncompetitive,  nonlimiting   agricultural systems.    Thus,  data  on the
presence  and effects of C02  fertilization and  water-use efficiency  phenomena
must be obtained from trees growing in unmanaged stands, in order  to hypothe-
size and  then  to reliably  simulate the  effects of carbon fertilization on
forests.

ACKNOWLEDGMENTS

      We express  our  deep gratitude to M. L.  Tharp,  who  programmed the simula-
tions  for  this  paper.    W.  M.  Post  and  V.  H.  Dale  provided  constructive
reviews.   Research  was  sponsored Jointly by the National Science Foundation
under Interagency Agreement Mo.  BSR84-17923,  A03, and the U.  S. Department of
Energy,  under   Contract  No.  DE-AC05-840R211K)0  with  Martin  Marieta  Energy
Systems,  Inc.   Publication No. 2781,  Environmental Sciences  Division, ORNL.

REFERENCES

Bartlein, P.J.,  I.e. Prentice,   and  T. Webb III.   1986.   Climatic  response
      surfaces  from  pollen  data for some  eastern  North  American  taxa.  J..
      Biogeogr.   13:35-57.

 Botkin,  D.B.,  J.F.  Janak, and  J.R. Wallis.    1972.   Some ecological  conse-
      quences of a computer model of forest growth. J. Ecol.   60:849-72.

 Dale,  V.H.,  M.   Hemstrom,  and J.  Franklin.    1986.   Modeling the  long-term
      effects  of  disturbances   on  forest   succession,  Olympic  Peninsula,
      Washington.  Can. J.  For.  Res.  16:56-67.

 Emanuel,  W.R.,  H.H.  Shugart,  and M.L.  Stevenson. 1985. Climate change andthe
      broad-scale  distribution  of  terrestrial  ecosystem  complexes.  Climate
      Change. 7:29-43.

 Hare,  F.K., and  M.K.  Thomas.  1979.  Climate Canada.  2nd ed.  Toronto: Wiley and
      Sons.
                                       35

-------
Johnson, A.H.  1983.  Red spruce  decline  in the  northeastern  U.S.: Hypotheses
     regarding the role of acid rain. J.  Air Pollut. Contr. Assoc. 33:1049-54.

Kramer,  P.J.,  and N.  Sionit.  1986.  Rising  CC^:  Expected  effect  on  tree
     development  and  what deficiencies  in knowledge  we  need to  remove.  In
     Rising Cp2 and changing climate:  Forest risks  and  opportunities,  ed.  W.
     E. Shandsr Washington D.C.: Conservation Foundation (in press).

LaMarche, V.C., D.A.  Graybill,  H.C.  Fritts, and M.R.  Rose.  1984.  Increasing
     atmospheric carbon dioxide: Tree ring  evidence for  growth enhancement in
     natural vegetation. Science. 225:1019-121.

Lemon, E.R. 1983.  COo and  plants:  The response  of  plants  to  rising levels of
     carbon dioxide. %oulder, Colorado:  Westview Press.

Manabe, S., and R.T.  Wetherald. 1986. Reduction in summer soil wetness induced
     by an increase in atmospheric  carbon dioxide.  Science. 232:626-28.

Mclaughlin, S.B.,  T.J. Biasing, L.K. Mann,  and  D.N.  Duvick.  1983.  Effects of
     acid rain  and gaseous  pollutants on forest productivity:  A regionscale
     approach. J.  Air  Pollut. Contr.  Assoc. 33:1042-49.

Mitchell, J.F.B. 1983. The  seasonal response of a general circulation model to
     changes  in C02  and  sea temperatures.  Quart.  J.  Roy.  Meteorol.  Soc.
     109:113-52.

Mitchell, J.F.B.,  and  G. Lupton.  1984.  A 4 X  COp  integration with prescribed
     changes in sea surface temperatures.  Prog. Blometeorol.  3:353-74.

Oechel, W.C.,  and B.R. Strain.  1985.  Native  species  responses  to  increased
     atmospheric carbon dioxide concentration.  In Direct  effects of increasing
     carbon dioxide on vegetation,  eds.  B.R.  Strain and  J.D.  Cure,  pp.  118-
     154.  DOE/ER-0238.  Washington,  D.C.: U.S. Dept. of  Energy.

Olson, J.S., J.A.  Watts, and  L.J. Allison.  1983. Carbon  in live vegetation of
     major world  ecosystems. ORNL/TM-5862.  Oak Ridge,  Tennessee: Oak  Ridge
     National  Laboratory.

Pastor, J., and W.M.  Post.  1985.  Development of a  linked  forest productivity
     soil  process model.  ORNL/TM-9519.     Oak  Ridge,   Tennessee:  Oak  Ridge
     National  Laboratory.

Plochraann,  R.   1984. Air pollution  and the dying forests  of  Europe.  Am.  For.
     90:17-21, 56.

Regehr, D.L.,  F.A. Bazzaz,  and W.R.  Boggess.  1975.  Photosynthesis,  transpira-
     tion and  leaf conductance of  Populus  deltoidea in  relation  to  flooding
     and drought.  Photosynthetica.  9:52-61.

Reichle, D.E., J.R. Trabalka,  and  A.M.  Solomon. 1985. Approaches to  studying
     the global carbon cycle. In  Atmospheric  carbon  dioxide and the  global
     carbon cycle,  ed.  J.R. Trabalka,  pp.  15-24.  DOE/ER-0239.  Washington,
     D.C.:  U.S. Department  of Energy.
                                      36

-------
Shugart, H.H.  1984.  A theory of forest dynamics.  New York: Springer-Verlag.

Shugart, H.H., and D.C.  West.  1977.   Development of  an  Appalachian deciduous
     forest succession model and  its  application to assessment of  the impact
     of the chestnut blight. J. Environ.  Management. 5:161-179.

Shugart,  H.H.,  and  D.C.  West.   1980.  Forest  succession models.  BioScience
     30:308-13.

Sionit, N., B.R.  Strain,  H. Hellmers, G.H. Riechers,  and C.H.  Jaeger.  1985.
     Atmospheric  C02  enrichment  affects  the   growth   and  development  of
     Liquidambar  styraciflua  L.  and  Pinus  taeda L.  seedlings. Can.  J.  For.
     Res.  15:468-71.

Solomon,  A.M.  1986.  Transient  response of  forests  to  C02-induced  climate
     change:  Simulation   modeling  experiments   in  eastern  North  America.
     Oecologia 68:567-79.

Solomon,  A.M.,  and H.H.  Shugart.  1984.  Integrating  forest-stand simulations
     with  paleoecological  records to examine  long-term  forest  dynamics. In
     State  and  change of forest ecosystems—Indicators  in  current research,
     ed.  G.I.  Agren,  pp.  333-56.  Rept.  13.    Uppsala,  Sweden:  Swed.  Univ.
     Agric. Sci.

Solomon,  A.M.,   and  M.L.  Tharp.  1985.  Simulation  experiments  with   late
     Quaternary  carbon  storage  in mid-latitude forests.  In  The  car&on cycle
     and  atmospheric C02:  Natural  variations  Archean to present,  eds.   E.T.
     Sundquist  and  W.Sr Broecker,   pp.  235-50.  Washington  D.C.:   American
     Geophysical  Union.

Solomon,  A.M.,  and T. Webb III. 1985.  Computer-aided reconstruction of  late
     Quaternary landscape dynamics. Ann.  Rev. Ecol. and  Syst. 16:63-84.

Solomon,  A.M.,  and D.C.  West. 1985.   Potential  responses of forests to  C02-
     induced  climate change.  In  Characterization of  information requirements
     for  studies  of COo effects;  Water resources,  agriculture,  fisheries,
     forests  and human-health, ed. M.R.  White,  pp.  145-70.  DOE/ER-0236.
     Washington,  D.C.: U.S. Department of Energy.

Solomon,  A.M.,  and D.C.  West.  1986.   Simulating forest  responses  to  expected
     climate  change in eastern North America: Applications  to  decision-making
      in the forest industry. In Rising CO., and changing  climate: Forest  risks
     and  opportunities,   ed.  W.E.   Shanm.  Washington, D.C.:  Conservation
     Foundation (in  press).

 Solomon,  A.M., H.R.  Delcourt, D.C.   West,  and T.J.  Biasing.  1980. Testing  a
      simulation  model   for   reconstruction   of   prehistoric   forest-stand
      dynamics.   Quat.  Res. 14:275-93.

 Solomon, A.M.,  J.R.  Trabalka,  D.E.  Reichle,  and  L.D.  Voorhees.  1985.    The
      global  cycle  of carbon.  In Atmospheric  carbon dioxide  and the  global
      carbon  cycle,  ed.  J.R.  Trabalka,  pp.   15-24.   DOE/ER-0239.  Washington,
      D.C.: U.S. Department of Energy.
                                       37

-------
Solomon, A.M.,  D.C. West,  and  J.A. Solomon.  1981.   Simulating the  role  of
     climate change  and  species  immigration  in  forest succession,  In Forest
     succession; Concepts and application,  eds.  D.C. West,  H.H.  Shugart,  and
     D.B. Botkin, pp. 154-177. New York: Springer-Verlag.

Solomon,  A.M.,   M.L.  Tharp,  D.C.  West,  G.E.  Taylor,  J.M. Webb,  and  J.C.
     Trimble.   1984.  Response  of  unmanaged  forests  to  CC^-induced  climate
     change;  Available  information,  initial  tests,  and  data  requirements.
     TR009, Washington,  D.C.: U.S. Department of Energy.

Strain, B.R.,  and  J.D.  Cure, eds.  1985.  Direct effects of  increasing carbon
     dioxide on  vegetation.  DOE/ER-0238. Washington,  D.C.: U.S.  Department  of
     Energy.

Trabalka, J.R.,  J.A. Edmonds,  J.M. Reilly,  R.H.  Gardner,  and L.D.  Voorhees.
     1985.   Human alterations of  the global  carbon  cycle  and  the  projected
     future. In  Atmospheric  carbon dioxide and  the global  carbon cycle,  ed.
     J.R.    Trabalka,   pp.   247-87.  DOE/ER-0239,   Washington,   D.C.:   U.S.
     Department of Energy.
                                      38

-------
Historical Changes in Forest Response to Climatic
Variations and Other Factors Deduced From Tree Rings
Harold C. Fritts
University of Arizona
Tuscon, Arizona USA
INTRODUCTION

     Hecht (1985)  defines  "climate"  as  a  time-transgressive phenomenon  being
the average state of the atmosphere  over  periods  of  25  to  30  yeartk-or  more.
While we have considerable knowledge  of the broad  characteristics of climate,
there  is much  less knowledge of  the major  processes  of climatic  change
(National Academy  of Sciences 1975).    Proxy  data,  i.e., substitutes  for
climatic  information,  can  span  time periods  before  instrumental  climatic
records were kept and thus are an  important source of information on the long-
term history of climatic variations (Hecht  1985).  Tree rings provide a unique
proxy  record  of seasonal  to century-long climatic variations  for  several
reasons.  First, usable trees can be found  in  all temperate  lands  and many
trees  are available  for replication.   Furthermore, the information obtained
from  these  trees   can be  dendrochronologically  dated  and  arranged  in  an
accurate time sequence.  Finally,  the ring  features can be measured easily and
combined  for many  trees  to  obtain a well-behaved time  series,  which  is
particularly  relevant   to   forest   response  as  ring   width   is  a  growth
measurement.

DENDROCHRONOLOGY AND DENDROCLIMATOLOGY PROCEDURES AND PRACTICES

      It  is  well known  that  yearly tree-ring width sequences,  called chrono-
logies,  have been  used  to  date  structures,  such  as  archaeological ruins,
historic  buildings, and early Dutch  paintings  (Anonymous 1977; Baillie 1982;
Trefil 1985).   A.E.  Douglass, an astronomer working in Arizona, is credited
with  developing tree-ring  dating  (1919,  1928,  1936) and  is  considered  the
 founder  of  the  discipline  of "dendrochronology"  (Webb 1983).  "Dendro"  is the
 root  word meaning tree and "chronology" means "time." The  discipline is most
 easily understood  as the systematic use  of  tree-ring  crossdating  to  study
 problems involving time and factors of the environment.   Crossdating was first
 used  to date beams or charcoal  fragments from  archaeological and  historical
 structures  in  the  North  American   southwest,   and  the  technique  provided


                                     39

-------
archaeologists with the most precise time control ever devised  (Douglass  1935,
1937).

     Crossdating  uses  the year-to-year  synchrony  of ring features associated
with  past fluctuations  in climate  to  place  each  ring  in  its  correct  time
sequence.   Various  discrepancies in ring synchrony  suggest where ring counts
may be  in error.  The  source of each error is deduced from  the ring structure
using  knowledge  of  tree  growth, and  the dating  is adjusted.   This tedious
procedure  continues  until  all  apparent  discrepancies  are  identified and
corrected for  every ring  in every  tree collected from the  site.   If this  is
done carefully, all rings will be assigned  to the correct year in which  they
were  actually  produced and the  data   can  be combined  to  obtain  an average
yearly response of the trees to variations in climate.

     The  science  that uses  dated  tree-ring sequences  to  reconstruct   past
climate (Douglass 1914; Schulman  1947,  1951,  1956; Fritts,  1976; Hughes et al.
1982) is referred to as dendroclimatology.  It is not as well known that  these
same  dated  tree-ring  sequences  can   be  used  to  study   various  ecological
problems; in these cases the term dendroecology is used.

     A  variety of  structural  characteristics of  tree-rings,  such  as width,
wood  density  (Schweingruber,   Braker,  and  Schar  1978b),   and  vessel   size
(Eckstein and  Frisse  1982), show variability  from one  ring to the next.  The
variations  in  ring  width  have  been  studied most  often  (Fritts 1976; Baillie
1982) because  width can  be observed and measured  easily from a finely sanded
surface by using a hand lens or dissecting scope.

     The  wood  can be  X-rayed  (Polge  1963,  1966, 1970), and the  image on the
exposed  film can be  scanned  to  obtain  detailed  ring  density measurements.
These, in turn, can be correlated with climatic variations as well as various
physical, chemical,  and  biological features  of  the  environment (Keller  1968;
Parker and  Henoch 1971;  Fritts  1976;  Huber  1976; Schweingruber,  Braker, and
Schar 1978a, 1978b;  Conkey, 1982a, 1982b).

     The  effects of nonclimatic  variations  on ring-width growth are minimized
by coring only trees  with characteristics  that  indicate  climate  was highly
limiting to growth.   Additional  random variability caused  by site differences
is controlled  by sampling and  averaging the effects  of  many trees  from  a
narrowly defined target site (Fritts et al.  1965; Fritts 1969; LaMarche 1974a?
LaMarche  1982; Norton  1979, 1983).   A narrowly defined target site helps  to
minimize  the differences  between tree microsites  which  could  obscure   that
portion of  the response  because  of  variations in macroclimate.   In American
work,  from  ten  to  forty  or  more of the  oldest  trees  with the  necessary
characteristics are  cored and two cores are usually obtained from each tree  if,
the site.

     The  samples  are  prepared  and crossdated  before performing  the desired
analysis.  When crossdating is complete, the dating is checked by the computer*
(Holmes   1983)   or  by  another  person,  the  rings  are  measured,  and  the
measurements are  standardized.   Standardization  identifies the slowly varying
growth changes in individual trees associated  with increasing age  and local
conditions  of  the  site  (Figure  1a).   These  changes  are estimated,  in   thi5
case, by  fitting  a  curve  or straight  line  to each dated and measured series-
The  width is divided  by the  estimate to obtain an index which is stationary


                                      40

-------
                      RING WIDTHS

                   CPN211
                      CPN101
                   sl  CPN102
                  CPN071
                  CPN072
                1B10 U» K90 1870 1MO 1910 WO WO
                                              CPN102
                                                    CPN071
                                                    CPN072
                                        «» ux UN two MOO mo an wo am
TREE INDICES
                                                                       un wn wo no aw «n
                                                                                              SUMMARY
                                                                                         CPN540
h Ata *\ft VMA A j Ai .. l\i i fVfcr

1 Jll ^i
^ >y^ rtfc, %
                                                                                             use tan two mo
      Figure  1.    The dated  ring widths are  transformed  into a  standardized chronology  by:    (a)  fitting a
curve or straight line  to the ring  widths from each  co$j, (b) dividing by  the  values of the fitted curve to
obtain the indices,  (c) averaging  the cores for each  tree to obtain the indices, (d) averaging the cores for
each  tree  to obtain  the tree  indices, and  (e) averaging the  tree indices to obtain the  chronology for the
site.

-------
over  time  (Figure  1b).   These  indices  can  then be  averaged for  the cores
within each  tree  (Figure 1c).   These in turn  are averaged for  all trees to
obtain  a  mean  chronology  for  a  species  and  site  (Figure  1d).    This
standardized chronology  reflects the  relative  variations in ring-width growth
associated with cliamte.   However, standardization must be applied carefully
because  in  certain circumstances  it  cannot   distinguish between  standwide
nonclimatic factors and  those due  to climate,   and a  linear or downward trend
in climate might be indistinguishable from age-related variations in growth.

LONG TREE-RING CHRONOLOGIES APPLIED TO ENVIRONMENTAL QUESTIONS

     The growth of trees from many  high-altitude or high-latitude  sites are
most  often  limited by low  temperatures, and many of these trees  may attain
great  age.    Ring-width  chronologies   from   these   trees  largely  reflect
temperature variations  (LaMarche  1974a,,  1974b,  1978;  LaMarche  and Stockton
1974; Schweingruber et al. 1978; Schweingruber,  Braker,  and Schar 1978, 1979;
Cropper and  Fritts 1984), although  other factors such  as snow  depth can be
important (Graumlich and Brubaker  1986).  Such  chronologies have been plotted
and  used  directly  as proxy  records  of  temperature variations and  change
(Figure 2).  However, LaMarche et  al.  (1984) found a  growth increase in high-
altitude trees from the  Great Basin,  U.S.,  beyond the  effects  they expected
from  temperature   trends.    They hypothesized  that   this could  be a carbon
dioxide fertilization effect.  Graybill  (1985)  is developing a more extensive
network  of  high   altitude   site  chronologies   that  ranges  from   the  Rocky
Mountains to the eastern  edge of the Sierra Nevadas for use in further testing
of this hypothesis.  In preliminary analyses that used upper treeline (3400 m)
data  from six  sites in the Great Basin  (P. longaeva)  and three from Colorado
(P.  arista ta).  the chronology  scores  on the  first   and  only  significant
principal component  for  each area demonstrated a similar  rise  (Figure 2) to
those reported by  LaMarche  et al.  (1984) and Graybill  (1986a).   In contrast,
the component scores of four other  Great Basin  chronologies (P. longaeva) from
relatively high altitudes (2600-2900 m), yet near the  lower altitudinal limits
of growth  for  the species,  demonstrated different growth  trends (Figure 2).
Further investigation is  required to understand the more precise relationships
of  tree  growth   in   all  of  these  high-altitude   sites  to  temperature,
precipitation,  carbon dioxide, and  other critical factors.

     The rings of  conifers  from their lower altitudinal limits  (Figure 2) in
semiarid western  North America are likely  to  reflect drought  resulting from
deficits of soil moisture and evaporative stress caused  by high temperatures,
wind, and intense  solar radiation at the tree  sites (Fritts 1976; Stockton and
Meko  1983).   The   interactions between  different climatic  factors  make these
chronologies  difficult  to  interpret,  although  generally  the  ring  width
variations can be  regarded as a  more  or less direct  response to soil moisture
due to precipitation variations with an inverse response to temperature.

CALIBRATION AND VERIFICATION

     Regression and related multivariate techniques can be used to relate many
climatic factors  to an  indexed  chronology  or  to  convert  the  indexed chrono-
logies into estimates of one or more climatic  factors.  The tree-ring data are
calibrated with instrumental  climatic measurements,  and the degree  of fit is
expressed as percent calibrated variance.
                                      42

-------
    COMPONENT  SCORES 1380-1983 (PCI) GRERT BflSIN UPPER TREELINE SERIES
  -3-
      1400
             1450
               1500    1550    1600    16SO    1700    1750    1800

COMPONENT SCORES 1380-1981 (PC13  GRERT BRSIN LOWER FOREST BORDER
   -3
                    I I  I I I  I | I  i 1 I  | I I  i i I  I i i 'i I i  i i i  I i i
      1400    14SO    1500    1550    1600    1650    1700    1750

     COMPONENT SCORES t380-1983 (PCU COLORflDO UPPER TREELINE SERIES
                                                                                      —2
                                                                      i  4 i |—r-r—1—i i  i i i'
                                                                         1900    1950
      Figure  2.  The first  principal component  scores  from 1380 to  1983 for  (a)
Great Basin  upper  treeline series,  (b) Great Basin lower forest border series,
and  (c) Colorado upper treeline  series (Graybill 1986).

-------
     The first type of calibration is called a "response function" because the
coefficients can be interpreted  as  the  response  to climate.  Cooper, Biasing,
and  Fritts  (1974) used  response functions  in a  CIAP  study to  estimate the
effect of a 2°-3°C temperature  decrease  and a 555  change  in precipitation on
ring-width  growth  throughout  arid  sites  in the  west.    A type  of response
function can be  used  to separate the effects  of  climate on growth from those
due to pollution or other possible agents of forest decline  (Cook in press).

     The second  type of  calibration  is  called  a  "transfer function."  Several
chronologies at different lags may  be used as  predictors of a climate-related
variable at  one  or more sites.   Least  squares techniques  are  used  to obtain
the "best-fitting" relationships over the calibration period (Lofgren and Hunt
1982).  A transfer function is  obtained with  coefficients that convert tree-
ring  chronology  information  into  estimates  of  the  calibrated  variable  of
climate.     The  reliability of  the  coefficients  of  the  equation  and  its
estimates can  be  tested  by withholding  some of the  observations  to  test
whether the  reconstructions for those particular years are correct.  This pro-
cedure is called "verification."  If the  verification tests are significantly
better than  expected by  chance,  the  reconstructions are considered a verified
result (Gordon 1982).

     After they are verified,  the verification and calibration  statistics may
be  compared  for different  models  to  help  select  which  reconstructions  are
best.  For  example, Cook and Jacoby (1977)  calibrated  tree-ring chronologies
with drought indices in  the Hudson  Valley,  New York,  verified the reconstruc-
tions with  independent  data,  and then  used the best verfied model  to recon-
struct past  drought.  They  also  used tree  rings  to reconstruct streamflow for
the Potomac River  (Good  and Jacoby   1983).   Stockton  and Jacoby (1976) used a
grid  of  chronologies  within  the Colorado  River Basin  to  reconstruct  Upper
Colorado long-term streamflow  trends.   Some other  dendroclimatic reconstruc-
tion studies include Briffa et al (1983), Conkey  (1982b),. Duvick and Biasing
(1981), Fritts,  Lofgren,and  Gordon (1979),  Garfinkel  and Brubaker  (1980),
LaMarche and Pittock (1982),  Rose,  Dean,  and  Robinson  et  al (1981), Stockton
and Meko (1983) and Graybill (1986b).

     A large grid of tree-ring chronologies can be calibrated with large-scale
variations  in  climate  over  a geographic  grid (Fritts  et al.   1971;  Fritts,
Lofgren,  and  Gordon  1979;  Stockton  and Meko 1975,  1983;  Lough and  Fritts
1985).  These  studies used canonical   regression  of principal  components  of
tree-ring chronologies  on  principal   components  of  climate,  drought,  or
seasonal averages of the Southern Oscillation index.

SPATIAL ANALYSIS

     A total  of 65 arid-site chronologies  were selected  (Fritts and  Shatz
1975) that spanned the  period  from   1600 to  1963  with a geographical coverage
extending from the North Pacific coastal  states  to the Black  Hills  of North
Dakota and  from  the  Canadian  Rockies  to  Durango,  Mexico.   Three  sets  of
climatic data  were selected for calibration with  the tree-ring chronologies.
The first two were arrays of seventy-seven data points for surface temperature
and  ninety-six  data points  for precipitation in  the  U.S. and  southwestern
Canada.  The third set was  an  array of  seasonal  sea-level pressure at ninety-
six grid points  from  100°E to 80°W  and 20°N to 70°N.   All  data for  the years
1901 and 1961 were complete.


                                      44

-------
     A  large  number  of  statistical  models  of  different  structure  were
calibrated  (Fritts  and  Lough  1985)  and  verified  (Gordon  1982).    Stepwise
canonical regression, modified  from Biasing  (1978)  (also  see Fritts,  Lofgren,
and  Gordon  1979;  Lofgren and  Hunt  1982),  was used  to calibrate  principal
components  of  growth with  principal components  of climate.   This  stepwise
analysis reduced  the  large  number of predictor principal  components  (fifteen
or  thirty)  to  one to  seven canonical  variates.   A  transfer function  was
obtained  and  applied  to the  tree-ring principal  components  to  reconstruct
seasonal temperatures and precipitation at each station and sea-level  pressure
at each grid point from 1602 to 1962.

     The estimates from the  two or  three models with the best calibration and
verification statistics  were averaged for each variable and  season,  and  the
average  of  the  seasonal  models  was  averaged   further  to  obtain  annual
estimates.   The  calibration and verification statistics  were  recalculated
using  the  seasonal data  and the annual instrumental  values.  Each  level of
combination showed  improvements in statistics above those  expected  by chance
[See Fritts and Lough (1985) for more discussion of the model  treatments].

     It  was  concluded  from  these  results  that   the  large-scale  regional
patterns of climatic  variation  were calibrated much better than  variation at
the  individual  grid points  or  stations (Fritts  and Lough  1985).   One could
take  advantage of  this  higher reliability  of the  large-scale  patterns by
examining  regionally averaged  reconstructions or   by  averaging  results  for
several  seasons  or  years.    In  the  following  examples,   the ^individual
reconstructions  have  been combined  and averaged over  space  or  time' to  take
advantage of the greater  reliability of the combinations.

     The  reconstructions for  the decade  1831-1840 are  used  in  Figure  3 to
illustrate  the spatial  reconstructions that  were  obtained  from  analysis of
spatial  growth  patterns.   The left-hand portion  of  Figure  3  is  a  map of
average  tree  growth for  1831-1840 expressed  as departures from the long-term
average  values.    The   upper  middle  and  upper   right-hand maps  are  the
reconstructed average temperatures for winter  and spring.   Those below are the
reconstructed total precipitation for winter and spring.  Above average growth
over most  of  the map  is transferred into  cool  or cold  winter  temperatures
especially  in the northern high plains with  spring  temperatures slightly above
average  for the  northwest and southeast.   Moisture  is reconstructed as much as
     above average  for winter and  spring  for  large areas of  the map.
      The average  annual temperature,  precipitation,  and  sea-level pressure
 were mapped  by  decade  from  1801  to  1850  (Figure 4).  The east-west differences
 in temperature and  the general wetness of the  1831-1840  decade is evident.
 This was actually the wettest decade that was reconstructed, and according to
 Edward Cook  (personal communication) the tree-ring data from the eastern  U.S.
 indicate that the wetness did indeed extend eastward.  The  pressure anomalies
 that were reconstructed suggest a southward displacement  of  storms  in  the
 North Pacific and enhanced  storm activity  from the North American southwest to
 eastern Canada.

      The maps in Figure 4 for other decades indicate  that  the  1800s and 1810s
 were generally  warm,  with drought  in much  of the  west  and wetter conditions in
 the east, although the verification statistics for the Atlantic and  Gulf coast
 indicated the  reconstructions  were unreliable  that far  east and south of the


                                       45

-------
               1831 - 184O
                                        -2
                                      100
                                               WINTER TEMPERATURE
                                               WINTER PRECIPITATION
                                                                                  SPRING TEMPERATURE
                                                                                                       100
                                                                                 SPRING  PRECIPITATION
     Figure 3.   Mean anomalies  in  tree-ring width  indices  and reconstructed  temperatures and precipitation
for winter and spring of  1831-1840.   Tree-ring data are normalized values multiplied by 10, calculated using
the 1601-1963  means and standard deviations.   The climatic data  are  departures expressed  as  °C  or % of the
1901-1970 mean values.

-------
            PRESSURE
TEMPERATURE
PRECIPITATION
  K>0f  120  UO  HBO 180  IftO  140  120  100 BOW
  WOE  120  140  160  180  1*0  UO  120  100 iOW
  tOOE  120  UO  ICO  I BO  16O  140  120   100  HO
  K»t 120  140  UO _ 110 _ 1*0. _ 140 120  100  BO
         -.6   O  .6  1-0
     120  140  160  180  160  140  120  100  BOW
                          1841-1850
      Figure  4.   The mean  reconstruction  sea-level  pressure (MB),  temperature
(°C),  and annual precipitation (%)  for decades  in  the  first half  of  the 19th
century plotted as departures  or  percentages of the   1901-1970 mean  values.
Shaded areas are warm and  dry anomalies.

-------
 tree-ring  grid.    There  was  cooling  in  the  1820s  and  drought  in  the
 southwestern deserts.   Temperatures were closer to the 20th century average in
 the 1840s, and  below  average  precipitation was reconstructed  for  most of the
 country.

      The  reconstructions were averaged for eleven regions  over the contiguous
 U.S.  and  southwestern Canada;  the averages  and  standard  deviations of  the
 annual  data  before 1901  were calculated  and the  differences between  these
 figures and the  20th  century data  were  calculated  (Figure 5)  to  examine  the
 question  of how typical the 20th  century statistics  are compared to those from
 the prior  three centuries.   The temperatures  from 1901  to  1970 have  risen
 0.20°  to  0.93°C for regions  5,  7, 8, 9,  10,  and  11, but  the  temperatures in
 the remaining  five  western regions have  fallen in  the other regions.   The data
 to  the extreme right  of Figure  5  indicate  that in  the  western states  the
 standard  deviations of both  temperature and  precipitation as  reconstructed
 were  to have declined  in  the  20th century.
              19.1
     Figure 5.   The  differences in climate between the 20th century and  three
prior  centuries averaged  within  11 different  regions in  North  America;  (a)
shows  the  change in means for 1901-1970 compared  to  the mean for 1602-1900;
(b) shows  the  percent  change in standard deviations for  1901-1961 compared  to
1602-1900.
     Figure  6 includes  a  plot of  the reconstructed  precipitation  for si*
western  regions  that  have been  smoothed  using  an  eight-year 50%  pass low-
frequency  filter.    The  horizontal  line marks  the  mean of  the  instrumental
record for  1901  to  1970.   The dots  on the right show  the smoothed average*
instrumental data with which  the  tree-ring  chronologies  were calibrated.  The
amount of similarity of  the two data sets  for 1901 to 1962 is proportional to
the variance  calibrated.   (The 1901  to 1905  and 1959 to 1963 periods include
end effects of the filter.)

-------
           E

          ULf
          Cti
                 1600
                                                1850
1900
                                                             1950
                                      YEAR
     Figure 6.  The regionalized  annual precipitation reconstructions for six
western  regions treated  with  a  low-pass digital  filter  with  a frequency
response of 50% at  periods of eight years and plotted as departures from the
1901-1970  averages.     Dots  on  the   right are   the  filtered,   regionalized
instrumental data used for calibration.

-------
     Region 3  is  made up of  eleven climatic stations,  including  the area of
the Great Salt Lake with which  the  smoothed reconstructions and climatic data
were highly correlated.  These reconstructions provide an enlarged data set to
evaluate the  present  high  level of  the  Great Salt  Lake,  which  exceeds  all
previous measurements.  It appears from this time series that precipitation in
this  region has  been below  the  20th  century  mean  since  1625.    It  was
reconstructed to have  been especially  high  in  the early 1600s, and therefore,
it  is  possible that  the  current high  levels  could become higher.   However,
this extreme  climatic condition  was  uncommon  over the  last 300 years,  and
therefore a  rise in  lake level,  while  possible,  is  not  the most  probable
outcome to expect.

     The reconstructions  allow spatial analysis of  climatic variations  for
time periods  when  the coverage  of  instrumental data was  inadequate.    For
example, large explosive volcanic eruptions can inject enough ash and gas into
the  upper  atmosphere  to  alter  the  global energy  balance and  consequently
decrease the  average  surface  air   temperatures  of  the Northern  Hemisphere
(Taylor, Gal-Chen, and Schneider 1980; Self, Rampino, and Barbera 1981).   Past
empirical studies of the effects of volcanic eruptions on surface climate have
been limited by the relatively small number of major eruptions occurring after
the  beginning  of  the  20th century  and the poor  coverage of  the instrumental
data prior to the 20th century.

     Lough and  Fritts (Submitted)  used the reconstructed temperature data to
test whether   there  was  a  significant spatial  response following  volcanic
eruptions.   The years  of major  eruption,  called key dates,  were selected from
the  historical  volcanic  eruption   chronologies  published  by  Lamb  (1970),
Hirschboeck  (1979/80)  and Newhall  and  Self (1982).  There  were twenty-six
volcanic events occurring in 24 years within the period 1602 to 1900 that were
suitable for the analysis.

     The average  temperatures for  the years associated with  the  selected  key
dates  were  calculated for  the  five years  before and  for  zero to  two  years
after  the  eruptions.   The  difference  (the average for  the  years  after  the
eruptions subtracted from the average  for the  years before  the eruptions)  was
then calculated for each station and mapped,  and  the 95% confidence level  was
calculated using Student's t-test.

     The volcanic events were first divided into  three groups according to the
latitude of the  eruptions  to  test  whether  this  influenced the  subsequent
climatic impact.  These  data suggested that a  large  part of  the U.S. appears
to  cool following low-latitude  volcanic eruptions,  but significant warming in
the  far western states is evident.

     The seasonal reconstructions of temperature  were  then  examined  using  key
dates from the low-latitude data set (Figure 7).  These data indicate that the
warming  reconstructed in the  west  is  most extensive  in  winter with 3&% of
stations  showing  significant   differences.    Significant  cooling  is  recon-
structed in spring for the central states (38? of the reconstructed points are
significant).    In   summer,  a  cooling is  reconstructed  east  of the  Rocky
Mountains while a warming is reconstructed in  the  far west,  including Nevada
and  the northern  Rocky  Mountains  of  the  U.S.   (61/6  of the  differences  are
significant).  The largest differences are centered over the Mississippi River
drainage.


                                      50

-------
                                Volcanic Effect: after-before
                         .20
                                  c) Summer lemperolufe Low lalilude
     Figure  7   Average reconstructed temperature differences (-C)  between  the



is significant at  the 95% confidence level.)
Source:   Lough and Fritts,  in  press.
                                          51

-------
     The  reconstructions of  temperature variations  over North  America were
used  to determine  that there  are significant  responses in  the temperature
patterns  forced by  large  volcanic  eruptions and  that  low-latitude volcanic
activity  seems  to have  the  most obvious effect.   Seasonal  variations in  the
major  centers  of  temperature  change caused  by  volcanic  activity  also  are
apparent.

CONCLUSIONS

     The  following  conclusions  can be made with  regard  to historical changes
in forest  response  to climate variations and  other factors  deduced from tree
rings:

     •  There are many types of proxy records of past climate.  Tree rings  are
        unique  in that  they  can be obtained  from most temperate and subpolar
        forests, and they can provide information on seasonal- to century-long
        variations in climate.

     •  The rings from old,  climate-stressed  trees are  particularly valuable
        for reconstructing climate over time  periods  before the instrumental
        record began or when the instrumental measurements were incomplete.

     •  The rings must  be dendrochronologically  dated to assure the correct
        time  control,  and  usually  standardization must be  applied  to  the
        measurements to obtain a mean chronology, which is a well-behaved time
        series with a strong signal of climate.

     •  These chronologies can be interpreted directly if one climatic factor,
        such  as temperature, is both  limiting  and linearly  related  to  the
        chronology index.  An interesting exception is shown, where increasing
        ring width of high-altitude trees from the Great Bas.in looks more like
        the rising  levels  of carbon  dioxide  than  the global  warming effect.
        More work is  needed  to  establish the  exact cause of increased growth
        in this case.

     •  Calibrations  of chronology  value  with  climate  predictors  produce a
        response  function  that can  be  used to  estimate  the  effect  of &
        specific climatic change on tree growth or  to remove from a chronology
        that variance related to climate to assess forest decline effects.

     *  Calibrations   of  climate   variations   with  tree-ring  chronology
        predictors produce a transfer function.  Independent climatic data  are
        usually reserved to  allow  for  verification of the  transfer function
        result.

     *  Many  applications  use  several  tree-ring  chronology  predictors   and
        reconstruct  one climate  series at  one  time.    More complex  models
        include many predictands and  predictors.   A study to reconstruct maps
        of temperature,  precipitation,  and sea-level  pressure  from arid-site
        tree-ring  chronologies   provides  a   data  source for  past  climatic
        variations and change.
                                      52

-------
     •   The reconstructions  of  temperatures  were  used  to  examine  whether  there
        is  any significant spatial response of North  American temperatures  to
        large volcanic eruptions.   A  large part of  the eastern and  central
        United States was  found to cool in response  to volcanic eruptions,  but
        significant warming occurred  in  the  western  states.  The extent  of
        this warming is greatest in the winter and  least in  the  summer.   This
        result,   although  based  on   indirect   dendroclimatic   evidence,   is
        important  because  it  suggests  that   previous  conclusions,   which
        identified  large-scale  average  temperature  decreases,   should  be
        modified  to  include regional-scale  warming at  least  in the  western
        United States.

ACKNOWLEDGMENTS

     I  would  like to thank Janice  M.  Lough, Martin  R.  Rose,  and  Richard  A.
Holmes  for  their help with the manuscript.  Some  of the research cited was
supported by  Research  Grants GA-26581; ATM75-1703U; ATM75-22378; ATM77-19216;
ATM81-15754 and  ATM83-19848 from  the Climate  Dynamics Program of the  National
Science Foundation.
REFERENCES

Anonymous.   1977.   Tales the  tree-rings tell.   Mosaic.    8(5),  Washington,
     D.C.:  National Science Foundation.
                                                                   »»
Baillie, M.G L. 1982.  Tree-ring dating and archaeology.  Chicago:  University
     of Chicago Press.

Biasing,  T.J.  1978.    Time  series  and  multivariate  analysis  in  paleo-
     climatology.    In Time  series  and econological  processes,  ed.  H.  H.
     Shugart,  Jr.,  212-26.   SIAM-SIMS Conference Series, No. 5. Philadelphia:
     Society for  Industrial and Applied Mathematics.

Briffa,  K.R.,  P.D.  Jones,  T.M.L.  Wigley,  J.R.  Pilcher and  M.G.L.  Baillie.
      1983.     Climate  reconstruction  from  tree-rings:    Part  1.    Basic
     methodology  and  preliminary results  for England.  Climatology. 3:233-42.

Conkey,  L.E.  1982a.  Eastern  U.S.  tree-ring widths and densities as indicators
     of past climate.  Doctoral Dissertation,  University  of Arizona, Tucson.
      Ann Arbor, Michigan:  University Microfilms.

Conkey,  L.E.   1982b.   Temperature  reconstructions in the northeastern United
      States.   In  Climate from tree rings, eds. M.K. Hughes, P.M. Kelley, J.R.
      Pilcher  and  V.C. LaMarche,  Jr.,  165-8.  Cambridge:  Cambridge University
      Press.

 Cook,  E.R.  Personal communication.

 Cooes,  E.R.    In  Press.    The use  and  limitations of  dendrochronology  in
      studying effects  of air pollution  on forests.   In  Proceedings of NATO
      Advanced Research Workshop;    Effects  of  acid deposition  on  forests.
      wetlands, and agricultural  ecosystems. May 12-17,  1985.  Toronto,  Canada.
                                       53

-------
Cook, E.R., and G.C. Jacoby, Jr. 1977.  Tree-ring-drought relationships in the
     Hudson Valley, New York.  Science.  198:399-401.

Cook, E.R., and  G.C.  Jacoby.   1983.   Potomac  River stream-flow since 1730 as
     reconstructed by tree-rings.   Journal  of  Climate and Applied Meteorology
     22(10):1659-74.

Cooper, C.F, T.J.  Biasing,  and H.C. Fritts.   1974.   Simulation models of the
     effects of climatic change on  natural  ecosystems.   In Proceedings of th<3
     Third Conference on the Climatic Assessment Program (CIAP), February 26-
     March  1,   1974.     Cambridge,   Massachusetts:     U.S.   Department  of
     Transportation.

Cropper,  J.P.  and   H.C.   Fritts.     1984.     A  36p-y_e_ar__ temperature  and
     precipitation  record  for  the  Pasco Basin  derived  from  tree-ring  data.
     Final  report  to   Pacific  Northwest   Laboratories,  Battelle  Memorial
     Institute, Contract No. B-G5323-A-E.

Douglass,   A.E.    1914.    A  method  of estimating rainfall  by the  growth  of
     trees.  In The Climatic Factor.  Publication No. 192.  ed. E. Huntington,
     101-22.  Washington, D.C.:  Carnegie Institution of Washington.

Douglass,  A.E.   1919.   A  study of the  annual rings of  trees in relation to
     climate and solar activity.  In  Climatic  Cycles and Tree Growth, Vol. I.
     Publication  No.   289.   Washington,  D.C.:     Carnegie  Institution  of
     Washington.

Douglass,  A.E.   1928.   A  study of the  annual rings of  trees in relation to
     climate.    Climate  cycles  and  tree growth.  Vol.   III.   Publication No.
     289.   Washington, D.C:  Carnegie Institution of Washington.

Douglass,   A.E.    1935.    Dating  Pueblo  Bonito  and  other  ruins  of  the
     Southwest.   National  Geographic Society contribution  technical paper.
     Pueblo Bonito Series 1.  Washington, D.C.:  National Geographic Society.

Douglass,  A.E.   1936.   A study of cycles.  In Climatic Cycles and Tree Growth;
     Publication No. 289.   Vol.  III.   Washington, D.C.:  Carnegie Institution
     of Washington.

Douglass,  A.E.   1937.   University of  Arizona  Bulletin 8(4), Physical Science
     Series 1

Duvick, D.N.,  and T.J.  Biasing.    1981.   A dendroclimatic reconstruction of
     annual  precipitation  amounts   in  Iowa  since  1680.     Water  Resourcgg
     Bulletin. 17:1183-9.

Eckstein,   D.,   and E.  Frisse.    1982.    The  influence  of  temperature and
     precipitation on vessel area and ring width of oak and beech.  In Climate.
     from  tree rings, eds.  M.K.  Hughes, P.M.  Kelly, J.R.  Pilcher,  and V.C«
     LaMarche Jr.,  12.  Cambridge:  Cambridge University  Press.

Fritts, H.C.  1969.   Bristlecone pine  in  the White  Mountains of California*
     growth and ring-width characteristics.  Papers of the Laboratory of Tregj
     Ring Research 4.  Tucson:  University of Arizona Press.


                                      54

-------
Fritts, H.C.  1976.  Tree rings and climate.  London:  Academic Press.

Fritts, H.C., and J.M. Lough. 1985.  An estimate of average annual temperature
     variations for North America, 1602 to 1961.  Climatic Change. 7:203-24.

Fritts, H.C., and  D.J.  Shatz.   1975.   Selecting  and characterizing tree-ring
     chronologies for dendroclimatic analysis.  Tree-Ring Bulletin.  35:31-40.

Fritts, H.C.,  G.R. Lofgren, and  G.A.  Gordon.   1979.   Variations  in  climate
     since  1602 as  reconstructed  from  tree rings.   Quarternary  Research.
     12(0:18-46.

Fritts,  H.C.,   T.J.   Biasing,   B.P.  Hayden,  and  J.E.   Kutzbach.     1971.
     Multivariate   techniques   for   specifying    tree-growth   and   climate
     relationships and  for  reconstructing anomalies in paleoclimate.   Journal
     of Applied Meteorology. 10(5): 845-64.

Fritts, H.C.,  D.G. Smith, J.W. Cardis,  and C.A.  Budelsky.   1965.   Tree-ring
     characteristics  along  a  vegetation   gradient   in  northern  Arizona.
     Econology.  46(4):393-401.

Garfinkel,  H.L.  and  L.B.   Brubaker.     1980.    Modern  climate—tree-growth
     relationships and  climatic reconstruction in  subarctic  Alaska.   Nature.
     286:872-4.
Gordon,  G.A.  1982.    Verification  of  dendroclimatic  reconstructions.    In
     Climate from tree rings,  eds.  M.K.  Hughes, P.M. Kelly, J.R. Pilcher, and
     V.C. LaMarche, Jr., 58-61.  Cambridge:  Cambridge University Press.
Graumlich,  L.,  and  L.  B.  Brubaker.    1986.    Reconstruction  of  annual
     temperature  (1590-1979)  for  Longmire,  Washington,   derived  from  tree
     rings.  Quaternary Research. In press.

Graybill,  D.A.  1985.   Western  U.S.  tree-ring  index .chronology data  for
     detection of arboreal response to increasing carbon dioxide.   Response of
     Vegetation  to  Carbon  Dioxide  Series.   No.  026.  Department  of  Energy,
     Laboratory of Tree-Ring Research. Tucson:  University of Arizona.

Graybill,  D.A.  1986a.   A network of  high elevation conifers  in the western
     U.S.  for detection of tree-ring growth  response to increasing  atmospheric
     carbon dioxide.  Abstracts of the International Symposium on Econological
     Aspects of Tree-Ring Analysis. 17-21  August 1986. Palisades, New York.

Graybill,  D.A. 1986b.  The reconstruction  of Salt River stream-flow, A.D. 740-
     1370.     Volume  5  in  "Environment  and  Subsistence,"  The  1982-1984
     Excavations  at  Las  Colinas,  eds.  D.A. Gregory and  C.  Heathington.
     Archaeological  Series  162, C.R.M.D. Tucson:  Arizona State Museum.

Hecht,  A.D.,  ed.  1985.  Paleoclimate  analysis and  modelling. New  York:  John
     Wiley and Sons.

Hirschboeck,   K.K.   1979/80.     A  new   worldwide   chronology  of  volcanic
     eruptions.  Paleogeography.  Paleoclimatology and Paleoecology. 29:223-41.
                                       55

-------
Holmes, R.L. 1983.  Computer-assisted  quality  control  in tree-ring dating and
     measuring.  Tree-Ring Bulletin. 43:69-78.

Holmes,  R.L.,   R.K.   Adams,  and  H.C.   Fritts.    In preparation.    Tree-ring
     chronologies of  western North  America:   California, Eastern  Oregon and
     Northern  Great  Basin.   Chronology  Series.  VI.  Tucson:    Laboratory  of
     Tree-Ring Research, University of Arizona.

Huber, F.  1976.   Problems d1interdatation chez le  pin sylvestre et influence
     du climate sur la structure de ses  accroissements  annuels.  Annales des
     Sciences Forestieres 33(2):61-S6.

Hughes, M.K.,  P.M.  Kelly,  J.R.  Pilcher, and  V.C.  LaMarche, Jr.,  eds.  1982.
     Climate from tree rings.  Cambridge:  Cambridge University Press.

Keller, R.  1968.   Des caracteristiques nouvelles pour  L'etude  des oroprietes
     mecaniques des bois: les composantes de la densite.  Annales des Sciences
     Forestieres.  25:237-49.

LaMarche, V.C.,  Jr.   1974a.   Frequency-dependent relationships  between  tree-
     ring  series  along   an  ecological  gradient   and  some  dendroclimatic
     implications.  Tree-Ring Bulletin. 34:1-20.

LaMarche, V.C.,  Jr.    1974b.   Paleoclimatic  inferences  from long  tree-ring
     records.  Science. 183(4129):1043-8.

LaMarche, V.C., Jr.   1978.  Tree-ring  evidence of  past climatic variability.
     Nature. 276:334-8.

LaMarche, V.C., Jr.   1982.   Sampling strategies.   In Climate from tree ringg>
     eds. M.K.  Hughes, P.M.  Kelly,  J.R.  Pilcher,  and V.C. LaMarche,  Jr., 2-6.
     Cambridge: Cambridge University Press.

LaMarche,  V.C.,  Jr.   and  A.B.   Pittock.    1982.    Preliminary  temperature
     reconstructions   for  Tasmania.   In Climate  from  tree rings, eds.  M.  K.
     Hughes,  P.M. Kelly,  J.R.   Pilcher,  and  V.C.  LaMarche,  Jr.,  177-85-
     Cambridge:  Cambridge University Press.

LaMarche, V.C., Jr. and C.W. Stockton.   1974.   Chronologies  from temperature-
     sensitive bristlecone  pines at upper treeline  in  western  United States.
     Tree-Ring Bulletin. 34:21-45.

LaMarche,  V.C.,  Jr.,  D.A.  Graybill,   H.C.  Fritts,  and  M.R.   Rose.    1984.
     Increasing atmospheric  carbon  dioxide:   Tree-ring  evidence for  growth
     enhancement in natural vegetation.  Science.  225:1019-21.

Lamb,  H.H.  1970.    Volcanic dust  in  the atmosphere;   with  a chronology  and
     assessment   of   its   meteorological   significance.      Philosophical
     Transactions of the Royal Society of London.  266(1178):425-533.

Lofgren, G.R.,  and  J.H.  Hunt.   1982.    Transfer  functions.   In Climate frog
     tree  rings,   eds.  M.K.  Hughes,   P.M.  Kelly,  J.R.  Pilcher,  and  V.C.'
     LaMarche,  50-6,  Cambridge:  Cambridge University Press.
                                      56

-------
Lough,  J.M.,  and  H.C.  Fritts.    1985.    The  Southern  oscillation and  tree
     rings:  1600-1961.   Journal of Climate and Applied Meteorology.
     24:952-66.

Lough,  J.M.,  and  H.C.  Fritts.   Submitted.   An assessment  of the  possible
     effects of  volcanic  eruptions on North American  climate  using tree-ring
     data, 1602 to 1900 A.D.   Climatic Change.

Mass, C., and S.H. Schneider.   1977.  Statistical evidence on the influence of
     sunspots and volcanic dust on  long-term temperature  records.   Journal of
     Atmospheric Science.  34:1995-2004.

National  Academy  of  Sciences.     1975.    Understanding  climatic  change.
     Washington, D.C.:  National Academy of Sciences.

Newhall, C.B., and S. Self.   1982.   The volcanic explosivity index (VEI):  An
     estimate  of explosive magnitude for  historical  volcanism.   Journal of
     Geophysical Research. 87(C2):1231-8.

Norton,  D.A.  1979.    Dendrochronological  studies  with  Nothofagus  solandri
     (Faoaceea) and Libricedrus bidwillii (Cupressaceae).  B.Sc. Hons. Thesis,
     University of Canterbury, Christ Church,  New Zealand.

Norton,  D.A.  1983.    A  dendroclimatic  analysis  of   three   indigenous   tree
     species, South  Island, New Zealand.   Ph.D.  Dissertation.   University of
     Canterbury, New Zealand.

Parker,  M.L.  and W.E.S.  Henoch.  1971.   The use  of Engelmann  spruce latewood
     density  for dendrochronological  purposes.    Canadian Journal  of Forest
     Research. 1(2):90-8.

Polge,  H.  1963.   Une  nouvelle  methode de  determination de la  texture du
     bois:   L'analyse densitometrique de cliches radiographiques.   Annales de
     L'ecole  Nat.  Eaux et  Forets  et de  la  Station  de  Recherche  et Exper.
     20:531-81.

Polge,  H.   1966.  Etablissement des courbes de variation de, la densite du bois
     par   I1exploration   densitometrique   de   radiographies   d'echantillons
     preleves  a  la  tariere  sur  des  arbres  vivants.    Annales des Sciences
     Forestieres. 23:1-206.

Polge,  H.   1970.    The  use  of  x-ray densitometric methods  in dendrochrono-
     logy.  Tree-Ring Bulletin. 30(1-4):1-10.

Rose,  M.R.,  J.S.  Dean, and W.J. Robinson.  1981.   Past climate of Arroyo Hondo
      reconstructed from tree  rings.  Arroyo Hondo Archaeological Series.  Vol.
     4,  School of  American Research Press.

Schulman,  E.  1947.   Tree-ring  hydrology  in  southern California.  Laboratory of
     Tree-Ring Research Series, No. 4. University of Arizona Bulletin.  18(3).

Schulman,  E.   1951.  Tree-ring  indices of rainfall,  temperature, and  river
      flow-     In  Compendium   of  Meteorology,   eds.   American  Meteorological
      Society,  1024-29. Boston: American Meteorological Society.


                                       57

-------
Schulman,  E.    1956.    Dendroclimatic Changes  in Semiarid America.   Tucson:
     University of Arizona Press.

Schweingruber,  F.H,  O.U.  Braker,  and E.  Schar. 1978a.  X-ray  densitometrie
     results  for  subalpine conifers  and their  relationship  to climate.   In
     Dendrochronology in Europe, B.A.R. International Series 51.

Schweingruber, F.H., O.U. Braker, and E. Schar.  1978b.  Dendroclimatic studies
     in Great Britain and  in  the Alps.  In Evolution of planetary atmospheres
     and the climatology of the earth,  369-72.   Paris:  C.E.R.N.

Schweingruber, F.  H., 0. U. Braker, and E. Schar. 1979. Dendroclimatic studies
     on conifers from central Europe and Great Britain.  Boreas.  8:427-52.

Schweingruber,  F.H.,  H.C.  Fritts,  O.U.  Braker, L.G.  Drew,  and E.  Schar.
     1978.  The X-ray  technique  as applied to  dendroclimatology.   Tree-Ring
     Bulletin.  38:61-91.

Self, S., M.R. Rampino and J.J. Barbera.  1981.  The possible effects of large
     19th and 20th century volcanic eruptions on zonal and hemispheric surface
     temperatures.  Journal of Volcanology and Geotherrnal Research 11:41-60.

Stockton, C.W.,  and G.C. Jacoby, Jr. 1976.  Long-term surface-water supply and
     stream-flow trends  in  the Upper Colorado River  Basin  based on tree-ring
     analysis.  University of California Los  Angeles Institute  of Geophysics
     and Planetary Physics, Lake Powell Research Project Bulletin. 18:70.

Stockton,  C.W.,  and  D.M. Meko.   1975.    A  long-term  history  of  drought
     occurrence   in  western  United  States  as  inferred  from  tree  rings.
     Weatherwise.  28(6):244-9.

Stockton, C.W.,  and D.M. Meko. 1983. Drought recurrence in the Great Plains a$
     reconstructed from  long-term tree-ring records.  Journal  of Climate and
     Applied Meteorology. 22:17-29.

Taylor, B.L.,  T.  Gal-Chen, and  S.H.  Schneider.  1980.  Volcanic eruptions and
     long-term temperature records:  An empirical search for cause and effect*
     Quaterly Journal of the Royal Meterorological Society. 106:175-99

Trefil,  J.S.  1985.    Concentric  clues  from  growth rings  unlock the  past-
     Smithsonian.   16(4):47-55.

Webb, G'.E.  1983.   Tree-rings  and  telescopes.   Review:   Masca Journal
     Applied Science Center for Archaeology) 2(6):191.
                                      58

-------
How Changed Weather Might Change
American Agriculture

Paul E Waggoner
The Connecticut Agricultural Experiment Station
New Haven, Connecticut USA
ABSTRACT
     Primary  production  of  food  for  man and animals  from solar  enejyjy  is in
crops, which grow outdoors  exposed to the weather.   Because crops make food
from C02, more C02 benefits them.   If weather  changes with changing C02,  the
benefit of C02 may be tempered.

     For specified changes  in weather, yields can be  calculated  from  plant
physiology or records of  past  weather and yield; for  specified  changes  in
weather and C02,  yields  can be  calculated  from plant physiology.  Changes in
weather  may  cause disproportionate  or  nonlinear  responses  as when  plants
freeze  or a  pest  intervenes,  making  probabilities  rather  than  averages
relevant.   A  small  absolute change  in probability  may be a  large relative
change  in probability  and  an  even  larger  ratio  of  relative  change  in
probability to relative change in weather.

     Because  irrigation uses  the difference between rain and evaporation,  the
change in the supply  of  irrigation water will  logically be relatively greater
than  the  change  in  rain,  especially  if the drier weather is warmer.   Only
anecdotal  history  may  prepare  us   for  the  ramifications  of  changes  in
weather.  Adaptations like migration,  commerce, and new  varieties, species, or
husbandry  may  temper  the  impact  of  changed,  especially  gradually changed,
weather.

     In  this  paper,  I  have drawn  upon my  chapter in  Changing  Climate.  NAS
Press  (Waggoner  1983), especially calculations of  Clarence Sakamoto described
therein, and upon my unpublished manuscript prepared for  "Genetic Agraria y
Sociedad," a conference sponsored by Fundacion Valencian de Estudios Avanzados
y  el  Capitulo Espanol  del Club de  Roma.   The  long  word  precipitation  is
replaced by the short word rain,  which here means all  forms of precipitation
                                     59

-------
INTRODUCTION

     Although  agriculture encompasses  animal  husbandry and  aquaculture,  the
growing  of  crops  holds  the  honor  of  first  place,  performing  the  primary
production or  transformation of  solar  energy  to  food energy for ourselves and
other animals.   Hence, if  the primary production of  crops  fails, all fails.
Crops have a further importance because they are peculiarly susceptible to the
projected  changes  in  COp  and  weather.     COp  is  the  raw  material  of
photosynthesis, which transforms solar energy to food energy.  Crops generally
stand unprotected  in  the weather,  and even  crops  seemingly  protected  from
drought by irrigation ultimately are affected  by rain and evaporation.  Thus,
although  the  projections  of  meteorologists  are  uncertain,  agriculturalists
reasonably ask:  "How will  crop  production  be  changed by a warming of 1°C and
a 10? decrease in rain and what can agricultural scientists do?"

     American  crop  production  is  important because  it feeds us.   It  is also
important to the world  because,  for  example,  it  produces approximately 70% of
the  world's  annual 500 million ton wheat crop and  25%  of the  world's  400
million ton corn crop.

RISING C02 SPEEDS PHOTOSYNTHESIS

     The direct effect of C02 concentration is upon photosynthesis, which will
be  speeded  by  increasing  C02  above  340   ppm.    When  crops  are  grown
experimentally   with   increased    C02   concentrations,    yields   increase
approximately  1/851  per ppm C02.   Although one might  expect  other factors to
limit the benefit of  C02,  it increases  growth  whether water  or  nitrogen is
deficient (Waggoner 1983).

     At  340  ppm, net  photosynthesis is  somewhat faster in  "C4  plants"  like
maize,  than  in  "C3  plants" like  wheat.   Maize  photosynthesis,  however,  is
saturated by  only  450  ppm  C02,  whereas that  of wheat  increases  to fully 850
ppm.   Thus,  the  increase in  photosynthesis per increase in C02  is somewhat
greater for  C3 plants  than  for  C4 plants.   Because more  C02 would logically
speed the photosynthesis  of  the  less productive  C3 more than that of the more
productive C4, the gap between them would lessen or even be reversed by rising
C02.  Such  turnabouts as  more aggressive C3 weeds in  fields  of C4 maize have
been suggested (Waggoner  1983).

     In dry weather, increased C02 has another benefit.  Increased C02 narrows
stomata, which decreases  transpiration  in crops  in the field (Waggoner et al.
1964).   Baker (1965)  found  that doubling C02 from  300 to 600 ppm decreases
transpiration about 20%.

WARMER AND DRIER WEATHER ALSO AFFECTS CROPS

     If the calculations of meteorologists are correct, crops will encounter a
warmer and drier environment as well as increased C02.  The indirect effect of
COp  upon  crops  via   the "greenhouse"  effect  and  changed  weather  can  be
estimated by the coefficients of multiple linear regression equations relating
past weather and yields.  These  equations,  which are associated with the name
of Thompson  (1969), were  employed  by Clarence  Sakamoto to estimate the effect
of a  1°C  warming and  10/J less rain.   In a linear regression  of the yield of
wheat upon past weather, the coefficient for hot days is the change in tons of


                                      60

-------
grain  per  hectare per  hot day.   For example,  from the  Red  River Valley  to
Nebraska the  regressions  of wheat yields upon  past weather have  coefficients
of about -0.003 T/ha  per  June day hotter than 32°C. For a  1°C warming  and  1Q%
less  rain,  the change  in yields  calculated  from the equations  from  the  Red
River  Valley  to  Oklahoma  is 0.04-0.18  T/ha or  2-12$  less  wheat  (Waggoner
1983).

     The physiology  of  the crop  and the  physics of evaporation provide  an
alternative  to history for predicting  the  effect of  changed  temperature  and
moisture upon yield.   Duncan et  al. (1967)  combined this knowledge  into  a
computer simulation  of crop  growth,  and again, Clarence  Sakamoto employed  a
simulator of  spring  wheat in North Dakota  to calculate  the effect of  weather
changes upon wheat yield.  Unlike  the calculation  of one change in yield  for a
locality and  the projected  change  in  weather from the regression  coefficients
encapsulating  many past  years of  weather and yield,  the simulator produced  a
frequency distribution of yields  because many past years of weather, with  and
without the  projected  changes,  were  fed  into the simulator.  The consequence
(see  Figure   1) of  the changed  weather  was  many yields  and  their  frequency
distribution skewed by a  higher  frequency of low yields and a decrease of  0.2
T/ha or 2 quintals/ha  in  the  median.   Although the simulated yields  are  lower
than actual  ones  and the change  in median  yield is somewhat greater than  the
change calculated by  the  regression  coefficients, the direction and  magnitude
of the changes are similar (Waggoner  1983).
                   60-
               ui
               oc
               LL
               H
               i
               oc
KXi AcwwIWwttwr


     Cfwngtd WMthw (+1*C,-10K prteip.l
     Figure  1.   Simulated yields of spring wheat  in North Dakota showing  the
possible  effects  of  changed  weather  accompanying  a  rise  in  C0?.     The
simulation used the actual weather during 19^9-80 to calculate yields and then
weather 1°C warmer and with  10£ less rain.   Ten quintals or q/ha equal a T/ha
(Waggoner 1983).
                                      61

-------
COMBINED EFFECTS OF C<>2 AND WEATHER

     In  the end,  the  advantageous direct  effects  of C02  must  be  weighed
against  the net  effect.    In  the  United  States,  the  predicted  direct and
indirect effects  seem to cancel  producing  a net of  zero.    In  more tropical
places where little  warming is predicted,  the benefit of  increased C02 upon
photosynthesis  would only be modified by changes in rain.   At northern margins
of crops, the warming of  the greenhouse effect plus increased C02 seem likely
to produce a net benefit regardless of rain.

     Citing  complications  omitted   from the   simple   calculations above  is
easy:  Less irrigation water,  shifty pests,  expansion onto  different soils at
margins,  change  probability of extremes,  and, in  the end,  ramifications  of
changed weather that can only be foreseen from anecdotal history.

Irrigation

     In relative  terms, the change  in supply of  irrigation  water will likely
exceed the change in  rain.   Because irrigation uses the small residue between
rain  and  evaporation  rather  than  rain  itself,   the  relative  change  of
irrigation  water  will  logically  be greater  than  the  change in  rain.   For
example,  if  runoff  were &5% of rain,  a  10$  decrease in rain would decrease
runoff by 10/15 or two-thirds.   Although a decrease in transpiration caused by
COp  narrowing  stomata  may  moderate the  expected change  in  runoff  from  a
watershed supporting much foliage (Idso and  Brazel 1984),  the projected warmer
and drier weather could substantially decrease  the  water, for example, in the
Colorado River  (Revelle and Waggoner 1983).

Pests

     The ravages of pests can  amplify  the direct  effects  of C02, temperature,
and humidity, changing  yield  disproportionately.   A student  of systems would
say  the  pests  made  the  effect of weather  nonlinear.   In Europe  the Irish
potato famine,  caused by  a mildew  encouraged  by wet weather, and  in America
the Southern corn leaf blight, caused  by another  fungus prospering in humid
weather,  exemplify amplified destruction by  the combination  of a new or shifty
fungus  and  favorable  weather.    Worldwide,  the  attacks   of  locusts  or
grasshoppers exemplify  insect  pests that amplify  the impact of weather (NAS
1976).

Margins

     Parry  and  Carter  (1984)  addressed margins  between  two  ecosystems  or
farming  systems.    They  distinguished  geographical   marginality   defined  by
physical factors; economic  marginality  where returns  barely  exceed costs; and
social marginality  where  people  are  forced  from  indigenous resources into
marginal economies.   Although one  can  cite examples  of all  these, maps make
geographical margins  easy  to  visualize.   Emanuel  and Shugart  (1984) mapped
movement of Holdridge Life-Zones  that  might  be caused solely by  the warming
from a doubling of C02.   They  show, for example,  the  northern boundary of the
cool temperate steppe moving  from the prairie  provinces of  Canada to central
Alaska and  the southern  boundary  of  the cool temperate  forest moving from
Illinois to Wisconsin.   Although such  a map does  not  encompass changed rain
and C02, it does  show margins where a  change  in  weather  will cause nonlinear


                                      62

-------
effects on crops.   For example, an  Illinois  farmer  may  continue to grow corn
with yields  changing  more  or  less  proportionally  with changes  in  weather,
while a Wisconsin dairy farmer now growing silage may, however, become a grain
farmer, experiencing a disproportionate change.

     Their maps illustrate that land  limits migrating margins.  Thus crops of
the dry warm temperate forest moving  northward in proportion to the change in
weather  would  encounter  a  nonlinear  change  when  the  migrating  margins
encounter the beaches of the Great Lakes.   Although  the  margins of soil types
are too subtle to  incorporate into the maps,  they too will affect movement of
crops.    For  example, in the future  fertile prairie  soils formed  in  zones of
moderate  rain  with a  summer  maximum  might  receive  the  rain  of  a  steppe or
forest.

Frequency Distribution of Rain

     The  regression  coefficients relating  past  yields  and  weather  show  how
many T/ha will be  lost  or gained  in proportion  to changes  in  the  weather.
Should the increments  be subtracted  from  a trend of  yields,  from a regional
average, or,  as Perry and Carter  (1984)  suggest,  from frequency distributions
of  yield  produced  by   annual  lotteries  with  the  weather?    Frequency
distributions were  illustrated  above  by simulated yields of  spring  wheat in
North Dakota.

     These  are  other  reasons   to   focus  on  frequency  distributions  and
probabilities instead of averages.  The  hardship  of  less food on t&e table or
less money  in the  bank  may grow  keener in  proportion  to  trends in average
weather.   The tragedies of  famine and  bankruptcies, however,  are  caused by
falling below a limit or threshold.

     Although the  frequency distribution of the rain  accumulated  over  a long
time  such  as  a  year  follows  normal  (Gaussian)   distribution,  frequency
distributions of rain for shorter periods are squeezed by the limit of zero on
the left  and  stretched by  a few downpours  on the right.  Thus  even  in humid
New Haven, Connecticut, the distribution of July rain is  greatly skewed toward'
large  amounts  although  the  driest July  in the  84  years had a full  22  mm,
making the mean of 107 mm far  above the mode of  72  (Figure  2).   In dry Great
Falls,  Montana, the mean July rain of 34 mm is nearly three times the mode.

     Recognizing that the normal distribution would not fit the skewed distri-
butions  of  rain,  Barger  and   Thorn   (1949)  employed  the gamma  distribution
function:

                      f(x)  = xg~1  exp(-x/b)  / bg / Gamma(g)

The amount of  rain is  x mm,  the scale is b mm, the  shape parameter is  g,  and
Gamma  is  the usual  gamma function.   The frequency f(x) per mm is 0 for x less
than 0.   If  g  is  between 0  and  1, the mode is 0; and  if  e    is greater than
1, the mode is b(g-1).   The mean is bg, and the variance  b^g mm.  Skewness is
2/(square  root  of g),  smaller  g  increasing skewness.   The  fit of  the gamma
distribution  function  or f(x)  to July  rain  in  New  Haven  is  illustrated in
Figure 2.
                                      63

-------
     t
     h
     •
     d
     t
     h
                             July precipitation in New Hav«n CT USA
                              HO            220
                                       Precipitation mm
                                            330
                                                           440
Haven.
     Figure 2.   The fit of the gamma distribution function to July  rain in New
      t
      h
      s
      d
      t
      h
               28
               21
14'
                Heao-48 with b-6.  o-B and sk«u».7
                             40   1.2       l.S
                               30             61
                                       Precipitation utn
                                                           123
     Figure  3.   The parameters b and  g of the gamma  distribution illustrated
by frequency  distributions for two  hypothetical rain  climates,  both with  mean
of 48.   If g is 8  and hence skewness  0.7,  the distribution  of frequency  f  is
nearly normal and  the mode is  42.   However, if g  is 1.2 and skewness  1.8,  the
distribution  is skewed to the right and the mode falls to 8.

-------
     The parameters  b  and  g are  illustrated  in  Figure 3  by the  f for  two
hypothetical  climates, both  with  mean  rain  of  48.  If  g  is 8 and  hence
skewness 0.7, the  distribution  is nearly normal and  the mode  is  42.  If g is
1.2 and skewness 1.8, however, the distribution is skewed to the right and the
mode is only 8.

     The probability F(x)  that  rain will be  less  than x mm is apt  to capture
the farmer's  interest.   Figure  4 shows that  in the  climate with a  mean  of 48
but a  nearly normal distribution,  the probability  of less than  29 is  about
0.1; whereas  in  the  climate with the  same  mean but  a skewed distribution,  it
is about five times as great.

     The f of Figure 3 can now be  seen in a new  light.  For example, in the
normal  distribution,  f is  the  frequency 0.0163  per season of  rain of  28.5-
29.5 mm.  The  other  meaning of f, however,  is  the  increment  per   mm in the
probability F in Figure 4.   If  we shift the  limit of 29 by  1, the probability
will  change  by  f.    Alternatively,  if  climatic  change  shifts  the  entire
distribution by 1 mm, the probability F of  less than  29  will change  about f or
0.0163.
     t
     h
     m
     d
     t
     h
            1000"
             750
aoo'
             230
                M»an»4B uith b-6.  Q-B and ak*w".7 or
                             4O   1.2      .1.8*
                -1
                 30             61
                         Precipitation
                                                           92
                                                           123
      Figure  4.   The probability  F  of rain less than a limit.   In the climate
 with a mean of  48 but a nearly  normal  distribution, the probability of less
 than 29 is  about  1/10 or  .100.   On the  other  hand, in the  climate with the
 same mean  but a skewed distribution, the  probability of  less than 29 is about
 .500.
                                       65

-------
     A farmer  may,  of course, be  even  more interested in the  relative change
in the probability  of drought.   Is  the 0.0163 change large or  small  relative
to the present probability of Figure 4 to which he has adapted?   Whereas the
change in probability for  1 mm change in rain  is approximately  f,  the  relative
change in probability is  f /  F.   Because  F  is much  less  than  1  for critically
small amounts  of  rain,  the relative  change f / F per mm  in  Figure 5 is larger
than the absolute change f per mm in Figure 3.   In the example,  the  relative
change in probability of  less than 29  is  (0.0163/0.1170)  =  0.14 per mm change
in rain.   Figure  5 shows  that  the relative change  f  /  F is greater  for more
severe droughts than for less severe ones.
                                     with b»6. Q*"8 and ak«w".7
      i
      /
      F

      1
              0.4
              0.3
0.2
              O.I
              0.0
                                           4O  1.2
                                                         1.8
                 -1.0
                3O.O           61.0
                         Pr*cipit*tion mm
                                                          92.0
123.0
     Figure  5.   The  relative change  f /  F  in probability  per mm change  in
rain.   In the nearly  normal  distribution and upper  curve of the  figure,  the
relative change  in probability of less than 29 is (0.0163/0.1170)  or  0.14 per
mm change in  rain.   In the skewed distribution with  its  greater probability F
of less than  29, the  relative change is less although  the f's at 29 in Figure
3  are  similar.   The  relative  change  f  /  F is  greater for  more  abnormal
droughts and less for more frequent ones.

     A meteorologist may  be  interested in still another  sort of change:  What
is  the  relative change  in the  probability of drought  for  a  given  relative
change  in  the mean  rain?  This relative-relative change is (mean * f  /  F)
dimensionless.  Because the mean  is normally much larger  than 1,  the relative-
relative change (mean * f / F) is larger than  the relative change f /  F, which
in turn is  larger  than the absolute change f.   In the example,  the change  of
probability of less than  29  is  0.0163 per mm, the relative change is  0.14 per
mm,  and  the dimensionless  relative-relative   change  is  48 times 0.14  or  6.7
[relative change in yield] /  [relative change  in rain].
                                      66

-------
     Skewness affects  these  changes.    Although the  skewed distribution  of
Figure 3, produced by  b=40 and g=1.2, has the same mean as  the  nearly normal
distribution that we have  been examining,  it produces different  changes.   It
has a  probability of  0.424  of rain  less  than 29  instead  of 0.117,  and  the
farmer has  adapted  to  drought.  Although  the change f is only  slightly less
for  the  skewed  than  for  the nearly  normal distribution,  the   relative  and
relative-relative changes are much less because the divisor  F is  larger.  That
is, compared  to the nearly  normal distribution,  the  skewed has  an  absolute
change f of 0.0123  instead of 0.0163 per mm, a relative change  f /  F of 0.03
instead of  0.14  per  mm and a  relative-relative  change  (mean * f  / F)  of 1.4
instead of  6.7.  That  is,  a  1-mm  decrease  in rain in the drier climate with a
skewed distribution  and  higher  probability  of  less  than  a given  rainfall
caused smaller relative changes in probability.

How Frequency Distributions Might Change

     Having examined the distributions themselves, we can consider alternative
ways that, say, a 1Q% reduction would occur.  Each future month might have 90%
of  the former rain.    Alternatively,  the rain  of each future month  might be
decreased by  the absolute amount  of one  tenth of the  mean.   Although the
alternative  decreases   of  future   rain would  produce  the  same mean,  the
frequency distributions and probabilities would differ.

     If the past July  rain in New  Haven is  changed to 90$  of the amount  that
actually fell in the month,  the  variance and mean are both  decreased,  and the
Probability  of  less than  25 mm rises from  0.031  to  0.042.   If, however,  a
tenth  of the mean  rain,  10.7 mm,  is subtracted  from each past  July  rain, the
variance is unchanged,  skewness increases, and the probability of less  than 25
nun  rises from 0.031 to 0.070.

     Different,  present  climates   suggest  which  alternative might  actually
occur  if  rain decreases.  Frequency  distributions  of  rain  during three weeks
 n  April  vary regularly along a transect  from Alliance, Nebraska to Wooster,
unio  (Barger, Shaw,  and  Dale 1959).   From drier to  more  moist,  the  mean
 ncreases  by  60£,   the variance   scarcely   changes  and skewness  decreases.
similarly,  among 360 localities around  the earth, the  coefficient of  variation
^ncreases as rain  decreases  below 500 mm  (Conrad 1941).    Thus the  second
  enapf° of steady variance and increasing skewness with decreasing  rain seems
j^f6 llkely than variance  and  mean  changing  in  step.   That  is, rain  seems more
  Ke-Ly to change by an absolute rather  than  a proportional amount.
     This  means,  in  the  example of  July  rain  in  New  Haven,  that the
          would a°out double  the probability of  less than 25  mm,  which  is  a
         change  of  1.2 and a  relative-relative  change of  12.
 pall To  examine the effect  of a  10£ decrease in the  July  mean of  dry  Great
 shal?'  ^ mm  can  be subtracted  from eacn  JulY with  tne  Provis°  that  none
 sta ri  be deoreased  below 1 .   This increases skewness but  scarcely  changes the
 to o 11? deviatlon-   Tne probability of  5 mm or less  is  increased from  0.068
 inst-   ,«      Because the  July  rain in  Great Falls was  decreased by  only 3.4
         °f the  10'7 in New Haven and because  :  chose the  critical  amount of 5
     PH
 lo  li?  °f  25  for Great  Falls,  the  relative changes are about the same in both
                                       67

-------
Probability of Consecutive Dry Periods

     Because consecutive  periods  of harmful weather may  exhaust reserves and
thus be  disproportionately  more harmful than  single periods,  the probability
of consecutive events of  less  than  a  specified rain in a period gives another
view of changing climate.

     If the probabilities of,  say,  dry June and  dry July,  are F(1) and F(2),
the  probability  of consecutive  dry  months  is F(T2)  = F{ 1)  * F(2)  if the
consecutive events  are  independent.   Then the  change of  probability  of the
consecutive  events  is  f(1)F(2)  +  f(2)F(1),  and  the  relative  change  is
f(1)/F(1) + f(2)/F(2),  which  is twice the relative change  for a single event
if the two distributions are the same.

     Examination of New Haven  and Great Falls  shows no evidence that June and
July rains  are  correlated.    Thus  the above  reasoning about  great  relative
changes  in  probabilities  of  consecutive events  caused  by small  changes  in
climate is valid.   For  example, decreasing the rain of each June and July in
New Haven by 5%  of  the  mean only  increases  the probability of less than 50 mm
in June from 0.21 to 0.27  and  in July from 0.17 to 0.21.  The probability of a
dry June followed by a  dry  July,  however,  increases from 0.036 to 0.071; that
is,  the  relative changes  in   probability  are about  0.3 for  June and  July,
individually,  but for consecutive months the relative change in probability is
fully  1.0.   Because  the  relative  change in  rain is  1/20 and  the  relative
change in  probability   is 1,  the relative-relative change   in  probability  of
consecutive dry months  is fully twenty-fold, which  far  exceeds even the large
relative-relative change of  six-fold in probability for June or July alone.

     The highlights about  frequency  distribution and changes in climate can be
summarized as follows;

     •  Although distributions  of large amounts  of rain such  as  annual sums
        are normal,  the  distributions of smaller amounts like monthly sums are
        skewed toward downpours and  fitted by the gamma distribution function.

     •  The frequency per mm  in  a  distribution  is also  the change in proba-
        bility per  mm change  in  rain,  and  the skewed  distributions  of, say,
        monthly sums have  relatively high frequencies  below the mean.

     •  While  the   frequency  distribution of  rain  in  a  dry  climate  has  a
        smaller  mean  than  in  a  humid  one,  the  drier  often  has  no  smaller
        variance and  a greater  skewness,  suggesting  a  change  to  a  drier
        climate  may be  caused  by a  decrease  in each  period of  an  absolute
        amount rather than a fixed percentage.

     •  The relative increase  in  probability  of drought  caused  by a  decrease
        in rain  will  be greater  than the absolute change  in  probability and
        much greater than  the  relative change in rain,  especially for abnormal
        droughts or for  consecutive  events like two dry months.

Frequency Distributions  of Yield

     Agriculturalists must,  of  course,  relate   climate  to  yield.   Because
extremes  of  yield  rather   than  averages  starve  and  bankrupt,  frequency
distributions must be examined here, too.

                                      68

-------
     Past  wheat  yields  are  a  heterogeneous  population  trending  upward,
especially since  the  1940s.   On  the other hand,  deviations of yield  from a
curvilinear trend  are not  distributed with  significant difference  from the
normal function.  Because weather and  crops differ over the earth, yields are
not perfectly  correlated  over  a  region,  and  there  is no correlation  at all
between distant regions (Waggoner 1979).   Thus the variance of regional annual
means decreases  as the size of  the  region  increases,  and for example,  the
variance of average annual yield of wheat in the United States is only about a
quarter of  that  of Montana  winter wheat  alone.   Thus,  commerce  moderates
extremes.

     How will  the probability  of a given yield  change with a  decrease  each
season in rain by, for example, mean/10?   From 1910-72 both winter and spring
wneat in Montana  (adjusted to  1972)  yielded  0.0025 T/ha less  for each  1  mm
Decrease in the April-September rain.   The estimate was obtained  by relating
yield to  year, year2  and rain  in Great  Falls  and  Miles  City  (correlation
Coefficient = 0.79).   Because there is no  evidence  that the relation between
Ifin  and deviafcion fr<«n the trend in yield with time is not linear from 100 to
  JJ m>  I  examined the  consequences of a  10% or 25  mm decrease in  rain  by
subtracting 25 mm x 0.0025 T/ha/mm or 0.0625 T/ha from each annual yield.   The
 ecrease is  3$ Of  winter and  H% of  spring  wheat  mean  yield.   This  small
  crease  in  yield is comparable  to  those  calculated  above  by  Sakamoto
 emembering that he included the effect of 1°C warming whereas the calculation
    Montana did not.
a m       the relative and relative-relative  changes  in  yield are calculated,
whp ?nification occurs as when rain was analyzed.  Thus the decrease in winter
0 OfiS increases  the probability  of,  say,  less  than  1.6  T/ha from  0.045 to
six? i  Which  is a  relative  change of 0.6  and a relative-relative  change of
  xtold  [relative change in yield] / [relative change in rain].

     Three lessons are suggested:

        Mean yields vary less over a wider region than over a small one.

        A ]Q% decrease in rain may decrease mean yield much less 'than 10£.

     '   The probability of a  given low yield may change  relatively  more than
        the mean, and the relative change in probability may be much more than
        the relative change in rain.
          FROM HISTORY SHOW THAT CALCULATIONS CAN BE NAIVE

     Although the  disproportionate responses of  crops to changes  in  weather
     ibed aKruro  *.M«u  ,,«  t-h^  innnoniAntal  chances   in  weather  may  cause
          above  teach  us  that  incremental  changes  in  weather  may
     —«B effects, calculation finally  fails  us because weather has  so many
anrt    ations within human affairs and our own reactions are  so unpredictable
p^ influential.   Our recourse is  history,  and  I  shall  illustrate by the Irish
 *mine  as  analyzed by Woodham-Smith (1962).

We   In the  beginning of  July  1845 the  potato  crop  promised well,  and the
pvi  er waa  hot  and  dry.  The weather changed  to gloom, a new  mildew was
H eaent and  a  single crop, the potato,  failed.   More ramifications than the
                                      69

-------
mildew of one  crop were  needed to transform gloomy weather into a decrease  in
the  1851 population  of Ireland from  the  expected 9 million souls to only 6.5
million.

     Between 1779 and 1841 the population had increased by  172$, encouraged  by
an incredibly  cheap  food,  the potato introduced  from America.  Turf provided
warmth,  and  miserable standards  encouraged  early marriage.   Land  had been
divided and subdivided.  The closely-packed population and frantic competition
for  land had been caused  by  the potato, and  in 1845, the  existence  of the
Irish people depended entirely on the potato,  a productive but dangerous crop
that could not be stored from season to season.  The  stage was set.

     Once the  tragedy  began  to play, ramifications  were  incalculable.   Reli-
gion as  well as  memory  of  conquest  divided the  Irish from  a  government  in
England that was passing through a financial crisis.  The doctrine of laissez^
faire made  the government nervous  that  too much  kindness  to the Irish would
corrupt them.  The government  did not assist with seed, did not encourage the
growth of other crops, and required  the  hungry to give up all possessions and
Join the army  of paupers to gain relief.  Believers in  free trade protected
grain  exports   with  soldiers   without   noticing   that   the  traders  were
inexperienced in  importing.  Laws enacted in 1848 and 1849 forced the sale  of
estates  on  a  depressed  market,   leaving  owners  impoverished,  creditors
penniless,  and tenants with  strangers for landlords.  Typhus administered the
coup de  grace  in Ireland, and immigrants carried the  disease to England and
America, where laws  were enacted  to  increase  the cost of passage, discourage
destitute immigrants, and turn back the diseased.

     History might have been different.   Diverse crops, fewer people per acre,
patient  creditors, louse control, and imported food might have  tempered the
affects of the change in weather and yields.  The ramifications of weather and
human  reactions  illustrated by  anecdotal history show  that straightforward
calculations are naive.
WHAT CAM AGRICULTURAL SCIENTISTS DO?

     Although refining  our  estimates of  the  projected change  and  its impact
are  easy to  suggest,  these are  spectator  sports,  and  society  may expect
agricultural scientists to be participants rather than spectators.

     Steady modification  of varieties and  amendment of soils,  especially at
the  arid  and  northern margins  of  regions,  is  surely   expected of  these
scientists.  They may be expected  to accomplish this continuously  by annually
exposing their experimental plots  to  the  weather rather than by logically but
slowly  unraveling  physiological  mechanisms  and engineering  genes to  adapt
varieties and husbandry.  They will be expected to quickly devise controls fot
shifting pests.   Someone must  understand the  commerce that  feeds,  from netf
regions  of  suitable  climate, the  populations  stranded  in regions where thS
climate produces fewer crops.  Agricultural scientists will surely  be expected
to aid rather than watch mankind's adaptation to an inexorable increase in
and its greenhouse effect.
                                      70

-------
 REFERENCES

    er»   D.N.   1965.     Effects   of  certain  environmental  factors  on   net
     assimilation  in  cotton.   Crop  Sci.  5:53-56.

  at>ger, G.L., and  H.C.S. Thorn.  1949.  Evaluation of drought hazard.   Agron.  J.
     ^1:519-26.	

  arger, G.L.,  R.H. Shaw,  and  R.F.  Dale.  1959.   Chances of receiving selected
     amounts  of precipitation in  the  North  Central  Region  of  the  United
     States.   First  Report to the  North Central  Regional Technical Committee
     on  Weather  Information  for  Agriculture.   .Ames,   Iowa:     Iowa  State
     university.

  °nrad, v. 1941.  Variability of precipitation.  Monthly Weather Rev. 69:5-11.

  Un°an, w.G.,  R.s. Loomis, W.A.  Williams,  and R. Hanau.  1967.   A model for
     simulating photosynthesis in plant communities.   Hilgardia 38:181-205.

   nUf}'  ^>R-'  and  H-H-   Shugart.   1984.    The Holdridge life-zone  classi-
      ication:     Climatic  change   and  the  distribution   of  terrestrial
     ecosystems.  In Climatic change, special issue:   Measuring sensitivity to
     SilJSatic__change (In press and cited by Parry and Carter (1984)).
Idao   ^ n
    'en       and   A.J.  Brazel.   1984.    Rising   atmospheric  carbon  dioxide
     concentrations may increase streamflow.   Nature  312:51-52.

 3 ^nr1 Academy of Sciences  (US).   1976.   Climate and  food.   Washington,
     °C:  MAS  Press.                           	

    y»  M.L.,  and T.R.  Carter.   1984.  Assessing the  impact of climatic change
     in cold regions.   Summary Report SR-84-1.   Laxenburg, Austria:   Intern.
     ^st. Appl. Systems Analysis.
Reveli«  n
     j!  R-R-»  and  P.E. Waggoner.    1983.  Effects  of  a  carbon  dioxide-induced
     •Llmatic  change  on water supplies  in  the   Western  United  States.  In
    Clanging  climate,  419-32.   Washington, DC:  NAS Press.

   mprr°n'  L-M.  1969.   Weather and technology  in the production of  corn  in  the
    U-S.  Corn Belt. Agron. J.  61:453-456.

  Sgoner,  P.E.  1979.   Variability  of annual wheat yields since  1909 and among
    "ations.  Agr. Meteorol. 20:41-45.

    ri?r>  P'E*  1983.    Agriculture  and  a  climate  changed  by more carbon
    aioxide.  In Changing climate, 383-418.   Washington, DC:  NAS  Press.

  Sgoner,  Pt   E>>   J
-------
Drought Policy Implications of CO2-lnduced
Climatic Change in the United States and Australia

Donald A. Wilhite
Center for Agricultural Meteorology and Climatology
University of Nebraska—Lincoln
Lincoln, Nebraska USA
INTRODUCTION

     Drought  frequently affects portions  of the United  States  and  Australia
 nd causes  substantial economic loss,  especially in the agricultural  sector.
government  has  come  to play a key role  in  both  countries  in  attempting to
witigate  the  impact of drought.  The  organizational structure  for  responding
 ° drought  used by  federal  and  state government  in  the United States has
evolved gradually since the  1930s.  Drought assistance is provided  by  federal
government  through  a  variety  of   emergency,  short-term  and   long-term
measures.    States  are  not  required  to accept  fiscal  or  administrative
responsibility for drought assistance.

     As a direct result of drought, the federal (Commonwealth)  government of
Australia faces problems similar  to those in the  United States.  No part of
Australia is free from drought, and most of the country suffers from frequent
occurrences of  severe  drought  (Foley  1957;  Gibbs and Maher 1967;  Reynolds,
       and Collins 1983).   Only  22 of the past 100 years have been free of
        (Anonymous  1983).    Much  of  Australia's  agricultural  land   is  in
         rainfall zones where  even a minor drought has  immediate economic
  percussions (Gentilli  1971; Heathcote 1967).   Hence,  Australian agriculture
    been forced to make significant  adjustments to  its precarious situation.

     The  Australian  government began  to  formulate  drought  programs  in the
     .  Both  federal and state governments have been actively  involved since
     in the evolution of an  organization to  administer drought assistance
Programs.   Although  the  philosophies  behind  Australian  and  United  States
dpought policy  are similar,  the administration of particular policies differ
considerably.    Both  have  been the target of  criticism from the scientific
community, government officials, and recipients of relief.
                                   73

-------
     This  paper  reviews,  evaluates,  and  compares  the  drought  programs and
policies  of  state   and  federal  governments  in   the   United  States  and
Australia.  Emphasis  is  placed on governmental actions during recent episodes
of widespread, severe drought—during 1976-77 in the United States and 1982-83
in  Australia.   Recommendations  are  offered  for  improving the  capability of
government in the United States and Australia to respond to drought.  Finally,
I  speculate  on   the applications of  COp-induced   climatic  change  on  the
formulation of clear and concise drought policy objectives and plans.

THE OBJECTIVES OF DROUGHT POLICY

     Drought policy has not been stated explicitly by government in either the
United States or Australia.  The underlying  question is:  should government be
involved  in  providing assistance  to  those  economic  sectors  or  persons that
experience hardship in times of  drought?   Because  of the frequency, severity,
and extent  of drought  in  the United  States and Australia,  governments have
elected  to  provide   assistance  through  a  wide  range  of measures.    These
measures are  the  instruments of  a de  facto  policy that has  evolved over the
past 50 years of reacting to rather than preparing for periods of crisis.  The
decision  of whether  or not  to  provide  aid  has  been  based  more often  on
political than economic reasoning.

     Without clearly  stated drought policy  objectives,  the  effectiveness of
assistance measures  is difficult,  if not impossible,  to evaluate.   I  propose
three objectives  for drought policy.   First,  assistance measures  should not
discourage  agricultural  producers,  municipalities,   and  other  grdups  from
adopting appropriate and efficient management practices that help to alleviate
the  effects  of   drought.    Second,  assistance  should  be  provided  in  an
equitable,  consistent,   and predictable  manner  to   all  without  regard  to
economic circumstances,  industry, or  geographic region.  Third, the importance
of protecting the  natural  and  agricultural  resource  base  must be recognized.
Although these aims  may  not be  achievable  in all cases,  they  do represent a
model against which recent  drought measures in the United States and Australia
can be evaluated.

GOVERNMENT RESPONSE TO DROUGHT:  THE  UNITED STATES

Mid-1970s Drought

     A  recent  episode  of  widespread,  severe  drought in  the United  States
occurred in the mid-1970s.   The  years  1974,  1976,  and 1977 stand out as those
in which the greatest economic losses occurred.  The impacts of drought during
these years were most serious in the  Great Plains and upper midwest states, as
well  as  in  the  far  west.    Although the  impacts were  most  critical  in the
agricultural sector,  the municipal,  industrial, and recreational sectors were
also affected.

Mid-1970s Drought Policy and Assistance Measures

     Although many programs  are  available to alleviate  economic  and physical
hardship  caused   by   natural  disasters,  only  a  few  of  these programs  are
designed  specifically for  drought.     In  1976-77,  sixteen federal  agencies
administered  forty  separate  drought  programs.   The  total   funds  allocated
through  these  loan  and grant  programs  during  1974-77,  plus  the costs  of


                                      74

-------
 administering  the  programs,  have been estimated at $7 to $8 billion  (Wilhite,
 Rosenberg, and Glantz  1986).

     Seven programs  accounted  for the vast majority of funds disbursed during
 the  mid-1970s drought.   The  most important  of  these  was the  Farmers Home
 Administration's  (FMHA)  Emergency Loan Program.  This program provided credit
 assistance to established farmers, ranchers, and agricultural operators when a
 natural disaster caused  physical  damage to property or resulted in severe crop
 Production losses.   During 1976-77 and the  first  eight months of fiscal year
 1978,  FMHA  made  more  than  92,000  loans  totaling  $3.23 billion  (General
 Accounting Office  1979).

     A second  major  program of the mid-1970s  was  the Small Business Adminis-
 tration's  (SBA)  Disaster Loan  Program.  SBA was authorized to make necessary
 and appropriate  loans  to victims  of  floods,  riots, civil disorders, and other
 catastrophes.    Two   types  of  loans  are  available  through  SBA:    physical
 disaster loans and economic  injury loans.   Congress appropriated $1.4 billion
 for SBA to meet the demands of farmers (General Accounting Office 1979).

     The Agricultural  Stabilization   and  Conservation Service  (ASCS),  a sub-
 agency  of  the  United  States  Department  of  Agriculture, administered  the
 Disaster Payments  Program.  Under this program, a farmer whose production was
 reduced by natural disaster to less  than  two-thirds of his historical average
 Production became  eligible for payment at  one-third of the target price level
 (ASCS  1976).   The total amount  of  funds  disbursed nationally  is  not knonh.
 However, in South Dakota, Nebraska, and Texas,  this program provided more than
 $600  million  in  disaster  payments  during  the  period  from  1974  to  1977
 (Wilhite, Rosenberg,  and Glantz 1984).

     Other  significant   programs during   the  mid-1970s  drought  were  the
 Emergency Fund and Emergency Drought  Programs of the Department  of Interior
 ($130 million), the Community Emergency Drought Relief Program of the Depart-
 toent of Commerce ($175 million),  and  FMHA's Community Program Loans and Grants
 ($225 million) (General Accounting Office 1979).

     States in the United States  do  not have  fiscal or administrative respon-
 sibility for  relief measures  under   conditions of drought or other  natural
 disasters.   This responsibility has,  since  the 1930s,  rested with the federal
 Government.  State governments  have  resisted attempts to bring  them into the
 Process (Wilhite,  Rosenberg, and Glantz 1984). State  arguments  against cost-
 faring on  drought assistance measures  has been  based  on  limited  resources
 and/or the inequality of available resources among states.

Valuation of the Mid-1970s Drought Response

     The mid-1970s federal and state  response  to drought  in the United States
£as been documented  and  evaluated elsewhere (General  Accounting  Office 1979;
 "Unite,  Rosenberg,  and  Glantz  1984).   The  latter  study demonstrated  that
Governments in the United States often responded  to drought  through  crisis
"^nagement rather than through proactive programs.  This  was true not only in
 the  mid-1970s  but  also  in  previous  episodes  of  widespread  and  severe
Bought.    In crisis  management  the  time to  act was  perceived by  decision
"fckers to be short.   Reaction  to  crisis often  resulted  in  the implementation
of  hastily   prepared   assessment  and   response   procedures  that  led   to


                                      75

-------
 ineffective,  poorly coordinated,  and untimely  response.    The  studies  cited
 above  suggest  that if  planning had  been   initiated between  droughts,  the
 opportunity  would have  existed  to develop  an organized  response  that  might
 have  more effectively addressed issues and  impacts.   The  limited  resources
 available  to government  to  mitigate the  effects of  drought  also  might have
 been allocated  in a more beneficial manner.

 GOVERNMENTAL RESPONSE TO DROUGHT:  AUSTRALIA

 The  1982-83 Drought

     The  1982-83 drought  was  confined  primarily  to  eastern  Australia,  but
 portions of  this  area  had been experiencing less severe droughts for a number
 of  years.    South  Australia and  New South  Wales,  for  example,  experienced
 droughts each  year  since  1976  and 1979,  respectively  (Reynolds,  Watson,  and
 Collins 1983).  The droughts preceding  1982-83 increased  the vulnerability of
 agricultural producers to additional severe drought.

     The  consequence  of  several consecutive  years  of drought  in  New  South
 Wales was that the number of sheep declined from a peak of about 73 million in
 the  1970s  to about 43 million  in  1983.   Cattle declined  from a peak  of 9
 million in  1976  to about 4 million  in  1983.   The  1982-83  wheat  crop  was
 reduced from the  normal  7  million  to  1.5  million  metric tons,  a  loss  of
 approximately  A$825  million  (New  South  Wales  Department  of  Agriculture
 1983).  The  agricultural  impacts  of  the drought in the  other eastern states
 was similar in magnitude to that in New South Wales.

 Recent Drought Policy and Assistance Measures

     States  have taken  a more active  role  in responding  to drought  than
 American  states have.   Nevertheless,  authority for  federal  involvement  in
natural disaster relief stems  from Section 96 of the Australian Constitution,
 in which the federal government is empowered to make payments to the states on
such  terms  and  conditions  as  the  Parliament determines  to  be  appropriate
 (Department of Primary Industry 1984).

     Before  1971, natural  disaster relief  and restoration was  provided  at a
state's request by joint federal and state financing through a wide variety of
arrangements.  These financial arrangements  were on  a  one-to-one cost-sharing
basis.  No limit was set on the level of funding that could be provided by  the
 federal government.

     In 1971 the Natural  Disaster  Relief Arrangements  (NDRA) were established
whereby states  were expected  to meet a certain base  level or  threshold  of
expenditures  for  disaster  relief  from  their own  resources   (Department  of
Primary Industry  1984).   Disasters  provided  for  in  this  arrangement  were
droughts,  cyclones, storms,  floods, and  bushfires.   These expenditure thresh-
olds  were  set  according  to 1969-70 state budget  receipts  and,  therefore,
varied between states.  The  original  base  levels ranged between A$5.0 million
for New South Wales to A$0.7  million for Tasmania.

     Under the  NDRA arrangements,  the  federal  government agreed  to provide
full reimbursement  of eligible  expenditures  after  the thresholds  for  state
expenditures on natural disasters were reached.   The NDRA formalized, for  the


                                      76

-------
first time, federal/state natural disaster relief arrangements.  When NDRA was
established,  a special set of core measures, i.e., federal government-approved
drought assistance  measures, had  evolved in  each state  on  the basis  of 30
years  of   government   involvement  in  disaster  relief.  These measures  were
Particularly   relevant  to  the  needs  of  each state  because  they had  been
Designed by  state  government in response to  their own disaster experiences.
The  formalization  of  NDRA  in  1971  resulted in an increase in  the number of
core measures eligible in each state for reimbursement under this arrangement.

     In June  1978 the Commonwealth government altered two features of the NDRA
(National   Drought  Consultative  Committee  1984).   First,  the state's  base
amounts were doubled  because  inflation  had  eroded  the real  value of  the
°riginal thresholds  and  the  number  of measures  eligible  under   these  had
increased.    Second,  the cost-sharing  formula  applied  to reimbursements  under
the NDRA was changed  to  a  three-to-one federal/state  ratio  for expenditures
above the base amount,   [Note:   State base  amounts  under the NDRA agreements
were increased significantly  in  1984 following the  1982-83 drought.  In most
cases these amounts doubled the 1978 figures (Keating 1984).]

     Table  1  shows the  state expenditures  for  drought aid   fr,om  1970-71  to
1983-84 under the NDRA.   The magnitude  of  these  expenditures is significant,
especially  when compared to the limited financial responsibility of states in
the  United  States.    The governments of New South  Wales,   Queensland,  and
eastern Australia spent the most under these arrangements.   The total for all
states  was  just  over  A$570  million.    Of  this  total, approximately  A$18€
Billion was spent during 1982-83  and A$120 million was spent during  1983-84.

     Federal  expenditures for drought  aid under  the  NDRA arrangements during
the period  from 1970-71 to  1982-83 are shown in Table 2.  During this period,
Payments to the states  were just under  A$370  million,  or about A$200 million
iess than  the total state expenditures.  The  largest  share of the assistance
was provided  to Queensland and New  South Wales.  Federal expenditures on other
natural disasters totaled about  A$315 million.   Queensland and New South Wales
We^e again  recipients  of the largest accounts.

     In addition  to  the  cost-sharing  measures described above, two  federal
drought assistance schemes were available during  the 1982-83  drought.   These
included the  Drought  Relief  Fodder  Subsidy  Scheme  and  the Drought  Relief
Interest Subsidy  Scheme  (National  Drought Consultative  Committee  1984).   The
fodder Subsidy Scheme provided  a  payment  to  primary  producers in drought-
declared areas to help defray the cost  of  fodder for sheep and cattle.   The
administrative costs of this  program were covered by the states.   The amount
of the subsidy was based  on 50%  of the  price  of  feed wheat and the nutritive
value of the fodder relative  to wheat.   The subsidy  was payable  on fodder
Purchased  after September  1,  1982.  This program was  terminated on June  30,
^'83.   Fodder purchased  after  this date  was  not eligible  for the subsidy.
M°Wever, under the NDRA arrangements with the states,  primary producers were
^plowed up  to six  months  to submit  claims  after  the  June 30  termination
<*ate.   Expenditures by the Commonwealth under this  program  were  about  A$104
million during 1982-83 and A$18 million through February of 1984.

     The Drought  Relief Interest Subsidy Scheme provided payments  to eligible
Primary producers to  cover all interest payments exceeding 12% per year.  These
                                      77

-------
           Table 1.  Expenditures in Australian States Under Natural Disaster Relief Arrangements, by Type of
                     Disaster, 1970-71 to 1983-84 (A$ Thousands) (National Drought Consultative Committee  1984)
00
DROUGHT

1970-71
1971-72
1973-74
1974-75
1975-76
1976-77
1977-78
1978-79
1979-80
1980-81
1981-82
1982-83
1983-84
(estimate)
New
South
Wales
'458
987
160
1,120
2,620
3,013
66,810
31,018
53,645
21,500
Victoria



	

1,626
1,228
1,422
34,796
8,100
Queens-
land
15,623
3,143




2,785
5,165
2,208
22,768
9,608
51,982
63,300
South
Australia



	


13,580
9,257
? ??*;
£,££D

27,380
Vtestern Northern
Australia Tasmania Territory
	 596 	

	 	 __ 	

•» n?i 	 	 --.
1 7 QQCJ 	 ,_ 	
j.ffjyy —
Q,U/U ~— — — ~ • — — —
1 5 «;fin , . , .
l£,DOU 	 	 	
5^081 295 	
12,653 1,282 	
' '
TOTAL
19,458
3,601
987
160
5,769
38,212
26,927
16,993
109,720
46,002
181,738
121,500
                           184,
101,

-------
             Table 2.  Comnonwealth of Australia Payments  Under Natural Disaster Relief Arrangements,  Estimated by
                       Type of Disaster,  1970-71 to 1983-84  (&$  Thousands)
                      Source:  National  Drought Consultative Committee  1984
vo
DROUGHT

1970-71
1971-72
1972-73
1973-74
1974-75
1975-76
1976-77
1977-78
1978-79
1979-80
1980-81
1981-82
1982-83
1983-84
(estimate)
New
South
Wales
450

38
114
779
1,458
743
42,447
14,554
32,557
11,800
Victoria



	

~716
399
173
-229

22,695
4,600
Queens-
land
13,632
1,502
46
	

3,091
2,942
1,224
14,780
5,162
37,297
45,300
South
Australia



	

12,350
5,430
-270
—7^7
18,368
4,300
Western Northern
Australia Tasmania Territory
	 •]/; 	 	
Av>




21 Til _
, US — — _______
15,269 	 	
f. fHfi 	 L 	 .
6QJJ - 	
fJ*L£ ————— _._ 	 .
1 •} C->^ 	 	 	
Ljfj£j —-——.___ ____ 	
2,239 267 	
7 7^1 --- ^^«
i ** inn finn 	

TOTAL
14,098
1,502
46
38
114
3,629
32,567
15,324
7,647
70,013
22,222
118,648
81,900
               Total
104,940
28,354     124,976
39,441
69,154
883
367,748

-------
payments applied to loans taken out for primary production on or before August
31,  1982,  and  for  carry-on  purposes  after  that date.    The  states  were
responsible  for  receiving  and  verifying  claims  under  this  program.   To be
eligible, producers could not have available financial assets in excess of \2%
of the total farm debt.  This program was terminated on December 31,  1983, but
producers were  given  12 months  to  submit claims from  the  date their drought
declaration  was  revoked or  from the date  of the  termination  of  the scheme,
whichever   came   first.     Expenditures   for  the  program,   not   including
administrative  costs,  were  about  A$3  million  in   1982-83  and A$23 million
through February of 1984.

Evaluation of the 1982-83 Drought Response

     The Livestock and  Grain Producers Association  (LGPA)  of  New  South Wales
strongly commended the  state and federal governments of  Australia "for their
positive  and  cost  effective  drought  assistance   measures  which   so greatly
contributed  to  the preservation  of the  national  livestock base  over recent
years to  enable  a more  rapid  post-drought recovery" (Anonymous  1983).   How-
ever, the Working Group  for  the  Standing  Committee  of the Australian Agricul-
tural Council  (1983)  concluded,  "With  the  exception of congressional finance
and  information,  existing  policy measures, including those introduced during
the current (1982-83)  drought, do not perform well  in achieving the objectives
of drought policy which  it  considered  important.   In summary,  the nearly $300
million of expenditures was not cost effective."

     These  contrasting  views  of  the  cost-effectiveness  of   recent drought
measures in Australia reflect the controversy that currently exists over state
and  federal involvement  in drought  aid.   Several other  studies  have  been
completed (National Farmers'  Federation  1983; South Australian Department of
Agriculture 1983; Stott 1983) and others are in progress (Minister  for Primary
Industry  1984;  Australian  Academies  of  Science  1984)  to  try to  solve  this
issue.  At stake is the future role that government will play in attempting to
alleviate or  mitigate the  hardships  caused  by  drought and,  possibly,  other
natural disasters as well.

     LGPA  based  its  conclusions  about  recent  assistance measures on  the
achievement of what  it considers to  be the first  priority of  drought aid in
Australia—the preservation  of  the national  sheep  and  cattle  herd.   Through
the  preservation of  these  resources,  farm  and  nonfarm income  was  able  to
recover more  quickly  than  after previous  episodes of  severe  drought.   LGPA
estimated that,  had  government  not intervened in  1982-83, 15 to  20 million
sheep would  have  been  slaughtered.   As a result,  post-drought recovery would
have been delayed  at  a cost to  the national  economy of A$500  million over  a
five-year period (Anonymous 1983).

DROUGHT POLICY COMPARISONS

     United States and  Australian drought  policies are compared  in  Table 3-
The   principal   policy   features   are   grouped   into    three   categories:
organization, response, and evaluation.

     Organizational features are planning activities that  provide timely and
reliable assessments, such  as  a  drought  early warning  system,  and procedures
for  a coordinated and efficient  response, such as drought declaration.  Thes®


                                      80

-------
              Table 3.   Comparison of Drought Policy Features:   United States and  Australia  Status as  of 1984
                        Features
                                              United States
                                             Australia
00
ORGANIZATION:

National drought plan

State drought plans

National drought early
warning system

Agricultural impact
assessment techniques

Responsibility for
drought declaration

Geographic unit
of designation

Declaration procedures

RESPONSE:

State fiscal responsibility
for assistance measures


State administrative responsi-
bility for assistance measures

Eligibility requirements and
provisions of drought assistance
measures

National crop insurance program

EVALUATION:

Post-drought documentation
and evaluation of procedures
and measures
None

In selected states

Joint USDA/NOAA
Weather Facility

Available, but generally
unreliable


Federal


County

Standard for all states,
varies by program/agency


Negligible, if any
                                                        No responsibility for
                                                        federal measures

                                                        Standard within programs
                                                        for all designated
                                                        counties

                                                        All-risk federal program
                                                        No routine evaluation by
                                                        government
Study in progress

Through NDRA agreements

Bureau of Meteorology


None available



State


Unit varies between states

Varies between states;
standard within states
Defined by NDRA agreements
up to base amounts, varies
by state

Defined by NDRA agreements
and by federal measures

Varies by state for NDRA core
measures, standard for federal
programs

Rainfall insurance feasibility
study in progress
                                    Routine evaluation by federal
                                    and state governments

-------
characteristics would  be the foundation  of  a national drought  plan.   Only a
few states  in the United States  have drought plans  (Wilhite  and Wood 1985).
State drought plans  exist only  in  a loose  form  in Australia under  the NDRA
agreements.

     Response features refer to assistance measures and associated administra-
tive procedures that are  in  place to assist  individual citizens or businesses
experiencing  economic  and physical  hardships because  of drought.   Numerous
assistance measures are  available in the United  States  but  few are intended
specifically for drought.  Relief arrangements  in  Australia  are, for the most
part, included under the  NDRA agreements.  An all-risk crop insurance program
has  been  evolving in  the United  States  since  1939 (Federal  Crop Insurance
Corporation  1980).     The Australian Bureau of   Agricultural  Economics  is
currently studying the feasibility  of a rainfall  insurance  scheme.   Hail and
flood insurance is provided by commercial insurance companies in some areas.

     Evaluation of organizational  procedures and  drought assistance measures
in the  post-drought  recovery period  is  the  third  category  of drought policy
features.    Governments  in  Australia have been more  conscientious  in  their
evaluation  of  recent drought   response   efforts.    In  the  United  States,
government does  not routinely  evaluate  the  performance of  drought  response
procedures or drought assistance measures.    An  evaluation  of  the  1976-77
drought response activities  was  made by the  General  Accounting  Office (1979)
at the request of the chairman of the Subcommittee on Environment, Energy, and
Natural Resources, the late  Congressman Leo  J.  Ryan.   Wilhite, Rosenberg, and
Glantz  (1984)  evaluated  governmental response to  the mid-1970s  drought under
sponsorship  of  the   National  Science  Foundation.    These  were  the  first
systematic evaluations of federal efforts  to  respond  to drought in the United
States.

     For government in the United States to improve its drought assessment and
response capability significantly,  progress  must  be  made in  four  key areas.
The  Australian experiences  suggest  that similar needs  exist within  their
drought assessment and response system.

     First, reliable  and timely  informational products  (advisories,  reports,
management  recommendations)   and  information  dissemination  plans  must  be
developed.  This has also been suggested as a high priority in Australia.  For
example, few can question the significance of more reliable  and timely infor-
mation  about   appropriate  drought management strategies.   Such  information
could reduce   the  effects of drought  as well as the  need for  government
assistance.   Campbell  (1973)  has  argued that  Australian  farmers have  not
exploited the available management strategies to their fullest.  Government or
the private sector should provide information to producers, not only about the
relative costs and benefits of different management strategies, but also about
the probability  of droughts of  various  duration  and intensity.   Government
must also more effectively inform potential  recipients  about the availability
and provision of drought assistance measures.

     Second,   impact assessment  techniques must be improved.    In  the  case of
agriculture,   which  is  usually  the  first   economic  sector  to  experience
hardships from drought, new  tools must be developed to provide decisionmakers
in government  and  business with  the types of information needed to identify
the onset and  termination of drought and  to  better understand the severity of


                                      82

-------
drought and  its  likely impact.   These tools would  be used by  government to
identify periods  of abnormal risk and to trigger various assistance measures.

     Third,  designation  procedures in  the  United States must  be centralized
under a single agency or committee with complete authority to determine eligi-
bility  for  all  assistance  programs.    Criteria  must  be  determined  before
drought occurs and must be  well  publicized  when drought occurs and applied
consistently to all affected states, counties, and localities.

     In Australia,  the declaration of drought areas is a state responsibility,
and procedures differ considerably between  states.   It may  not be feasible to
standardize  procedures between  the states because of  the  large precipitation
gradients that exist over much of  the  country.   In the United States, drought
declaration  decisions  are  a federal  responsibility,  considered  at  a state's
request.  Declaration procedures vary between agencies and,  at times, between
Programs and within agencies.  Drought  policies  on revocation  of declarations
"lust be better defined  in  both countries and take into account the lingering
effects of drought.

     Finally,  assistance measures must  be developed  before drought occurs,
i*e'» a  proactive  approach  must  be taken  to  avoid   the  delays  in program
formulation  and  congressional  approval that  occurred  in  the  United  States
Curing the  mid-1970s.    Programs  should  be administered by a  single  agency
through the mechanism  of an interagency  committee in  which federal agencies
with  responsibility  in  drought   assessment  and  response  are  represented.
Representatives of  the  affected states and/or  regions should be included in
 he  membership  of  this committee.    Assistance  measures  must address  the
specific problems associated with drought.

     Another question  deserving considerable attention in  the  discussion of
national  drought   policy   is  the   degree  of   fiscal  and   administrative
Responsibility that states  should have in support of assistance measures.  The
ftustralian approach  of sharing  the costs  of these programs  has  been  quite
successful and may be applicable in the United States.   Such an approach would
j^low states to have greater fiscal and administrative  control  over assistance
""easures.   These measures  could also be tailored  to reflect the unique water
SuPply problems and specific drought-related impacts of each state.

     More attention should  be  directed  to  the development  of  assistance
J-asures that  encourage  producers to  incorporate appropriate levels of risk
r5nagement in individual farm plans.  Recipients  of  drought  aid would benefit
  om  knowing   in  advance  what  types  of assistance  will  and  will not  be
£ °vided.    Generally,  Australians prefer  assistance   in  the  form  of  loans
su°aUse recipients retain the flexibility to use  the money  in  a way that best
tj|lts their  farming situation; that is, farm management decisions remain with
0 e  farmer.   Loans  also have  an important  secondary  effect:    farmers  can
Co*1  nue to spend  at  relatively normal levels and the economy of neighboring
  ^unities  is not disturbed substantially.   Equity  requires that  loans be
  de available to all.   The  Australian government has concluded  that feed
* aerves and  freight subsidies  for water  and feed  can discourage adopting
             risk  management   techniques.     These   measures   promote  soil
            by keeping livestock  on  the land during periods when the vegeta-
     is severely  stressed.

-------
IMPLICATIONS OF CLIMATIC CHANGE ON DROUGHT POLICY

     Recurring periods  of severe drought frequently  affect large portions of
the United  States and  Australia.   Past efforts by  federal and state govern-
ments  to  respond  to  these  events  have been  largely  ineffective  and poorly
coordinated.      Predictions  of   climatic   change   caused   by   increased
concentrations of carbon dioxide (Ct^) and other gases, such as  fluorocarbons,
in  the  atmosphere are  cause  for  concern.    These  predictions  have  been
discussed extensively  elsewhere  (National  Academy  of Science  1982;  Hansen,
Volume  1;  Manabe  and  Wetherald  Volume  1.)    These  changes   in  climate may
substantially alter existing  regional  water  supplies, leading to an increased
frequency in the occurrence of severe drought.

     Many mathematical  models  have  been  used  to  predict  the effects  of
increasing COp and other  trace gases  on changes in the temperature and preci-
pitation regimes  of global  and  regional climates.   For  example,  Manabe and
Wetherald (1980) have used a  model  of global  climate to test the effects of a
doubling and  quadrupling of  the  preindustrial  carbon  dioxide  level.   Their
results can be  summarized as follows.   First,  the temperature in the surface
layers of the atmosphere  will increase  by about 3°C in the latitude zone from
approximately  35°  to   50°N.    Second,  precipitation  will  increase  in  the
latitude zone between approximately  12°  to 37°N and decrease in the zone from
37° to 50°N.    Third,   evaporation  will increase  slightly at  all  latitudes.
Fourth, a  net  increase will occur  in  available  water  between 12°  to  37°N
latitude but  a  net decrease  will occur between 37°  to 50°N.   Finally,  minor
changes in  soil moisture will occur south of  37°N latitude but will decrease
significantly in the latitude zone between about 37° to 47°N.

     If these predictions are accurate, we  can infer  that much of the major
food-producing areas of the  United States,  Europe, and  the  Soviet Union may
become drier  and  less  productive,  while other areas  may become wetter.   A
decline in  available water  supplies will be especially critical for marginal
agricultural zones, e.g., the Great Plains of  the  United States.  The messag6
seems clear—some regions will be winners,  others losers.  However, regardless
of  the direction  of  the change,  many economic  sectors  will  be  affected»
including agriculture  and forestry,  transportation, energy,  recreation,  and
health.  In addition,  these changes will influence  the formulation of publi°
policy and alter demographic patterns.

     In Australia,  as   in  the United States,  the  predicted  climatic  change^
will be highly  regional in character.   The signs  point to an increase in fcne
intensity,  duration, and southern penetration of  the present summer rainfall
regime, except along the  southern coast, and  to a  decrease in winter rainfall
in the southwest  (Pittock  1983).  The  major  agricultural regions of the soufc*1
central portion of the country are expected to experience no change in current
winter rainfall,  and rainfall in  the southeast  may  increase from 10/t  to 3^1"
These  predictions may  appear more  favorable  than  they  actually are  sine6
summer rainfall is of high  intensity  and results in high runoff.  Evaporation
rates during the summer are also quite high.

-------
     Large portions  of  Australia are  considered  marginal for  agriculture.
Therefore? any  change  in the  current climate that  results in a  decrease  in
available  water supplies  will  have  substantially  greater  impacts  on  the
economic   sectors  noted  above  than   will  occur  in  the  United  States  with
comparable changes.

SUMMARY AND CONCLUSIONS

     The   purpose  of  this  paper  is  to  compare  recent  drought  policy  in
Australia  and  the United  States and to offer  recommendations for policy change
*n the United  States.   Four critical  needs were  identified:   (a)  reliable and
timely informational products  and dissemination plans  that  provide  producers
With better  information about drought, alternative  management  strategies,  and
Bailable  assistance measures;  (b) improved  assessment  techniques,  especially
 |J  the agricultural sector,  for  use by  government to  identify periods  of
^normal  risk and  to  trigger  assistance measures;   (c)  administratively  cen-
t^alized  drought declaration  procedures that  are   well  publicized and  con-
Sl3tently  applied;   and  (d)   standby   assistance   measures  that  encourage
Appropriate  levels  of  risk  management  by  producers  and are  equitable.,  con-
 istent, and  predictable.  These  measures  must not  discriminate  against  good
 arm managers.  Although  aimed at governments  in the United  States,  most  of
Jtese recommendations  will be applicable to  drought  policy in  other  countries
as well.

     Governments in the  United States  have  responded   to  drought by  crisis
"^agement rather  than  risk management.    This approach  has  been  grossly
 n®ffective.   Several  recent  studies  have  addressed  the  issue  of  drought
^o;Licy, or lack  of  it,  in the United  States  and  have concluded that  we should
 °W move   toward drought  planning  with  the  aim  of   improving  its  efficiency.
 ne development  of  a national drought plan is proposed  as  an effective way  of
 •"Plementing  these  recommendations in the  United States.    In Australia,  two
 Atonal drought committees are considering the  benefits of a national.drought
Policy that would be the basis  for a  drought plan.   The U.S. National  Climate
 fnnhas  recently supported  the establishment of a  national drought  plan
       on  Atmospheric  Sciences and  Climate  1986).   A recent call  for  the
           of national drought response plans has  also come from the World
     rological Organization (1986).

f    An appropriate  question  to ask  at  this  point is:  should we have a  plan
^ dealing with the impact  of drought?  To  answer  that question,  let  us  pose
0 other question.   Have  previous  approaches  been successful?   This question
th? ke &nswered in  terms of the  drought policy objectives  raised earlier  in
  13 Paper.  The first  objective was  to  determine  whether the current approach
.   Policy  encourages adopting  appropriate  and efficient management  practices
ou *nsure  against abnormal risk.  It would appear that  it  does not.  In fact,
...^ent  policy  often   discourages  wise   risk-management   decisions    by
      era.   For example, tax  incentives  encourage the  plowing  of marginal
       When drought occurs, farmers often receive  assistance for the  losses  of
      where such losses were  inevitable.

(jn.  The  second objective was  to determine whether  drought  policy  in  the
ha  ed States is equitable,  consistent,  and predictable.   Previous  studies
H  ® shown that  it  has not been so.   In fact,  the  opposite has been  true  of
   t drought response  efforts.   A national drought  plan  would  help to  rectify


                                      85

-------
 this  situation  by  focusing  attention  on  the  policy  objectives  and  on  efficient
 means to  achieve them.

      The  third  objective was  to assess whether  the  current approach recognizes
 the  importance of protecting our  natural  and agricultural  resources.    The
 current  approach  appears to  recognize  the  need, but  assistance measures  are
 often implemented  in  such  an ineffective and untimely manner  that this objec-
 tive  has not been realized.  A  national drought  plan  would  promote greater
 recognition and preservation  of natural  resources.

      A national drought plan  would encourage states  to take a  more active role
 in  planning for drought.    In fact, drought  planning  should be coordinated
 between  the states and  federal  government.   In  the  past,  most  states have
 played a  passive role, relying almost exclusively on the federal government to
 come  to  the assistance of  residents  of  the drought-affected  area.   Although
 federal  government has accepted  this role, improving  government  response to
 drought   requires  a  cooperative  effort.   States must  develop  their  own
 organizational  plans  for  collecting,  analyzing, and disseminating information
 on drought  conditions.  Cost-sharing  of  drought assistance measures should be
 pursued as a means of involving state government in drought assistance.

     The  evidence presently available indicates that increasing concentrations
 of COp  and other gases are  likely  to result in changes  in  climate  that will
 significantly alter  regional water supplies,  at  times  intensifying  existing
 water  management  problems.   For  drought-prone  regions,  more  logical  and
 systematic planning for short-term, drought-related water shortages 'today may
 provide  future  generations  with  strategies that  are  appropriate for  a  new
 climatic  regime.


 REFERENCES

 ASCS.   1976.   Annual  Report. Nebraska.  Lincoln, Nebraska: U.S. Department of
    Agriculture.

 Anonymous.  1983.   LGPA submits priorities for government assistance in future
    drought situations.   Livestock and  Grain  Producer 6, no.  12  (December )'•
    1-3 •

 Australian  Agricultural  Council.  1983.  An  evaluation  of  existing  drought
    policies  given the  current  drought experience.    Report  by  Standing
    Committee on Agriculture Working Group, Canberra.

 Australian  Academies  of Science.    1984.   National  strategy for  drought^
    background and  objectives.  Notes for joint study  of Australian  Academic3
    of Science.   Prepared by Garth Paltridge.  Aspendale, Victoria, Australia1
    CSIRO.
Board on Atmospheric Sciences  and Climate.   1986.   The  national climate
    gram;    Early  achievements  and  future  directions.   Washington,  D.C
    National Academy Press.
                                      86

-------
Campbell,  K.O.   1973.   The future role of agriculture  in  the  Australian  eco-
    nomy.   In The environmental, economic and social  significance  of drought.
    J.  V.  Lovett,  ed.  Sydney:  Angus and Robertson.

Department of Primary Industry.  1984.  Review  of  the Natural  Disaster Relief
    Arrangements.   Prepared for the National  Drought Consultative,  Canberra.

federal Crop Insurance  Corporation.    1980.   An inside look at  all-risk  crop
    insurance.   Washington,  D.C.:  Federal  Crop Insurance Corporation.

p°ley,  j.c.  1957.   Droughts  in  Australia:    Review of records  from earliest
    years  of settlement to  1955.  Bull. No.  43.  Melbourne,  Australia: Bureau
    of  Meteorology.

General Accounting Office.   1979-   Federal  response to the 1976-77 drought:
    What should  be done  next?   Report  to the  Comptroller General.  Washington,
    D.C.:  Government  Printing  Office.

^ntilH,  J., ed.   1971.   Climates of Australia and  New Zealand.   Amsterdam,
    Elsevier.

Gifebs,  w.J., and  J.V. Maher.  1967.  Rainfall deciles as drought  indicators.
    Bull.  No. 48.   Melbourne,  Australia:  Bureau  of  Meteorology.

Heathcote,  R.L.    1967.  The effects of past drought  on  the  national economy.
    In  Report of the A.N.2.A.A.S. Symposium on  Drought.  Melbourne,  Australia:
    Bureau of Meteorology.

      I,  P.J.   1984.   Payments to or  for the states, the Northern  Territory
    and local government  authorities   1984-85.  Budget Paper  No. 7.  Canberra,
    Australia: Treasurer of the Commonwealth  of  Australia.

      , S.,  and  R.T.  Wetherald.   1980.   On the distribution of  climate change
    resulting from an increase in  C02  content of the atmosphere.  Journal of
    Atmospheric  Science. 37:99.

 lnister for Primary  Industry.   1984.   Report of National Drought Consultative
    Committee meeting.  Media  release.   March 30,   1984, Canberra,  Australia:
    Ministry for Primary Industry.

 ational  Academy of  Science.   1982.   Carbon dioxide and  climate:    A second
    assessment.  Washington, D.C.:  National Academy  Press.

 ational Drought Consultative  Committee.   1984.   Drought assistance—financial
    arrangements.   Notes  from meeting, March  28,   1984. Canberra,  Australia:
    National Drought  Consultative  Committee.

  tional  Farmers'  Federation.   1983.   Drought  policy.   Canberra,  Australia:
    National Farmers'  Federation.

  w South  Wales  Department of Agriculture.   1983.   Drought policies.  Sydney,
    Australia: New South Wales  Department  of  Agriculture.
                                      87

-------
 Pittock,  A.B.   1983.   The carbon dioxide problem and  its  impact.   Meteorologl
     Australia,  February.

 Reynolds,  R.G., W.D. Watson, and D.J. Collins.   1983.   Water  resources  aspects
     of  drought  in  Australia.    Water  2000:    Consultants  Report No.   13-
     Canberra, Australia:  Australian Government  Publishing  Service.

 South Australian Department of Agriculture.   1983.  Rural  adjustment:   Interim
     report  on drought  relief  measures.   Submission  to Industries  Assistance
     Commission  Inquiry,    Adelaide,  Australia:  South   Australian  Treasury
     Department.

 Stott,  K.J.   1983-   An  economic assessment  of assistance  measures  for  the
     1982-83  drought  and  for  future  droughts"InternalReportSeriesT
     Victoria, Australia: Department of Agriculture.

 WESTPO.   1977.   Directory  of  federal drought  assistance.  Washington, D.C.:
     United States Department of Agriculture.

 Wilhite,  D.A.  1983.    Government  response  to the mid-1970s drought:   With
     particular  reference to  the  U.S.  Great  Plains.    J. Climate  and Applj.
     Meteorol. 22:40-50.

 Wilhite, D.A., N.J. Rosenberg,  and M.H. Glantz.  1984.  Government response _tg
     drought  in  the United  States:   Lessons  from  the  mid-1970s.   Parts 1^4•
     Final   Report   to   the   Climate  Dynamics  Program,  National  Science
     Foundation,  Progress Report 84-1  to 84-4.  Lincoln,  Nebraska:  Center for
     Agricultural Meteorology and Climatology, University of Nebraska—Lincoln-

 Wilhite,  D.A.,  N.J. Rosenberg,  and M.H.  Glant2.    1986.   Improving  federal
     response to drought.  J. Climate and Appl. Meteorol. 25(3):332-342.

Wilhite, D.A., and D.A. Wood.   1985.  Planning for drought:  The role of state
    government.  Water Res.  Bull. 21:31-38.

World Meteorological Organi2ation.   1986.   Memo to  Permanent Representatives
    of Members of WMO from G.O.P.  Obasi.  May 14,  1986.  Geneva, Switzerland:
    World Meteorological Organization.

-------
An Assessment of the Potential Economic
Impacts of Climate Change in Oklahoma

Ellen J. Cooter
Oklahoma Climatological Survey
Norman, Oklahoma USA


BACKGROUND

  2  The  State of Oklahoma has an area of  approximately 177,816  km2  (68,655
*  )  and  is  located  in  the  southern  Great  Plains region  of  the  United
        it has a population  of  three million people  and  its economy is  based
 ? oil,  natural gas,  and  agriculture.   Major  agricultural activities are
 ivestock  and  winter  wheat  production.     Other  important  agricultural
Commodities include  sorghum,  cotton, hay,  and some corn.  The average freeze-
Iree  period  ranges  from 181  days  in  northwestern Oklahoma  to  217. days in
s°utheastern regions.   Precipitation ranges  from 52  cm (20.42  in)  in north-
Jjestern  Oklahoma to  138 cm  (53.75 in) in southeastern parts of the state.
 Urrently,  the region  is  suffering under unstable markets for both energy and
a?ricultural products.   A projected climate  change  (Hansen,  Volume  1; Manabe
    Wetherald,  Volume  1)  would cause additional  stress  which  would  have
       implications for the economic future of the state.  Therefore,  as part
   our state-mandated  mission,  the Oklahoma Climatological  Survey  performs
        ^analyses of  the impact of weather  and climate on  food production
       (Cooter and Haug  1986).   As climatologists we  would like to be able to
        our clients  with  climate information  that  is relevant  to  their
           In  the present  case,  the agricultural community,  or at  least the
                 who  represent the community, tell us what is  "relevant" to
     through the structure  of their crop-yield models.   The present analysis
   Potential agro-economic impacts resulting from a hypothetical change in the
             over  Oklahoma will begin by  describing the climate-change hypo-
j esi
  ^a°ts on estimated production according to the model are presented, followed
    a, followed by a brief description of the crop-impact  analysis models.
   more qualitative  discussions  of  potential  impacts  from  the  hypothesized
lo      change  on supplemental  irrigation requirements,  pest and  pathogen
r Sses,  rate  of  maturity,  and  field  work  days.   The  summary  concludes with
 Commendations  for  further  research development.   Although this  analysis
 WUSes on changes  in precipitation, we plan to address  temperature changes in
      research.
                                   89

-------
CLIMATE CHANGE HYPOTHESIS

     We begin  by  assuming that  a  representative hypothesis  for  the southern
Great  Plains  is  a  10%  decrease  in  precipitation.    Because  we are  most
interested in agricultural impacts, we  assume  that  this climate change occurs
during the Oklahoma growing season, roughly April 1  through September 30.  The
next task is to determine  how  this 10/1  change might be distributed throughout
the  growing  season.   It  is  reasonable to  assume  that not every  storm event
will be modified by the same fraction since rainfall can result from a variety
of  environmental  instabilities.    As  a  first guess,  observations of storm
development  and  behavior  in southwestern Oklahoma suggest  the precipitation
changes illustrated in Table 1, which assumes that storms exceeding 12 mm (0.5
in)  would not  be changed, while storms with  less  than 6 mm (0.25 in) would
decline  30$ to  50%.    This  hypothesis  implies  that  those  synoptic  scale
features  that   control   southwestern  Oklahoma's precipitation pattern  will
expand northward and eastward.  If this  categorical  hypothesis is applied to
each storm across the state  over many growing seasons, anticipated precipita-
tion changes range from  a  loss of  70  mm (2.75 in) in southeastern Oklahoma to
39 mm (1.52  in) in northwestern Oklahoma  with an area weighted state precipi-
tation change of 57 mm (2.21 in).

PRODUCTION IMPACT MODELS

     Two types of agricultural crop-yield models were available to address the
hypothesized climate change  impacts.  The first—and  most widely used because
of  its  modest  data and  computing  requirements—is  the statistical regression
model.   Regression models represent statistical  modeling of past point  or
regional data.   The second type of model  is geophysical plant-process simula-
tion, or a  carbon-cycle  model, so  named  because  they are usually "driven" by
the  photosynthetic process.   Plant process models  are designed to mathemati-
cally simulate physical  processes of  plants.   Although  these relationships
derive from  or may  be  verified  by  field  or laboratory observations, they are
not  bound  to existing or  past conditions and, given  existing soils and crop
varieties, are able to  respond in a realistic fashion  to  new environmental
conditions.  Plant  responses to altered  levels  of  any given atmospheric con-
stituent could be used in similar models by analyzing data collected in growth
chamber experiments.  Verified plant process models were not  available for all
of  the selected study crops;  therefore, a family of plant process and regres-
sion models  was  assembled.  The selected crops and  the  models used in their
analysis are presented in Table 2.

     Each regression model predicts  crop  yield as a  function of time, tech-
nology,  and  weekly weather  at  predetermined  dates  of critical phenological
stages.   Cultural and  technological  influences are  assumed  to  be contained
within trend and error terms.   Thus,  once the model has been constructed, the
only inputs required for executing  the  model are weekly averaged weather.  The
plant process  models are somewhat more  complex,  which is  reflected  in the
number and  detail  of  required  input  (see  Table 3).   The model estimates crop
yield in  terms of total biomass and  grain.   In addition, the model estimate^
dates of critical growth stages.   Input data are readily available across the
United States, and reliable yield and phenology estimates have  been made usinjj
these models on  scales  as large as  the  U.S.  corn belt.   For instance, 19*5
corn belt  production  estimates  were  within 5% of  final  USDA figures.   Esti"
mates of  critical phenological  events  such  as  silking and  maturity  were


                                      90

-------
           Table 1.   The Distribution of a 10$ Decrease in Growing
                     Season Precipitation by Storm Category
          24-Hour Precipitation          Precipitation Decrease

                          April,  May,  and June

                rain <_ 6 mm                        50%

                6 mm < rain £ 12  mm                25%

                rain > 12 mm                    No Change


                         July,  August,  and September

                rain £ 6 mm                        30#

                6 mm < rain < 12  mm                25%

                rain > 12 mm                    No Change
          to be within one  week  of observed values across  the  region  (Botner
et al.  1986).
        INPUT DATA

     The weather  data selected for use in this analysis between  1960  and  1984
    from the  National  Weather Service cooperative  observing  network.  These
    daily data  which  include maximum and  minimum  temperatures  and  24-hour
    ipitation (Sladewski 1986).   Daily observations of solar radiation which
an^  required  by  the plant process models are not available  from  this  data set
    therefore were simulated  using  a statistical  solar  radiation generator
         French,  and  LeDuc  1983),  based on  the work  of Richardson  (1982).
     moisture, which  is  required  by the cotton  regression  model,  was estimated
      the process model soil  water budget (Jones and Kiniry  1986).   Seventy-
~^ven weather data locations  were  selected  across  the  state,   one  site  to
  Present each of Oklahoma's  seventy-seven counties.

      RESULTS AND ANALYSIS

     The potential  impact  of  the hypothesized  change  in  rainfall  on  crop
         i  is  simulated by  altering  24-hour  precipitation  at  each  county
           All other  weather inputs remain as before.   For  each season  and
     year,  the   difference   between  modeled   crop  yield  under   natural   and
                                     91

-------
Table 2   Study Crops and Direct  Impact  Models Used to Assess the Potential  Impact
          of a 10$ Change in Growing Season Precipitation on Food  and  Fiber  Produc-
          tion in Oklahoma.
 CROP
                              MODEL
 Winter Wheat
  Corn
  Sorghum
  Hay
  Cotton
         CERES  - Wheat
         J.  T.  Ritchie and C.  A.  Jones
         Institute of Water Research
         Michigan State University
         East Lansing, Michigan

         CERES  - Maize
         C.  A.  Jones and J. R. Kiniry
         Texas  A&M University  Press
         College Station, Texas

         DY = 2004.6 + 276.7 * PPT24 + 372.A * PPT34

              - 33.6 * TEMP35  + 318.6 * PPT36

where:      DY  = estimated detrended yield (kg/ha)
         PPT24  = weekly precipitation for week 24 (planting)
         PPT34 = weekly precipitation for week 24 (heading
                 to dough)
        TEMP35 = weekly average temperature for week 35
                 (dough)
         PPT36 = weekly precipitation for week 36 (dough)

         DY = -0.40 + 0.34 * PPT27 + 0.27 * PPT33

where:      DY = estimated detrended yield (mt/ha)
         PPT27 = weekly precipitation for week 27
                  (2nd cutting)
         PPT33 = weekly precipitation for week 33
                  (3rd cutting)

         DY = -82.6 + 41.1 *  SM28  -  56.1 * PPT45

where:      DY  =  estimated detrended yield  (kg/ha)
          SM28  =  soil moisture for week 28  (planting)
         PPT45  = weekly  precipitation for week 45
                  (boll opening)
                                       92

-------
                      Table  3.   Plant  Process  Model  Input

§OIL_PARAMETERS:

         Soil Albedo
         Stage 1 Soil Evaporation Coefficient
         Drainage Coefficient
         Runoff Curve Number
         Soil Layer Thickness
         Soil Water Contents
         Root Distribution Weighting Factor

HEATHER PARAMETERS:

         Daily Maximum Temperature
         Daily Minimum Temperature
         Precipitation
         Solar Radiation

SggETIC PARAMETERS:

         Growing Degree Days from Seedling Emergence to End of Juvenile Phase
         Photoperiod Sensitivity Coefficient
         Growing Degree Days from Silking to Physiological Maturity
         Potential Kernel  Number
         Potential Kernel  Growth Rate

               PARAMETERS:
         Fertilizer
         -  fates
         -  form
         -  depth
         -  dates  of application

         Irrigation
         -  rates
         -  method
         -  dates  of application

         Pesticide  or  Herbicide
         -  type
         -  rate
         -  method of application
         -  dates  of application

         Cultural Practices
         -  sowing date
         -  plant  population
         -  sowing depth
                                     93

-------
modified precipitation regimes  is  declared  to  be the direct impact of climate
change on  yield.   This  difference is multiplied by  the  harvested acreage in
the  county that the  weather station  represents,  to determine  the  impact on
production levels.   County  production is then multiplied  by the average crop
price of the  1984  season to arrive at the  direct  impact  (in dollars) on each
commodity  during a specific growing  season.   Figure 1 depicts  the potential
total direct  impact  in  dollars of  precipitation  changes  on all  five  study
crops during one season,  1981.  It also depicts the geographic distribution of
precipitation modification  impacts in hundreds of  thousands of  dollars.   The
contours  represent  smoothed estimates  of  county  production  value  changes
across space.   If  we  assume that  each county is roughly 2,590 km2 (1,000 mi2)
in area, then  the  value  at any point on  the contour represents  the potential
impact  of  precipitation  modification on  an  area  of 2,590 km   (1,000  mi )
surrounding the point.   For example,  a contour  value  of 5.0  represents  a
production loss of $500,000 per 2,590 km2 (1,000 mi2).

     This  procedure  is  repeated  for each  year  between  1960  and 1984.   The
result  is  a distribution of potential direct  impacts  to  agriculture on the
Oklahoma economy under  1984  cultural practices, technology,  crop varieties,
and  commodity   prices,  and  throughout 25  weather  years;    the  impacts  are
summarized  in  Table  4,  which  indicates  that the  largest  impacts  on  yields
would be  in winter  wheat  [mean  annual  impact  of  325  kg  per  ha  (5 bu per
ac)].   As  a result of the  extensive wheat acreage  in Oklahoma, winter  wheat
also sustains the  largest production  value  impact.   A preliminary examination
of expected price response to an average annual decrease of nearly 703 million
kg  (26  million bu)  of wheat suggests only slight increases  in whea't  prices
[price behavior per  2,703 million kg  (100  million  bu) change in supply  taken
from Womack  (1980)].   On the  average, under the stated  climate change  hypo-
thesis and over 25 model years, a  direct  impact loss of $90.44 million to the
State of Oklahoma  can be expected.   Losses for  a  particular year could  range
from $39 million to $158 million.

     The  potential   impacts  of   climate  change   on  specific  agricultural
activities  such as  irrigation  water demand  can also be  addressed.  In the
simple example  completed for this  analysis  of four  far  western counties which
irrigate corn,  we  computed  an  average growing season irrigation  increase of
5.2JI.

     Other potential agricultural  impacts are  more  difficult to  quantify than
weather-determined  crop  yield  and  irrigation requirements.    These  include
changes in  pest or pathogen activities,  crop maturity rates, and field work
days.

     In the case of  both pests  and pathogens,  damage is usually greatest when
plants are  in  a stressed or weakened  condition.   Rainfall changes at certain
critical times during the growing season could increase plant vulnerability to
infestation by  increasing plant moisture  stress, which would  be important to
irrigated  as well  as dryland agriculture.   Irrigated fields  provide a  haven
for  weeds  and  pathogens  which thrive when  such  fields are flooded or sprayed
(Hatfield  and   Thomason   1982).    Irrigation  decreases  canopy  temperatures»
increases  soil  moisture, and,  consequently, increases the  likelihood of pest
and  pathogen infestation.   A climate  change  could  increase the necessity for
flooding and spraying.   By  increasing the number of these supplemental appli-
cations,  damage   resulting  from  pests  and  weeds  may  be  increased.   I*1


                                      94

-------
vo
Ol
           Figure 1.  Potential  direct negative impacts  of a 10? decrease in growing season precipitation on 1981
      Oklahoma production  value  of wheat, corn, sorghum, hay, and cotton  in  hundreds  of thousands of dollars per
      county (after Cooter and Haug 1986).

-------
Table 4.  Potential Direct Impacts (Losses) Resulting from a W% Decrease in Growing Season
          Precipitation on Oklahoma Crop Production (1984 Acreage and Price, 25 Weather
          Realizations) (After Cooter and Haug 1986).
mean
s.d.
median
maximum
minimum
mean
s.d.
median
maximum
minimum
mean
s.d.
median
maximum
minimum
Wheat
kg /ha bu/ac
-324 -4.9
-125 -1.9
-343 -5.1
-576 -8.6
-133 -2.0
Wheat „
10 mt 10 bu
-698 -25758
-268 -9911
-738 -27242
-1239 -45739
-286 -10547
Wheat
-86.3
-33.2
-91.3
-153.2
-35.3

CHANGES IN YIELD
Sorghum Cotton
kg/ha bu/ac kg/ha Ibs/ac
-64 -.9
-21 -.3
-63 -.9
-99 -1.5
-29 -.4
-Sorghum-
ID"3 mt 10
-12 -427
-10 -373
-11 -418
-18 -666
- 5 -198
VALUE OF
Sorghum
-1.1
- .9
-1.0
-1.7
- .5
-3 -2.6
-1 - .8
-3 -2.4
-5 -4.4
-1 -1.3
CHANGES IN PRODUCTION
. Cotton.
bu 10 mt 10 Ibs
-448 -985
-143 -314
-417 -917
-745 -1660
-215 -473
PRODUCTION CHANGES (1984
(Million of Dollars)
Cotton
-.5
-.2
-.5
-.8
-.2
TOTAL DIRECT IMPACTS
(Million of Dollars)
mean -90.4
s.d. -32.9
median -96.9
maximum -158.0
minimum -39.2
Hay
mt/ha tons/ac
-.02 -.01
-.02 ,-.01
-.02 '-.01
-.04 -.02
0 .00
3 Hay 3
10 mt 10 tons
-14 -15
-11 -12
-18 -19
-35 -38
0 0
Prices)
Hay
-1.3
-1.0
-1.6
-3.2
0
Corn
kg/ha bu/ac
-558 -8.9
-545 -8.7
-531 -8.4
-1710 -27.3
+326 +5.2
- Corn -
10 mt 10 bu
-11 -445
-11 -434
-11 -420
- 9 -363
+ 7 +260
Corn
-1.4
-1.4
-1.3
-4.2
+ .8

-------
response,  the number, strength or quantity of herbicide and pesticide applica-
tions could also change.

     The precipitation efficiency (the amount  of water available  for direct
Plant use)  is  also  important.   This measure is  determined by the  degree to
which rainfall penetrates  through the plant canopy and  its subsequent infil-
tration into the soil. Under our climate change hypothesis, heavy storms that
result in  runoff  are not  affected.   Light rains, up  to 6 mm (.25  in)  in 24
hours, are greatly affected.   This could be detrimental in two ways.   First,
Vgry light rainfall  (trace to 1  mm) on a well-developed plant canopy  usually
does  not   reach  the  soil  surface  before  absorption  or  evaporation  takes
Place.   There may also  be  beneficial  cooling; but  in  general,  these  rains
increase canopy humidities without much direct benefit to the plant.  A 30/t to
50^ reduction in precipitation would  imply  that more  of  the light rain storms
Would be too light to increase soil moisture.

     The development  and  spread  of pathogens could  also change.   Very  light
rainfall (and  even heavy  dew) creates  the  kind of humid environment  that is
conducive to pathogen development.  A heavy rainfall  can deter development by
locking or  washing  fungal  spores  from the  plant leaves.    Lighter rainfall
c°uld thus increase  the duration of conditions favorable for pathogen growth
and development.   The result would be  an  increase in the  rate of  successful
establishment and  spread  of pathogen populations.

     The potential impact  of climate change  on the rate of crop development
       be addressed directly in this analysis.   The plant process  models could
 Q  used  if  temperature  change  information were  available.   The  estimated
lmPact  of  increased temperatures  would  differ  for  various  thermal  unit
Models.    With a  model  that  is   simply  a deviation  from   a  fixed   base,  the
^crease in thermal  time  (and subsequent decrease in calendrical time)  could
^e  linear.   With  more  sophisticated thermal unit models, . the  rate  of crop
Development would  progress  on a  sliding scale  that peaks  at some  optimal
temperature and  decreases  to either side.    One benefit from  lengthening
Crowing seasons  is  a decrease   in  the  likelihood  of  early or  late  frost
Damage.     Oklahoma  fruit   crops  are  occasionally  damaged  by  late   spring
"rosts.    Crops harvested  in  the  fall,  such as  cotton and corn,  sometimes
8uffer losses as the  result of early fall frosts.

     The final indirect impact of a modified precipitation  regime  to  be cen-
tered here is change in  field  work days.  Although a  variety of  conditions
can influence  whether a  day  is  available  for  field  work  or not,  this  study
Considers trafficability  as represented by soil  moisture  to be  the dominant
"actor.   Under the present  hypothesis, no change  in  field-day availability is
exPected to result from changes  in the number  of  rainfalls.   Only  historical
r*infalis are  modified.    A field work  day is  defined  as  one in  which soil
"^isture (as computed by  the  plant process  models)  is at BQ% or less of water
c^Pacity available to the  plant.   A crop moisture budget was run at selected
Rations across the state,  with and without climate change.   The difference in
 he number  of work  weeks  between the  two budgets at  a particular location
^Presents  the potential   impact of climate  modification  on  field  days.
*|esults indicate that one  would expect an increase of from  three to  seven work
days per growing season.   Whether these impacts are  economically  significant
ino if  thev  are benefits or  disbenefits  remains  to be seen  (Cooter and Haug
1986).
                                      97

-------
     Up to this point, our analysis has dealt primarily with the assessment of
the direct  impacts of hypothesized climate  change on Oklahoma's agricultural
economy.    These  are impacts  that  can  be measured  directly  in  terms  of
commodity, such  as dollar changes  in uhe quantity of  food,  fiber, or energy
produced  or  consumed.   There  are  other  impacts as well,  which  are called
indirect, or  stemming-from,  effects.    These effects can  be  assessed through
the use of an input/output (I/O) economic model.   I/O models generally consist
of transaction matrices,  which are tabular  statements  of  the dollar value of
production,  and  the  "trading  relations"  among  the various  sectors of the
regional  economy.   "Multipliers"  for  changes in  sectoral  transactions can be
derived  that  estimate the  impacts  upon  the economy when changes  in natural
resource  supplies affect  the  production of other  sectors of the economy.  The
model  can be  used  to  relate  the producing sectors  systematically to the
resources and  the consumers  on the  economy (Grubb  1960).   The input/output
model  selected  for use  in  this  analysis  is taken  from  Little and Doeksen
(1968) and was applied in W. Cooter (1984).

     Two  types of multipliers  were developed by  W.  Cooter  (1984)  to address
the  crops modeled in   the  present  study.   The first  are  called Type  I
Multipliers and represent the  impacts for  a $1  change in production for crops
processed by  the local  crop-processing sector.   These crops  include winter
wheat, corn, sorghum,  and cotton.   Type  II  multipliers represent the impacts
of a  $1   change  in crop production for crops sold as  feed or  forage to the
livestock sector.  Hay production changes would utilize Type II Multipliers.

     Table 5  summarizes  the indirect  economic  components.   Each mean is the
potential  long-term  (24-year)  average annual  costs  (losses)  arising  from
hypothesized climate  change  precipitation  modification  given  that each year's
weather occurred  while  the  agricultural sector employed  1984 crop  varieties,
acreage,  and technology.  Table 6 summarizes the  state-level  findings of this
study.

        Table 5.   Average Annual Indirect Impacts  (Losses) Attributed
                  to a 1055 Decrease in Growing Season Precipitation
                  (After Cooter and Haug 1986).
Agriculture
Output
Final
Demand
Personal
Income
Gross
Taxation
State
Product
mean
s.d.
median
maximum
minimum
327.7
121.3
< 342.6
577.2
141.5
176.9
65.6
185.3
311.7
76.1
113.4
41.9
119.2
200.1
48.5
20.8
7.7
21.4
36.5
9.3
171.5
63.5
179.7
302.3
73.8
                                      98

-------
Table  6.  Summary  of  the  Average  Annual  State-Level  Potential  Direct  and
          Indirect Impacts  Resulting  from  a  10$  Decrease in  Growing Season
          Precipitation on the Oklahoma Economy.
Impacted Sector (Activity)
ppecipitation (area weighted)

Winter Wheat

Corn
Cotton

Hay

 est and Pathogen Damage Control

     of Maturity

      Work Days

Ippigation Costs

Irpigation Applications

 rrigation Water Demand

Sts*e Output

      Demand

         income
Average Annual State
    Level Impacts
      State Product
- 57 mm (-2.21 inches)

- $86,240,000

- $1,380,000

- $1,090,000

- $480,000

- $1,270,000

       *

       0

+3 to +5 days

+ $45,441 »»

+ .40 **

+.41 ac-in «»

- $327,700,000

- $176,900,000

- $113,400,000

- $20,800,000

- $171,500,000
*» putative analysis only
   amall region analysis
                                      99

-------
 RECOMMENDATIONS FOR FURTHER RESEARCH

     Further studies using plant process models to assess potential direct and
 indirect  impacts  of hypothesized  climate changes are  clearly warranted.   A
 logical  first  step would  be  the  incorporation  of General  Circulation Model
 (GCM)  estimated  weather  inputs.   The  advantage  of  using  these data  is the
 additional confidence in the physical "sense" of the precipitation and temper-
 ature  changes  which the models utilize it provides.   The GCMs  also have the
 capability  to  produce  radiation  estimates  which  would  be  valuable  in many
 plant  process  models.   However,  some  roadblocks  are perceived  with some GCM
 products.  Such roadblocks are space resolution (the GCM data are, at present,
 produced on too large a grid  to be input "as-is"  to plant process models) and
 the lack  of maximum and minimum temperature  estimates.   Both of these objec-
 tions  can  be,  and in some cases have  been,  overcome  through the supplemental
 use of a variety of statistical modeling techniques.

 CONCLUSIONS

     This  research  demonstrates   the   value  of  plant   process   models  to  a
 regional climate  change assessment.   Using these models  for winter wheat and
 corn, as well as regression models in the cases of hay,  cotton and sorghum, we
 estimate  that   a 10%  decrease  in total  growing season precipitation  could
 result  in average  annual direct  losses  to  the  Oklahoma  economy  of  $90.44
million.  Over 25 model years, losses could range  from $39-2 million to $158.0
million.   When indirect  impacts  are  included in the  analysis, we estimate
 that, on the average, gross state  product  could be reduced  by $171.5 million,
 ranging over a 25-year  period from $73.8  to  $302.3 million.   Impacts of cli-
mate change on irrigated agriculture do not  appear (in  this  analysis) to play
 a significant  role.   Field work  days  also do not appear to be  significantly
 affected by the hypothesized precipitation changes.   Pest and pathogen losses
and changes in rates of crop maturity  can  be expected but have not been quan-
 tified.   Even  without  these latter  two impacts,  we have demonstrated  that a
 10$ change  in  growing  season precipitation  could  result in  severe stress on
 the economy of the  State  of Oklahoma.   In view of the  possible  magnitude and
 variability of these   impacts,   the  development  of  policy  alternatives,  in
 conjunction with  improved estimates of reasonable environmental futures  and
 expected plant responses,  would seem to be appropriate.


 REFERENCES

 Botner, D.M. et al.  1986:   The CERES-MAIZE model:   1985  operational test for
     the U.S.  cornbelt.  Columbia:  Cooperative  Institute for Applied  Meteo-
     rology (CIAM).

 Cooter,  E.J.,  and  J.H.H.  Haug.  1986.   An  economic  evaluation of  proposed
     weather modification  programs in Oklahoma. Operational  Weather Modifica-
     tion, vol. 17.  Norman:  Oklahoma Climatological Survey.

 Cooter, W.S. 1984.  The economic impact of climate:   An  analysis  of the impact
     of climate on the Oklahoma economy.  The Economic Impact of  Climate, Vol.
     17.  Norman:   Oklahoma Climatological Survey.
                                      100

-------
Gpubb,   H.W.   1960.  Estimating  and  using quantitative  models  to  plan  and
     evaluate public  sector programs  in Texas.   In The  Economic  Impact  of
     Climate  3:1-22.   Norman:  Oklahoma Climatological Survey.
          J.L., and  I.J.  Thomason eds.   1982.   Biometeorology  in  integrated
     rest management.   New York:  Academic Press.

H°dges,  T.,  V.  French, and S.K. LeDuc.   1983.  Estimating  solar  radiation for
     plant simulation models.   Paper  presented at the Sixth Annual Workshop of
     the Biological  Systems Simulation Group,  University of Illinois.

Jones,  C.A.,  and J.R. Kiniry, eds. 1986.  CERES-MAIZE:   A  simulation  model of
     maize growth and development.   College Station,  Texas: A&M  University
     Press.

Llttle,  C.H., and  G.A. Doeksen.   1968.  An input-output  analysis  of  Oklahoma's
     economy .   Technical Bulletin T-12U.  Stillwater: Agricultural  Experiment
     Station, Oklahoma State University.

 i
-------
Climatic Change—-Implications for the Prairies

*• B. Stewart
Regional Development Branch
Agriculture Canada
Ottawa, Ontario Canada
f   This  paper  describes  the impact of  possible  climatic  change resulting
c On>  increased  C02  warming on estimated dry matter yields  for  spring wheat
  ops in Saskatchewan,  Canada.  Data generated by  the GCM modeling experiments
 * fche  Goddard  Institute for Space Studies  (GISS) for a  doubling  of atmos-
      C02 concentration are compared to the 1951-80 climate norms.  Climate
      projected by  the  GISS  model  would  increase  the annual temperature  in
     chewan by  an  average of 4.7°C and precipitation by 1555.   The  growing
,      length  would be  increased by  an average of  48 days, advancing  the
.Sinning of the growing season by about 2 to 3 weeks and extending the fall
.  rvest by about 3 to 4 weeks.  Precipitation during the growing season would
In increased by an average  of 15/S; however, analysis using the Palmer Drought
On ex suggests that Saskatchewan would become more drought-prone.   The impact
t_ yields is estimated using a generalized  crop growth model by modifying the
Pepferature and  precipitation  input  data   in relation  to the 1951-80  norms
Uo     Results suggest that, in the absence of direct C02 effects,  produc-
iti uln Saskatchewan would  be  reduced by 16?  to 26%.  Assuming a 15% increase
to ?hot°synthetic capacity  as  a direct effect of doubling of C02,  in addition
by ^e increase in temperature and precipitation,  production  would still fall
fto  to 1^-  Anv decrease in precipitation  from current levels would signi-
*fea?tly reduce yields and  production.   To avoid  midsummer  drought,  farmers
   •Likely to shift  to  fall-sown crops.

PRODUCTION

    As long as  man has cultivated  crops   the returns on  his  endeavors have
     SubJect  to tne vagaries of  weather  and climate.   The last  few  years
    ioly  demonstrate  this feature  with  regard  to drought  impacts  on  the
   RPles-   For example, the devastating drought in 1985 has been estimated  by
      ^*)  to  nave  cost  Prairie  farmers  $231  million in  terms of  cash
    Pts from the loss  in crop production and  to have increased feed costs and


                                    103

-------
destocking  of  beef cattle.   Cash receipt  losses  carried over  into 1986 and
1987  are  expected to  add a further  $545  million and  $53.2  million, respec-
tively, for a total cost  of approximately  $829 million.  The "dirty thirties"
(1933-37)  and   1961  are  other notable  years  when  the  vagaries  of weather
severely affected agriculture and the economy in the prairies.  These year-to-
year  variations are  for  the  most  part random and generally  unpredictable.
They are part of  the  normal cycle of weather events forming the basis for the
agricultural zonation  of  crops we see  in  place today—for  example, the hard
spring wheat crop that currently dominates prairie agriculture.  This crop has
been bred for and is well adapted to the prairie region.

     Wheat is the most  important  cereal  grain crop in  the prairies and indeed
in Canada.  It is grown more extensively and produced in greater quantity than
any other  crop.   Total harvested area of  wheat  in  the prairies  in 1984 was
12.8   million   hectares  and   total  production   was   17.5   million   tons.
Saskatchewan produced 9.6 million tons, Alberta 4.3 million tons, and Manitoba
3.6 million  tons  (Saskatchewan Agriculture  1985).   The dollar  value of this
crop  in  terms  of cash  receipts was  worth  $3.89  billion  to  the  prairies and
exports of this crop contributed approximately $2.3 billion to Canada's inter-
national balance of trade.   The importance  of wheat  to  the prairie region and
to  Canada  as  a  whole certainly justifies  an examination  of  the possible
effects of climatic change on prairie wheat production.

     Long-term climatic changes that produce distinctively different climatic
regimes from the  current norm  may have  significant influence  on crop yields
and subsequently on the geographic zones in which  crops can be grown (Bootsma
et al. 1984; Parry, Carter,  and Konijn 1984).  Such a change could result from
the increase in carbon dioxide  (CC^)  concentration in the atmosphere which has
been occurring  at a fairly  rapid rate since the  beginning  of  the Industrial
Revolution in the  1860s (Keeling  et  al.,  1976a, 1976b).  A growing consensus,
supported  by  a  number of  studies  investigating  the  effects  of  increasing
atmospheric COp concentration,   is that a general warming in the global climate
can be expected (Manabe and Wetherald 1980;  Manabe and  Stouffer 1980; Mitchell
1983; Hansen et al. 1983).

     The projected warming  for a doubling  of C02 which varies  from 1.5°-4°C
(Bach  1986) could  have major repercussions for agriculture.   For example, it
has  already been estimated  in  the United  States  that  a 1°C  temperature
increase would  shift  the  corn  belt  175 km  further northeast (Newman 1980); a
similar shift  in Canada  is postulated  by  Bootsma et  al.  (1984).   Williams
(1975a, 1975b)  and  Williams and  Oakes  (1978) have estimated  similar effects
for wheat  and  barley  in the  Canadian prairies.   More recently,  Rosenzweig
(1984) has  examined  the  change in crop zonation  in  the North  America wheat
belt  as  a  consequence   of  a  possible  doubling  of   COp  concentration.
Rosenzweig's results suggested  that  the northern boundary  of the winter wheat
belt, which currently parallels the  mean minimum January  temperature of -13°C
in the United States,  would shift north and east into Canada.

     Outside of the  work  by  Bootsma et  al.  (1984),  Blackburn and Stewart
(1984) and  Williams  et al.  (1986),  little has been  done to  investigate the
impact  that climatic  change might  have on  yields  in Canada.    This  paper
attempts to examine this  question,  looking at  the potential  impact on spring
wheat  production  in  Saskatchewan.   Results presented  here represent part of
                                      104

-------
the Canadian  contribution  to a  cooperative  study  sponsored  by the  Inter-
national Institute  for Applied  System Analysis  (IIASA)  (Parry,  Carter,  and
Konijn 1984).   Details of the  entire  Canadian study  are presented in Williams
et al. (1986).

     The effects  of  climate  change  on  spring  wheat  dry  matter yield  and
Production  are examined using a simple  crop  growth model  developed by Stewart
(1981).   Climatic change postulated for a  doubling of atmospheric C02 concen-
tration  is analyzed  using temperature and  precipitation  data derived  by  the
general  circulation  model developed  by  Hansen et al.  (1983) at  the Goddard
Institute  for  Space  Studies  (GISS).   The  effects  of  climatic change  are
discussed  in  relation  to the  current "normal"  climate,  where  "normal"  is
defined  as  the average climate  for  the  1951-80  period.   Dry  matter  yields
derived  for the 1951-80 normal period  are used as a reference for establishing
the change  in  production.

HETHODOLOGY

         Phenoloeical Models
     The procedures  used  to  estimate  spring  wheat  phenological dates  and
     s are briefly described  here.   A detailed discussion  of the yield model
   Provided  by Stewart  (1981) while  Robertson  (1968)  and  Williams  (1975a)
        more detailed  discussion  of the  phenological  model.   Calculation of
       wheat yields  is based  on  the  methodology  developed  by  the FAO (1978)
    uses tabulated  results from  the  de  Witt  (1965) photosynthesis model to
      e "constraint-free" yields.   Yield  estimates assume a sigmoidal  cumula-
      growth  curve  with  development  incremented  up  to  the  number  of  days
         for the  crop  to mature.   Net  dry matter biomass  production  (B ) is
           as a function of the gross  biomass production (B   ) capacity of the
    »  determined by its photosynthetic response to temperature and radiation,
 .  maintenance  respiration   coefficient   (Cm),   calculated  using a   method
 (N?Crlbed by McCree CI974)» and tne number °* days required to reach maturity
  '•   This  relationship is expressed as:

                   Bn = 0.36B  /(1/N + 0.25 CT)                      Equation 1

 K .   T°  estimate  N  the   biometeorological  time  scale  model  developed  by
 1 Dertson (1968)  is  used.   Robertson's model  basically describes  the  pheno-
 , Sical development  of "Marquis" spring wheat as a function  of temperature and
  otoperiod in the form:

     S2
     1    [(a1  (Lra0)  + a2 (Li-ao)2)(b1(Tmaxl-bo) +  b2(Tmaxi - bQ)2
        ..
           •f b3(Tmini-b0)  + b4(Tmini-bo)2)] = 1                      Equation 2
      Li  is fche Photoperiod (duration of daylight in hours)  on day i,  Tmaxj^ is
      Kimum air temperature on day  i, Tmin^  is  the minimum air temperature on
       S1  ls  tne  date of  a  phenological stage  in the development  of wheat
       maturity  and SP  is the  next stage,  and aQ to  a2 and  b1  to  b^  are
      cients.
                                      105

-------
     Five  phenological  phases  are  considered  in  the  model:    planting to
emergence, emergence to  jointing,  jointing  to heading, heading to soft dough,
and  soft  dough  to  ripe.    For  each  stage a  different  set  of  a and  b
coefficients is used.

     The  date  the crop  ripens  or matures  is derived by  continuing the  sum-
mation  from  the  planting  date  through  all five phases.   N  is  then derived
as:   IEND -  ISTART +  1, where  ISTART and  IEND are  the  Julian dates that the
crop is planted and matures, respectively.

     The  beginning of  the growing  season  length  (GSL)  or planting  date is
calculated as the date the smoothed mean minimum air temperature first exceeds
5°C in  the spring.   This represents, with a 5Q% probability, the average  date
for the last spring and  first autumn  "killing  frosts"  (-2.2°C)  when using
averaged 30-year  climatic  norms  data (Sly and Coligado 1974).   In determining
this date  the monthly  temperature data  are  first  converted  to  daily values
using the Brooks  (1943)  sine-curve technique.  The  earliest planting date is
then  derived by  computer  interpolation of  the  first  day  the  minimum  air
temperature  reaches  5°C.  Similar criteria are used  to determine the end of
the growing  season in the autumn.   If an  estimated fall  frost occurs before
the crop  reaches  a maturity level of 4.8 the crop  is assumed  killed and the
yield component is set equal to zero.

     Crop dry matter yield (B ) is then derived as:

                           By = Bn x HI,                            Equation 3

where:  HI  is  the harvest index, defined as  that  fraction of  the net biomass
production that is economically useful, i.e.,  the grain component.

     In this study the work of Major and Hamman (1981) at Lethbridge, Alberta»
for Neepawa wheat is used to calculate harvest index values.  Using their  data
it was  found that the  harvest  index  was inversely related to moisture avail3*
bility.   That  is,  if moisture  is  limited,  a  higher percentage  of the  crop
biomass  is   converted  into  yield than  if moisture  is not  limited.    This
relationship is expressed  by using the ratio  of  actual  evapotranspiration fc°
potential evapotranspiration (AE/PE).   If  the value of  AE/PE  is greater
0.75,  the value of HI is set equal to 0.35.  As AE/PE decreases below 0.75
value of  HI  is  increased linearly to a maximum value  of 0.52.   This value
reached when AE/PE has declined to approximately 0.36.  In the situation
a frost occurs  before  the crop reaches  a maturity  value greater than 4.8
less than 5.0,  HI is extrapolated linearly  from 0.0  to 0.35, respectively.

     In this study PE  is calculated using  the Penman  (1948) method and AE **
derived using a combination soil moisture  budgeting/split canopy evapotran»*
piration model.   The former  involves using  the techniques described by Balerj,
Dyer,  and Sharp (1979) and the  latter the work of Ritchie (1972).  Details o •
the procedures  used  to calculate  both  AE  and  PE are  provided  by SteW3r
(1981).
     Values  of  By  computed  by  equation  (3)  are constraint-free  or gene*' ~
potential yields  and neglect  the effects  of  yield-reducing factors such  ,
moisture  stress;   weeds,   pests,  and  diseases;   climatic   effects   on  y *  <
components,  yield  formation,  or quality of produce; and  field            '
                                      106

-------
por the  purposes of  this  study, values  of B   were corrected by  a moisture
stress yield-reducing  factor  (MSF)  to give  values  of  estimated  dry  matter
yield (Bvo), in the form:
        y«

                         Bye = By x MSF                             Equation 4

All other yield-reducing factors are assumed negligible.

     Moisture  stress  is derived using  an expression  relating  the  relative
yield decreases to the relative evapotranspiration deficit in the form:

                      MSF = (1 - ky(1 - AE/PE))                     Equation 5

where k   is an  empirically derived  coefficient  for  crop-yield response  to
""nature* deficit.  For  srin   wheat  the value of k.. is  set  to 1.1  based  on
         deficit.   For  spring  wheat,  the value of k.. is set  to  1.15  based on
    work of Doorenbos  and Kassam (1979).             y

     The procedures  for  estimating  dry-matter  yields presented  above  are
      ed to  evaluate  tne  long-term crop  production  capability or  potential
      optimum  management  conditions  on a  continental scale from  standard
         information.    The input  data  required  include  long-term  monthly
  n      of temPerature>  precipitation,  incoming  global  solar  radiation,
  "dspeed,   and  vapor  pressure.    These  data  are  normally  available  from
  servation networks  or, alternatively,  can be derived  using  simple empirical
 "

             Input
30  The monthly  climatic data used in the yield calculation  are based on the
stafear avera8e  1951-80  Canada normals  (Atmospheric Environment Service, 1982)
^tion data.  These data,  converted by LeDrew et al.  (1983)  to a 100 km x 100
Co eQual  area  grid  system  covering  the  land mass area  of  Canada,  were
de Verted  to cover  the  Saskatchewan crop districts,  following  the  procedure
subd?lbed  by stewarfc  (1981).   The  agricultural  region of  Saskatchewan  is
yie?rided  into twenty administrative areas,  or crop districts,  for which crop
of h, data  are published annually  by the  provincial government.   The  locations
Flo,    °entroid  and reference number for  each crop district  are  given  in
•  8u*es 1 to 7.

gen °aily  information  for all  climatic  parameters except precipitation  were
of Rrated fr°m  the monthly data using  the sine-curve extrapolation  technique
                 similar to  that  described  by Williams  (1969,  1975a,  1975b),
                 e,  and Sheppard (1980)  and  Dumanski  and  Stewart  (1981).
             data were  converted to weekly  values and then adjusted  so  that
             is  received over  a  three-day  period  as  follows:   60%  the first
        the second, 10£ the  third, and Q% for the  remaining four days.   Daily
         n was used in  simulating the  crop-water balance.
         methodology used  in  this study  was designed  to evaluate  long-term
   4      ds under  optimum management  conditions.  As such, there are no  long-
     xPerimental  results available  for comparison in Canada.  However,  there
     ort-term  experiments available that can  be used.   Model estimates were
                                     107

-------
compared  to  experimental  work  undertaken  by  Major  and  Hamman  (1981)  at
Lethbridge, Alberta, and by Onofrei (1984) at six locations  in Manitoba.

     To compare model estimates  with  observations  at the Alberta and Manitoba
sites climatic  data including the  mean (Tmean), maximum  (Tmax),  and minimum
(Tmin)  air temperatures;  precipitation  totals;  and  data  on  soil moisture,
observed  for  each  year  were obtained.   For  the  Alberta site  only,  monthly
observations  were  available, while for  the Manitoba  sites daily  data were
obtained.   Results of the  comparison of estimated  yields to observed values
are given  in  Table 1.   Results  show  that  the  model  is within 155& of reported
values.
Table 1.  Comparison of Model Estimates to Observed Experimental Yields in
          Alberta and Manitoba
Year Location


No. of
Sites

Dry Matter
Yield (kg/ha)
Model Observed
Model/
Observed

1976
1977
1982
1983
Albertaa
Albertaa
Manitobab
Manitobab
1
1
6
6
3495
2454
3547±659
2666±607
3098
2623
3967±440
3098+763
1.13
0.94
0.98
0.86
a Data from Major and Hamman (1981) at Lethbridge, Alberta, for Neepawa
  Wheat.

b Data from Onofrei (1984, personal communication) for six sites in Manitoba
  (Beausejour, Winnipeg, Woodmore, Mariapolis, Bagot and Teulons) for
  Glenlea Wheat.
     These results  are  remarkably  good  considering  that  the model  was
designed for use with data  for  individual years,  but rather, for application^
employing monthly  data averaged  over  several years.   The  equations  used *°
compute crop biomass and yield  employ  averaged growing season information, a"?
opposed to the actual day-to-day values that would be used in a model design6*
for real-time application.  For this reason the model does not simulate
growth and photosynthetic activity on  a  daily basis as would be the case
sophisticated models, nor was  it intended  to  do  so.   It was designed
to simulate what happens to the crop biomass productivity for averaged
season  conditions.   As  a  consequence,  many  environmental  factors affective
crop  growth  and  productivity   are  unaccounted  for  in  the  model  framewor^'
Nevertheless, it  is a  physically based  model using the  broad  principles °
biomass production, and the needed data for large area application are readi-"
available.
                                      108

-------
     It is also  emphasized  that the  comparison outlined  in  Table  1  involves
sxperimental  yields that represent  the potential or maximum yields that can be
obtained under optimum management practices.   They do not represent the yields
obtainable under  current commercial  conditions which  are considerably  less
than the potential.  For example, reported commercial  yields  were 59% and 60%
°f the  values given  in  Table 1 for  experimental  wheat  yields  for  1982 and
1983,  respectively, in Manitoba.  Similarly,  in Alberta the ratio was 16% and
7°fc»  respectively,  for 1976 and 1977.   For  this  reason,  all  scenario yield
estimates are expressed in  terms of  percent of normal,  where normal represents
the model estimate  derived  using the  climatic data averaged  for the 1951-80
Period.    it  is assumed that the effects  on  yields of  variations in  climatic
c°nditions are  the same for  both  commercial  and  experimental  spring  wheat
Production.     It   is   also  assumed  that  existing  technology  and  climatic
tolerances of the existing  spring wheat cultivars grown in Saskatchewan remain
unchanged.

^Sdjflcation  of Input  Data  to  Simulate Climate Change

     Climatic change is simulated in this  study using the GISS 2 x CCU general
Circulation modeling results of  Hansen  et al.  (1983).   Theoretically derived
™°nthly mean  temperature and precipitation data computed for a doubling of C02
°y the GISS  model  were obtained from Bach (1986).  Temperature  and  precipi-
ption data for the Saskatchewan study area were obtained in  the form of a 4°
~;atitudinal   by  5°  longitudinal grid square  framework  for   a  control  case
 ^Presenting  the current climate and for a 2  x C02 climate.  For the purposes
   this study the differences  in temperature  between the control (1 x C02) and
  x C0p climates were used  to adjust the temperature data from normal, while
 lie ratio of  the  2 x C02  climate to control  was used to adjust  the monthly
Capitation  totals from normal.   For  simplicity, this scenario  is referred
 ° as GISS1.   Three, additional  scenarios using  various combinations  of the
 ISS1  data will  also be  discussed in the following sections.   These scenarios,
 eferred  to as GISS2,  GISS3,  and  GISS4,  will  be defined later.

at.   !n all, 9 GISS grid points cover the Saskatchewan study  area.   These were
tj 50°. 54°,  and 58°N latitude and  100°,  105°,  and 110°W longitude.  Tempera-
f re and  precipitation data for these points  were  plotted and mean  adjustments
t r each  crop district  were  interpolated from the  mapped results.   Monthly
^?erature  and precipitation changes  averaged for  the entire Saskatchewan
Pal!   ltupal  area  are listed in  Table  2.   To estimate maturity  and biomass
GlSJtoeters  in  the  model,  values  for Tmax and  Tmin are required.   Since only
Vf  Tmean data were available,  it was  assumed that any  change  in  Tmean
^ected  Tmax and  Tmin  equally  (i.e.,  if Tmean  increases  by 1°C,  then both
   * and  Tmin  are assumed to increase by 1°C).

ad   Climatic  change  for all scenarios is  simulated  in  the  yield  model  by
Busting the 1951-80 monthly norms  data  with the  corresponding  temperature
ij1.Precipitation  data  obtained  for each scenario.   All other  climatic data
  put required by the yield model are fixed at  the 1951-80 normal level.
                                      109

-------
  Table 2.  Monthly Temperature and Precipitation Adjustments Used  in  the
            Yield Model to Represent the GISS 2 x C02 Scenario
            Month                    Temperature3       Precipitation13
January
Febuary
March
April
May
June
July
August
September
October
November
December
6.1
5.6
4.8
4.1
3.7
3.3
3.3
3.7
4.6
5.3
5.9
6.3
1.29
1.34
1.24
1.17
1.15
1.15
1.13
1.05
0.99
1.12
1.26
1.30
            AVERAGE                       4.7°C              1.15
a All  1951-80  monthly  temperature data were adjusted by addition of these
  values.  Values  represent  the  difference  between  GISS 1 x C02 and GISS
  2 x C02 estimates.

b All 1951-80 monthly precipitation totals were adjusted by multiplication
  by  these  values.  Factors  represent  the  ratio  of GISS 2 x C02 to GISS
  1 x C02 model estimates of precipitation.
                                      110

-------
CLIMATIC WARMING—WHAT DOES IT MEAN?

     If Saskatchewan were  to  undergo the projected  climatic  change suggested
by  the GISS  model,  how  would  the new  climate  compare with  the  present
climate?  The following discussion attempts  to  outline  some of the changes we
might expect.

     Table 3 outlines the computed growing season  start and end dates for the
1951-80 norms  and the GISS  scenario.   As shown,  the growing season start (GSS)
varies from May  17  to  May  27 and the growing season end (GSE) occurs between
September  10 to September 15.   Average  growing  season climatic conditions for
the 1951-80 normals  period  in Saskatchewan  are illustrated  in Figures  1a to
^a-   Figure 1a outlines the  GSL available  for crop growth; Figure  2a,  the
thermal resources  available during  the growing season, expressed  in degree
days above 5°C (DD5);  Figure  3a,  the available moisture in terms of precipi-
tation occurring during  this  growing period; and  Figure 4a,  the evaporative
demand or  ratio  of  precipitation to  potential evapotranspiration  (precipi-
£ation/PE) that existing  crops are adapted to.   These figures basically define
the growing season as being relatively  short, warm,  and dry.   The average GSL
       from 100 to  120 days,  generally decreasing in a south to  north direc-
       The  exception to  this  pattern is in the southwest  corner  of the study
     where  the higher  elevation of  the terrain,  particularly  in the Cypress
    s, results  in  the GSL being of  similar  duration  to that  of  the  more
"J°rtherly  agricultural  area of the province.  DD5, as presented in Figure 2a,
Allows the pattern of the growing season length,  with  the greatest amount of
heat in the central part of  the  study area (1400  DD5) and  the  least in the
n°rth (1100) and southwest  corners  (1200).   Also,  as shown in Figure 3a, the
s°uthwest  corner  of  the  study area  is  the  driest, with total growing season
Precipitation   averaging  slightly  less  than  180  mm.    Values  increase  to
Rightly more than 240 mm  in  the northestern part of the  study area.   Figure
^ highlights the dryness of  the region indicating  that the  prebipitation it
Deceives is enough  to  supply only 35 %  of the  evaporative  demand in  the dry
8°uthwest, slightly more  than  50% in  the east, and  60% in the north.

     Given that the existing growing season climate  in  Saskatchewan  is short,
      and dry  at  present,  what sort of average change might  we  expect  given
    warming projected by  the GISS model?  Table  2 outlines the monthly temper-
  ure  and  precipitation adjustments  for  Saskatchewan.    As shown,  monthly
JemPeratures vary  from an  increase  of  3.3°C  in  June  and July  to  6.3°C in
         and January.   Altogether,  the   warming suggested by the  GISS model
       increase   the   annual   temperature  in  Saskatchewan  by  about  4.7°C.
     ermore,  precipitation is  also  expected  to  increase.   Monthly increases
15n8ing from 5%  to  15/t during  the summer, to a slight decrease in September of
.*» to an increase in the winter of slightly more  than 30J6 are projected.  In
 erms of annual  precipitation  totals,  levels  are  projected to  increase from
  * to 18* above the  current  level.   The projected May to August total which
      up 51% to 57* of the current annual  value  will not  change (i.e., 50% to
                                     111

-------
Table 3.  Estimated Changes in Growing Season Start and End Dates (Julian)
          and Growing Season Length (Days) by Crop District in Saskatchewan
          for the 1951-80 Normal Period and GISS 2 x C02 Temperature-Increase
Crop
District
la
Ib
2a
2b
3an
3as
3bn
3bs
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
9c
9b
Prov
Ave.
Growing Season Start
1951-80 GISS
140
138
140
178
139
142
139
145
144
137
140
144
139
138
139
138
147
140
146
145
141
±3
118
117
119
117
118
123
122
125
126
124
128
126
118
117
120
118
129
124
128
127
122
±4
Growing Season End
1951-80 GISS
256
258
256
258
258
254
257
251
251
258
257
253
257
258
256
257
252
257
252
253
255
±3
285
288
286
288
287
284
283
282
280
276
287
285
287
286
282
284
284
286
284
284
284
±3
Growing Season Length
1951-80 GISS
117
121
117
121
120
113
119
107
108
122
118
110
119
121
118
120
106
118
107
105
116
±5
168
172
168
172
170
162
162
158
155
153
170
160
170
170
163
167
156
163
157
158
164
±6
_^*
141 » May 20           255 » September 12
                                   112

-------
       a) 1951-80 Normals
110
o>
                                       100
                                       110
                                        120
          108
                                    102
b) GISS 2xCO2
                                                    __   .
                                                     	1	^	-
                                                                                        160
   Fl8Ure '•
                                                         r * «

-------
          a) 1951-80 Normals
 1300
1200
                         105W
102
                     b) GISS 2xCO2
                                                                                              1900
                                                    2000

                                                               108
105W
102
         Figure 2.  Variation in Growing Season Degree Day Totals Greater Than  5°C  in
                    Saskatchewan Derived for  the  1951-80 Normal Period and GISS 2 x C02 Scenario

-------
a) 1951-80 Normals
A                       I    •   I
                    H     /  i
                      /    /*'
                    JL-JLj-J
  108          105W          102
o>
I  •    / 3BS        1    • 1
L_/'        -7     )     •"
  	CH	J\	1— —i-
       108           105W          102
Figure 3.  Variation in Growing Season Precipitation Totals (mm) in Saskatchewan
          Derived for the 1951-80 Normals and GISS 2 x?£Q2 Scenario

-------
ov
               01
                       a) 1951-80 Normals
10SW
                                                       102
                                     b) GISS 2xCO2
                                                                     o>
105W
102
                    Figure M.  Variation  in  the  Growing  Season Ratio of Precipitation to Potential
                               Evapotranspiration (Pipe) Derived for the 1951-80 Normals and GISS 2 x C02 Scenario

-------
     Figures 1b, 2b, 3b,  and  4b outline the change  in GSL,  DD5,  and precipi-
tation received  during  the  growing  period and  the  precipitation/PE  ratio.
Comparing Figures  1a and 1b,  2a and  2b, and  3a and 3b  reveals  little  change
from the  current  isoline patterns.   For example,  the length of  the growing
season will increase by  an  average of 48 days or by 40$ to  50%;  the increase
will be distributed relatively uniformly across the province.  The increase in
GSL will  affect both the GSS and GSE.   The  GSS will be advanced an average of
19 days and will range  from  April  27 in the southeast  to  May  9  in the  north.
Similarly,  the GSE  will be extended by  an average  of 29 days in the fall from
October 7 in  the  southwest,  October  11  in  the north,  and October  15  in the
southeast.   In conjunction with the increase in GSL,  the higher  temperatures
wiH also  augment  the  total  available  heat  by about  800 DD5 or by  6Q%  to
™J.  In  both cases, the largest changes will  take  place in  the northern half
and the southwestern corner of  the agricultural area;  the least  impact is in
the central and southeastern parts.   In essence, the  effect of  the projected
GlSS warming on the climate of  Saskatchewan would be  tantamount  to a shift of
tlle climate  in Nebraska  north  to Saskatchewan.   Given a climate change of this
"^gnitude what sort of  impact might we expect to see  on spring wheat produc-
     in the prairies?  The following  sections outline  the impact  on maturity,
     ,  and production potential.

RESULTS

     In analyzing the GISS scenario,  an attempt is made to  assess the  likely
e£fects on  spring  wheat  yields and  production  resulting  from   the  implied
shifts  in long-term average climate.   It is recognized,  however,  that  spring
wheat is  not the only crop  that would be affected by  these  perturbations and
fhifts.   Assessing  the  impact on all crops, however,  is  beyond the  scope of
Chis report.

     In discussing  the  impacts of (X^-induced climatic change  on spring wheat
fields  the following sections outline  the effect of  the temperature changes on
£he ability  of spring wheat to mature,  the effect of temperatures and precipi-
 ation  on yield, and subsequently,  on  provincial crop production.

       n Spring Wheat Maturity
ft    As  shown  in Table  4,  the  average  time required  for  spring wheat  to
e, as  determined by the  Robertson (1968)   biometeorological   time scale
 Op 1951-80,  ranges from 86  to  98 days with the lower values observed  in the
          and central crop  districts (Figure 5a).   Comparing these  maturity
          ts  with  tne available DD5 (Figure 2) shows the  close correlation of
    length of time  required to reach maturity  and the total heat  available.
          the DD5  of spring wheat from planting  to ripening indicates  that
       wheat  requires from  1000 to 1100 DD5 in Saskatchewan.  These  results,
      ignoring  the  effect  of day length,  indicate  that  the amount  of  heat
       d  for  wheat  to mature is basically the  same  throughout the agricultural
Of   in Saskatchewan.  The key factor affecting wheat  development is  the  rate
at> "eat accumulation,  and  as can be seen from  examining Figures 2a and 4a,  the
      with the longest maturation  time  requirement correspond to  the coolest
        Conversely,  the  warmer the  temperature,  the faster the  spring wheat
                                     117

-------
Table 4.  Average Temperature (°C) Difference from the 1951-80
          Normal Period from Planting to Maturation for Spring
          Wheat in Saskatchewan for the GISS 2 x C02 Scenario
Crop
District
la
Ib
2a
2b
3an
3as
3bn
3bs
4a
4b
5a
Sb
6a
6b
7a
7b
8a
8b
9a
9b
1951-80
Maturation
Time
(Days)
88
87
86
86
87
88
89
92
96
98
89
92
87
87
88
89
93
87
95
94
Normal
Mean
Temp
CC)
17.1
17.1
17.2
17.3
17.5
17.0
16.9
16.7
16.3
15.9
16.8
16.2
17.1
17.1
16.9
16.7
16.0
16.7
15.8
15.7
GISS
Maturation
Time
(Days)
84
83
82
82
82
82
82
82
84
84
84
81
83
82
82
82
79
80
82
82
Mean
Temp
(°c)
1.0
1.1
1.3
1.3
0.9
1.4
1.5
1.5
1.8
1.7
1.0
1.8
1.1
1.0
1.3
1.2
1.9
1.5
2.1
2.1
                               118

-------
                    a) 1951 -80 Normals
vo
                                                     95
                   	f-	1	,_
      b) GISS 2xCO2
;   .-A I     3BS             .«*     I   \
l     /              •***          /IA I
L-*~»- ------- h ____ A-J
                                                                                                       102
                  Figure 5.   Variation in Average Spring  Wheat  Maturation  f-ime  (Days)  in Saskatchewan
                             for the 1951-80 Normals and GISS  2  x  C02 Scenario

-------
     Derived  maturation  requirements  for  the  GISS1   scenario  are  shown  in
Figure 5b; the difference  from  the norm is given in Table 4.  The primary  im-
pact of the  large-scale warming associated with  GISS  is to reduce maturation
time for current spring wheat varieties to the 79- to 84-day range, a decrease
of 4 to 14 days.  However, unlike the current maturation requirement where  the
longest requirement  is in  the  northern  part  and southwestern  corner  of  the
agricultural area, with the GISS  warming this pattern  tends  to be reversed.
That is,  the  northern  region has  the  shortest  requirement,  79 to 80 days,  as
opposed to 82 to 84  days  in the south and central parts.  A further impact of
the warming suggested by  GISS is  that  the difference in maturation time would
tend to disappear as the region becomes much more homogeneous  (i.e., currently
there is  a  12-day  range in maturation from 86  to 98 days;  for GISS the range
decreases to  5  days,   79  to 84  days).   Both  effects,  in   addition  to   the
temperature increase, are augmented by the advance in the planting date of  two
to  three  weeks  and  the coincident greater  increases  in  northern districts'
daylengths (the  biometeorological  time  scale  considers both  temperature  and
daylength in determining crop development).

     In Table 4 the  average  temperature  experienced  by the crop from planting
to maturity  is  given for the 1951-80  norm and GISS1 scenario.   As  shown,  at
present the  temperature range  in  Saskatchewan  is about 1.5°C from  the north
and southwest to the south central part (i.e.,  16°-17.5°C).  Data presented  in
Table 4 for  the GISS1  scenario reveal two interesting  features.   First,   the
range in mean temperatures over the course of spring wheat maturation through*
out  Saskatchewan   is reduced,  tending  to make  the  agricultural area  more
homogeneous  (i.e.,  range  in  temperature  with  GISS1  condition  is  about
0.8°C).  Second, and most interesting,  the effective temperature increase that
the  wheat crop  is exposed to  ranges  from 1.0°-2.1°C, not the  average 3.38C
increase  shown   in  Table  2.    In this  instance  the effective  temperature
increase  is greatest in the  north and  southwest parts of Saskatchewan and  the
least in  the southeast  and  central areas.  The greatest temperature increases
are coincident with  the areas with the  largest  reduction  in maturation time-
The reason the  full  temperature increase  (i.e.,  3.3°C)  is  not experienced  by
the  crop  is  the advance  in the growing  season  planting date by  about three
weeks.

Impact of GISS Temperature and Precipitation Changes on Yield

     Figure 6a shows the impact on spring wheat yields in relation to the nori"
for  the   projected  GISS  changes  in  temperature  and  precipitation  (GISS1)'
Results suggest that the overall effect of warming would be a  general decreaS^
in  yields.   The  southern  area would be less affected  than the  north wit-j
yields remaining  within 2Q%  of current  levels,  whereas in the north, yiel
reductions of 25% to 35/5 could be expected.
     To estimate  the overall  impact  on  total  spring  wheat  production,
average extent  (hectarage)  of crop  districts  in Saskatchewan  for  the
1961-79 were used (Table 5).   The  production potential for each crop
was calculated  by  multiplying the derived  yield by the  average  extent.
total provincial production potential  was then derived  by summing the p
tion values for all crop districts.  The results for all scenarios in relati°n
to the 1951-80 computed total are given in Table 6.
                                      120

-------
                  a) GISS1 - T,P
N>
     b) GISS2 - Tonly
            -P (1951-80)
                                                         at
I  »   I   m*"a
Ll.\_	-_"*_
       108          105W
                    Figure 6.  Variation in Spring Wheat Yields of the 1951-80 Normal for the
                              GISS1 and GISS2 Climate Scenarios

-------
           Table 5.  Saskatchewan Crop District Extent (ha, thousands), Yield
                     (kg/ha), and Production (tons, thousands) of Spring Wheat
                     Average for 1961-79
Crop
District
1A
1B
2A
2B
3AS
3AN
3BS
3BN
4A
4B
5A
5B
6A
6B
7A
7B
8A
SB
9A
9B
Extent
(1000 ha)
306.06
217.03
308.34
421.33
460.96
246.74
332.32
456.52
162.57
254.29
419.41
375.04
577.49
436.30
419.39
311.16
190.30
280.21
283.75
191.20
Yield
(kg /ha)
1555
1679
1533
1719
1480
1458
1383
1480
1232
1443
1695
176Q
1582
1526
1678
1700
1789
1768
1655
1713
Production
(1000 tons)
476.15
364.41
472.72
724.40
682.16
359.62
459.65
675.58
200.23
367.00
710.93
660.22
913.33
665.74
703.56
528.91
340.44
495.29
469.59
.327.50
     TOTAL
6,650.41
1593
10,597.33
      Table 6.  Change in Saskatchewan Spring Wheat Production from the
                1951-80 Normals for the GISS 2 x C02 Scenario


                       1951-80     GISS1     GISS2     GISS3    GISS4

% OF 1951-80 Normal      100        84.3      74.3      94.4     86.5
GISS1 - Temperature and precipitation

GISS2 - Temperature only, precipitation held constant at 1951-80 level

GISS3 - Temperature, precipitation and 15? increase in photosynthetic capac

GISS4 - Temperature and 15? increase in photosynthetic capacity - precipita*
        tion held constant at 1951-80 level
                                      122

-------
  Table  7.   Differences in Temperature (T) and Precipitation (%P) from the
            1951-80 Normal Period for 1933-37, 1961  and GISS 2 x COp
            Scenario for a Cross Section of the Saskatchewan Study Area
            Averaged for the Four Months May,  June,  July,  and August
Crop
District
1A
3BS
4B
7A
7B
9B
1951-80
T(°C)
17.46
15.29
14.72
15.83
15.72
14.45
P(mm)
248.7
195.5
176.5
195.8
210.4
247.8
1933-37
T
-1.3
+ 1.2
+2.3
+ 1.0
+0.1
+0.3
%P
0
88
78
55
75
70
1961
T
+0.3
+2.5
+3.5
+ 1.8
+ 1.3
+ 1.7
*P
38
50
57
45
64
61
GISS
2 x COp
T
+3.4
+3.4
+3.4
+3.4
+3.4
+3.4
%P
115
115
115
115
115
115
     GISS1  results  suggest  that  the  yield  reductions  in Saskatchewan  wjjuld
occur in spite of  the projected  precipitation  increase of approximately  155&
      current  levels.   This precipitation increase is more  than  offset by the
        effects associated with  the  higher  temperatures.

     From  a historical point  of view the  projected  GISS warming  presents an
   ^resting contradiction,  since historically,  when  temperatures have  been
int
                         ,                     ,
*oove  normal,  precipitation has  had a tendency to  be below normal.   Table 7
^lustrates  this  point clearly,  using data for the  period  1933-37  and 1961 as
 Samples.    Both  represent years  when major  drought was  experienced  in  the
 anadian  prairies.   Results also  show that,  historically,  temperature devia-
 l°ns  have tended to vary considerably throughout  the  region.

    The  GISS  data on the other  hand indicate a relatively uniform  change in
     temperature and  precipitation.   Because of this, it  is of  particular
    rest  fco have  some  indication  of  the  contribution  of the  precipitation
   rease  projected  by the GISS model to  the  estimated  impact  on spring wheat
 "c
yields
      and production.
    This  was  undertaken by rerunning the  yield  model with the GISS  tempera-
    Adjustment  only (GISS2); precipitation was  held  constant at the  1951-80
f'   The  results (Table  6)   indicate  that potential  production would  be
t.!:Uced overall  by  1651 for GISS1  and 26* for GISS2.  In other words, a further
       n  of  10^6 in spring wheat yields  could be anticipated if the  climatic
       predicted  by GISS occurred but precipitation  remained at the  1951-80
       The 3Patial pattern for  GISS2 (Figure 6) is quite similar to  that for
^es     igure  6a),  which reflects the fact that the additional  10Jt reduction
      ng  from ignoring the GISS precipitation  increases is spread uniformly
  °ughout Saskatchewan .

    The  above  results  further  suggest  that  precipitation  will  have  to
         significantly  above  the \5%  level  projected  by the  GISS model  to
  e      production at current levels.  If  the precipitation stays the same or
   eases, given the warming  suggested  by GISS,  spring wheat production  would
          sharply from current levels.


                                     123

-------
Impact of Increased COo on Photosynthesis

     In the  above discussion the  effect of temperature  and precipitation on
spring  wheat  production  has  been  considered while  the  direct  effects of
increased C02  on  photosynthetic capacity have been  ignored.  Various studies
have  suggested a number  of possible  beneficial  effects  arising  from the
increase in C02 concentration on plant productivity.  Kimball (1983), from the
results of a  literature review,  suggests,  for  example,  that  yields   would
increase by approximately 33% with a  doubling of C02 concentration.  Gifford
(1979), Lemon  (1983) and Aston  (1984)  have indicated that improved efficiency
in the use of plant moisture would be another direct effect.

     In Kimball*s  review, much  of the data  reported was obtained from growth
chamber experiments which  are  notorious  for oversimplifying  the  study of
environmental  factors  on  plant  productivity  in  comparison to  actual   field
studies.   In  growth  chamber studies,  such as those  carried out  by Gifford
(1979)  and  Sonit, Hellmers,  and Strain  (1980)  to  investigate  the effect of
elevated  C02  levels  on crop   growth  and  yield,  light,  temperature,  and
humidity,  levels  were  kept constant  in  simulating day  and  night conditions.
In controlled  studies of this sort,  plant stresses are generally minimized to
isolate or study  the effect of  a  particular environmental parameter.  Conse-
quently, the  results often  differ considerably  from  actual field conditions
where   the   diurnal  light,   temperature,  and  moisture  levels   fluctuate
considerably  from day to day and throughout  the course  of a  crop's growth-
For this reason the increase in  productivity reported by Kimball  (1983) can be
viewed  as unrealistically high.   Reported field  experiments for spring  wheat
support this contention.  For example, experiments by Krenzer and Moss (1975)
and Havelka,  Wittenbach, and Boyle  (1984)  found that for  wheat crops  grown
under  field  conditions  with elevated C02  levels  the  effect  on  dry-matter
yields was about half the magnitude suggested by  Kimball (1983),  i.e., 15% f°r
Krenzer and  1'\%  for  Havelka.   In the  former,  results  were obtained  with a
doubling of C02 concentration, while in the latter, results  were obtained wit*1
a 4 x C02 increase.

     In this  study  the possible direct  effects of  an  increase  in  C02 °n
increased photosynthesis  and moisture use efficiency are considered.    UsW&
the results  of Krenzer  and  Moss (1975)  and  Havelka, Wittenbach,  and  Boyle
(1984)  as  a  basis,  photosynthetic  capacity  of  the spring  wheat  crop was
increased  by   15$  to  simulate direct  effects on  plant  growth  (GISS3).   ^e
results are  presented  in Figure 7a  in  combination  with the GISS temperature
and precipitation  increase.   The results  show  that,  in  spite of the increa«^
in productivity associated with  elevated COp levels, provincially,  that  a 1?>
increase  in   photosynthetic  capacity  woula  not  be  enough  to  overcome tb
adverse effects   of  elevated temperatures and moisture  stress.   Product!0
would still decrease by  approximately  6%.   Results indicated that an increaS
in  photosynthetic capacity  of   approximately  20% to  25% would  overcome
temperature and moisture effects.

     As shown  in Figure 7a,  the  yield  pattern is  similar to  that described
the GISS1  scenario.    Overall,  the  \5%  change  in  photosynthetic
increases  the  provincial  dry-matter production potential  by about  10%
                                      124

-------
to
Oi
                  in
                  in
a) GISS3 - T.P
        - 15% T Photosynthesis
                        108
                          100
                                                      102
     b) QISS4 - T
             -15% f Photosynthesis   -\~
                                                                    en
                                                                           4A
                                                                                   3BS
/     I  I"         .3AS             1A
-—/^J.	'  X	1, J
       108            105W10~°~    "  102
                       Figure 7.   Variation ^Spring Wheat Yields  (% of Wc^mal) in Saskatchewan
                                   for the GISS3 and GISS4 Climate Scenarios

-------
 the  GISS1  level.   Again, yields  are  more affected in the north with a  20%  to
 30%  decrease while  in  the south  central part of  the  province yield changes
 range  from a slight decrease  (10%)  to a slight increase (15%).   Assuming the
 15%  increase  in  photosynthetic  capacity and  the GISS  temperature  increase
 only,  and  holding precipitation constant  at  current levels— GISS4 ,  as shown  in
 Figure 7b — the  effect of precipitation  is a reduction of about 10% in  yields
 in comparison to GISS3.

     The above  results  show the  impact of potential long-term average climate
 change on  spring wheat  production in Saskatchewan.   In  general,   the results
 suggest that the  overall impact  would  decrease spring wheat production by  6%
 to 26%.

 The Effect  of Increased Drought

     The  slight decrease  in  production, outlined  above,  can  be attributed
 primarily  to an advance in the growing  season  start (planting date) of about
 three  weeks and associated higher  light levels,  the latter  enhancing crop
 photosynthetic  activity.    The GISS model predicts that the  region  would  be
 only slightly drier.   However,  the question  one might ask  is  how  might a
 climate  change  of  the sort  projected  by  the  GISS model  affect  drought
 potential   in  the  region  (i.e.,  the   frequency  and  severity   of  drought
 events)?    Drought  is  a   critical   element of  concern  today and  must   be
 considered  in any future climatic change scenarios.  The stability of prairie
 agriculture  can be  badly shaken  if drought  frequency  and severity increases-
 The  last  five  years  clearly  illustrate  the impact  that  drought  has  had  on
 prairie  wheat  production   and,   subsequently,  on  the  economy.    Given  the
 projected GISS  climate  change, what  sort of  effect might we expect in drought
 frequency and severity?

     Williams et al.  (1986) have  examined  this feature  using  the GISS model
 projections.   In their study, drought  frequency  and  duration  changes were
 examined using  the well-known  Palmer  Drought Index (Palmer  1965).   Using this.
 index,   drought  is  taken as  the  interval of time,  generally on the  order  of.
months, during which  the actual soil  moisture supply at a given place consis*
 tently  falls short  of  the  climatically  expected  or  climatically  appropriate
moisture supply (Williams et al.   1986).   The severity of drought is a function
 of both duration and magnitude of the moisture shortfall.
     In the Williams  et al. (1986) study,  various  Palmer Drought Index
values were characterized in terms of the deviation  from the 1951-80 norrna1
climate, as follows:   greater than +6 — severe  wet  spell, +4 to +6 — extreme!?
wet, +2  to +4— wet spell, +2 to  -2~near normal, -2  to  -4— dry  spell,  -4 fc°
-6 — drought, and  less than -6 — severe  drought.   A  value of -4  to  -6 over a
period of several months is generally reflected in lower  crop yields and watef
supply problems  while a severe drought (PDI < than  -6)  has serious econowi<3
consequences because of water shortages and potential crop failure.
     Results of  the PDI  analysis in  Saskatchewan  by Williams  et  al.
suggest  that  the GISS  warming would  lead  to  a more  drought-prone
primarily because of increased evapotranspiration associated with the
temperatures.  The  existing variability  of dry  and  wet spells, however,
remain  unchanged.    Specifically, Williams  et  al.   (1986)  suggest  that  tl1
following might occur:


                                      126

-------
     •  The frequency  of drought months  (below -4)  would be  increased  by a
        factor of 3 (3/6 to 9/5).  If the GISS temperature increase occurred and
        precipitation  remained  at current  levels,  as simulated in  the  above
        results, drought month frequencies would increase by a factor of 10.

     •  Drought duration would be longer and more severe.

     •  The return period  of drought  (below  -4.0)  and  severe  drought  (below
        -6.0)  events  would be halved (see Table 8).

     The results of Williams' analysis would suggest that the relatively minor
long-term yield  changes  derived from  the  modeling  exercise presented  in the
sections addressing  the impact  of temperature  and precipitation changes  on
yield and the  effect  of increased COp on  photosynthesis could  be  somewhat  of
an over-simplification.  Indications are that  fluctuations  in yield  from year
to year could  be quite significant with drought years  becoming more frequent
and more severe.   At  the same time good years would  also occur.   Ultimately,
ln a highly unstable environmental setting  such  as  that projected  by the GISS
Cample, the current spring wheat cropping  system would most certainly  be put
J-° the test in terms  of the farmers'  ability to cope financially with goofl and
bad years,  assuming that price is not the problem (as is the case today).

       g Crop  Boundaries
     In the analysis presented above,  the effects on spring  wheat production
     discussed.   No attempt was made to look at changes  in crop zonation, that
 3 1  the movement of major  crop  boundaries or  replacement  of  certain types of
crops by others as a result of the  environmental  changes incurred.  Recently,
 °senzweig (1984)  has  assessed  the implications  of the  GISS results  for  a
 oubling of  COp  concentration  on  North American  wheat  zonation using  the
simple environmental  criteria listed  in Table  9.    These criteria  describe
       accurately  the  current  zonation  illustrated in  Figure  8 for  North
America.

     Defining the growing season as the number of days  between  the last frost
*n the spring and the first frost  in  the  fall  (0°C)  and applying the  environ-
mental criteria  listed  in Table 9  to the GISS data,  the wheat  zonation  map
lllustrated in Figure  9 was derived.   In this analysis,  the GISS data were
        over  an  8° latitudinal  x   10° longitudinal  grid system  covering  the
     mass area of  North America.   As shown, the effect of  the  GISS climate
       would  be  the  expansion  of  the winter wheat region  in  the  northern
       States  north into  Canada and  the extension of the  fall-sown  spring
      region northward and eastward.  The results in relation to moisture, at
      in  terms  of annual  requirements,  appear to  be  generally  adequate  for
      production.   However,  they  do not  take  into  account  the change  in
anriP°rative demand» which  would be increased  with  the elevated  temperatures
 ncl increased vapor pressure deficits  during  the growing season.

     Rosenzweig's results  indicate that  the climatic  change postulated  for
     would be conducive to the shift of the winter wheat belt in the U.S. into
    Canadian prairies.   They do not imply, however,  that  spring wheat produc-
     in Canada would be replaced by winter wheat.   Due  to the -very large grid
          (very large  area)  used, changes  in  the temperature and precipitation
                                     127

-------
    Table 8.  Return Period in Years for Palmer Drought Index Values of
              < -4.0 (Drought) and < -6.0 (Severe Drought) Derived for
              Selected Stations in Saskatchewan
STATION
Yorkton
Kindersley
Swift Current
Moose Jaw
Regina
Prince Albert
North Battleford
Saskatoon
Hudson Bay
Broadview
Estevan
-4.00
(drought)
1951-80 GISS
7.0
19.9
8.6
8.0
6.6
8.5
9.5
9.5
10.0
6.5
6.8
4.0
5.6
5.3
5.0
4.2
5.3
5.7
6.0
4.4
4.1
4.4
-6.00
(severe drought)
1951-80 GISS
19.0
35.0
28.0
23.5
27.0
24.0
30.0
31.0
34.0
15.0
19.0
10.0
17.0
15.0
13.5
10.0
13.5
16.0
17.5
13.5
8.5
11.2
Source:  Table 2.5, Williams et al. (1986).
                                      128

-------
    Table 9.   Wheat Environmental Requirements Used in Classification of
              Wheat-Growing Regions of North America
     Length of growing season (days)                                 90

     Growing degree units per growing season                      1200
       Minimum and base temperature                                4°C
       Maximum temperature                                        32°C

     Mean minimum temperature in January
       Spring wheat                                             <-12°C
       Winter wheat                                             2-12°c

     Vernalization requirement
       Winter wheat - at least one  mean monthly surface           <5°C
       temperature
       Fall-sown spring wheat - mean  monthly temperature          >5°C
       for all months

     Annual Precipitation (mm yr~')
       No wheat grown                                            M200
       Soft wheat                                             760-1200
       Hard wheat                                                0-760
         Dry moisture conditions                                 0-380
         Adequate moisture conditions                          380-760
Source:   Rosenzweig (1985)
                                     129

-------
  SOFT
 WINTER
  HARD
FALL SOWN
 SPRING
                                         MAJOR WHEAT GROWING AREA
    Figure 8.  Major  Wheat-Growing Areas of North America
                Source:   Rosenzweig (1985)
                               130

-------
         IvMv/Xv!] HARD WINTER


         pO^I SOFT WINTER
   j HARD FALL-SOWN SPWNG


j';-:  SOFT FALL-SOWN SPWMG
Figure  9.  Simulated North American Wheat  Regions  Using the
            GISS-GCM
            Source:  Rosenzweig  (1985)
                                131

-------
fields shifted the wheat classifications in several grid squares just under or
over  the  limit into  the  next category, and  in this  situation  some wheat of
each type might be present.

     In the case of winter wheat expansion into Canada, the key to this  is the
change  in  mean  minimum  winter  temperature.    Currently,   low  temperatures
affecting survival during  winter are the major  constraint.   However, techno-
logical  improvements  in  varieties  and production  techniques over  the last
decade appear  to  be  overcoming  the winter  survival problem.   For example,
since 1976 winter wheat hectarage in the prairies has  increased from virtually
nothing  to  over  500,000 hectares   in  1985   (Statistics  Canada 1986).   This
expansion has  also  benefited  somewhat from  the climatic  warming  currently
underway.  Hansen  et  al.  (1981),  for example, have found evidence that  global
temperatures have  risen about 0.2°C since  the middle of  the 1960s.  Similar
results have  been recorded  by Shewchuk  (1984)  for  climatic  analysis  of the
last two decades in Saskatoon, Saskatchewan.  As a consequence, if the warming
trend continues,  it  is  not unreasonable to expect  that  the shift from  spring
wheat to  winter  wheat will  continue,  assuming new markets  will  be  found and
exploited.   As to whether  winter  wheat  would entirely  replace  spring wheat
production,  only  time will  tell.    Certainly if summer  droughts  become more
frequent  and  severe,  logic  would suggest  that  this  would  be   the  case.
However,   future  market  conditions,   improvements   in  technology,  and  the
possible  development  of  more  heat- and drought-resistant  varieties  in the
future will ultimately determine this.

CONCLUSION

     Results from  this  study  suggest  that  the possible  changes  in  the long-
term  climate   of  Saskatchewan  resulting from the  GISS  general  circulation
modeling  experiments  for a  doubling of atmospheric  COp  would  increase the
annual growing season  in  the  prairies by an average of 48 days  and would
increase precipitation  from  11/5 to  14/t.  In  conjunction  with the increase i°
growing season length agricultural  planting dates would  be  advanced by about
three weeks and  the  fall  harvest  period  would  be  extended by about four
weeks.    In   total,   the  growing  season   climatic  conditions  throughout
Saskatchewan would become more homogeneous.

     In  spite  of  the  increased precipitation and  enhanced  COp   effect  on
photosynthetic capacity,  the  impact  of  the  GISS  COp  climate change  woul^
generally reduce  spring wheat  yields   and production potential by  5%  to 20*
depending on whether  projected GISS precipitation  increases  were  attained °f
not.   Any change  in  variability or reduction in  precipitation  from current
levels could reduce production significantly.

     Analysis  of  the  projected GISS climate  change  for  Saskatchewan suggest-'
that the prairies would become more drought-prone with droughts occurring wittJ
greater  frequency and  severity.    The effect of  this  is   likely  to  be  ajj
increase  in the   variability  in  yields and production  between years.    V.
probable  consequence  of this  situation in  all likelihood will be a shift °*
the winter wheat belt from the U.S. into Canada.   This is a likely conclusi^
strictly  from  the magnitude  of  the  climate change  postulated  by  the GlS\
model,  which  estimates  a  shift  of  the  present   climate   of  Nebraska  fc
Saskatchewan.  A  shift to fall-sown crops in  the  existing agricultural
would enable  farmers to  take  advantage  of  increased fall  and  early


                                      132

-------
moisture levels.  It would enable crops to develop and mature before the onset
°f drought  conditions  in  June and  July, which  would  be  most damaging  for
SPF ing-sown crops.

ACKNOWLEDGMENTS

     The  author would   like  to  express  his  appreciation  to  a  number  of
individuals for their invaluable  contributions to the above  project.   Thanks
are extended to W.J.  Blackburn  and C. Stewart  for  reviewing this  manuscript,
to D. Murray for help in preparing the  figures,  to R. Muma  and J.  Hardy  for
Preparing the  computer  programs and generating  the required data,  and  to L.
Tneriault for the  typing of this manuscript.


REFERENCES

Aston,  A.R. 1984.    The  effect of doubling atmospheric C02  on stream flow:  A
     simulation.   J. Hydrol.  67:273-280.

Atmospheric Environment  Service. 1982.   Canadian climate normals 1951-80. Vol.
     1-6.   Downsview,  Ontario, Canada:  Environment Canada.

      W. 1986.  In  press.   GCM-derived climatic scenarios of increased atmos-
     pheric C02 as a basis  for  impact  studies.    In  Assessment of climatic
     impacts on agriculture   in  high  latitude  regions.  I,   ed.  M.  Parry,  T.
     Carter, and N.  Kanin.   Laxemburg,  Austria:  International  Institute  for
     Applied Systems Analysis.

Baier,  W.,  J.A.  Dyer,  and W.R.  Sharp.  (1979).  The  versatile soil  moisture
     budget.   Tech. Bull.  87:52.   Ottawa:  Land Resource Research  Institute,
     Agriculture Canada.
Bi
 Aaekburn,  W.J., and  R.B.  Stewart. 1984.   Crop production potential:   Esti-
     mating the effects  of climatic change.   Agrologist.  13(2) :4-6.

 °°tsma,  A., W.J.  Blackburn,  R.B.  Stewart,  R.W. Muma,  and  J. Dumanski.  1984.
     Possible  effects of  climate  change on  estimated  crop  yields in  Canada.
     Tech.  Bull. 1984-9E.  Ottawa:  Research Branch,  Agriculture Canada.

 r°°ks,  C.E.P.  1943.  Interpolation tables for daily values of  meteorological
     elements.   Quart. J.  Roy. Met. Soo.  69(300): 160-62.
         C.T. 1965.   Photosynthesis of leaf canopies.   Agricultural  Research
     Report  663.    Centre  for  Agricultural Publications  and  Documentation,
     Wageningen.  pp 57.

         , J., and  A.M.  Kassam.  1979.   Yield response to  water.   Rome:  F.A.O.

          j.f and  R.B.  Stewart.  1981.   Crop  production potentials for  land
     evaluation  in  Canada.    Tech.  Bull.  1983-13E.  Ottawa:  Research  Branch,
     Agriculture  Canada.
                                     133

-------
F.A.O. 1978.  Report on the agro-ecological zones project.  I, Methodology and
     results from Africa.  World Soil Resources Report 48. Rome: F.A.O.

Gifford,  R.M.  1979.   Growth  and  yield  of C02  enriched wheat  under water-
     limited conditions.  Aust. J. Plant Sci. 6:367-78.

Hansen,  J., D.  Johnson,  A.   Lacis,  S.  Lebedeff,  P.  Lee,  D.  Rind,  and  G.
     Russell.  1981.     Climate   impact   of   increasing  atmospheric  carbon
     dioxide.  Science. 213:957-66.

Hansen,  J.,  G.  Russell,  P.  Stone,  A.  Lacis,  S.  Lebedeff,  R.  Reidy,  and  L.
     Travis.  1983.    Efficient  three-dimensional global  models  for climate
     studies, Models I and II.  Mon. Wea. Rev.  111:609-62.

Havelka, U.D.,  V.A. Wittenbach,  and  M.G.  Boyle.  1984.  C02 enrichment effects
     on wheat yield and phenology.  Crop Sci. (24):1163-68.

Keeling,  C.D.,  J.A.  Adams  Jr., C.A.  Ekdahl Jr.,  and  P.R.  Guenther.  1976a-
     Atmospheric  carbon  dioxide  variations  at  the  South  Pole.    TeUH5
     28(6):552-64.

Keeling,  C.D.,   R.B.  Bacastow,  A.E.  Bainbridge,  C.A.  Ekdahl,  Jr.,  P.R-
     Guenther,   and   L.S.   Waterman.   1976b.     Atmospheric   carbon  dioxide
     variations at Mauna Loa Observatory, Hawaii.  Tellus 28(6):538-51.

Kimball, B.A. 1983.   Carbon dioxide  and agricultural  yield:  An assemblage and
     analysis of 430 prior observations.  Agron. J. 75:779-88.

Krenzer  Jr.,  E.G.,  and  D.N.  Moss.  1975.   Carbon dioxide enrichment effect3
     upon yield and yield components in wheat.   Crop Sci. 15:71-4.

LeDrew,  E.F.,  R.  Dumancic,  J. Peivowar,  B.   Tudin, and  D.  Dudycka.  1983-
     Development of a  100 km  grid  square climatic data base.   Report prepare"
     for  Agr.  Canada,  Contract Number DSS 245U.0.1416-2-0754,  April  19&3-
     Ottawa: Agriculture Canada.

Lemon, E.R., ed.  1983.  C02 and plants.  AAAS Selected Symposium 84.  Boulder'
     Colorado:  Westview Press, Inc.

Major, D.J.,  and W.M.  Hamman.  1981.   Comparison of sorghum with  wheat  an
     barley grown on dryland.   Can. J. Plant Sci. 61:37-43.

Manabe, S., and R.J. Stouffer. 1980.  Sensitivity of a global climate model fc
     an  increase  in COj,  concentration in  the  atmosphere.   J.  Geophys.
     85:5529-54.

Manabe,  S.,  and  R.W.  Wetherald. 1980.   On  the  distribution  of climate
     resulting from an  increase in COP content  of  the atmosphere.  J.
     Sci. 37:99-118.

McCree, K.J. 1974.  Equations for the rate of dark respiration of white
     and  grain sorghum,  as functions  of dry weight,  photosynthetic rate
     temperature.  Crop Soi.  14:509-14.
                                      134

-------
Mitchell,  J.F.B. 1983.   The seasonal response of  a  general circulation model
     to changes in  C02  and  sea  temperature.    Quart.  J.   Roy.  Met.  Soc.
     109:113-52.

Newman, J.E.  1980.   Climate change  impacts  on  the  growing season of the North
     American corn  belt.   Biometeor. 7(2): 128-42.

Onofrei,  C.  1984.  Personal  communication.  University of Manitoba,  Winnipeg,
     Manitoba.

palmer, W.C.  1965.   Meteorological  drought.    U.S.  Department of  Commerce
     Research Paper No.  45,  Washington,  D.C.

Papry,  M.L.,  T.R.  Carter, and  N.  Konijn.  1984.   Climate  impact analysis  in
     cold  regions.   Proc. XIX Conf. of  Finnish Geographers, Oct. -27-28, Oulu,
     Finland.   Nordia  l8(2):67-79.

Penman, H.L.  1948.   Natural evaporation from open  water,  bare  soil  and grass.
     Proc.  Roy.  Soc. London.  A.  193:120-45.

Ritchie,  J.T.  1972.   Model  for predicting evaporation  from  a  row  crop  with
     incomplete cover.   Water Res.  8(5) : 1204- 13.

R°bertson,  G.W.  1968.    A  biometeorological  time  scale for  a cereal  crop
     involving  day  and night temperatures  and photoperiod.  Int. J.  Biometeor.
     2:191-223.
          ,  C.  1984.   Potential  C02  induced climatic effects on North American
     wheat-producing  regions.  Climate Change 7:367-89.

Saskatchewan Agriculture.   1985.    Agricultural  statistics 1984.   Statistics
     Section,    Economics    Branch,   Saskatchewan    Agriculture,    Regina,
     Saskatchewan .

 "ewchuk,  S.R.  1984.   An  atmospheric  carbon dioxide review and  consideration
     of  the mean  annual temperature  trend at Saskatoon.  Saskatchewan.    SRC
     Technical  Report No.  160.   Saskatoon,  Saskatchewan:  Saskatchewan Research
     Council.

  y»  W.K.,  and M.C.   Coligado.  1974.    Agroclimatic maps  for Canada  derived
     data:   Moisture  and critical  temperatures  near freezing.   Tech.  Bull.
     (81), Ottawa:  Research  Branch,  Agriculture Canada.

 °nit»  N.,  H.  Hellmers, and B.R.  Strain.  1980.   Growth  and yield of  wheat
     under C02  enrichment and water  stress.   Crop Sci. 20:687-90.

  atiatics  Canada.   1986.   Field  crop  reporting.   Series  No.  8.   Ottawa,
     Canada.
Sf
  ev*art,  R.B.  1981.     Modeling methodology for  assessing  crop  production
     Potentials in  Canada.    Tech.  Bull.  (96).  Ottawa:    Research  Branch,
     Agriculture Canada.
                                     135

-------
Tung,  F.L.  1986.    Personal  communications.    Regional  Development Branch,
     Agriculture Canada, Ottawa.

Williams, G.D.V. 1969.  Applying estimated temperature normals to the zonation
     of the Canadian Great Plains for wheat.  Can. J. Soil Sci. 49:263-76.

Williams,  G.D.V.  1975a.   An  assessment  of the  impact of  some hypothetical
     climatic  changes  on cereal  production in Western  Canada.    Proc.  Confj.
     World Food Supply Changing Climate.  Sterling Forest, N.Y., 1974.

Williams,  G.D.V.  1975b.   An  assessment  of the  impact of  some hypothetical
     climatic  changes  on cereal production  in  western Canada.    Proc.  ConJU
     World Food Supply in Changing Climate. Sterling Forest, N.Y., 1974.~~~

Williams,  G.D.V.,  and  W.T.  Oakes.  1978.   Climatic  resources for  maturing
     barley and wheat  in Canada.   Essays  on meteorology and climatology.   In
     Honour of Richmond W. Longley.   In Studies in Geography Monograph 3.eds.
     K.D. Hage and  E.R. Reinelt, 367-385.   Edmonton, Alberta:  Department of
     Geography, University of Alberta.

Williams, G.D.V.,  J.S. McKenzie, and  M.I. Sheppard.  1980,    Mesoscale  agro-
     climatic  resource mapping by computer,  an example  for the  Peace  River
     region of Canada.  Agr.  Meteorol. 21:93-109.

Williams, G.D.V., H.K. Jones,  E.G. Wheaton,  R.B.  Stewart,  and  R.A.  Fautley-
     1986.   Estimating the impacts of  climatic change on agriculture  in  the
     Canadian prairies,  the  Saskatchewan  case  study.  In  Vol.  1  - Assessment
     of climatic impacts on agriculture in high latitude  regions.   Laxenburgi
     Austria: International Institute for  Applied  Systems Analysis.
                                     136

-------
Potential Effects of Greenhouse Warming
on  Natural Communities1

Robert L Peters
World Wildlife Fund and the Conservation Foundation
Washington, DC USA

Joan D. S. Darling
Consulting Ecologist
Springfield, Virginia USA
ABSTRACT

     Previous  natural  climate changes have  caused  large-scale  geographical
shifts,  changes  in  species  composition,  and  extinctions  among  biological
communities.   If the widely predicted greenhouse  effect  occurs,  communities
wUl  respond  in similar ways.   Moreover,  population reductions and habitat
Destruction due  to other human activities will  make  it  difficult  for species
r° shift  ranges  in  response to  changing  climatic  conditions.   This  paper
ldentifies  some groups of  species  at risk,  including  coastal,  species  and
refonant  populations   near   the  extremes   of the  original   species  ranges.
Survival of many species  will depend  upon  greater management responses than
currently envisioned,  including transplantation  and in situ management.

     As I did stand my watch upon the hill,
     I look'd toward  Birnam,  and anon,  methought,
     The wood began to move.
        Macbeth, Act  V Scene V ,

     Current human development and population trends suggest  to  all  but  the
 er>y  optimistic  that  by the next century  most other surviving  terrestrial
j|Pecies may  well be  relegated  to small  patches  of their original habitat,
Patches  isolated by  vast  areas  of  human-dominated urban  or agricultural
 ands.    Without heroic measures. of  habitat  conservation  and  intelligent
g^gement, hundreds   of thousands of  plant and animal  species could become
 *tinct by  the end of this century  (Myers  1979; Lovejoy  1980), with more to
o°Uow £n  tne  next.   This  diminution of biological diversity will have major
 °naequences for human society.
1
   Adapted by the authors  from BioScience  35:707-17, December  1985.
   c°Pyright American  Institute of Biological Sciences 1985.  Published
   with permission.
                                    137

-------
     Many species will  be  lost because no habitat  reserves are set aside for
them, but  even those within  reserves  will be threatened  by  a combination of
genetic and ecological  events (Diamond 1975; Soule and  Wilcox 1980).  Recent
investigation into these events has  provided  insight  into  how reserves should
be designed and  managed (Frankel  and Soule  1981; Schonewald-Cox et al. 1983?
Soule  and  Wilcox  1980).    But although  the significance of  future climate
change to species survival has been independently mentioned by several authors
(Ford  1982; Worse  and McManus  1980;  Wilcox   1980),  little attention has been
given to the  impact  on  biological diversity  of  an  increasingly likely event!
global C02-induced  climatic  change,  due  to  the greenhouse  effect.   If the
greenhouse warming  occurs,  it  will  pose  a  new and  major threat  to species
within  reserves,  species  already   stressed  by  the  effects  of  habitat
fragmentation.

     Our understanding  of  how atmospheric composition affects global climate
is still  in   its  infancy,  but an  increasing  body  of knowledge  suggests that
several  types  of change   affecting  the  survival  of  species—including  a
substantial  global  increase  in  temperature,   a  widespread  alteration  of
rainfall patterns, and  perhaps a  rise  in  sea level—may  be  caused by risinS
concentrations  of COo  and  other  anthropogenic polyatomic   gases  (Hoffmaflt
Keyes, and  Titus  1983;  Machta  1983;  Manabe, Wetherald,  and  Stouffer  198K
National Research Council  1983; Schneider and Londer 1984).

     This  paper  identifies  problems  caused by  climate  change that  affect
biological communities, examines  the particular  difficulties  faced  by species
in  biological  reserves,  and  suggests  management  options.    Although  we
recognize that dealing  with  short-term  extinction  threats alone will strai*5
the resources of conservationists, we  feel that  the possible negative effect^
of global warming could be so severe that conservation plans should be amended
to reflect  knowledge of  climatic effects  as soon  as   it becomes  available^
Decisions about  the  siting and design of  reserves, and  assumptions about
much  management will  be  needed  in  the  future, must  reflect  the increa
economic and biological demands of global warming.

PATTERNS OF CLIMATIC CHANGES

     Continued  burning  of  fossil fuels,  with a  possible contribution
progressive deforestation,  is  increasing  atmospheric CCU concentration
could reach double the concentration in 1880 within  the next 100 years (t
et al.  1981;   MRC  1983; Schneider  and Londer  1984).   The  concentration  JJ
additional—greenhouse gases,  notably methane  and  chlorofluorocarbons-  wi
also  increase  significantly as the  result  of human activities  (Machta
Ramanathan et al. 1985).   Because  the  greenhouse  gases  absorb some  of
upward infrared  radiation  from the ground, preventing its escape  into
the  lower  atmosphere  will  grow  warmer.   There  is still a great  deal
uncertainty  about   the  greenhouse   process,  and   predictions  depend
assumptions about future trends in fossil  fuel  use,  the precise nature -   gf
carbon cycle, and the complexities of atmospheric  interactions.  Nonethel^ ^
most  experts  agree  that globally  the  climatic  average  could  warm  by 1
U.5"C by the  end of  the next century (NRC  1983).  Moreover, this change
likely be two or  three  times  greater at  the  poles (Schneider  and Londer
see Figure 1a for one model's predictions).
                                      138

-------
 H WETTER

   DRIER
eof Ure  1*  ^   Global  patterns  of  surface  temperature  increase,  as
tone? by the Goddard Institute for Space Studies (GISS) model (Hansen et
    iQQr\degrees C'  (b)  Global conges in moisture patterns (Kellogg and
                                 139

-------
     A change  of this magnitude  is  large compared with  normal fluctuations.
For  example,  an  increase  of  only  2°C over  the  current  average  global
temperature would make the planet warmer  than  at any  time in the pas-t 100,000
years (Schneider and Londer 1984).

     Furthermore, although CC^ doubling would not  be reached  for  some  time,
transient  temperature  increases occurring  before  doubling  is  reached  might
still have significant  impact on biological  systems.    Indeed,  if climatic
models predicting  the  greenhouse effect  are correct, warming distinguishable
from  normal climatic  variation  should  occur  within  the  next  10-15  years
(Hansen et al.  1981; Madden  and  Ramanathan  1980)  and  may, in fact, already be
observable (World Meteorological Organization 1982).

     As  important  to biological  communities as  temperature  change  itself is
that the projected increases  in  temperature would  cause widespread changes in
precipitation patterns (Hansen et al. 1981; Kellogg and Schware 1981; Manabe,
Wetherald,  and Stouffer 1981; Wigley et  al.  1980).   For  many species, a change
in water availability  would  have greater impact than  temperature  changes of
the order predicted (e.g., Neilson and Wullstein 1983).

     Although  precise  regional  predictions  of future  precipitation patterns
are  yet  to  come,   some  attempts  have  been  made  to estimate  large-scale
changes.  For example, in their model of future rainfall patterns, Kellogg and
Schware  (1981)  suggest  that  the American Great  Plains  may experience as much
as a  40£ decrease in  rainfall by the  year 2040 (Figure  1b).   In some  areas
increased  evaporation   caused  by   increased   temperature  could  exacerbate
regional drying (e.g., Manabe, Wetherald, and Stouffer  1981).

     A rise  in sea  level resulting from thermal  expansion of  sea  water and
melting  of glaciers and  polar  ice  caps  has been  widely discussed  as  well*
although such  estimates  of a rise vary.  NRC  (1983)  has estimated a possible
increase of 70  cm over  the next century;  another study  projects a most likely
rise of  between 144 cm and  21?  cm  by 2100 (Hoffman, Keyes,  and Titus 1983)-
If the western  Antarctic ice cap melted, which  is  highly uncertain, rises of
up to 5-6  m might occur  over the next several  hundred years  (Hansen et al-
1981; NRC  1983).

     In  addition,   the  warming  trend  may   alter   the  ocean's  vertical
circulation, causing change in the upwelling patterns that sustain many marine
communities (Frye 1983; Kellogg  1983).

     Finally,  increased  atmospheric  C02 may result in  more acidic, nutrient-
poor  soils  (Kellison  and Weir  1986).   It  may  also   change  photosyntheti0
efficiencies, growth rates,  and  water requirements  of different plant species
in different  ways (NRC  1983), thereby  altering competitive outcomes (Strain
and Bazzaz 1983) and possibly destabilizing natural ecosystems.

THE SPECIAL CASE OF BIOLOGICAL RESERVES

     Such  changes  in  important environmental parameters that  determine the
range of species would affect nearly all species,  but  the consequences would
be most  dire for those  restricted  to  reserves  or  sharing characteristics °*
species  restricted  to reserves, notably  limited range, small population, an
genetic  isolation.    Populations  within reserves,   such  as national


                                      140

-------
national forests, and  wildlife refuges, will typically  be  remnants of larger
original  populations  reduced  through  overharvesting  or   habitat loss  and
therefore will be subjected  to a  variety of threats more serious to them than
to larger and more widespread populations.

     Whenever  the  area  that  an  original  community of  species occupies  is
reduced,  as  when   a  reserve  is  created  and  the  land  surrounding  it  is
developed,  some  species are lost  (Diamond 1975;  Terborgh and  Winter  1980;
Wilcox  1980).   Some disappear rapidly  because  the reserve does not  include
necessary resources;  others  are  lost  because  any  large-scale  environmental
change can  cause  extinction  if the  population is  too  localized; some vanish
because of inbreeding and genetic drift.

     As these  environmental  and genetic factors combine  to cause the  loss of
some  species,  readjustment  of   mutualistic,  parasitic,  competitive,  and
predator-prey relationships  among the  remaining species  must take place,  most
likely causing  the  loss of  still others (e.g.,  Paine  1966).   Climate change
thus  brings  new  pressures,  including  physiological  stress  and  changejs  in
competitive   interactions,  to bear on  reserve species  already stressed  by  a
disequilibrated community.   A common result of these climate-induced pressures
Would be  further  diminution  of species' ranges  and population  sizes,  which
Would  in turn accentuate  the various environmental  and  genetic  effects
associated with small populations,  perhaps leading to extinctions.

     Not only can  the isolation of a population within a reserve surrounded by
Altered, unsuitable habitat  mean it would receive  little numerical  or  genetic
Augmentation  from  any  populations outside  the reserve,  but the converse  is
also  true.     Isolated  reserve  populations  could  not  respond  to  changing
climatic  conditions within   the  reserve  by  colonizing other   "islands"  of
habitat outside the reserve where the climate is  suitable.

     Reserve species, which would generally be geographically localized,  would
be more  likely to  experience  intolerable  climatic changes throughout  their
Ranges than  would more widespread species.   For example,  a  tree species  whose
entire range  falls  in  an area due to undergo regional drying  is  more  at  risk
than  one  whose larger range  includes  areas  outside  the  desiccation  zone.
^rther, remnant populations  in reserves may represent only a  fraction of the
8ene  pool originally present  in  the  species  as  a  whole  (Frankel and  Soule
^981).   Diminution of  a species'  range could mean  the loss of populations
adapted to  particular  climatic  conditions, decreasing  the genetic material
that both nature and humans have to  work with.

     A climatic change  would often improve conditions for a  particular  species
at one  margin of  its  range  and worsen  conditions  at the  opposite.   Reserve
P°pulations   located  near  a  margin  where  conditions are deteriorating  would
therefore be more  threatened  than  ones at  the  opposite  end  of the  range
figure 2).

COMMUNITIES  RESPOND TO CLIMATE CHANGE

     In the  past,  entire biomes have shifted in response to  global temperature
Changes no  larger than  those that may occur during  the  next  100 years (Baker
J983;  Bernabo and  Webb  1977; Butzer  1980;  Flohn  1979;  Muller  1979;  Van
"evender  and  Spaulding  1979).  In  general,  when  temperatures have risen,

-------
     (  /FORMER\  V
     \  xRESERVE/  \


oI/SL  ^x^	.}
                                                    r

     Figure 2.   How climatic  warming may turn biological reserves into for"16,
reserves.     Hatching  indicates:    (a)  species  distribution  before
habitation, southern limit, SL, indicates southern limit of species range;
fragmented  species  distribution  after   human  habitation;   (c)   spe
distribution after  warming.
                                     142

-------
species colonized  new habitats  toward  the  poles,  often  while  their  ranges
contracted  away  from the  equator  as  conditions  there  became  unsuitable.
Equatorial organisms thus expanded their ranges into areas previously tenanted
by temperate ones,  while temperate organisms did  the  same  in some areas that
had previously been the domain of boreal communities.

     During several Pleistocene interglacials, for example,  the temperature in
       America  was  apparently 2°-3°C higher  than  now.    Osage oranges  and
Pawpaws grew near  Toronto,  several hundred kilometers north  of their present
distribution; manatees swam in New  Jersey;  tapirs  and  peccaries  foraged  in
Pennsylvania;  and  Cape  Cod  had  a   forest  like  that of  present-day  North
Carolina (Dorf 1976).  Other significant changes  in  species'  ranges have been
°aused  by  altered   precipitation  accompanying   global  warming,   including
exPansion  of prairie  in  the American Midwest during a global warming episode
aPproximately 7000 years  ago (Bernabo and Webb 1977).

     Although Pleistocene and past Holocene warming  periods were probably not
    to elevated C02  levels,  researchers have predicted that,  if the proposed
    induced  warming  occurs,  similar  species  shifts  would  also occur,'and
   etation belts  would move  hundreds of  kilometers  toward  the poles  fFrye
     ; 300 km is a reasonable estimate based  on models  (Miller, Dougherty, and
        1986)  and  on the  positions  of  vegetation  zones  during  analogous
        periods in the past  (Dorf 1976; Furley et  al. 1983).
     Although both  the fossil  record and  current distributions  demonstrate
     many  species  have been  able  to shift  successfully  in response  to such
 •"•iniate changes,  many others have not, either because their rates of migration
*6re too  slow or  because geographical  barriers like  oceans,  mountains,  or
^eas of inappropriate soil type prevented their reaching  suitable habitats.

     For example,  a  large, diverse  group of plant  genera, including water-
J? eld  (Brassenia).   sweet  gum (Liquidambar).   tulip tree  (Liriodendron).
?^8nolia  (Magnolia),  moonseed  (Menispermum),  hemlock (Tsuga).  arbor  vitae
ir-Syja), and  white cedar  (Chamaecyparis), had  a circumpolar distribution  in
 jje Tertiary.   But  during  the Pleistocene ice ages, all went extinct in Europe
  *le surviving  in North  America.    Presumably,  the  east-west orientation  of
aH°!?, barriers  as the  Pyrennes, Alps, and the  Mediterranean,  which  blocked
           migration,  was  partly  responsible  for  their  extinction  (Tralau
         In the  case of  reserve species, human  modification  of  surrounding
        will create barriers of agricultural or urban  land  which  will  be Just
    effective  as mountains  or  oceans  in preventing  colonization  of  other
         areas.

w    If global warming of 2°-3°C did  occur by the end  of  the next century,  it
w uld  be  very   rapid  compared  with  some  prehistoric  changes  of  similar
Ia8fitude*   In  contrast,  the change  to  warmer  conditions  at  the end of  the
- 5t ice age, considered  rapid,  spanned  several thousand  years  (Davis 1983).
a* pate  of  change  has  profound  significance  for  species  survival,  for even if
^oo   le  land is  Preserved for a  species to shift  to,  extinction may  still
  °up if present  habitat becomes  unsuitable faster than  new habitat can be
8pe  Tne  fossil  record  shows  that  dispersal  rates  have  been  crucial  to
   °ies'  ability to  colonize  suitable  habitat during  past climate  changes.
                                     143

-------
For  example,  warm-temperate  plant species  were  pushed south  out  of Great
Britain  and  Ireland  by  cold during  the  Pleistocene.    As  the  temperature
increased,  these plants  later moved northward again, but  only some .dispersed
rapidly  enough  to  reach Great Britain  before rising sea  levels separated it
from  the European continent,  and  fewer  could  colonize Ireland  before that
island  was separated  from  Britain (Cox,  Healey,  and  Moore  1973).   Other
species  that  thrived in  Europe during  the cold periods,  but could not survive
the conditions in postglacial forests,  could not extend their ranges northward
in time and became extinct except in cold, mountaintop refugia (Seddon  1971).

     If  estimates  of  a several-hundred-kilometer  poleward  shift in temperate
biotic belts  during the  next  century are  correct,  then  a localized population
now  living  where temperatures are near  its  maximum thermal  tolerance would
have  to  shift northward  at  a rate  of several  kilometers  per  year  to avoid
being left behind in areas too warm for survival.  Although some species, such
as plants propagated by spores or dust seeds, may be able to match these rates
(Perring 1965), many species  could  not  disperse  fast enough to compensate for
the expected  climatic change  without human assistance,  particularly given the
presence of dispersal  barriers.   Even wind-assisted dispersal may fall short
of the mark for  many  species.  For example,  wind  scatters  seeds of the grass
Agrostis hiemalis.  but  95?  fall  within   9  m of  the  parent plant  (Willson
1983).  In the case of the Engelmann spruce, a tree with light, wind-dispersed
seeds,  fewer  than  5% of  seeds  travel  even  200  m downwind,  leading  to  an
estimated migration rate of between 1 and 20 km per century (Seddon 1971).  An
extreme case  is the double coconut  (Lodoicea maldivica), whose giant seed can
"only fall  off the tree,  and if the  tree  grows  on a  slope, roll downhill"
(Willson 1983).

     Although animals are mobile,  the  distribution of some  is limited by the
distributions of particular  plants; their dispersal rates  would thus largely
be determined  by  those of co-occurring  plants.    Behavior  may often restrict
dispersal even of animals physically capable of large movements.   For example*
dispersal  rates  below 2  km/year  have  been  measured  for several  species  of
deer, and many tropical  deep-forest birds simply do  not  cross even very small
unforested  areas   (Diamond  1975).   On  the  other hand,  some  highly  mobil6
animals,   particularly   those   whose  choice  of   habitat  is   relatively
unrestrictive, may  shift rapidly.   Several  authors (see  Edgell  1984)  have
suggested, for instance,  that climate change caused major range shifts in some
European migratory waterfowl  in this century.

     Figure 3 illustrates the difficulties to be  faced  by  a population whose
habitat  becomes  unsuitable due  to climate  change.    Propagules must  run  an
obstacle course  through  various  natural  and  human-created  dispersal barriers
in a  limited  amount of time  to reach habitat that will  be  suitable under tfce
new climatic regime.   Dispersal ability will be crucial.

     Because  species shift at different rates in  response  to climate change»
communities may disassociate  into their component  species  (Figure  4).  Recent
studies  of  fossil  packrat (Neotoma spp.)  middens  in the  southwestern United
States show that  during the  wetter, moderate climate of  22,000-12,000
ago,  there was  not  a  concerted  shift  of  communities.     Instead,  s
responded individually to climatic  change,  forming stable,  but by present-
standards,  unusual assemblages   of plants   and  animals   (Van  Devender
Spaulding  1979).   In  eastern  North  America,   too,  postglacial communitie


                                     144

-------
      OT
      E
   300
   250
   200
   150
   100
    50
                                         SUITABLE
                                         HABITAT
           UNSUITABLE
           SOILS
                                          URBANIZATION
                    DESERTS
               OCEANS
AGRICULTURE
MOUNTAINS
                             \
     I
          "STARTING
           POPULATION
                                    SL
                                                                  SL
                                                                      1
     Figure  3.  Obstacle course to be run  by species facing climatic change  in
a human-altered  environment.   To  "win,"  a population  must track its shifting
^iraatic  optimum and reach suitable habitat north of the new southern limit  of
the  species range.    SL1   =  species  southern range  limit  under  initial
°°nditions.   SL2  = southern limit after  climate change.   The  model  assumes a
Want species consisting  of a single population, which has  its distribution
Determined   solely  by   temperature.   After  a  3°C  rise  in  temperature the
Population must have shifted 250 km  to the north to survive,  based on Hopkins
Dloclimatic  law  (MacArthur  1972).   Shifting will occur  by simultaneous  range
contraction  from the south and expansion by dispersion and colonization to the
        Progressive shifting depends upon propagules that can  find  suitable
        to  mature and  in  turn  produce  propagules  that  can colonize more
        to  the north.   Propagules  must  pass around natural and  artificial
 "Stacles  like mountains,  lakes,  cities, and  farm  fields.    The  Englemann
 Pruce  has an estimated,  unimpeded  dispersal  rate of  20 km/100  years (Seddon
 *'1).  Therefore,  for  this species  to "win," colonizing habitat to the  north
 r the  shifted hypothetical  limit  would require a minimum of 1,250 years.
                                    145

-------
                              original  A + B

     Figure 4.   (a)   Initial distribution of  two species, A  and B,
ranges largely  overlap.    (b)    In  response to  climate change,  latit
shifting  occurs  at species-specific rates, and the ranges disassociate.
                                     (46

-------
were often  ephemeral associations of  species,  changing as  individual ranges
changed (Davis 1983).

     An alternative  to  latitudinal  shifting,  even  for  species  that cannot
disperse rapidly,  is to change altitude.  Generally, a short climb in altitude
corresponds  to  a  major  shift in  latitude:  the  3°C cooling  of  500 m  in
elevation equals roughly  250 km in  latitude (MacArthur 1972).   Thus, during
the middle Holocene when temperatures in eastern North America were 2°C warmer
than at .present,  hemlock  (Tsuga  canadensis) and  white pine  (Pinus strobus)
were found  350  m  higher on  mountains  than  they are  today  (Davis 1983).
Similarly,  species  that  could not  shift  poleward rapidly  enough during  a
future  warming  trend  to  track a  climatic  optimum  might  be  able  to  find
sanctuary on mountains.

           Biological Mechanisms
     Climate change  might cause  local extinction  in two  interrelated ways.
    is physiological:  the climate of  a formerly habitable area changes so it
^°  longer   corresponds  to  a species'  physical  tolerances.    The othet-  is
interspecific:   climate  change  alters  interactions,  such  as  predation  or
competition, so that a formerly successful species  is eliminated from an area
where it could  physiologically survive.

     Numerous examples of  temperature's direct influence on species' distribu-
tion and  survival  exist.    The direct  range-limiting  effects of excessive
warmth include  lethality,  as in  corals (Glynn  1984),  and  interference with
^Production,  as  in  the large  blue  butterfly,  Maculinea arion  (Ford  1982).
"oisture extremes exceeding  physiological  tolerances also  determine species'
aiatributions.   Thus, the  European range  of the beech  tree (Fagus sylvatica)
®nds to the south where rainfall  is  less than 600 mm annually (Seddon  1971),
*nt  only  be  threatened by  competitors naturally  occurring
  w   a  reserve>  but  tneY ""ay also feel Pressure  from invaders that find the
bamhcll«atic  regime to their liking.  For example,  Melaleuca quinquenervia. a
         e  Ausfcralian eucalypt, has  invaded the Florida Everglades,  forming
   3e  wonotypic stands where  drainage and frequent fires have  disturbed the
        marsh  community  (Courtenay  1978;  Myers  1983).   Such  invasions may


                                      147

-------
become commonplace in response to large-scale climate changes, and controlling
them  is  one  of  the major  concerns of  reserve  managers  (Goigel  and Bratton
1983).

     The underlying  physiological  adaptations of most  species  to climate are
conservative, and it is unlikely that  most species could evolve significantly
new tolerances in the time  allotted to them by the coming warming trend.  The
llama, for  example,  has  water turnover  rates  as  low  as those of its relative
the  camel,  even  though  the  llama  has  lived  in  cold,  wet  environments for
several  million   years   (MacFarlane    1976).     Indeed,   the   evolutionary
conservatism  in  thermal  tolerance  of many  plant  and  animal species — beetles,
for example (Coope 1977) — is the underlying assumption that allows us to  infer
past climates from faunal and plant assemblages.

     In contrast, some  invertebrates have  apparently  adapted when introduced
into  new  thermal  environments.     Several species  of   freshwater  tropical
invertebrates accidentally introduced into temperate waters survived initially
only  in  artificially heated  waters,  such  as  power plant  outflows,  but were
later found spreading into nonheated sites (Aston 1968; Ford 1982).

A Reserve Scenario

     Because  the  ecological  ties  binding a species to  its environment are so
complex,  the preceding physiological, interspecific, and genetic factors  would
combine to affect reserve populations confronted with climatic change.

     Imagine  a hypothetical  situation  where a  single  oxlip  population  &
confined within a British reserve,  excluded from  the  reserve's dry sites by a
competitor,  the  dog's mercury.   Then,  because  of  global  climatic  changes*
rainfall decreases within the reserve, allowing dog's  mercury to displace the
oxlip from an increasing number of its traditional sites.
     At the  same  time,  the ecological  relationships of other  species in
reserve are also changing, and some of these affect the oxlip.  For example* a
previously  rare,   second   competitor  of   the  oxlip  undergoes  a  populati°n
explosion following the extinction of its  major  predator.   In addition, a
insect herbivore introduced by humans finds the oxlip to its liking.
     As  the  oxlip  population  becomes  smaller  and  more  fragmented
physiological stress,  competitive  exclusion, and  increased  predation,
environmental catastrophes and  the inevitable genetic  deterioration of
populations take their toll.   Because the reserve population has been
from  other  populations  outside  the   reserve,   its  genetic  composition  **
relatively homogeneous to begin with and thus lacks the genetic variability  6
cope with the environmental threats.  Moreover, no propagules from outside &1
reserve  can  bolster  or  reestablish   populations where  the  oxlip  become5
extinct.

     When  the oxlip  disappears, other  reserve  populations,  such  as inse?fl
herbivores, that depend on the  oxlip will  also decline.  Even a decrease *^,-
the oxlip  population that falls short  of extinction may  cause  extinction °
species depending on it for food.
                                      148

-------
     Although this scenario is hypothetical, such complex interplay leading to
extinction can be seen today.  For example, the two southern subspecies of the
northern flying  squirrel,  Glaucomys sabrinus  fuscus and G.s.  coloratus,  are
°onfined as glacial relicts  to several boreal populations  in  the Appalachian
Mountains.   They are increasingly endangered by both human-caused habitat loss
and encroachment  into  their  range by the  southern  flying  squirrel,  Glaucomys
solans,  which outcompetes them in the deciduous forests that are replacing the
boreal conifers  in  retreat because of  harvesting and climatic  warming.   The
endangered subspecies  are further  threatened by  a nematode  parasite,  which
kills them  but not the  southern flying  squirrel,  its primary  host  (Handley
1979).   Additional climatic warming  may  expand  the range  of  the  southern
flying squirrel  at  the  expense  of the  northern subspecies,  hastening  their
decline.

Soggnunities at Risk

     Although many reserve communities  would suffer from changing climates, we
theorize   that  the  following  types  of   species   and  communities   may,  be
Particularly affected  by  warming  trends over the next 100 years:

     Peripheral Populations.   Populations  located  near the edge  of a species
range that is  contracting  in response  to  climate change would  be  at greater
ri-sk than those at the center or  an  expanding edge (see Figure  2).

     Geographically  Localized Species.   Even if  their  populations were large,
sPecies  whose geographic  range  is small to begin with,  such  as  many reserve
JPecies,  would  be less  likely to have any  populations  in areas of  suitable
"abitat  after  a climate  change  than  those  whose  distribution is more  wide-
spread (Beardmore 1983).   Island species are a special case of geographically
Restricted  species.  If the latitudinal  migration required  of them exceeds the
size  of  the  island,   a  climate  change would  leave  little  alternative  but
e*tinction.   However,  climatic changes on  oceanic islands might  be  relatively
     because  the sea would  moderate  the  air temperature.
     Genetically  Impoverished  Species.    Species  that are  reduced to  small
Populations  or  whose  ranges are  severely  curtailed may  lose  the  genetic
            including  ecotypes  adapted  to  particular  climatic  conditions,
       to successfully respond  to climatic change.   Thus,  projected  climate
       provides yet  another  reason  to  retain  as much  genetic diversity  as
Possible  within a species.

     Specialized  Species.    Such  species  are  generally  less  tolerant  of
 °°logical change because,  by definition, some aspect  of  their  life requires a
       range  of  environmental  conditions, conditions that might not  exist
       the ecological  perturbations  of  a  major climatic change.   Often  the
         of a specialist  is  tied  to  the survival  of one  or a few  other
     e3)  as  in tne Everglades  kite  ( Rostrhamus  sociabilis) , which  depends  on
   . apple snail (Pomacea  caliginosa)  as its single  food source.   The snails
^   themselves localized  in  range,  and a  decrease in their abundance due  to
 *7ing of the  Everglades has  threatened  the kite with  extinction in the United
       (Bent 1961).   Future  saltwater  incursion into the swamps or decreases
            could further  threaten the kite.
                                     149

-------
    _ Poor  Dispersers.    During  past  periods  of  climatic  change,  different
species expanded their ranges at highly  individual rates.   For example, sugar
maple (Acer saccharum), hickories  (Carya  spp.),  oaks (Quercus spp.), and elms
(Ulmus  spp.)  spread northward  rapidly  in  eastern  North  America  during the
postglacial  early  Holocene.   Chestnut  (Castanea  dentata)  spread  much more
slowly, apparently  because  its  self-sterility made  it  difficult to establish
by  seed (Davis  1983).   The  increasingly  disjunct distribution  of suitable
habitat may make it very difficult for species not adapted for colonization to
spread to new areas if the climate changes.

     Annuals.   Another interesting  possibility  is that annual  and perennial
species would differ in  their ability to persist  in reserves when confronted
by  climatic  change.   Complete reproductive  failure  in  a  given  year  by  an
annual  species  within  a reserve  spells local  extinction  unless propagules
either  remain  dormant  until  a  more  favorable  year or  arrive from  sources
outside the reserve.  Because many annual species are efficient dispersers and
colonizers, with  long-lasting propagules,  these  strategies may  succeed.   A
perennial  with  equal  dispersal  abilities, however,  has  an advantage over
annuals because the parent population  can often  survive conditions unsuitable
for the  establishment  of young (e.g.,  Banus  and  Kolehmainen  1976), possibly
for a number of years,  until conditions become favorable for reproduction.

     Whatever the case with annuals, some evidence suggests that species that
depend on annual hosts run a  greater risk of  local extinction than those that
depend on  perennials.   For  example, Ehrlich et  al.  (1980)  found that popula*
tions of the checkerspot  butterfly Euphydryas editha relying  on annual plant
hosts   apparently   suffered   a  higher   rate  of   local   extinction   during
climatically unfavorable years than did those  relying on a perennial host.

     Montane and Alpine Communities.   Because mountain  peaks are smaller than
bases, as  species  shift  upward  in response to warming,  they typically occupy
smaller and smaller areas, have smaller  populations,  and  may thus become more
vulnerable  to  genetic  and environmental  pressures.    And because mountain
populations are relatively isolated from other populations of the same specif
on other mountains,  recruitment  and recolonization would be difficult excep*
for highly mobile  species.    Species originally  situated  near  mountaintop3
might have  no habitat  to move  up  to and  may  be  entirely replaced  by
relatively thermophilous species moving up from below (Figure 5).  Examples
past extinctions attributed to  upward shifting  by  communities include alp
plants once living on mountains in Central and South America, where vegetati0
zones  have shifted  upward by  1000-1500  m  since the  last  glacial
(Flenley 1979;  Heusser 1971).

     An  interesting  analogy  to  alpine species  are  those  species  living
other types of cold refugia that would also  shrink as the climate warmed.
example, the northern  Gulf of California contains  a  fauna  distinct from
of  the  southern  gulf.    Several   endemic isopods  survive  in  the  norfc"1'
apparently because a cold  local  climate protects  them  from the  tropical
predators that occur throughout  the rest of the gulf (Wallerstein and
1982).  If the climate warms, however,  these  fish  may extend their range
the cold refugium.
                                      150

-------
        a
c    Figure  5.   (a)  Initial altitudinal distribution of three species,  A,  B,
2*   (b)   Species distribution after a 500 m shift  in altitude  in  response to a
s c rise in temperature (based on  Hopkin's  bioclimatic law; MacArthur  1972).
     es  A becomes  locally extinct.  Species B shifts upward,  and the  total
     ifc  occuPies decreases.   Species C becomes fragmented and restricted to a
        area,  while species  D  successfully  colonizes  the  lowest altitude
                                     151

-------
      Arctic  Communities.   Because temperatures in arctic  regions  may increase
 more  than  in areas closer to the equator (Hansen et  al.  1986),  arctic species
 possibly  may undergo  greater physiological  and  competitive stress-.   On  the
 other hand,  many  arctic  species  have adapted  to  withstand very  large annual
 fluctuations in  temperature, and  so  a sizable  temperature  change  may  be
 tolerated.

      Coastal Communities.    Many  coastal  species,   like marine  mammals  and
 birds,  depend on  the  rich food  sources supported by coastal upwelling.   The
 coastal communities  they  belong  to  may be disrupted  if, as  has  been  suggested
 (Frye 1983;  Kellogg  1983), upwelling  patterns  are  altered by global  warming.
 That   changes  in upwelling  may  provoke  widespread  disruption   has   been
 demonstrated by recurrent  El  Nino events (e.g., Duffy 1983).

      If  those predicting  sea-level rise are correct, much coastal  habitat,
 like  saltwater  marshes and inlets used  by  nesting  birds, may be  inundated or
 eroded.   With no  development, coastal communities would  shift upland as  the
 sea rose, but human  development  of  land above present high  water  may preclude
 this.   In a  study for  the EPA, Kana, Baca,  and Williams  (1986)  concluded  that
 losses  of  wetlands '  around Charleston,  South Carolina, could be  severe __ 40/f-
 80%— in the  face  of  a  rapidly rising sea level, but  they  will be  even  worse--
 approaching  100#--if  bulkheads  are  built  to  protect the  area  that  is  now
 highland .

     Freshwater lowlands  along the  coast would also  be likely to suffer  from
 the intrusion of salt water.  The  cypress  trees of  the  U.S.  Gulf Coast,  f°r
 example, do  not  tolerate  salt water, yet  they grow only slightly above  sea
 level  (Titus,  Henderson,  and  Teal  1984).   (See Titus,  Volume  1  and Park  et
 al.,  Volume  4 for additional  discussion of the impacts  of sea-level rise  oO
 costal marshes . )

WHAT THIS MEANS FOR MANAGEMENT

     Preventing global warming would  be the most environmentally  conservative
 response.   Granted,  this  would be difficult,  not only because fossil fuel
will  increase as  the  world's population  grows, but also  because effect
action would demand a high degree of international cooperation.  If efforts
prevent global warming fail,  however,  and  if global  temperatures continue
rise,  then ameliorating the negative effects  of climatic  change on biological
resources will require substantially  increased  investment in reserve
and management.
     To make intelligent plans for siting  and  managing reserves, we need
knowledge.   We  must  refine  our ability to  predict  future  conditions  irl
reserves.  We also need to know more about how temperature, precipitation, &l
concentrations, and  interspecific interactions determine  range limits (e.g-f
Picton  t984;  Randall  1982)  and,  most  important,  how  they  can  cause loca* .
extinctions.  Adequately understanding the influences of climate on populatl0"
dynamics may require long-term studies of reserve populations, studies  simil*.
to Ehrlich1 s two decades of research on checkerspot butterflies  (Ehrlich 19^'
Ehrlich et al.  1980).

     In addition to basic research, reserves that suffer from the stresses °
altered  climatic  regimes  will  require  carefully  planned  and increasing^


                                      152

-------
intensive  management  to  minimize  species  loss.    For  example,  modifying
conditions  within  reserves  may  be  necessary  to  preserve  some  species,
depending on  new moisture patterns,  irrigation  or  drainage  may  be  needed.
Because of  changes  in  interspecific interactions,  competitors  and predators
foay need to be controlled  and  invading species weeded out.  The goal would be
to stabilize  existing community  composition  by forestalling  both succession
and  habitat deterioration,  much  as  the  habitat  of  Kirtland's   warbler  is
periodically burned  to maintain pine woods (Leopold  1978).

     If  such  measures  are  unsuccessful,  and  old  reserves  do   not  retain
     sary thermal  or  moisture characteristics,  individuals  of disappearing
species may have to  be transferred to new reserves.  For example, cold-adapted
ecotypes or  subspecies may  have  to  be transplanted  to reserves  nearer the
Poles.   Other species may  have  to be reintroduced  in reserves where they have
Become  temporarily extinct.   An unusually severe drought, for example,  might
cause local  extinctions in areas where a species ordinarily could survive with
Animal management.    Such transplantations and  reintroductions, particularly
involving complexes   of  species,  will  often  be   difficult,  but  applicable
technologies are being developed (Botkin 1977; Lovejoy 1985).

     To the  extent that we can still establish reserves, pertinent  information
about changing climate  and subsequent ecological  response should  be  used in
^ciding how  to  design and  locate them to minimize the effects  of changing
temperature  and  moisture.   In  many areas  of the  Northern  Hemisphere,  for
         where northward shifts  in climatic zones are  likely,  it  makes sense
   locate reserves as near the northern limit  of a  species' range as possible,
   her  than farther south, where  conditions are likely  to become  unsuitable.
Again,  plans to reserve  certain shallow alkali lakes in the Great Plains for
the endangered piping plover,  Charadrius melodus (Chipley 1983),.could perhaps
incorporate   information  on  potential  effects  of  the  future  decreases  in
Precipitation  that may occur  in this area (Kellogg  and Schware 1981).

     It is  often  suggested that reserves might best be placed in areas of high
sPecies endemism,  like  the  presumed  Pleistocene  refugia of South  America,
which are often interpreted as areas where many species successfully survived
an
-------
     Maximizing the  size and number  of reserves would  enhance the long-term
survival of  species.   In large reserves,  species would have a greater chance
of   finding   suitable   microclimates   or  of   shifting   altitudinally  or
latitudinally.   If  we  could  increase the  number  of  reserves so  that each
species and community type were represented  in more than one reserve, we  would
increase the  chance  that if  the  climate  in a  reserve  became unsuitable, the
organisms within it might still survive elsewhere.

     Flexible zoning around reserves could preserve an option to shift reserve
boundaries in the future, as, for example, by trading pasture land for reserve
land.   The  multiuse, multizoned  biosphere reserves now being  set  up in some
countries,   such as   India  (Saharia  1986),  provide  models  of  the sort  of
flexibility needed.

     The unique situation of each reserve will challenge managers and planners
to produce further ideas  for  maintaining  biological diversity,  and their task
will be made  more  difficult by how fast  changes are likely to  occur.    If we
wait until we  can  predict exactly which parts  of the  world will be wetter or
drier, for example,  it will  be  too  late—too late to begin the time-consuming
task  of setting  up  alternative reserves,  too  late  to  begin studying the
effects of  climate on competitive  interactions, too  late to  identify  those
species most vulnerable to climatic change.

     If we are  concerned with setting up  reserves  and maintaining biological
diversity—not just to eke out  another 50 years or so of species survival but
to preserve some remnants of  the  natural  world  for the year 2100 and beyond-*
we  must begin  now  to  incorporate information about  global  warming,  as  i
-------
Banus, M.D., and S.E. Kolehmainen.   1976.   Rooting and growth of red mangrove
     seedlings from  thermally stressed  trees.   In Thermal  Ecology  II,  ed.
     G.W.  Esch  and  R.W.   McFarlane,   pp.   46-53.    Springfield,  Virginia:
     National Technical  Information Service.

Beardmore,   J.A.    1983.    Extinction,  survival,  and  genetic  variation.   In
     Genetics and  conservation,  ed.  C.M.   Schonewald-Cox,  S.M. Chambers,  B.
     MacBryde,  and  W.L.  Thomas,  pp.  125-51.     Menlo  Park,   California:
     Benjamin-Cummings Publishing.

Beauchamp,  R.S.A.,  and P.  Ullyott.   1932.   Competitive  relationships between
     certain species of  fresh-water triclads.   J.  Ecol. 20:200-08.

Bent,  A.C.    1961.   Life histories  of North  American  birds of prey,  Part  I.
     New York:  Dover Publications.

Bernabo,  J,C., and  T. Webb  III.    1977.    Changing  patterns  in the  Holocene
     pollen record  of northeastern North  America: a  mapped  summary.   guat.
     Res. 8:64-96.

Botkin,  D.B.  1977.   Strategies for the reintroduction  of species  into damaged
     ecosystems.   In  Recovery  and  restoration of damaged ecosystems,  eds.  J.
     Cairns,   Jr.,    K.L.    Dickson,    and    E.E.    Herricks,    pp.    241-60.
     Charlottesville:  University  Press of  Virginia.

Butzer,  K.W.  1980.   Adaptation to global environmental  change.   Prof. Geogr.
     32(3):269-78.

         R.   1983.   Preserve design and migratory species.  Nat.  Conservancy
     News.  33(5):25-26.

c°ope, G.R.  1977.   Fossil coleopteran assemblages as  sensitive  indicators  of
     climatic changes during the  Devensian   (Last) cold stage.   Philos.  Trans.
     R.  Soc. Lond. B.  280:313-40.

c°urtenay,  W.R.  Jr.   1978.  The introduction of exotic  organisms.   In  Wildlife
     and America,  ed. H.P.  Brokaw, pp.  237-52.   Council  on Environmental
     Quality.   Washington,  D.C.:   U.S.  Government Printing Office.

     C.B.,  I.N.  Healey,  and P.O.  Moore.  1973.  Biogeography.   Oxford,  United
     Kingdom:   Blackwell  Scientific Publications.

     , M.B.    1983.    Holocene vegetational  history  of the  eastern  United
     States.  In Late-quaternary  environments of  the  United  States.   Volume
     2.    The  Holocene.  ed.   H.E.  Wright,  Jr.,   pp.  166-81.   Minneapolis:
     University of Minnesota Press.

°lamond,  J.M.   1975.   The  island  dilemma:  Lessons of modern biogeographic
     studies for  the design of natural  preserves.   Biol.  Conserv. 7:129-46.

 Ot>f»   E.     1976.     Climatic  changes  of   the   past  and  present.     In
     galeobiogeography;  Benchmark  papers  in  geology  31. ed.  C.A. Ross, pp.
     384-412.   Stroudsburg,  Pennsylvania:Dowden,  Hutchinson, and  Ross.
                                     155

-------
Duffy, D.C.  1983.  Environmental uncertainty and commercial fishing:  Effects
     on Peruvian guano birds.  Biol. Conserv. 26:227-38.

Edgell, M.C.R.   1984.  Trans-hemispheric movements of Holarctic Anatidae:  The
     Eurasian wigeon (Anas penelope L._1  in North America.  J.  Biogeogr. 11:27-
     39.

Ehrlich, P.R.    1965.   The  population  biology  of the  butterfly,  Euphydryas
     editha: The structure of the Jasper Ridge Colony.  Evolution 19:322-36.

Ehrlich, P.R.,  D.D.  Murphy,  M.C.  Singer, C.B. Sherwood,  R.R.  White, and I.L.
     Brown.   1980.   Extinction,  reduction,  stability,  and  increase:   The
     responses   of  checkerspot  butterfly  (Euphydryas)  populations  to  the
     California drought.  Oecologia 46:101-105.

Flenley, J.R.  1979.  The equatorial rain forest.  London:  Butterworths.

Flohn,  H.    1979.   Can  climate  history  repeat  itself?   Possible climatic
     warming and  the case of  paleoclimatic  warm phases.   In  Man's impact _Qg
     climate,  eds.    W.  Bach,  J.  Pankrath,  and  W.W.  Kellogg,   pp.  15-28.
     Amsterdam:  Elsevier Scientific Publishing.

Ford, M.J.   1982.  The changing climate.  London:  George Allen and Unwin.

Frankel, O.H.,  and M.E. Soule.   1981.   Conservation and evolution. Cambridge,
     United Kingdom: Cambridge University Press.

Frye, R.   1983.   Climatic change  and  fisheries management.  Nat. Resources_Jj.
     23:77-96.

Furley, P.A., W.W. Newey, R.P. Kirby,  and J.  McG. Hotson.  1983.  Geography_J>£
     the biosphere.  London:  Butterworths.

Glynn,  P.W.    1984.    Widespread coral  mortality  and  the 1982-83   El  Nino
     warming event.  Environ. Conserv. 11(2):133-46.

Goigel,  M.,  and  S.  Bratton.    1983-    Exotics   in the  parks.    Natl.  ParkJ
     Jan./Feb.  1983:25-29.

Handley, C.O.,  Jr.   1979.  Endangered mammals.   In Endangered and threatened
     plants  and   animals  of  Virginia.  ed.  D.W.   Linzey,   pp.  483-621'
     Blacksburg,   Virginia:     Virginia  Polytechnic  Institute   and  State
     University.

Hansen,  J.,  D.  Johnson,  A.  Lacis,  S.  Lebedeff,  P.  Lee,  D.  Rind,  and  G*
     Russell.     1981.    Climate  impact  of  increasing  atmospheric  carbon
     dioxide.  Science. 213:957-66.

Hansen,  J.,  A.  Lacis,  D.  Rind, G.  Russell,  I. Fung, and  S.  Lebedeff.  1986-
     Evidence for future warming: How large and when.   In C02> climate
     and forest  management  in  the  United States,  eds. W.E.   Shands and
     Hoffman.  Washington, D.C.:  Conservation Foundation (in press).
                                      156

-------
Heusser, C.J.    1974.   Vegetation  and  climate of  the southern  Chilean  lake
     district during and  since the  last  interglaciation.  Quat.  Res.  4:290-
     315.

Hoffman, J.S., D.  Keyes,  and J.G. Titus.   1983.  Projecting  future sea level
     rise.   Washington,  D.C.:  U.S.  Environmental Protection Agency.

Kana, T.W.,  B.J.  Baca,  and  M.L.  Williams.    1986.   Potential impacts  of sea
     level  rise  on wetlands around  Charleston,  South Carolina.   Washington,
     D.C.:   Environmental  Protection Agency.

Keliison, R.C.,  and  R.J.  Weir.  1986.    Selection and breeding  strategies  in
     tree  improvement  programs  for   elevated   atmospheric   carbon  dioxide
     levels.    In CQp,  climate change,  and forest  management in  the  United
     States,  eds. W.E. Shands and  J.S.  Hoffman,  Washington,  D.C.: Conservation
     Foundation (in press).

Kelloggt W.W.   1983.   Impacts of  a C02- induced climate change.   In  Carbon
     dioxide, eds. W.  Bach, A.J. Crane,  A.L.  Berger,  and A. Longhetto,  pp.
     379-41.   Dordrecht, Netherlands:  D.  Reidel.

Kell°gg» W.W.,   and  R.  Schware.    1981.     Climate   change  and  society;
     Consequences  of  increasing   atmospheric   carbon   dioxide.     Boulder,
     Colorado:   Westview Press.

Leopold, A.S.  1978.   Wildlife and  forest practice.   In  Wildlife and America.
     ed.  H.P.  Brokaw,   pp.   108-120.      Washington,   D.C.:     Council   on
     Environmental Quality,  U.S. Government  Printing  Office.

Lovejoy, T.E.  1980.  A  projection of species  extinctions.  In The global  2000
     report  to  the president;   Entering  the  twenty-first  century.   Washington,
     D.C.:   Council on Environmental Quality and  the  Department of State,  U.S.
     Government  Printing Office.

         T.E.   1985.   Rehabilitation of  degraded tropical rainforest  lands.
     Commission  on Ecology  Occ.  Pap. 5.  Gland,  Switzerland:   International
     Union for  the Conservation of Nature  and  Natural  Resources.

         , R.H.   1972.  Geographical ecology.  Mew York:   Harper  & Row.
          , W.V.   1976.   Adaptation  to  climatic zones and  the ecophysiology  of
     ruminants.   In Progress in animal biometeorology.  ed. H.D. Johnson, pp.
     425-33.   Amsterdam:  Swets and  Zeitlinger.

      ,  L.   1983.   Effects of non-COp  greenhouse gases.   In Changing  climate.
     pp. 285-91.   Washington, D.C.:  National Academy Press.

     n,  R.A., and V.  Ramanathan.    1980.   Detecting  climate  change due  to
     increasing carbon dioxide.  Science. 209:763-68.

      ,  S.,  R.T.  Wetherald, and R.J. Stouffer.   1981.   Summer dryness due  to
     an  increase of atmospheric C02  concentration.  Clim.  Change 3:347-86.
                                     157

-------
Miller,  W.F.,   P.M.  Dougherty,  and  G.L.  Switzer.    1986.    Rising  C02 and
     changing  climate:    Major  southern forest  management implications.   In
     C0p»  climate  change, and  forest management  in  the United  States,  eds.
     W.E. Shands and J.S. Hoffman.  Washington, D.C.:  Conservation Foundation
     (in press).

Muller, H.   1979.   Climatic changes  during  the  last  three interglacials.  In
     Man's impact on climate, eds.  W.  Bach,  J. Pankrath, and WW. Kellogg, pp.
     29-41.  Amsterdam:  Elsevier Scientific Publishing.

Myers, N.  1979.  The sinking ark.  New York:  Pergamon Press.

Myers,  R.L.     1983.   Site  susceptibility  to  invasion by  the exotic  tree
     Melaleuca quinquenervia in southern Florida.  J.  Appl. Ecol. 20:645-58.

National  Research  Council  (NRC).    1983.    Changing  climate.    Washington,
     D.C.:  National Academy Press.

Neilson,  R.P.,  and  L.H. Wullstein.    1983.   Biogeography  of  two  southwest
     American oaks in relation to atmospheric dynamics.  J. Biogeogr.  10:275-
     97.

Norse,  E.A.,  and  R.E.   McManus.    1980.   Ecological  and living resources-
     biological diversity.    In  Environmental quality;   The  Eleventh Annual
     Report of  the Council  on Environmental Quality,  pp.  31-80.   Washington,
     D.C.:  U.S. Government Printing Office.

Paine,  R.T.    1966.    Food web  complexity  and  species  diversity.   Am.
     100:65-75.

Perring, F.H.   1965.   The advance  and  retreat of the British flora.   In
     biological significance of climatic changes  in Britain,  eds. C.J. Johnson
     and L.P. Smith,  pp.  51-59.   London:  Academic Press.

Picton, H.D.    1984.   Climate  and the  prediction of reproduction'  of  three
     ungulate species.   J. Appl.  Ecol.  21:869-79.

Ramanathan, V., R.J. Cicerone,  H.B. Singh, and  J.T. Kiehl.   1985.   Trace
     trends  and their  potential  role  in  climate change.   J.  Geophys.___
     90:5547-66.

Rand, A.S. 1964.  Inverse relationship  between temperature and shyness in
     lizard Anolis lineatopus.  Ecology 45:863-64.

Randall, M.G.M.  1982.    The  dynamics of an  insect population  throughout
     altitudinal  distribution:   Coleophora   alticolella  (Lepidoptera)   *
     northern England.   J. Anitn.  Ecol. 51:993-1016.

Saharia, V.B.   1986.   Strategies  and  guidelines for management planning  0*
     sanctuaries  and  national  parks  in   India.    In  The  development	
-------
Schneider, S.H., and  R.  Lender.   1984.  The  coevolubion  of climate and life^
     San Francisco:  Sierra Club Books.

Schonewald-Cox, C.M.,  S.M. Chambers,  B. MacBryde,  and  W.L.  Thomas.    1983.
     Genetics and  conservation.   Menlo  Park, California:   Benjamin-Cummings
     Publishing.

Seddon, B. 1971. Introduction to biogeographv.  New York:   Barnes and Noble.

Soule,  M.E.,   and  B.A.   Wilcox.   1980.   Conservation biology.  Sunderland,
     Massachusetts: Sinauer Associates.

Strain, B.R., and F.A. Bazzaz.   1983.  Terrestrial plant  communities.   In CQ2
     and plants, ed.   E.R.  Lemon,  pp. 177-222.   Boulder,  Colorado:   Westview
     Press.

Terborgh,  J.,   and B.  Winter.     1980.    Some  causes  of  extinction.    In
     Conservation  biology ,  eds.   M.E. Soule  and B.A. Wilcox,  pp.  11^-133.
     Sunderland, Massachusetts:  Sinauer Associates.

Terborgh,  J., and  B.  Winter.   1983.   A  method for siting parks and reserves
     with special  reference to Colombia and Ecuador.   Biol. Conserv. 27:45-58.

Titus,  J.G.,  T.R.  Henderson, and  J.M.  Teal.    1984.   Sea  level rise  and
     wetlands  loss in  the  United   States.   National  Wetlands  Newsletter.
     6(5):3-6.

Tralau, H.   1973.   Some quaternary  plants.   In   Atlas of palaeobiogeography.
     ed. A.  Hallam, pp.  499-503.   Amsterdam:  Elsevier Scientific Publishing.

van Devender, T.R., and W.G. Spaulding.  1979.   Development  of vegetation and
     climate in the southwestern United States.   Science.  204:701-710.
           ,  B.R.,  and R.C.  Brusca.  1982.   Fish predation:   A  preliminary
     study of  its role  in the  zoogeography  and evolution  of shallow  water
     idoteid  isopods (Crustacea:   Isopoda:   Idoteidae.)   J. Biogeogr.  9:135-
     150.
      ,  T.M.L.,  P.D. Jones, and P.M. Kelly.   1980.   Scenario  for  a warm,  high
     C02 world.   Nature.   283:17.

wUcox,   B.A.    1980.    Insular ecology  and  conservation.    In  Conservation
     biology,  eds.  M.E.  Soule and  B.A.  Wilcox,   pp.  95-117.    Sunderland,
     Massachusetts:  Sinauer Associates.

^Ucox,  B.A.,  and D.  Murphy.    1985.  Conservation  strategy:   The  effects  of
     fragmentation on  extinction.   Am. Nat.  125:879-87.

Alison, M.F.    1983.   Plant reproductive  ecology.   New  York:   John Wiley  &
     Sons.

      Meteorological Organization.   1982.   Report  of the JSC/CAS:   A meeting
     of   experts  on detection   of possible  climate change  (Moscow, October
     1982).  Geneva, Switzerland:   Rep.   WCP29.


                                     159

-------
WATER RESOURCES

-------
The Effects of Climate Change on
the Great  Lakes

Stewart J. Cohen
Canadian Climate Center
Downsview, Ontario Canada
INTRODUCTION

     This paper presents a  few tentative answers,  and a  lot  of questions,
regarding CCU-induced  climatic  change and  its  potential  impacts  on  the  bio-
Physical  ana  socioeconomic  environments  of the  Great Lakes  region.    The
research  involves  atmospheric sciences,  hydrology, and  a  wide range  of other
fields,   including  agriculture,   forestry,   wetlands   ecology,   fisheries,
shipping,  energy,  tourism and recreation,  economics,  and political  science.
Some of  these  impacts  directly link to climatic change.  Other impacts result
from changes  in the environment caused by  climaMc change,  and so indirectly
•Unk to  climate.   These  impacts include  spinoff effects on water distribution
systems,  regional employment  patterns, personal  income, and costs of goods and
services.

     The  purpose of this discussion is to present a status report,  a review of
what we  know (or think we know) about future  climate and its  impacts  on  the
Great Lakes, and a listing of areas where more  research is needed.

CLIMATE AND HYDROLOGY
     Climatic fluctuations have a significant effect on Net Basin Supply  (NBS)
    lake levels. In this discussion, it is assumed  that for the entire basin,

                        NBS = P(lake)  - E(lake) + R

Where P(lake)  is lake  precipitation,  E(lake) is  lake evaporation,  and R is
runoff  from  land.   P(lake)  is estimated from shoreline stations.  E(lake) is
®atimated from a mass transfer model, using ship observations 'of lake  tempera-
?!Ure» wind speed, and air  temperature.  R  is obtained using the Thornthwaite
     balance  approach.   Existing  diversions and  groundwater are assumed to
     minor importance at this scale.
                                    163

-------
     Examination  of the  historical  record  shows  that  air  temperature was
relatively high  during the  1930-60 period,  and  has been  cooler  since  then.
Precipitation has been  high  since  1940 (Quinn  1981),  except in 1963 when low
rainfall  led  to a  sharp  reduction  in  streamflow and  lake levels.   We will
return to this later.

     Climate change scenarios, based  on general circulation models (GCM) of a
doubled C02 environment, indicate higher air temperature and precipitation for
the basin.   Preliminary calculations  indicate  a significant  decline  in mean
NBS to  1963-65  levels, because  of projected increases  in  E(lake), for  which
the increase in precipitation would be  insufficient  to match.   These computa-
tions, however,  are highly dependent  on assumptions  about wind speed over the
lake and lake surface temperatures which affect the dew point.

     Figure 1 shows  the Great Lakes study  area.   Calculations were performed
using two scenarios  of C02-induced climatic change obtained  from GCMs:
(Goddard  Institute  for  Space  Studies) and  GFDL (Geophysical  Fluid Dynamic3
Lab). Grid points are shown on the map.

     Calculations of E{lake)  for  Lake Erie  for GISS and  GFDL are  shown in
Figure 2  as  S and L,  respectively.    The  various S and  L,  scenarios  include
changes in wind speeds  and  vapor  pressure  (VP) as follows:  G (normal wind and
VP),  1  (80?  normal  wind), 2 (GFDL wind scenario),  3  (110?  normal VP), 4  (9<#
normal  VP),  5  (GFDL  wind  scenario  and  110?  normal  VP),  and 6  (GFDL wind
scenario and 90?  normal VP). A mass  transfer  model was  used  in which  E(lake)
is  directly  related  to  wind  speed  and  the  magnitude of  the  lake-air VP
gradient  (VPD).   Higher  VP in   the   overlying  air,  i.e.,  higher  relative
humidity, would reduce  VPD,  thereby reducing  E(lake).  The reverse would occur
with lower relative humidity.
     Data  for  calculations  of  present  normals  were  obtained  from  the
archive of ship  data (M),  located at  the  Canadian Climate Centre, DownsvieW*
Ontario.   Two  estimates using  land  stations are also shown  (Al,  A2).   These
require the  use  of  lake/land ratios  to estimate lake data.  The newer esti-
mates (A2) use stability-dependent ratios of wind speed, VP, and mass transfer
coefficients, and also  account  for changes  in ice cover.   The  new values are
closer to those using ship data than the old estimates.
     Results show significant  increases  in  E(lake),  especially for GISS,
if we  assume present  normal wind  speeds  and VPD  (GS in  Figure  2).   If ^
change  the  wind speed and  VPD,  the results  change.   The  GFDL  wind scenari
(decreased wind speeds in fall, increases and decreases in other months) lead
to higher E(lake),  but not as high as  in  the calculation with present norffla
winds.   Lower  VPD  (S3, S5, L3, L5), due to higher  surface UP, would actual1'
reduce E(lake)  to below present normals.  The greatest departure from presen
E(lake)  is  obtained  with  reduced VP and either  present normal winds (S4, L4'
or the GFDL wind scenario (S6, L6).
                                                                            GS
     Similar results are  obtained for the  entire Great Lakes (Figure 3)-   ^
shows  greater  increases  than  GL.  S4,  S6,  L4,  and L6,  the scenarios wh i°
include  lower  VP and  either  present  normal winds or GFDL  winds,  lead to fcn
largest  increases in E(lake),
                                     164

-------
"li-v-t	
 J^cr-V :
iJfVSo/Mie          I   Jj\ i'.:N,plgon.O
jjcj^sso*          \   PI  v  -
; v-f ,4 '    "1	i-V-f-rk"-;	
 -7	JU
                                   	o	
                                                                QUEBEC
        -	
      Vi WISCONSIN n	
    X±rf
        /!
                 /

                     T >.  j ,'  :  f  i        'VA  : i  ;:
                     i  \ iC  La(ie V i   M CHIGAN   i \ ./ ';
                                        	[-r
                                          LaK»  • ft!
                                         S . C/a/rJ/*^
                                        Wndsor/T"!]]
fi	  | \ i   iMt  \ i
V        I    T,\ Michigan'H'
; \	•	J|   I   / 1
;   VQ    i   K  -   7i
               /
            P
         •—;j	
        V
      ILLINOIS |\f5^Aic
             :'  i      > /'
     		1--4~
'Jr
                  INDIANA
               9D'l
L...
"** x*^
i  x**x
                                         	U-U
                                                                            -%7i
                                                                            K
                                                                ^	4-

                                                                0
                                 -i  ^*7  '  *l 1
                              --  TX   • Georgian  	!	„*

                                 jLade^Hufon I'ji
                                 •K--'^-<	i— ""
                                 1 \  \     :  Toro
           r f--\-/*~--,^,

           f&Styt
                iTT
                c6.'
                                Ra?;in ^*— -^r
                            i"NEW'VORK  V   j
                               i   — i    '   rn
                                         -J-
PENNSYLVANIA

	r	V""
                       tt u   ff* t»  H' M  C* H   M* K   «'«  H* K  •> M
                                                                               I
     Figure  1.   Great  Lakes  Study Area for GISS and  GFDL General
                 Circulation  Models
                                      165

-------
                           CHANGE FROM MAST)
70 -

50 -
40 -
3O -
2O -
1O -
0 -
-10 -
-20 -











^
^




Y/

-n
y,
'/,






Y/
'/,




\/<
YA
7~\
Y/
'/,
Y/
'/,
y/
y,









y/
'/,
y,
'/.
y/
'/,
y/
'/,

















/
Y,


&

Y,






Y,
//
Y,
'//
Y,
//
Y,








'/,
'/,
y
V
'/
y,
'/
y,



        A1  A2  M  GS  S1  S2 S3 54  S5  S6 CL L1  L2  L3  U4 L5 L6
                               SCENARIO
                        J! CHANGE FROM MAST
 Figure 2.   Lake Erie Evaporation for Various Scenarios
                         (% CHANGE FROM MAST)
    80
    70
    60
    50
    40
                                                    V,
    30

    20
    10
   -10
-20

-30
               TA

                   V,
                              V,
y,
                                         %
                                                         ,
                                                  /
         M  GS  S1  S2  S3  S4 S5  S6  CL LI  L2  L3 L4  L5  L6

                               SCENARIO
                      1771 % NORMAL (MAST)
Figure 3.   Great Lakes Evaporation for  Various Scenarios
                           166

-------
     The  above  calculations  are  highly  speculative because  of a  number of
factors, including the following:

     •   GCMs have coarse  resolution, and  at  present they  do not  include
         mesoscale lake effects.

     •   The  evaporation "scenarios"  are  based on  crude assumptions  about
         changes in lake temperatures, ice cover, and wind speed.

     The calculation of changes in NBS is also influenced by assumptions about
p(lake).  If we  assume that P(lake) = 0.92 P(land), as indicated by the Inter-
national  Field  Year  for  the Great  Lakes (IFYGL) study  in 1972  (Wilson  and
Pollock 1981), and that precipitation  from  the  GISS  and GFDL models represent
P(land), we get the results  shown  in Table  1.   Actual evapotranspiration from
the land area, snowmelt, runoff, and soil moisture deficit were obtained using
the Thornethwaite  water  budget  model.   Both scenarios  point  to warmer condi-
tions with  lower  snowmelt and  runoff,  and  to higher soil  moisture deficits,
despite  the  increased  precipitation.    Significant  decreases  in  NB$'  are
Projected, but it  is  apparent that  assumptions  about wind  speed (which affect
*ake evaporation estimates)  have  a significant influence on the results.   In
addition,  estimates  of  present and  future water  consumption  are uncertain
vCohen 1986a,b).

     Similar  difficulties  exist  when  assessing  C02-induced  climatic  change
°ver the land portion of the  basin.   The  coarse resolution of GCMs forces the
climatologist to interpolate when estimating changes  in snow cover  or impacts
°n crop productivity,  forest fire  hazard,  and  outdoor  recreation.   A recent
study by Crowe (1985)  used present normals from operating  stations to estimate
snowfall changes for  173  grid points in southern Ontario,  although there are
only  H  grid points   for  GFDL  and  6   for  GISS  in  or  near  the  region.
Interpolation using present  normals forces us  to assume  that  local  synoptic
c°nditions  will  not  significantly change  in  the  future  (e.g.,   timing  and
frequency  of  weather   types).    Will  future  regional  climate  include  fewer
6Pisodes of  frontal  passage  and  increased  convective  activity?   Would  that
      the  spatial  distribution of precipitation within the  region?

     One approach  to  answering  questions  about future regional  climate  is to
     at historical analogues.   For  example, what  were conditions during 1963-
   when lake  levels and NBS were almost as low as those predicted  in  Table 1
   gure 4)?   Over the  land  area,  air temperatures  at  Sault Ste.  Marie  were

-------
     Table 1.  Effects of Climatic Change Scenarios on Annual
               Water Balance of the Great Lakes Basin
 CLIMATE
 CONDITION                    GISS                 GFDL
Temperature               +4.3 to +4.8 C       +3.1 to +3.7 C
  Change

Precipitation                  +6.4?                +0.855

Actual Evapo-                 +18.1?                +6.7?
  transpiration

Snowmelt                      -45.9?               -35.8?

Runoff                        -10.9?                -8.2?

Soil Moisture                +116.4?              +166.2?
  Deficit (summer)

NBS (present normal           -20.8?                -18.4?
  winds)

NBS - consumptive             -28.9?                -26.4?
  use (2035 proj.)
NBS (80? winds) -4.1?
NBS (80? winds) -11.8?
-consumptive use
(2035 proj.)
^4.0?
-11.7?
Source:  Adapted from Cohen (1986b).
                                168

-------
                                    (m3/aec)
«2
El
            8.5



             8



            7.5



             7



            6.5



             6



            S.5
                       !
-------
                                     (millimeters)
                195O
                PRECIPITATION
                           197O
                                      1990
 year
GFDL
                                                 2010
                                                   GISS
                                                            2030
                                                              MEAN
Figure  6.   Annual  Precipitation,  Great Lakes Basin.   Adapted from
            Quinn  (1981  and personal communication) and Cohen (1986a)
                JAN 61
                                  (.1951-1980 NORMAL)
80 -
70 -
60 -
50 -

40 -
30 -
20 -
10 -

0 -
-10 -
-20 -
-30-
-40 -









U






t
••
,
-


I








n


f|

!
B :








ft
Ji













1

•







•
D »,
n_
:T : :: :
'
J . , u
*•




1
•


1
• : xjf
\ i






1 9

1
n
n if
:l
\
j




*
•



*
•




                       JAN 62
                                JAN 63
                                        JAN64
                                                JAN65
                                                        JAN66
                                        DATE
                                  [771 ANOMALY
    Figure 1.   Monthly Precipitation  Anomalies,  Windsor,  1961-1966
                                    170

-------
      E
           go
         -10 -
         -20 -
         -30-
         -40 -
         -50-
         -60
            JAN 61
                                (1951-1980 NORMAL)

'

pi
a
n :
:i:

1
;


*mm
f
:



n


1
JAN62
JAN63
                  JAN64
                                               JAN65
                                                        JAN 66
                                __^   DATE
                                ZZI ANOMALY
   Figure 8,   Monthly Precipitation Anomalies, Thunder Bay, 1961-1966
          100
            JAN 61
                                (1951-1980 NORMAL)
        111111
         JAN63
                                ZZI
                                      JAN64
                  DATE
                ANOMALY
                  JAN65
JAN66
Pigure 9.  Monthly Precipitation  Anomalies, Sault  Ste.  Marie,  1961-1966
                                   171

-------
                                  (1965-1983 MEAN)
  O
1 -







-1 -


-


-2 -
_
-3 -




"7~\
U-gULd











PI
/
V\ /,
/ /
s f7~\ x
^
/
2









/
'*
y
'A
'.
'*
',
r.
f.
/.
/
£
/
/
/
/
/
/
/
/
/
X
^
;,- Ld- LdLd^ -a—
•y
IZJ









JAN65 JUL65 JAN66 JUL66
DATE
\7~7\ ANOMALY
   Figure 10.  Monthly Lake Temperature Anomalies, Lake Superior, 1965-1966
     It appears  from the  above  information that  the 1963-65 episode  of
levels was initiated  by  below-normal precipitation in 1963.   NBS returned t°
normal levels because of cool temperatures in 1965, and perhaps,  high rainfal*
in the central part  of the region.  The  cooler  conditions would have reduced
evaporation losses.  For example, in 1965, Lake Superior evaporation was 3& ^
(1%) below the 1965-84 normal.

Future Research Needs

     A workshop on climate impact assessment in  the Great Lakes, sponsored
the Canadian Climate  Program,  was  held in February  1985  (Timmerman and
1985).  Recommendations for future research included the following:

     •  More direct measurements and better estimates of P(lake), E(lake)f
        cover, and overland runoff

     •  More research on thermal structure of the lakes and of air-lake inter"
        actions

     •  Establishment of a regional climate monitoring system
                                      172

-------
     •  Research on  air  and water quality,  including  pollution  from overland
        runoff, atmospheric  deposition,  and  release  of toxic chemicals from
        the lakes to the atmosphere.

     Although the use of climate change scenarios from GCMs  was endorsed for
climate  impact  research  in the  Great Lakes,  it was  recommended  that such
scenarios be used with caution because of  deficiencies  in  simulating boundary
layer  processes  on  a  regional  scale grid,  which  is smaller  than  the  GCM
grids.   Projections  of  frequencies of  synoptic types (e.g., dry  spells)  are
also  needed.   However,  this is  a rapidly  changing field  of  research and,
hopefully, we  may  be able  to  see regional  scenarios from  GCMs  in  the near
future.

IMPACTS

Status Report

     The Great Lakes basin  is  a  highly  industralized international  basin;  a
       producer  of  hydroelectric  power,  crops,  and  wood products;  and  an
important transportation  corridor. Some of these activities  are  directly tied
to the lakes themselves  (Figure  11).   Others,  such as winter recreation,  are
not  dependent  on  the  lakes  directly,  but  might experience  major  impacts
because  of  changes  in  climate  over  the  land portions of   the  region.   In
Edition, certain activities, such as  agriculture,  may experience increases in
demand for water, which  could  lead to further declines in lake  levels.   Such
Demand could originate  from outside  the  basin, thereby  making  diversion  an
important issue of the future,  Just as it  is  under present conditions of high
•Uke levels.

     A pilot study on  impacts  of  CO^-induced  climatic change on  Ontario  was
Decently  completed.    It  was  coordinated  by  the  Ontario  Region  of  the
Atmospheric   Environment  Service  (AES),  under sponsorship   of  the  Canadian
cUmate Program.  Participants  included academics,  government  scientists, and
Private consultants.   The pilot study, which used  the GISS scenario, examined
Qlimate system  components and resource uses  (Table  2).  Upon completion of the
individual sector  studies,  a workshop  was  held  in November 1985  to  quali-
tatively evaluate interdependencies,   identify  possible mitigation  strategies,
    provide  recommendations  for future research.

     A number  of  assessments  of climate  system  components were  presented
       3).  The decline  in  NBS calculated by  Inland Waters Directorate  (IWD)
   similar (but not  identical)  to results presented  earlier in this  discus-
   n,  although  the  modeling   techniques  were  different  (Shiomi   1985).
Deterioration in water quality is anticipated  because of higher lake  temper-
atures.  Closed marshes may dry  out,  while open marshes would migrate  to  the
Uevi lake levels.  Significant decreases  in  snowfall-and snow cover  season were
[Injected by  Crowe  (1985).    Deterioration  of air  quality  is  anticipated
lonause of increased intrusions of air  masses from the United States  (Wilson
'"5).   Note that most  of  the  above judgements  are qualitative,  with  the
 Xception of  NBS, lake levels, snowfall, and  solar  energy.

     Impacts  on resource uses  are listed in Table 4.   Some  of  these  impacts
^e indirectly  caused by climatic change,  in that the change in NBS and lake
 6veis are the  causal factors.    In some cases,  impacts   were quantitatively
                                     173

-------
                             GREAT LAKES IMPACTS STUDY
                     Direct Effccl on
                    »mlci recreation
                    winter ruud maintenance
                    «inter Miiim darru|!C
                    summer kinrm & llmid damage
                    fureMry. CorcM lire*
                    wildlife
                     Direct Effect on
                  Lake Levels and Flows
                     IAE.  A»"ter haliinccl
                     Direct Effect on
                   Water Withdrawals
                    and Consumption
                        (all HCIUK)
                                            New Lake Levels
                                               and Flows
                                            (Direct and Indirect Effect*)
                    Direct Effccl on
                     • electricity demand
                        (AC. Ill)

                     • in) fit* demand
                        (HI)

                     • crop type and yield
                     Direct Effect on
                   External Demand for
                         Water
   Additional
     Water
  Withdrawals
and Consumption
  Diversion
                       • road accidents
                       • insurance
                       • pipe and sewer
                        maintenance
                        (existing)
                       •, retail
                      • Hydro.
                        Production
                      • Shipping, ports
                        navigation,
                      • Fisheries
                      • Summer
                        recreation
                      • Shoreline
                        properties
                       1 Pipe and Sewer
                        services (new)
                       1 Sewage
                        Treatment
                                                                    services
• Irrigation pipe
  and sprinklers
• Wells
• Import/export
  agr.
• Import/export
  elec.
• Import oil/gas
• Electricity prod.
  from nuclear,
  coal, oil
                      CHANGES
                          IN
                         THE
                      REGIONAL
                      ECONOMY
                          S
                      employment
                         I/O
      Figure  11.    Interconnected  components  of  climate  impacts  and
responses  within  the  Great Lakes  region.     A  =  change,  T  =  temperature,  p  *
precipitation,  E =  evaporation,  AC =  air  conditioning,  HT =  space  heatingi J/
=  inputs/outputs (Cohen 1986a).
                                                 174

-------
Table 2.  List of Individual Sector Studies
TOPIC QUANT/QOAL
CLIMATE SYSTEM
Streamflow
Net Basin Supply
Lake Levels
Water Quality
Wetlands
Snowfall
Air Quality
Solar Energy
RESOURCE USE
Electricity Demand
HT and AC Demand
Hydroelectric Power
Shipping
Agriculture
Tourism & Recreation
Municipal Water Use
Forestry
Health

QUANT
QUAL
QUAL
QUANT
QUAL
QUANT

QUANT
QUANT
QUANT
QUANT
QUANT
QUANT
QUANT
QUAL
QUAL
$ PARTICIPANT*

IWD (with AES)
IWD
Wall et al.
AES
AES
AES

$ U. of Windsor
AES
$ U. of Windsor
$ U. of Windsor
$ U. of Guelph
$ Wall et al.
AES
CFS
workshop
* AES = Atmospheric Environment Service
CPS = Canadian Forestry Service
!WD = inland Waters Directorate
                    175

-------
    Table 3.  Impacts of GISS Scenario of C02-Induced Climatic
              Change on Climate System Components in Ontario
CLIMATE SYSTEM COMPONENT
                IMPACT
Net Basin Supply
    Cohen 1986a)

  -23.8* with consumptive
    use (-28.955 in Cohen
    1986b)

Lake Levels
  Mich-Hur:  -0.59 m/-0.83 m
  Erie:      -0.44 m/-0.68 m
  Ontario:       (N/A)

Ice Cover

Snowfall

Snow Cover

  -6 to 10 weeks (Wall et al.)

Solar Energy

Water Quality

Air Quality
-15.3* overall (-20.8$ in
Superior:  -0.22 m/-0.30 m
reduced to zero (Erie?)

-80 to 140 cm (Crowe)

-4 to 6 weeks (Crowe)



little change

less dilution

reduced cold season
  pollution and acid shock

increased intrusion of
  polluted air masses
  northward

increased wet deposition

increased local and
  regional episodes
                                  176

-------
              Table 4.   Impacts of COp-Induced Climatic Change on the
                        Economy of Ontario (Allsopp and Cohen 1986)
  RESOURCE USE
            IMPACT
     ANNUAL
   GAIN/LOSS
($ MILLION CDN)
Electricity Demand
(U. of Windsor
1986)
     Heating and
Cooling Demand
(Bhartendu and
Cohen 1986)

Hydroelectric Power
(U.  of Windsor
1986)
Great Lakes
Shipping (U.  of
Windsor 1986)
 griculture (U.  of
       1985)
Reduced winter space heating
  demand=6533 to 7840 GWh.
Increased summer air cond.
  demand=162 to 216 GWh.
Net reduction=6371 to 7624 GWh.

Reduced space heating demands
Increased summer cooling demand =
  +7*.

Reductions due to lower lake levels
  and flows, except at St. Mary's
  River.  Losses at De Cew, Sir Adam
  Beck, and Robert H. Saunders power
  plants.  Total loss=2220 to 4165
  GWh, depending on Lake Ontario
  regulation .

Reduced cargo per vessel due to lower
  channel depths.  Losses are high-
  est if coal shipments greater than
  present.  Losses do not include
  delays at canal locks due to poten-
  tially heavier traffic and occur
  despite reduction of ice cover and
  longer shipping season.

Reductions in crop yield due to in-
  creased heat and moisture stress;
  below potential crop yield due to
  technological improvements.  This
  considers positive impacts of ex-
  tended growing season and northern
  expansion, but does not consider new
  crops or irrigation.
  +99 to +118
 (reduced costs
 to provincial
   utility)
  -34 to -65
  • 10 to -27
    ($US)
    -107
                                     177

-------
                                   Table 4.   Cont.
  RESOURCE USE
          IMPACT
     ANNUAL
   GAIN/LOSS
($ MILLION CDNJ
Tourism and
Recreation (Wall et
al.  1985)
Municipal Water Use
(Cohen 1985)
Forestry (Stocks
1985)
Health, Air, and
Water Quality
Greater use of parks and campgrounds
  due to longer summer and shoulder
  seasons.  Indirect gains would
  occur in local retail sectors.
  Eight parks were studied.

Ski areas in southern Ontario could
  be eliminated due to reduction in
  snow season.  Small loss or gain
  could occur in the north, depend-
  ing on success of competing faci-
  lities elsewhere (e.g., Laurentians)
  local retail sectors (e.g., Colling-
  wood).

Effects of changes in lake levels,
  streamflow, and water quality on
  shoreline properties and wetlands.
  Possible negative impacts on closed
  marsh.

Increased demand of 5 litres per
  capital per day during May-
  September, due to warmer summer
  season.  This may also result in
  increased demand for treatment
  services, and accelerate the ex-
  pansion of surface water distri-
  bution systems, and replacement of
  wells in some areas.

During the transition from boreal
  forest to new hardwood forest
  types, there would be increased
  damage due to disease, insects,
  and fire, and reduction in winter
  logging operations.  Possible bene-
  fits could result from C02 enrich-
  ment.

Increase in heat stress.  Decrease in
  cold stress.  Poorer air and water
  quality in summer?
                                                                         -50
                                      178

-------
estimated,  so  these  results  are  presented where  available.    Impacts  are
largely  negative,   particularly  in  agriculture,   winter   recreation,  and
hydroelectric power production.

     However, reductions in energy demand represent a significant saving.  The
lower electricity demand actually  overshadows  the loss in hydroelectric power
production.   More  research  is  needed  to  determine  the economic  effects of
reduced demand  for  heating oil  and  natural gas, as  well  as increased demand
for summer air  conditioning.   The latter may  be underestimated because it is
based on  a weak  correlation  with temperature,  and does not  include extreme
events.   Although per household  energy  consumption should decline, population
growth  should still  lead  to  an  increase   in  total   demand  by  the  mid-21st
century.

     Some of  these results require additional  discussion.   The loss projected
for agriculture represents the effect of  COp-induced  climate change on future
crop yields  with  future  technology.   Technological change  is  expected to
increase yields  by  66£ between  1981  and   the  mid-21st century.    The GISS
scenario would reduce this gain  to 60%.   That  6% loss in anticipated yield is
equivalent to $107 million  (Canadian)  in 1981  dollars.   If precipitation was
higher than  the  GISS scenario, yields  would increase further.   Lower preci-
pitation  could   lead  to  yield  decreases  below  present  production due  to
moisture stress.   Overall,  it  is projected that many areas in southern Ontario
would not  be able  to  support grain  corn,   particularly  soils that  are well
drained  or prone to droughtiness.  Some of  this  stress could be alleviated by
irrigation, so the availability of irrigation water will become more important
in the future than it is at present.

     Water use   by  municipalities  will  also  be  influenced  by  C02-induced
°Umate  change.   Results were based on regression models for 17 cities in the
reSion.   The study only considered changes  in  summer  temperatures and did not
include  possible feedbacks  due to future increases in  demand for piped surface
Water by agriculture, recreation facilities, and other  users  of groundwater.
"^e increased demand is  considered a loss, since  costs to municipalities would
Probably increase,  and  under  present pricing  arrangements,  such  costs  would
n°t be completely met by revenues collected  from  consumers.

     Forest resources would undergo a  transition from boreal  to  mostly  hard-
w°od species.  During the transition, damage to  the existing boreal forest by
disease,  insects, and forest fires would increase.  Reductions  in snow and ice
°°ver may  reduce damage to mature trees,  but  could  increase desiccation of
Jeedlings,  affect life  cycles of  insects and  animals,  introduce  an  earlier
J°rest  fire  season,  and   interfere  or  constrain winter  logging  operations
QePendent  on  snow cover  or  frozen ground.

     In  summary,  a wide range of topics has  been considered,  but a great deal
    work  remains  to be done,  particularly in  areas where no  quantitative
         has  been attempted,  such as air and water  quality  changes, and their
           impacts on resource uses.   We will  return to  this in  the  Future
        Needs section.
                                     179

-------
     During the November 1985 workshop, three types of strategies were consid-
ered  that  could mitigate  the consequences  of  climatic change:   preventive,
compensatory, and substitutional.

     Preventive strategies include removal  of COp from emissions.  Since  this
is  a global  problem,  it  was  felt  that any  action  here would  have little
effect, though it might encourage other jurisdictions to follow suit.  Compen-
satory  strategies  involve altering   resources  to  meet  human  needs  (e.g'»
dredging to avoid draft restrictions for commercial shipping), and a number of
these are  listed  in  Table 5.    Substitutional  strategies focus  on adapting
human needs  to  the "changed"  resource (e.g.,   new  ship design for shallower
draft).  These  are also included in Table  5.   This list  was selected from a
longer  list  of  possibilities.   No quantitative modeling has been  done to
evaluate the effectiveness of these strategies,  and this too will be discussed
in the following section.

Future Research Needs

     As indicated  in the  previous  discussion,  a number  of  important topic3
require further  study.   For  atmospheric  scientists and  hydrologists,  these
include incorporation of regional scale processes (e.g., lake effect snowfall)
into GCMs and projections  of synoptic type  frequencies  (e.g.,  frequencies of
dry spells) in a  "greenhouse  effect" climate.   The present data base could &e
improved if  there were regular  direct measurements over  the lake surface of
air  temperature,  water  temperature,   wind  speed,  relative  humidity,  and
precipitation.   All these data  are  needed  to improve  estimates of projected
lake evaporation, NBS, air quality,  and frequencies of extreme events.

     A wide range  of  quantitative impacts studies are  still needed to provide
a more  complete picture of biophysical  and socioeconomic  effects.   Priority
areas  appear  to  be  health,  air  quality,  water  quality,  forest  growth
(including pests and  disease),  forest  fires, wetlands,  fisheries, land trans-
portation,  water  consumption  during  dry  periods,  and  air  conditioning  de-
mand.  Additional  work  is  needed to  determine  secondary economic effects  tha6
might result  from changes  in agriculture,  outdoor  recreation,  lake shippin8'
space heating demand,  and  hydroelectric power production.  An  analogue study
of  the  1963-65  period may  be  of  benefit  in  identifying ripple/multiplie£
effects of low lake levels on various aspects of the region's economy, such a
real  estate   values,   changes  in water  prices,   insurance,  and  recreati0
activities in wetlands and shoreline areas.
     A number  of mitigation strategies were  considered in  the  previous
tion.   The decision  to  implement  any or  all  of  these  requires  addition*
research on  their  feasibility,  costs, benefits,  and possible  side effect3^
Since  many of  these  have  political  implications,  it  is better  to Perf°L
cost/benefit and other kinds  of analyses as  early  as possible, so that tn
issues can be discussed in public before action is needed.
                                     180

-------
         Table 5.  Strategies for Mitigation of Impacts of
                   C02~Induced Climatic Change in Ontario
    RESOURCE
      USE
   COMPENSATORY
    STRATEGIES
  SUBSTITUTIONAL
   STRATEGIES
Residential
Heating and
Cooling

Great Lakes
Shipping

Hydroelectric
Agriculture

Tourism and
Recreation

Municipal Water
Use

Forest Resources
Health, Air, and
Water Quality
Regulate Lake Levels
Extend Season

Use Available Water
Irrigation
Shorter Rotation
Intensive Management
Genetic Adaptation

Improve Water Treatment
Reduce Contaminants
                              New Technology
Redesign Ships
Modern Technology—
  Retrofit
Conservation
Other Energy Sources

Northern Extension

Diversify Operations
  and Activities

Conservation Tech-
  nology

Reforestation
                                 181

-------
CONCLUSION

     Recent efforts in climate impacts research has  led  to  increased awareness
of climate-environment and climate-society  interactions.  There have been  more
questions  than answers,  and researchers  have been  extremely cautious about
discussing  results because  of the  numerous  uncertainties  about  COp-induced
climate  change,   future  technological   change,   and  the  various  modeling
procedures employed in the  research.   Some have said that  impacts  research  is
premature,  and that we  should wait  until these  uncertainties  are resolved.
However,  it will  take  considerable  effort to  develop  improved  methods and
models.     Impacts  research  conducted   today  will  better  prepare  us   as
researchers  to assess  these  impacts when  climate  modeling  reaches  a  more
advanced stage of development.

ACKNOWLEDGMENTS

     My sincere thanks  to G. Irbe for providing  the data on  lake  evaporation
calculated from and based stations.
REFERENCES

Allsopp,  T.R.,  and  S.J.  Cohen   1986.   C02-induced  climate change  and ifcs
    potential  impact  on  the  province  of  Ontario.    Proceedings,   American
    Meteorological  Society  Conference on  Human Consequences  of 1985 Climat®
    and Climate and Water Management, Asheville, August 4-8.

Bhartendu, S., and  S.J.  Cohen   1986.   Impact of C02-induced climate change on
    residential heating  and cooling  energy  requirements  in  Ontario, Canada-
    Unpublished.

Cohen, S.J.   1985.   Projected  increases in municipal water  use in the Great
    Lakes  due to  COp-induced climatic  change.  Paper presented  at  America11
    Meteorological   Society  Fourth   Conference   on   Applied   Climatology'
    Scottsdale, Arizona.

Cohen, S.J.   1986a.   Climatic change,  population growth, and their effects °n
    Great Lakes water supplies.   The Professional Geographer, in press.

Cohen, S.J.  1986b.  Impacts of C02-induced climatic change on water resource
    in the Great Lakes basin. Climatic Change, in press.

Crowe, R.B.  1985.  Effect of carbon dioxide warming scenarios on total
    snowfall and length  of  winter snow season  in  southern Ontario.   Cana
    Climate Centre Report No. 85-19.

DPA Group Inc., with Concord Scientific  Corp.   1986.    C02-induced cliffl?0..
    change  in  Ontario:  interdependencies  and  potential resource  and
    economic strategies.  Prepared for Atmospheric Environment Service-
    Region.

Quinn, F.H.  1981.  Secular changes in annual and seasonal Great Lakes PreCe$
    itation, 1854-1979, and their implications for Great Lakes water
    studies.  Water Resources Research.  17(6):1619-24.
                                      182

-------
Shiomi, M.
    (IWD).

Stocks, B. J.
    (CFS).
             1985.   Climatic  change  and  its  impact on  Ontario.  Unpublished
               1985.   Climate change in Ontario—forestry impacts. Unpublished
Timmerman,  P., and A. P. Grima.   1985.   Climate impact assessment in the Great
    Lakes basin.  Institute for Environmental Studies, U. of Toronto, Pub.  No.
    EM-7.

University of Guelph.  1985.  Socioeconomic assessment of the implications of
    climatic change  for  food production  in Ontario.   Land  Evaluation Group,
    University School  of  Rural   Planning   and  Development.    Prepared  for
    Atmospheric Environment Service.

University of Windsor.   1986.  Social  economic assessment of the implications
    of  climatic  change  for  commercial  navigation  and  hydroelectric  power
    generation in  the  Great Lakes-St.   Lawrence  River  System.  Great  Lakes
    Institute. Prepared  for Atmospheric Environment Service.
wall,  G.,  R.  Harrison,   V.  Kinnaird,  G.
    Climatic   change  and  its   impact  on
                                            McBoyle,  and  C.
                                            Ontario  tourism
    W» J. .LlLlCt V X O   ^*HCLll£\rf  t*«**»  • ir w  	j.	r— ^  va«  v» j W«AA A»-f  WV/U1 X OI11  Cl
    Prepared  for Atmospheric Environment Service-Ontario Region


Wilson,  E.    1985.   Climate  change   impacts on air  quality:
    assessment.  Unpublished (AES).
                                                              Quinlan.   1985.
                                                              and  recreation.
                                                                 a  qualitative
         J.W.,   and  D.  M.  Pollock.     1981.     Precipitation.     IFYGL-The
    International Field Year  for  the Great  Lakes,  eds. E.J.  Aubert  and T.L.
    Richards.   Ann Arbor:   U.S.  Department of Commerce.
                                     183

-------
 Climatic Evolution and Variability in
 Dryland Regions: Applications of History to
 Future Climatic Change
 Sharon Nicholson
 Florida State University
 Tallahassee, Florida USA
     Climate is an  environmental factor that  has  influenced  the  course of
      activities throughout history.   It  is also a highly variable  character-
 istic of the human  environment.  Dramatic changes, such as the Ice  Ages, have
 occurred on geological  time scales;  less  extreme but nevertheless significant
 ^actuations have  occurred  during  recent centuries.   In the past, climatic
 change resulted from "natural" (i.e.,  geophysical factors) such as volcanic
 Activity,    ocean-atmosphere   feedback,   or  changing   patterns   of   solar
 Radiance.   Now a wealth of  evidence suggests that mankind's impact  on the
 Atmosphere  rivals  natural  effects,  and  such  factors  as- increased  carbon
 Dioxide or  atmospheric  trace  gases  may  produce  changes  of a  magnitude
 Unequaled  during recent history.

     If the  projected  changes occur—notably  a global  warming—it  is  the
 Climatic transition zones, such as the polar margins or the semi-arid regions,
• ?hat will  undergo the greatest  climate changes.  In the high latitude margins,
 -eroperature  ia  the  most critical variable,  but  in  the  semi-arid lands, water
   the limiting factor,  and  the changing water resource issue has the greatest
        in  these  regions.   This paper  summarizes  the  history of rainfall
        in dryland  regions to  provide a basis  for  assessing the magnitude of
        changes that might be expected from human  impact on the atmospheric
 j,    In some parts of  the  world,  knowledge of  past  weather  and  climate  is
 <-adiiy derived from written  records  and from other  methodologies,  such  as
 • ee-ring  analysis.   In  Europe,  for  example,  instrumental records  go  back
      centuries and  detailed information  for early times can be derived  from
    >rical  archives, weather chronicles,  records of river flows and freezing
       wine harvests, and  a  variety of other indicators providing local infor-
 th     both seasonally and annually.    In  many dryland regions,  especially
   86  in developing  countries, information  is  scarce  and few long-term or
                                   185

-------
Table 1.   Types  of  Data  Useful for  Historical  Climatic
             Reconstructions,  Particularly with Respect to Africa
                             I Landscape descriptions

     1  Forests and vegetation: are they as today?
     2  Conditions of lakes and rivers:
       (a) height of the annual flood, month of maximum flow of the river
       (b) villages directly along lakeshores
       (c) size of the lake (e.g.  as indicated on map)
       (d) navigability of rivers
       (e) desiccation of present-day lakes or appearance of lakes no longer existing
       (/) floods
       (g) seasonally of flow; condition in wet and dry seasons
     3  Wells, oases, bogs in presently dry areas
     4  Flow of wadis
     5  Measured  height  of lake surfaces (frequently given in travel journals, but
       optimally some instrumental calibration or standard should accompany this)

                II Drought, famine and other agricultural information

     1  References to famine or drought, preferably accompanied by the following:
       (a) where occurred and  when occurred: as precisely as possible
       (b) who reported it; whether the information is second hand
       (c) severity of the famine or drought; local or widespread?
       (d) cause of the famine
     2 Agricultural prosperity:
       (a) condition of harvest
       (b) what produced this condition
       (c) months of harvests, in both bad years and good years
       (d) what crops grown
     3 Wet cultivation  in regions presently too arid

                           III Climate and meteorology

     1  Measurements of temperature, rainfall, etc.
     2 Weather diaries
     3. Descriptions of climate and the rainy season: when do the  rains occur, what
       winds prevail?
     4 References to occurrence of rain, tornado, storm
     5 Seasonality and frequency of tornadoes, storms
     6 Snowfalls: is this clearly snow or may the reporter be mistakenly reporting
       frost, etc?
     7 Freezing temperatures, frost, hail
     8 Duration of snow cover  on mountains (or absence)
     9 References to dry  or wet years, severe or mild winters, other unusual seasons
                                   186

-------
       500 BC-100 AD
        200-500AD
        LOW  LEVEL
i—i—i—i—i—i—r
                                                         NILE
                                         I—I—I—I
               5m
                                                        LAKE
                                                        CHAD
                   I—I—I—I—I—I—I—l—I—|—\—i
                     1000   1200   1400   1600   1800  2000
                                   YEARS A.D.

     Figure  1.  Fluctuations of Lake Chad and the Nile in Historical
                Times (Maley 1981; Nicholson 1976)
             records  exist.   The principal indicators of weather and  climate
     e  1)  that are useful  in dryland regions include landscape  descriptions
vegetation  and water bodies);  drought,  flood,  and harvest information;  and
        and  weather  descriptions (Figure 1).   Typical  sources are  settlers'
     s,  archives,  travellers'  and  explorers'  journals, geographical  texts,
     written and oral histories and local chronicles.  When carefully  assess-
    these  sources  provide rough estimates of rainfall fluctuations  that  have
        in  recent  centuries.   Although,  with the  exception of  lake-levels or
        W|  thig ig not a Direct assessment of changes  in the water balance or
   Available water  resources, some general inferences can  be made (Nicholson


ftf   figure 2 illustrates long-term fluctuations in the sub-Saharan regions of
le!iiQa  (*••*•• fche  Sahel  and neighboring regions); these are generally paral-
197ftd by fluetuations along the northern margin  of the  Sahara (Nicholson 1976,
4y i    ^79,  1980),   For  southern  Africa,  less  long-term  information  is
flj^lable but as of about 1700, the evidence suggests that fluctuations in the
$ah  d re8i°ns of  southern Africa more  or less  paralleled  those  south of the
   ara  (Nicholson   1981).   Wetter conditions prevailed  during  the  last  few
                                     187

-------
centuries B.C.  and  into  the first century of  the Christian  era.   A  second
humid episode occurred during the  ninth  through  thirteenth centuries.  More
recent fluctuations  were  of  lower magnitude and included wetter conditions in
the sixteenth through eighteenth centuries and again in  the  late  nineteenth
century (c. 1870-95) and long and severe droughts during the  1820s and  1830s
(Figures 3, 4,  and  5).   Long intense  droughts  in  the Sahel also occurred in
the 1680s and c. 1738-56.  Throughout  most of Africa, the present century has
been relatively  dry  compared with the historical past.
          MEDITERRANEAN
          MIDDLE EAST
          INDIA
          SW US/MEXICO
          MEDITERRANEAN
          MIDDLE EAST
          US GREAT PLAINS
          PERU
                                                                    WETTER
                           SUB-SAHARAN AFRICA
                         (NORTH AFRICA SIMILAR, SOUTHERN AFTER 1700)
1
500 BC
1 	
i
0
i
500 AD
H 	 f
i
1000
	 1 	
1500
1 	 1 	

2000
— I 	 1
     SUB-ATLANTIC
SCANDIC    NEOATLANTIC       NEOBOREAL
                      PACIFIC        RECENT
     Figure  2.   Long-term rainfall  fluctuations  in  sub-Saharan Africa (1
-------
     The fluctuations described  for  the  sub-Saharan region appear to be rela-
tively synchronous  with the major  episodes of climate  recognized  first from
pollen sequences  and  later from radiocarbon dating  of indicators of environ-
mental changes (Wendland and Bryson  1974).   These episodes and their approxi-
mate timing  are  indicated  in Table  2.   The humid episode south of the Sahara
that ended  at the beginning of the Christian era  roughly corresponds to the
Sub-Atlantic  period.    The  second  humid  episode  corresponds  to  the  Neo-
Atlantic, a  period  of warmer conditions over much  of the northern hemisphere
during medieval  times.   The most recent extended period of more humid condi-
tions in the  sub-Saharan region  occurred during  the Neo-boreal, a period also
called the  "Little  Ice Age."   It is  known that the  episodes  in Table 2 are
probably hemispheric  and perhaps even global fluctuations,  thus the  general
correspondence  with  sub-Saharan  rainfall   fluctuations  is  not  surprising.
Similar fluctuations probably occurred in other dryland regions, but extensive
historical material has only been collected and synthesized  into  a climatic
chronology for the United States.
           Table 2.  Approximate Dates of Global Climatic Episodes
                     Source:  Wendland and Bryson 1974
                  EPISODE                          DATE
                SUB-ATLANTIC                 500 B.C. - 300 A.D.
                SCANDIC                      300 A.D. - 800 A.D.
                NEO-ATLANTIC                 800 A.D. - 1250 A.D.
                PACIFIC                     1200 A.D. - 1550 A.D.
                NEOBOREAL                   1550 A.D. - 1850 A.D.
                RECENT                      1850 A.D. to present
                                     189

-------
                          LAKES/RIVERS
   RAINFALL
                                                                                     HARVESTS
                   2m
                                                                                   SOUTHERN ALGERIA
                                                                                    iiiffliiiiin,  nil
                                                                                   NIGER BEN)
                                                                                           Hup
                                                                                            ii  in
                                                                                   CENTRAL NAMIBIA
                                                                                       IS!! I
                                                                                   SOUTHERN NAMIBIA
                                                                                     . II JIMIIIII
                                                                                      iTlIT
                                                                                   NORTHERN NAMIBIA
                      1880  1890  1900 1910   »2O
189O  1900 1910   1920
                                                                                     1880   1900  1920
     Figure  3.   Trends  of African Climate Indicators (Rainfall,  Rivers,  Lakes,  Harvests),  1880 to  1920.
Harvest quality, good = above  the axis, poor =  below the axis; rainfall r and river discharge d expressed as
%  of the mean  (r  or d} or % of standard departure; lake levels,  annual mean surface height, in meters (from

-------
          c. 1820-1840
c. 1870-1895
c. 1895 -1920
Figure 4.  African  Rainfall for the Periods 1820 to 1840, 1870 to  1895,  and 1895 to 1920.
           (+ is above  normal,  0 is normal, - is below normal)
           Source:  Various historical materials (Nicholson 1978).

-------
                                TANGANYIKA   I2
  1700   1725   1750    1775   1800   1825   1850   1875   1900   1925    1950
Figure 5.   Variation of  African Lake Levels Since  1700  (Nicholson  1978)
                                   192

-------
     The remainder of this paper will provide a brief sketch of general trends
in  dryland  regions  of  the  United  States,  South  America,  India,  Asia,  and
Australia.  Note, however, that the remaining discussion is usually based on a
few indicators or investigations for each region in contrast to the chronology
for Africa, which was  derived from several years  of research using literally
hundreds of sources.   Thus,  the following descriptions  of historical climate
fluctuations  in  other  dryland regions  should  be viewed  as merely  sketchy
summaries of  a  few  readily available  documents;   the  fluctuations described
     be verified by extensive research and compilation of much more evidence.
     The two  major dryland  regions  of  the United  States include  the Great
Plains and the  deserts  of the Southwest.   Detailed  knowledge  of the last 400
Years (Figure 6) is derived  from  tree-ring  analysis  (Stockton  and Meko 1983);
archaeological studies and a few tree-ring series provide information for much
longer periods  (Lamb  1977).   The settlements and  the pollen record from Mill
Creek, Iowa (Bryson and Baerrais  1965) suggest  that  relatively wet conditions
Prevailed in the Great Plains  from  c.  700 to 1200 A.D. (Table  3).  The Reople
°f the  region  were  farmers  inhabiting  a  land  with  dense  stands  of  trees,
forest animals  such as  deer, and tall grass prairies.   Within a few decades,
drier conditions  set  in  and  lasted  until c.  1400.   The settlements  were
Abandoned,  the prairies and  woodlands  gave  way  to  short-grass  steppe, and the
dominant animal herds were plains animals,  such  as bison.   Tree-rings suggest
Better conditions again prevailed  c.  1400 to 1650,  but  relatively dry condi-
tions characterized  the  period c.  1650 to  1900.    The  present  century  is
comparatively wet  but probably also  warmer than  the few centuries prior  to
it.   Similar  fluctuations  occurred  in the arid  Southwest  and  Great Basin and
Probably in  Mexico  (Lamb 1977,  1982;  Leopold,  Leopold,  and  Wendorf 1961;
"eGhee 1981).

     For most other dry regions of  the world, much less  information is avail-
     for the longer time period.   In  Peru,  Ecuador,  and  the Galapagos Islands
     e 4) some archaeological evidence exists of dry  conditions c. 600 to 1000
   ., wetter ones c.  1000 to  1400, then drier conditions  until  c.  1650.   The
    conditions c. 1500-1650  are also  suggested by  an analysis  of ships' logs,
diaries, and  other  historical  information.    This  material  also  indicates
Better conditions c.  1650 to 1850,  and conditions  similar  to  the present ones
conunencing  c. 1850.   Not  enough material is available to  generalize climatic
fluctuations  beyond  a  century or  two  for  other regions.    In Arabia,  the
Mediterranean, and the Middle  East,  rainfall  fluctuations  probably paralleled
those for sub-Saharan Africa (Figure 2) (Lamb 1977, 1982; Rosenan 1963;  McGhee
'"I).  Only scanty evidence in the form of lake level fluctuations (Figure 7)
and  dunes is available for the U.S.S.R.  and Central  Asia,  as well as one very
General  drought chronology for the  Ukraine.  For  the desert regions of India
and   China,  little information is  available in  western  literature although
pt>esumablyr  given  the long  historical traditions in  both countries and  the
Very   early  interest  in  meteorology  in both,  extensive information  could
pt%obably be found.
                                      193

-------
        ID.
        •10-
                        1780   WOO   ttSO
                          YEAR
IMO   1*50
                                               4-0
                                         YEAR
Figure 6.  Regionalized Annual Temperature and Rainfall Reconstructions  for  Six Regions of the Western
           United States, Based  on Tree  Ring Analysis (dots on right indicate instrumental data).
            Sourcev  ^ritts et al.,  unpublisned manuscript^.

-------
     Table 3.  Long-term Climatic Fluctuations in the Dryland Regions of the
               United States (see Bryson and Baerrais 1968; Lamb 1977, 1982;
               McGhee 1981; Leopold et al. 1963; Fritts et al. 1982;  Stockton
               and Meko 1983).
U.S. GREAT PLAINS
              c. 700 - 1200 A.D.   WETTER
              c. 1200 - 1400
              c. 1400 - 1650

              c. 1650 - 1900
       Settlements at Mill Creek
       Farming
       Trees, tall grass prairie, deer

DRIER
       Mill Creek settlements abandoned
       Decrease in oak and other tree
         pollen
       Increase in bison herds - plains
         animals
       Short-grass steppe

WETTER

DRIER  (except c. 1720-30, c. 1820-40)
U.S. SOUTHWEST/GREAT BASIN/CALIFORNIA VALLEYS
              c. 500 - 600 A.D.    ONSET OF DRIER CONDITIONS

              c. 800 - 1100 A.D.   PROBABLY WETTER
                                          Erosion instead of sedimentation
                                          Dry farming
              c. 1100 - 1300 A.D.  DRYING
                                          Irrigation replaces dry farming
                                          Settlements abandoned
                                          Tree rings
              c. 1620 - 1900 A.D.  DRIER THAN PRESENT
                                      195

-------
Table 4.  Long-term Climatic Fluctuations  in Peru, Ecuador, and the
         Galapagos Islands (see Pejml 1966; McGhee  1981;  Lamb 1977,
         1982).
c.  600 - 1000

c.  1000 - 1400



c.  1400

c.  1500

c.  1500 - 1650

c.  1650 - 1850

c.  1850 to now
DRY (Ecuador coast)

WET (Ecuador coast)

    (Wells in use, agriculture)

ONSET OF DRIER CONDITIONS

LAKE TITICACA FELL SEVERAL METERS

DRY

WET (Ships'  logs, lakes, diaries)

DRY
                  USSR/CENTRAL ASIA
        UKRAINE  DROUGHT
        LAKE SAKI
         (CRIMEA)
       'V
         10 m
             T   CASPIAN  SEA
        I   i  I   i   |  r |   i   |   i  i   i   i   i  i   >   i
      200  400  600  800  1000  1200  1400  1600 1800
                            YEARS A.D.
Figure 7.  Variations of Asian Lakes Compared with a Drought Index for
          the  Ukraine
          Source:  Lamb 1982, 1977; Buchinsky 1963
                             196

-------
     At first glance,  the coincidence  of wetter  episodes  in approximately the
nineth through thirteenth centuries  in Africa, the U.S. Great Plains and arid
southwest,  and Peru suggests some broad synchroneity  in rainfall fluctuations
in  the  earth's dryland regions.   A more detailed look  at  recent centuries1
suggests,  however, that this  is  not the case.   Figure 8 summarizes rainfall
fluctuations and drought from c.  1750 to the present for central  Chile, the
U.S. Great  Plains,  Peru, and India.  Very few generalizations can be drawn for
all four regions.   Some periods  do,  however,  stand  out.  In  the 1820s and
1830s (a period of continental  drought in  Africa),   rainfall  conditions  in
central  Chile  were generally good while in the drier  regions of the U.S. that
Period tended to be both cold and wet (Figures 6  and 8).
CENTRAL CHILE  rainfall

L    IIIBll   III   i I      --UII Ml inlijl in
                                               II  ill  Hill
                                      T
                                           mri
 US. PLAINS  drought
                                  II
                                                       k
      PERU
  heavy rains

      INDIA
    drought
              -
                                JLi
1
Jllj
I
                                                    3 EVENTS
11
                                                                       11
               —I—i—|—i—r
       1750          1800
                                 1	1	1	1	1	1	1	1	1	1	1	1	1	1
                                     1850           1900           1950
     Figure  8.    Rainfall  fluctuations  and  drought  occurrences  in  Central
ChUe,  Peru,  the United States,  Great Plains and India (U.S. drought,  length
of marker indicates  drought  severity; India,  drought  years  indicated;  Peru,
""ackers  indicate  occurrences of heavy  rainfall,  longer markers  indicating
Particularly  strong  rains;  Chilean  rainfall,  length  of  marker  indicates
         magnitude or departures from normal).
     Major  fluctuations that occurred at the end of the nineteenth century can
     be traced to many regions.   In  many  of the  semi-arid  regions  of Africa
         during the period c.  1870-1895 averaged 2Q%-$Q% above  the mean for
    present century.   The situation changed very abruptly in the  1890s and
         decreased rapidly  in  many  regions, a  trend culminating  in  severe
         in the 1910s.  With the possible exception  of the 1950s rainfall in
    -arid Africa was probably never again as copious as in the late nineteenth
                                     197

-------
century.   Kraus  (1955), using  data  primarily from  India,  Australia,  and the
Mediterranean, observed that  this  trend  was  a general  one  throughout the
tropics and it was also evident in parts of the United States (Wahl and Lawson
1970).  Data for a few selected regions are given in Figures 9 through 14.  In
many cases, wetter conditions returned in mid-twentieth century.

     It is difficult to draw conclusions from such sketchy material, but a few
observations can be  made.   First, changes  in  water  availability on a histor-
ical time scale can be documented for many dryland regions and these are often
of  great  enough  magnitude  to  have  been  highly significant  for the regions'
population.   In  Africa,  for example, rainfall  in areas  such as the Sahel was
probably 20-30? above the present mean within recent centuries,  but conditions
even drier  than  the current ones  have  also occurred.   In  semi-arid parts of
the United States, similar  changes  in rainfall and  streamflow in order of 20-
30% have  occurred.   Some changes  have  occurred rather  abruptly,  others as a
gradual trend.   In notable  examples, not only  the  magnitude of rainfall has
fluctuated, but  the variability  in  time and  also   the  seasonal distribution
have been  affected.    The  historical fluctuations  of rainfall  have not been
spatially uniform, even within  the dryland  regions.   Nor is there any consis-
tent pattern of  increased  or decreased  rainfall in  association with apparent
global  temperature changes.    Changes  in  rainfall  might  be  accompanied  by
changes in  temperature,  which  also  influence  the  water balance via evapora-
tion.   This  is  a more significant effect in  higher  latitudes and  in some
regions, notably the  U.S.  Great Plains, reduced rainfall  is often associated
with higher temperatures.  In tropical dryland regions, both historical info*"*
mat ion and model results suggest  that changes  of temperature are lik-ely to be
small.

     It  is  even more  difficult  to  generalize  about  changes  of  rainfall  °r
water resources that might  occur as  a result of human-induced climate change*
such  as  the  projected global warming  due   to  increased  carbon  dioxide-
Although  there  is  fair agreement  among   estimates  of  global  temperature
changes, the model  projections  of changes  in  rainfall  and  soil moisture s°°w
little consistency in results (Figure 15).  Moreover, the trends of climate in
the  historical  past  show  that  calculations  must  be regionally specif*0'
because few trends can be generalized,  and this is beyond the state-of-the-a?
for climate models.  Another approach is  to use past conditions as analogs
a global  warming;  the Altithermal of c.  6000  B.P.  is often used  as such
analog.  At that time rainfall was considerably higher than at present in
low-latitude dryland regions but  relatively dry in  others,  such as the
cultural regions of  the United  States (Figure  16).   The suggested changes
rainfall  do resemble  model projects  for  a doubling  or  quadrupling of ^ 2
(Figures 17 and 18), and are markedly similar to rainfall changes, as can be
be established,  during the warmer Neo-Atlantic period  in  the  nineth thrjujj,
thirteenth centuries.  Nevertheless, in many areas similar changes of rain£frm
also  occurred during   periods  of  globally  reduced  temperatures,  and  *
predictions cannot be made.
                                      198

-------
           EAST RAJASTHAN
     1875
1900
1925
1950
Figure 9.  Fluctuations  of Rainfall (%  Departure from Normal) in Arid
          Regions of India, 1875-1955
          Source:  Rao  and Jagannathan 1963
                               199

-------
                 B90  1900  19(0  1920  I93O  I94O  I9SO  I960  I97O  I98O
                 I89O  I9O  I92O  I9X>  I94O I960  I9«O  I97O I9BO  I97O
                                                          ORMAL
                                                          MCMMAL
                   IflSO 1900  190  l»20  1930  »40 t»«  1900
                                                           I960
Figure 10.   Five-year Moving  Averages of Annual Rainfall of  Arid
              Zone of  India
              Source:   Krishnan 1977
                                      200

-------
                                          I40'E
                                                     ISO-E
             40- -
Figure  11.  Changes  in Mean Annual Rainfall  (mm)  Between the Periods
            1881-1910 and 1911-19^0 (Positive  Values Denote Increase),
            Source:   Gentilli 1971
                20-S -
                                     iOUILPIC
                                     \  MITCHELC
                                      V
                                       iCUNNAMULL*
Figure  12.   Shift in the Climatic Belts Over  60  Years.   (Position of
             Boundaries in 1881-1910 Shown by  Solid Line,  1911-40 by
             Dashed Line)     Source:  Gentilli  1971
                                  201

-------
                           DECADES
 troo im wo* me 100 UJQ IMP ma mo IOTP IMP
                       cm
lite «w it 10 i«w I*M i«« i»so

ROI
f


• •*


NICOSIA
BEIRUT
to
O
10 TEL AVIV

.— -



JERUSALEM
ALEXANDRIA
CAIRO

•— •



y
\



^



'






s



s



7

^


\





^


s*

SCALE S TIMES L
^•E





x^




S

**^

s
K



^



--•




\
\



/
ioa
w DECADES
toio mo ino IM* uw IMO irro toot 10*0 i*» 1*10 i*x I*M IMO if


taMI

5s-




\

/
_

4-

^

•^x

^
/

^^

~""""

s,

1 	 L 	 . 	 • 	 1__. M
xT^T^s-^
--J *
t — i i r — i — i N
^^


\





\











S

••••

•^^^

^^. 1 -^
/
|/
•B^M

— *

^^

I0t "

1 	 ' ' 1 	 1— 1 	 1 	 3 to
^mj



S


^


s










^s


^


\

^i i^
/
s
^

•^i


s
V


I^HOM

/
f

s,
V

^-«


i^mm

•-»

Y,




X

^^



ARQER f"*!^ 	 js^ I/T"*

, 	 . X*»_. 	 . 	 .



















•^•M
.
/






H
70
M
M
10
M
M
90
40
M
10
3
M 10








1 I 1 1 1












M K1._L
^










/
r^

^k

_

f
/

^

\

/
















^


\

/
/

— <-l

— >_

\














^


\

^^m

•^
S*

^

S


















^

s

'

"^

«^*


V
\

^
10 1130 mo 1040 lisa iun tin ma ma noo 1*10 i*» ine luo »
TEMPERATURE f
H
l«.0
isj ROME
110
n.s
>•• ATHENS
I7.S
1*0 NICOSIA
IM
JU BEIRUT
11.0
17.5
17.0 JERUSALEM
14.5
11.0
»J ALEXANDRIA
M.0
31.5
TI.O CAIRO
30.5
3*.5
3».o KHARTOUM
Jt.l
la
                                                                                                        DECADES
KHARTOUM

-------
Figure 14.  Precipitation (^) and temperature (°F) deviations of the
            period 1850 to 1870 from the 1931-60 climatic normals:
            A. Winter; B. Spring; C. Summer,  D.  Early Fall
            Source:  Wahl and Lawson 1970
                                203

-------
 .
 SON
—I	1	1	1	1	'    '    '    ' -04
 70   50   30  ION  10530  50  70  90  9i
                                                         ION  IOS  30   50   70  90S
                      I.


                      10


                      08
                   <  °2


                      0


                     -02


                     •0.4
                            \j
'-v
                      90N   70  50  30  ION  IQS  30  50	7090

                                      Latitude
Figure  15.   Change in  Zonal Mean Precipitation  Rate Simulated by Six GCMs
             for Doubled  C02 (See Schlesinger and  Mitchell  1985 for Detail3'
                                       204

-------
•ON
40N
403-
•03
                    	•	•' '"*'"" '""I	••••
      iJ Welter than now

        Drltr than now
        Butzer, 1980
•OS
Figure  16.   Two Paleocliraatic Reconstructions  of the Altithermal (c.  1500
             to 800 years ago)
             Source:   Kellogg and Schware 1981.
                                   205

-------
             90*
              80°  -
             10°
                                LONGITUDE
Figure 17.   Simulation of Change in Soil Moisture (cm) for Doubled
            Source:  Manabe and Wetherald 1975
                               206

-------
ho
O
               30TN -
               30TS -
            Figure 18 .  Changes  in Model Soil Moisture for June to August for Quadrupled C02

                        Source:   Mitchell 1984

-------
     The  changes  of rainfall and water resources  could  be gradual or abrupt;
 they may  involve changes of mean conditions, variability about the mean  (i.e.,
 the reliability of water resources from year to year) or seasonality.  Even  if
 only  a change of mean  conditions results,  this will  drastically  alter the
 frequency  of what  is now  perceived as  "extreme"  events  (e.g.,  droughts  or
 floods) which have  severe  impact  on  populations (Mearns,   Katz, and Schneider
 1984).   Perhaps  the best lesson of  climatic  history is  that agricultural and
 economic  systems must be flexible  enough  to adapt to changing conditions and,
 in the face of potential water scarcity, systems must be designed that require
 minimum use of resources.
REFERENCES

Bryson,  R.A.,  and  D.A.  Baerrais  1965.   Climatic  change  and the  Mill Creek
     culture of Iowa.  Journal  of the Iowa Archaeological Society.   15-16:  1-
     358.

Buchinsky,  I.E.  1963-    Climatic  fluctuations  in   the  arid  zone of  the
     Ukraine.  91-94 Symposium on changes of climate with special reference _to
     arid zones.  Rome:  UNESCO.

Gentilli, J. (ed.).   1971.   Climates of Australia and New Zealand.  The World
     Survey of Climatology. JJ3.  Amsterdam:  Elsevier.

Kellogg, W.W., and  R.  Schware.    1981.   Climate  Change and Society.  Boulder,
     Colorado: Westview Press.

Krishnan, A.   1977.   Climatic changes relating to desertification in the arid
     zone of Northwest India.  Annals of Arid Zone Research.  16 (3):302-309-

Lamb, H.H.   1982.   Climate.  History  and the Modern World.   New York: Methuen
     and Company 376 pp.

Lamb, H.H.    1977.    Climate  Present.  Past  and  Future.  Vol. 2.    New York:
     Methuen and Company, 802 pp.

Leopold, L.B., E.B. Leopold,  and  F.  Wendorf.   1961.  Some climatic indicators
     in  the  period A.D.  1200-1400 in  New Mexico.   Symposium on  changes_o£
     climate with special reference to arid zones.  Rome:  UNESCO, 265-268.

Maley,  J.,  1981.     Etudes  palynologiques   dans le  bassin  du  Tchad  et
     paleoclimatologie de  1'Afrique  nord-tropicale de  30000 ans  a
     actuelle.   Travaux et documents de L'O.R.S.T.O.M. Paris.

Manube,   S.,  and  R.T.  Wetherald.    1975.   The  Effects of  Doubling  the  C°2
     Concentration on the Climate of  a  General Circulation Model.   Journ-
     the Atmospheric Sciences.  32:3-15.

McGhee,  R. 1981.  Archaeological  evidence  for  climatic change during the
     5000 years.    Climate  and History,  eds.  T.  Wigley,  M.  Ingram,  and
     Farmer,  162-179.   Cambridge:  Cambridge University Press.
                                      208

-------
Mearns, L.O.,  R.W.  Katz,  and S.H.  Schneider,  1984.  Extreme high-temperature
     events:     Changes   in  their   probabilities  with   changes   in  mean
     temperature.   Journal of Climate and Applied Meteorology.  23:1601-1613.

Mitchell,  J.F.B.,  1984.    The  Effect  of  Global  Pollutants  on  Climate.
     Meteorological Magazine.  113:1-15.

Mooley,  D.A.,  and  G.B.  Pant.   1981.   Droughts in  India over  the  last 200
     years,  their  socio-economic   impacts   and  remedial  measures for  them.
     Climate and History,  eds.  T.  Wigley, M.  Ingram,  and G. Farmer,  465-478.
     Cambridge:  Cambridge University Press.

Nicholson,  S.E.   1976.    A  Climatic  Chronology for  Africa:    Synthesis  of
     Geological,    Historical,   and   Meteorological  Information  and   Data.
     Unpublished Ph.D. dissertation, University of Wisconsin, Madison, 324 pp.

Nicholson, S.E.,  1978.    Climatic  variations  in the Sahel  and  other African
     regions during the  past five  centuries.   Journal  of Arid  Environments.
     1:3-24.

Nicholson, S.E.  1979-  The  methodology of  historical  climate reconstruction
     and its application to Africa.  Journal of African History.   20:31-49.

Nicholson, S.E. 1980.   Saharan climates in historic times.  The Sahara and the
     Nile, eds. M.A.J. Williams and H. Faure.  173-200.  Rotterdam:  Balkema.

Nicholson, S.E.  1981.   The historical  climatology of  Africa.    Climate and
     History, eds.  T. Wigley,  M.  Ingram, and G, Farmer,  249-270.  Cambridge:
     Cambridge University Press.

pejml,   Karel.    1966.   "0  Kolisani  Klimatu  v  Historicke.  Dobe no  Zapadnim
     Pobrezi    Jizni  Ameriky."      On   the  western   coast   of   America,
     Hydrometeorologicky Ustav,  Prague (in Czech),  82 pp.

     K.N., and P.  Jagennathan.   1963-  Climatic Changes  in  India.  Symposium
     on  changes  of Climate  with   special  reference  to  arid  zones.  49-63.
     Rome:  UNESCO.

         N.,  1963.   Climatic fluctuations in the  Middle East  during the period
     of instrumental  record, 67-72.   Symposium on changes  of   climate  with
     special  reference to arid zones.   Rome:   UNESCO.

Sohlesinger,  M.E,   and  J.F.B. Mitchell.    1985.   Model  projections of  the
     equilibrium  climatic response  to increased carbon dioxide.   The Potential
     Climatic Effects of Increasing Carbon  Dioxide,  eds. M.C. McCracken  and
     F.M.  Luther,   81-148.   Washington,  D.C.:  Department of  Energy.

Stockton,  Charles W.,  and  D.M. Meko.   1983.  Drought Recurrence  in the Great
     Plains as Reconstructed from  Long  Term Tree-Ring  Records.   Journal  of
     Climate  and  Applied  Meteorology.   22:17-29.

Street-Perrott,   F.A.,  and   S.   Harrison   1985:     Lake-Level   Fluctuation.
     Paleoclimate  Analysis and Modeling, ed. A.D. Hecht.   Mew  York:   J. Wiley
     &  Sons.
                                      209

-------
Taulis, Enrique, M. 1934.  De la Distribution des Pluies au Chili.  "Materiaux
     pour  1'  Etude  des  Calamites,"  No.   33,   3-20,   Geneva  (Societe  de
     Geographic).
Wahl, E.W., and T.L. Lawson.   1970:  The climate of the mid-nineteenth century
     United States  compared  to the current  normals.   Monthly  Weather Review
     98:259-265.

Wendland,  W.M.,  and R.A.  Bryson.    1974.   Dating Climatic  Episodes of the
     Holocene.   Quaternary Research.  4:9-24.
                                      210

-------
 Response of Lake Levels to Climatic
 Change—Past, Present,  and Future

 F. A. Street-Perrott
 Tropical Palaeoenvironments Research Group
 School of Geography
 Oxford, UK

 M.A.J. Guzkowska, I.M. Mason, and C.G. Rapley
 Milliard Space Science Laboratory
 University College
 London, UK
     Historical  and  geological  evidence  clearly  demonstrate that  the  water
levels of lakes  in Jtorth  America and elsewhere  have  varied significantly,  on
time scales  of 1-1CT years,  in response to climatic changes (Street and Grove
1979;  Street-Perrott and  Roberts  1983;  Street-Perrott  and Harrison  1985;
Harrison and Metcalfe 1985).  The largest fluctuations have, been experienced
by closed (terminal) lakes.   A closed,  sealed lake (one that has no signifi-
cant  surface or  subsurface   outflow)  will attain  equilibrium  with  the
prevailing climate when:

                     Z = AL/AB =  P, and E, > ER
(Street-Perrott and  Harrison 1985).                                        B

     Hence,  a lake's equilibrium  area, depth,  and  volume  will increase  in
Response to  an  increase  in  precipitation,  a  decrease  in  evaporation,  or
both.   Because  they integrate  the  climate over their  basins,  lakes are
excellent indicators  of regional hydrological changes.

     However, it  is  important to know how rapidly an individual lake reaches a
new equilibrium after a change in climate.   For a step change  in  climate,  it
is possible  to define a characteristic equilibrium response time,   T   For a
closed,  sealed lake,  this is given by:                              eg

                      T   = A./((dA./dL)  (E. -  P.)),
                       eq    L    L       L    L

where  req   is the  time  taken  to  achieve  1  -  1/e (63% of the  equilibrium
response) and L is lake depth  (Mason et al.  1985).  Calculated response  times
    some important lakes are given in  Table  1.   These show that many closed
      can be expected to exhibit a fast equilibrium response to hydrological
                                   211

-------
Table 1.   The Equilibrium Response Times of Selected  Lakes
LAKE

Pyramid (USA)
Mono (USA)
Great Salt Lake
(USA)





George
(Australia)




Eyre
(Australia)
Valencia
(South America)
Chad
(Africa)

Abiyata
(Ethiopia)
Palaeolake
Abiyata
(-Ziway ShaJa)
Winnipeg (Canada)
Data from: Harding
TIME

1840-1920 AD
1900-1930 AD
1850-1874 AD
1875-1899 AD
1900-1924 AD
1925-1949 AD
1950-1974 AD
1975- 1985 AD
1958- 1963 AD
1963-1968 AD
1968-1973 AD
1973-1978 AD
1978-1983 AD
1974-1975 AD
1977-1981 AD
1967-1971 AD
1972-1976 AD
1969-1974 AD
9kyrBP
2.5 kyr BP

S.D., 1962, 1965;
AL
(106 m2)
540
227
6279
6196
5093
4978
4698
5908
153
153
135
148
146
9790
350
22500
20500
212
2690
1590

Bye J.A.T.
dAt/dL
(106 m2/!
5.3
1.5
579.4
566.0
706.8
709.0
745.3
671.2
4.7
7.7
42.3
42.1
48.7
800.0
8.75
1712.0
1563.0
14.7
22.8
18.0

fiLaL 1978;
EL-PL
m) (m/yr)
1.03
0.82
1.23
1.29
1.3
1.34
1.3
1.09
0.19
0.55
0.89
0.67
1.26
1.47
1.06
1.93
1.98
1.41
1.02
1.18

Tetzlaff G. & B;
^
(yr)
99
187
9
8
6
5
5
8
171
36
4
5
2
8
38
7
7
10
116
75
2
yt J.A.T., 1978; *
F.A., PhD. thesis, 1979; Vuillaume G., 1981; Lewis, W.M., 1983; Lund L.V. et al.. 1984;
Bureau of Meteorology, 1985; Bureau of Mineral Resources, Geology & Geophysics, A.C.T:, 1985:
P.A.  & Diaz H.F., 1985; South Australian Engineering and Water Supply Department, 1985;
Waters Directorate Canada, 1985; Stauffer NE., 1985; Mason et al.. in press;
                                  212

-------
changes resulting either from  increasing  levels  of atmospheric trace gases or
from other causes operating  on a similar  time scale.   For open lakes such as
Lake Winnipeg (Table 1),  Teq  is given by
                                teq  =  AL.dL/dD,

where D is the discharge through the outlet.  In such cases,  T    tends to be
very short.                                                    eq

     This  rapidity  of  response suggests  that  monitoring lake  levels  and/or
lake  areas could  be a  highly  cost-effective  way  of keeping  track  of  the
Regional hydrological  impact  of increasing  levels  of greenhouse gases.   The
number of  lakes  that can be  measured in the field  is  restricted  by accessi-
bility.  It would be possible,  however,  to measure  the water levels and areas
°f lakes remotely from  satellites  using a  combination  of radar altimetry and
an imaging instrument.   These observations would supplement  and be validated
by ground-based measurements,  but would  add greatly  to  the coverage of '"remote
°P inhospitable areas.   It would also  be possible  to  carry  out frequent and
regular  sampling,  whose  quality  could   be   controlled  for  accuracy  and
°onsistency.

     The radar altimeter on Seasat  was capable of measuring lake levels with a
Precision of + 10 cm RMS.   ERS-1,  due to be launched  in  1989,  will carry two
sensors  that   could  provide  the  necessary  measurements:    an  along-track
scanning radiometer (ATSR)  and a radar altimeter  (RA).   With a 35-day repeat-
track period,  the RA should be  able to monitor  approximately 100 closed lakes
°f more than  100  km2  and many  more  smaller lakes (Guzkowska et al.   In press).

     Historical data on  lake-level  fluctuations in  the United  States suggest
that changes  of quite large amplitude can be expected  to  occur in response to
future climatic change.  Between 1963 and  1986,  for example,  Great Salt Lake
lncreased in  depth by  5.9  m.   It expanded in area from 2590  km2 to over 6000
***   (Arnow  1984  and the New  York  Times.  April  28,  1986),  causing  enormous
damage to lakeside property, industry, and  communications.   Over the last 150
^ars, some lakes have fluctuated  in depth by  nearly  30 m (see Figure  2 in
Street-Perrott and  Harrison  1985).    Other  things being equal,  since  the
6
-------
 orbital  configuration  during  the  early  and mid-Holocene  gave  rise  to an
 increase in annual net  radiation  of the same order as that expected to  result
 from a doubling of C02,  although  the seasonal distribution of the forcing was
 different.   One  important  way of  validating  the climatic  predictions made
 using  general  circulation   models   (GCMs)   is  to  use  the   same  models  for
 paleoclimatic simulations,  which  can then be  tested  against  geological data.
 The early and mid-Holocene are particularly interesting in this respect.

      A high-priority  task for the  future is  to develop  topographically  and
 hydrologically realistic  models  of  individual  lake  basins  that can  use  the
 output of  GCMs or other climate models as  input  (see  Cohen,  this  volume).
 This  approach will  permit   more  rigorous  comparisons between  paleoclimatic
 simulations and geological  data.    It  will also enable predictions  of future
 climatic change to be translated  into quantitative measures  of surface-water
 availability,  provided that a rigorous method can  be  devised  to cope with the
 mismatch in scale  between the two  types of model.

 REFERENCES

 Arnow,  T.  1984.   Water-level  and water-quality changes  in  Great Salt  Lake,
      Utah,  1847-1983.   U.S.  Geological  Survey Circular.  913.

 Australian  Bureau  of  Meteorology.    1985.   Reports  of  monthly  and
      rainfall  and  evaporation for  Lake  Eyre.

 Bureau of Mineral  Resources, Geology & Geophysics,  A.C.T.  1985.  Water levglj.
      area,  precipitation and evaporation  data for Lake George.  N.S.W.

 Bye  J.A.T.  et al.  1978.   Bathymetry of  Lake Eyre.   Trans.  R. Soc. S-Aus^
      102; 85-88.                                        ----

 Guzkowska,  M.A.  J.,   et  al.    In  press.   The  prospects  for hydrologies-1
     measurements  using  ERS-1.    In  Proceedings  of   the  Conference  on_j5S
     Parameterization  of Land-Surface Characteristics.  Rome, December  1985.

 Harding, S.T.  1962.  Evaporation from Pyramid and Winnemucca Lakes, Nevada.
     J. of  Irrig.  and Drainage Div. . Proc.  Am.  Soc.  Civil Engineers.  88,  "°'
     j£-y-                                                     -


 Harding, S.T.  1965.  Recent variation in  the water supply of the western  Great
     Basin Archives Ser.  Rept. No.  16,  Water  Res.  Center Archives,  Universi^
     of California.

Harrison, S.P., and S. E.  Metcalfe.   1985.  Spatial variations in lake level3
     since  the last glacial  maximum in  the  Americas north of  the equatx""'
     Zelthscrift fur Gletsoherkunde und GlazialpeolOKie 21:1-15.
Inland Waters  Directorate  Canada  1985a.   Historical  Water  Levels
     Manitoba,  to  1983.  Inland  Waters Directorate,  Water  Resources  Bran°n'
     Water Survey of Canada, Ottawa, Canada.

Inland Waters  Directorate  Canada.    1985b.   Historical Streamflow
     Manitoba,  to  1984.   Inland  Waters Directorate, Water  Resources
     Water Survey of Canada, Ottawa, Canada.


                                      214

-------
 Kay,  P.A.,  and  H.F.  Diaz  (eds.).    1985.   Problems  of and  prospects  for
      predicting  Great  Salt Lake Levels.  Appendix  1.  Papers  from  a  Conference
      held   in  Salt  Lake  City,  March  1985,  Center for  Public  Affairs  and
      Administration, University  of Utah.

 Kutzbach,  J.E. and  P.J.  Guetter.  In press.   The  influence of changing orbital
      parameters  and surface boundary conditions on climate simulations for  the
      past  18,000 years.  Journal of the Atmospheric Sciences.

 Lewis,  W.M.  1983.   Water budget  of Lake  Valencia, Venezuela.   Acta -Cient.
      Venezolana.  34:248-51.

 Lund,  L.V.,  et  al.   1984.    Background  Report   on Mono  Basin  Geology  and
      Hydrology*,  Los Angeles Department of Water  and Power, Aqueduct Division,
      Hydrology Section.

 Mason,  I.,  et al.  (in prep.) The remote  sensing  of lake levels and areas  for
      climate research.

 Mason,  I.M.,   C.G.  Rapley,  F.   A.  Street-Perrott,  and M.  A.  J. Guzkowska.
      1985.  ERS-1 observations of lakes  for climate research.   In Proceedings
     of  the EARSeL/ESA  Symposium  on European Remote  Sensing Opportunities.
     Strasbourg. 31 March - 3 April 1985 (ESA SP-233, May  1985).

 South Australian  Engineering  and Water  Supply Department.   1985.   Lake  level
     data for Lake Eyre.

 Stauffer, N.E. 1985.   Great  Salt Lake Water  balance model,  168-178, In: Pro-
     blems of and prospects  for predicting Great  Salt  Lake  Levels,  eds. P.A.
     Kay and  H.F.  Diaz.   Papers from a  conference held  in  Salt Lake  City,
     March  1985,  Center  for  Public Affairs and Administration,  University of
     Utah.

 Street,   F.A.   1979.     Late  Quaternary  Lakes   in the  Zlway-Shals  Basin.
     Ethiopia.  Ph.D.  thesis  University of Cambridge, Cambridge,  U. K.

 Street,  F.A. and A.T.  Grove.    1979.   Global maps  of lake-level fluctuations
     since 30,000 yr BP.   Quaternary  Research 12:83-118.

 Street-Perrott,   F.A.  In  press.    The response   of lake  levels  to  climatic
     change:  Implications  for the  future.   In Proceedings  of  NASA  Workshop on
     Climate-Vegetation Interactions, Goddard Space Flight  Center,  January 27-
     29,  1986.

Street-Perrott, F.A., and S.P. Harrison.  1985. Lake levels and climate recon-
     struction.   In Paleoclimate analysis and modeling,  ed. A. D.  Hecht, 291-
     340.   New York: Wiley.

Street-Perrott, F.A.,  and N.  Roberts.   1983.   Fluctuations in closed lakes as
     an  indicator of past atmospheric circulation patterns.   In  Variations  in
     the  Global Water Budget, eds. F.A.  Street-Perrott, M.A.  Beran, and R.A.
     S.  Ratcliffe, 331-345.  Dordrecht:  D.  Reidel.
                                     215

-------
Tetzlafkfk, G.,  and J.A.T.  Bye,  1978.   Water balance  of Lake Eyre  for the
     flooded period January  1974-June  1976,  Trans. R. Soc.  S.  Aust..  102:91-
     96.

Vuillaume G.  1981.   Bilan  hydrologique mensuel  et modelisation  sommaire  du
     regime hydrologique du Lac Tchad.   Can.  O.R.S.T.O.M.,  ser.  Hydrol.,  vol.
     XVIII, 23-72.
                                     216

-------
 Regional Water Resources and Global Climatic Change
 Peter H.GIeick
 Energy and Resources Group
 University of California, Berkeley
 Berkeley, California USA
 ABSTRACT

     Concern over  changes  in global  climate caused by  growing  atmospheric
 concentrations  of carbon dioxide and other trace  gases has increased in recent
 Years as our understanding  of atmospheric dynamics and global climate systems
 has  improved.   Yet,  despite a growing  understanding of  climatic processes,
 "fcny  of the  effects  of  human-induced  climatic  changes  are  still  poorly
 understood.  The  most  profound effect  of such climatic changes may  be  major
 alterations   in   regional hydrologic  cycles  and changes  in regional  water
 availability.   Unfortunately, these are among  the least understood impacts.

     This paper discusses the applicability of modified water-balance methods
 for evaluating  regional hydrologic  impacts  of global climatic changes.   Such
 Methods  offer  considerable  advantages  over  other  hydrologic  methods for
 identifying  the  sensitivity of  regional  watersheds  to future  changes  in
 temperature,  precipitation,  and other  climatic variables.   Furthermore,  such
 Methods can  be  combined with  information from both general circulation models
 ofv the climate  and with hypothetical scenarios to generate information on the
 Water-resource  implications of plausible future climatic changes.

     Water-balance modeling techniques  modified for assessing climatic impacts
 "ave been developed and tested for a  major watershed  in  northern California
 using climate-change  scenarios from both state-of-the-art general  circulation
 Models and from a series of  hypothetical scenarios.   Results of  this  research
 Su8gest  strongly  that  plausible  changes  in  temperature  and  precipitation
 °aused by a  doubling  of atmospheric carbon-dioxide concentration would  have
^Jor impacts on  both  the timing and magnitude of runoff and soil  moisture  in
 ln»portant agricultural areas.  Of  particular importance are predicted  patterns
 °f summer soil  moisture drying that are consistent across  the entire  range  of
 '•Gated scenarios.   In addition, consistent  changes in the timing  of  runoff—
 Specifically, significant  increases  in winter runoff and decreases in summer
                                    217

-------
 runoff—raise  the  possibility of major difficulties for future water-resource
 planning.

 INTRODUCTION

     Concern over  the  possible impacts of changes in global climate caused by
 increasing  atmospheric concentrations  of carbon dioxide and other trace gases
 has  grown  in  recent  years as  our understanding of  atmospheric  dynamics and
 climatic  systems has  improved.    Despite  a better understanding  of climatic
 processes,  however,   the  possibility  of  major   alterations  in  regional
 hydrologic  cycles  and  subsequent  changes   in regional  water availability are
 among the least understood impacts.

     One  of the  most  useful tools  for evaluating  global-averaged climatic
 changes due to increased CC^ concentrations  is  the  general circulation model
 (GCM) of  the  climate.   Yet  state-of-the-art GCMs,  though much advanced over
 early versions,  are  still limited in  their ability  to incorporate details on
 small-scale hydrologic processes or regional climate.   As a result, alterna-
 tive  techniques  for  evaluating  regional  hydrologic  consequences must  be
 developed,  tested, and applied.

     One attractive method for looking at impacts of climatic changes on water
 supplies  involves  combining  regional  hydrologic  modeling (water  balance)
 techniques  with information  on  plausible  climatic  changes from  both hypo-
 thetical  scenarios  and from  state-of-the-art GCMs.   This  method  can produce
 information on the sensitivity of  water availability in regional watersheds to
 changes in  temperature and precipitation.   Appropriate water-balance modeling
 techniques  have  been developed  and  tested  for  a major watershed  in northern
 California  (Gleick 1986b).

     Results  of this  research  suggest strongly that  plausible  changes  i°
 temperature and  precipitation  caused  by  increases  in atmospheric  trace EaS
 concentrations  would  have major  adverse  impacts  on  both the  timing
magnitude of runoff  and soil moisture in  important  agricultural  areas.
of the  most important of  the changes  in runoff and  soil  moisture are
across  a  wide range  of climate-change  scenarios.   These  hydrologic results
will have significant  implications for future water-resource planning and f°r
 international environmental and political behavior.

THE PROBLEM

     As fossil-fuel  use  and  industrial  development  expanded  over  the la^
century, the atmospheric concentration of carbon dioxide and other radiative-^
active trace gases has also risen.  Only within the last two decades, however»
have serious scientific efforts investigated the geophysical ramifications-

     Climate affects most  of  the world's environmental conditions—the supP1'
of  food  and   water,   the  need  for  shelter,  the  accessibility  of  roiner!!Bi
resources,  the distribution  of flora  and  fauna,  and so on.   Even snorfc'"t'^e
variations  in  climatic conditions  can cause enormous  human suffering.   *-,
possibility that global  climate may change permanently as  a result of n  ^g
activities  must  be cause  for substantial   international concern,  and pernap
alarm.
                                      218

-------
     Although we can presently conceive of ways in which global climate may be
affected by our actions, we  are  unable to see clearly either the direction of
changes in climate or the societal impacts of such changes.  Because 'we cannot
conduct controlled experiments on  the  entire  planet,  we must attempt to model
climate  and  climatic   changes — an   imprecise   alternative  because  of  the
complexity of  the  global climate  system.   Because of  the  many intricate and
intertwined phenomena that make  up the climate, much  of the effort of trying
to understand the climate system has focused on the development of large-scale
computer models.   The  most  complex  of  these  are detailed,  time-dependent,
three-dimensional numerical  programs  that  include atmospheric motions,  heat
exchanges,   and  important  land-ocean-ice  interactions.     These  models  are
typically  referred  to  as  "general  circulation  models"  or "global  climate
models" (GCMs).

     GCMs have permitted us  to begin to evaluate  some of the implications for
climatic patterns  of  increasing  concentrations  of atmospheric  gasea; and  a
consensus  is  beginning  to   form  about both  the  direction and magnitude  of
certain major impacts,  such  as  increases  in  global average temperatures and
changes in the intensity of distribution of the global hydrologic cycle.

     General circulation models  are large  and expensive  to operate.   While
they are  invaluable  for identifying  climatic sensitivities  and  changes  in
global  climatic characteristics,  they have two particular limitations to their
usefulness  to  researchers   interested  in  more  detailed  climate  impact
assessment: GCMs  cannot provide much  detail on  regional  or  local  climate
impacts, nor  can  they  provide much  detail on  surface hydrology.   For  these
Reasons, new  methods  must be  developed that  can  incorporate  information  on
both hypothetical and predicted climatic  changes in  order to determine how
future   global  changes may  affect  regional  water   resources   and  water
availability.

METHODS FOR REGIONAL HYDROLOGIC STUDIES OF CLIHATIC CHANGE

     Recently there   have  been  initial  attempts   to   evaluate the  regional
hydrologic  implications of   climatic  changes  (Schwarz  1977;  Stockton  and
B°ggess 1979; Nemec  and Schaake  1982; Revelle and  Waggoner  1983;  Flaschka
     Rind and Lebedeff 1984). These early  works  provided the first tentative
evidence  that  relatively   small   changes  in  regional   precipitation  and
evaporation patterns might result in significant, perhaps critical,  changes in
regional water availability.

     Before realistic estimates of changes in regional water  availability can
be calculated, however,  improvements must be  made  in  several areas.   Among the
"^st important characteristics of regional hydrologic assessments should be a
f°cus on  short time-scales  (i.e.,  months  and  seasons,  rather  than  annual
averages);  the ability  to incorporate both hypothetical climatic changes and
£he increasingly detailed assessments of  regional  changes produced  by GCMs;
ttle  use of  methods  that  produce  information  on  hydrologically  important
Vapiables,  such as changes in soil moisture  and runoff;  and the incorporation
of   the   complexities    of   snowfall   and   snowmelt,   topography,    soil
°naracteristics, and natural  artificial storage.

     One  of  the  most   promising  methods  for  doing   regional   hydrologic
 Saessments of  global climatic  changes is  the use  of water-balance  models


                                     219

-------
modified  for conditions  of changing  climate (Gleick  1986a).   Water-balance
methods  are  useful  in  diverse  watersheds.    They  can  evaluate  changes  in
vegetative  cover,  snowfall and snowmelt rates; characteristics of groundwater
recharge  and withdrawal;  and  monthly  and seasonal  responses.   They can  be
developed   and   run  on  existing  generations  of   state-of-the-art  general
circulation models.

A Water-Balance Model for  Climatic Impact  Assessment

     We developed  a water-balance model to evaluate  the  capabilities of  such
models for  climate impact  assessment.   This model was then tested and used  to
evaluate  hydrologic  impacts  of  changes  in   climate in  the most   important
hydrologic  basin   in  California  and one of  the  most important  in  the United
States—the Sacramento Basin (see Figure 1).

     The Sacramento  Basin  provides  over 30% of the total runoff for  the state
of California,  including  almost all of  the water  used  for agriculture in the
Central  Valley—one  of  the  most  productive  agricultural  regions of  the
world.    Moreover,  the  water  resources  of  this basin  are already heavily
subscribed—hence  any climatic change that decreases total water availability
or significantly  changes  the  timing of  soil  moisture and runoff would affect
the social and physical environment of the region.  Details of the development
of  the model,  the  modifications of  the  model  for use under  scenarios  of
changing climate,  and the  statistical  verification of the model are  presented
in Gleick (1986a,  1986b).

     To determine  the effect  of changing climate on the water  resources  of
this region,  we developed  a series  of temperature and precipitation  scenarios
and used  them to  drive the water-balance model.    For the  purposes of  this
study,  both  purely  hypothetical   climate-change  scenarios  and   scenarios
developed for general circulation model  output were chosen for analysis.  The
hypothetical  scenarios  of  temperature  and precipitation  changes  were chosen
after  reviewing  state-of-the-art estimates   of  future  changes  in  climatic
conditions.   The  GCM precipitation  and temperature  scenarios  were developed
after discussions  with leading climate modelers in the United States and after
a review of model  capabilities and  design.  These scenarios can be summarized
as follows:

       •  Ten hypothetical scenarios  involving  combinations of  plus 2°  and
          plus 4°C and +2Q%, +10/5, 0%, -10%, and -2Q% changes in precipitation

       *  Eight scenarios  of temperature  and  precipitation  changes predicted
          for   this  general   region   by  three   state-of-the-art   general
          circulation  models:     the  Geophysical  Fluid  Dynamics  Laboratory
          (GFDL)  model  (Manabe   and  Stouffer  1980; Manabe,  Wetherald,  and
          Stouffer  1981),   the Goddard  Institute  for  Space Sciences (GlSS)
          model  (Hansen et al.  1983,  1984),  and  the  National  Center  f°r
          Atmospheric Sciences Community Climate Model  (NCAR CCM) (Washington
          and Meehl  1983,   1984).

     None of  the  hypothetical  or GCM-derived  scenarios  includes  decreases  i|j
average  monthly   temperatures,  because   of  the  consensus  in  the  climat
community  that  increasing  concentrations  of carbon dioxide and other trac
                                      220

-------
     HYDROLOGIC STUDY AREAS
                  OF
             CALIFORNIA
                    SACRAMENTO BASIN
                                        _CR_	:
Figure 1.  The Scramento Basin, California
                221

-------
gases will lead to increases in surface air temperature on both a global and a
regional scale.   This consensus has been  expressed as a 95/J probability that
an "equivalent doubling" of atmospheric carbon dioxide—the  injection into the
atmosphere of  CC^ and other trace gases  such that the climatic effect of the
combined gases is equivalent to the climatic  effect of a doubled concentration
of C02  alone—will result  in  an average  global warming of between 1.5° and
5-5°C,  with  a most  likely  temperature increase  of 3.0°C  ±1.5°C   (Dickinson
1984).   Current  GCMs suggest  that  broad  regional  temperature  increases will
exceed  these  values,  particularly in polar  latitudes.   Temperature increases
of 2° and 48C, however,  are reasonable expectations for the region  considered
in this paper.

     Global  average  precipitation  is  predicted to  increase 5-]Q%  under  an
equivalent  doubling  of atmospheric  carbon  dioxide,  due  primarily  to  the
increase  in  global   average  temperature  (and, hence,  evaportranspiration)
(Manabe and Wetherald  1975, 1980).   Regional variations  are also expected to
be significant,  with  evidence for both increases and decreases  occurring  in
different regions.  Revelle and Waggoner (1983)  evaluated changes of ±10% in a
series  of  U.S.  watersheds,  and  both  Nemec  and Schaake (1982)  and Flaschka
(1984)  chose  scenarios  of  ±10-25$.    For the  purpose  of  this study,  five
precipitation  scenarios  were  evaluated:    no  change  in  monthly  average
precipitation, increases in monthly average  precipitation of W% and 20%, and
decreases in  monthly average  precipitation  of  10£ and 20%.   Because actual
precipitation changes may exceed  these  values,  the scenarios studied here can
be considered  conservative  in that they  explore  the  sensitivity  of water-
resources characteristics to changes in precipitation that are well within the
realm of possibility.

     Data from the three GCMs were obtained after a series of discussions with
leading climate modelers.   It  must be  noted  that individual grid-point data,
made available by each of  these  modeling groups,  do  not represent realistic
predictions  of  the  expected  climate  at these grid  points.    The  current
generation of  GCMs does not permit detailed regional estimates of climatic
changes because of limitations on computer time and  speed,  model resolution,
major  physical  parameterizations,  and existing  data  sets.   GCM modelers
understand these  limitations.   Nevertheless,   the  researchers  at  the three
climate centers  agreed  to  provide grid-point data so that  we  might evaluate
(and confirm or dispute) hydrologic effects  seen in general  circulation mode}
results and help  to  gain a better  understanding of both  the differences and
similarities among the  models, and the sensitivity of hydrologic  systems  fc£
climatic changes.  Thus, while the scenarios  developed  by the GCMs should not
be treated as a more likely description of the future than any other scenario*
they do offer insights  into  both  the capabilities of GCMs and their estimates
of hydrologic responses.

     The differences  among  the  GCM scenarios  also provide  some advantage3'
First  of all,  they  highlight  some  of   the  limitations of GCM  grid-p°inj
estimates and GCM model resolution.   Second,  by evaluating different predict6"
temperature and  precipitation  scenarios,  we can  evaluate  a  wide  range  °
climatic changes.  Third, by incorporating these scenarios,  it may be p
to identify areas  in  which consistent  changes  in soil moisture or  runoff
be obtained despite widely  varying precipitation and temperature inputs.
result  would  indicate  areas  of  important hydrologic  sensitivity  and
certainly be worthy of additional attention.
                                     222

-------
     The eighteen  scenarios (summarized  in Table  1)  were used  to  drive the
water-balance model  of the  Sacramento Basin  and  to estimate  the  effects on
available soil moisture and  runoff.   For  every scenario, a new 50-year record
of monthly average  temperature and precipitation was  created  by applying the
hypothetical  changes  to  the  50-year  historical  record  of  monthly  average
temperature and precipitation in the Sacramento Basin.  These data inputs were
then used  to drive  the water-balance model,  producing a 50-year  record of
monthly runoff and available soil moisture.  These data were then averaged, to
produce  long-term  average   monthly  and  average  seasonal  runoff  and  soil-
moisture results.

RUNOFF AND SOIL MOISTURE RESULTS

     Major hydrologic  changes  resulted from the eighteen scenarios,  including
some changes that are  consistent  in  their direction in every scenario despite
significant  differences  in  the  original   precipitation   and  temperature
inputs.  These changes include alterations in the magnitude of runoff,and soil
moisture,  as  well  as important  changes  in  the  timing  of  runoff -and  soil
moisture.

HYPOTHETICAL CLIMATE-CHANGE SCENARIOS

     Significant  changes  in  runoff patterns  were  observed   for  all of the
hypothetical scenarios.   On an annual basis,  the direction of the  change in
runoff  from  the  different  temperature  and  precipitation  scenarios  was
unsurprising:   temperature increases alone  led to  decreases  in annual runoff;
temperature increases  combined with  increases  in precipitation of W% and 20%
resulted in  increases in annual  runoff;  temperature  increases combined with
decreases  in  precipitation  of 10£  and 2Q% resulted  in decreases  in annual
runoff.

     Because shorter term hydrologic changes are of greater interest to water-
resource planners than  annual  average  changes,  both  seasonal and  monthly
impacts were studied.  Two "seasons" were evaluated—winter (assumed to be the
sum of December,  January, and  February runoff) and  summer (assumed  to be the
sum of June, July, and August  runoff).  These assumptions are consistent with
most GCM analyses of seasonal  climatic variables.    They also  correspond well
to actual seasonal conditions  in  the Sacramento Basin,  which  receives much of
its precipitation during winter months.

Changes in Average Summer and Winter Runoff

     Summer runoff in  all the  hypothetical  scenarios is reduced significantly
and consistently  when compared to summer runoff in the base  case.   Although
the reduction in runoff is most pronounced in those runs where monthly average
temperature  is  increased  and  monthly  average  precipitation  is  reduced,
Deductions in summer  runoff  are also  evident  when  monthly average  precipita-
tion is  increased  significantly.    The most  dramatic  example  of  this  is  a
Deduction  in  summer  runoff  of nearly 50%  when  monthly  average temperature
increases  4°C and  monthly  average precipitation   increases  20%.    Even  an
increase in  the  monthly average  temperature  of  only  2°C  combined  with an
increase  in  monthly  precipitation  of  2056  does  not increase total summer
                                      223

-------
                      Table  1.  Climate-Change Scenarios


Hypothetical Climate-Change Scenarios1

Change in Monthly Temperature (°C)        Change in Monthly Precipitation

       T + 2°                                           No Change
       T + 2°                                           - 10$
       T + 2°                                           - 20$
       T + 2°                                           + 10$
       T + 2°                                           + 20$

       T + 4°                                           No Change
       T + 4°                                           - 10$
       T + 4°                                           - 20$
       T + 4°                                           + 10$
       T + 4°                                           + 20$

General Circulation Model Climate-Change Scenarios1


Geophysical Fluid Dynamics Laboratory (Princeton. New Jersey)

       Temperature Changes Only
       Temperature and Relative Precipitation Changes
       Temperature and Absolute Precipitation Changes

Goddard Institute for Space Sciences (New York. New York)

       Temperature Changes Only
       Temperature and Relative Precipitation Changes
       Temperature and Absolute Precipitation Changes

National Center for Atmospheric Research (Boulder.  Colorado)

       Temperature Changes Only
       Temperature and Absolute Precipitation Changes
1  The temperature and precipitation change scenarios  in  this table were used
   to drive a water-balance  model of a major hydrologic  basin.   See text
   details.
                                      224

-------
runoff.  Table  2 summarizes the percent  reductions  in summer runoff from the
ten hypothetical scenarios.   Figure 2 plots  the percent  changes  in average
summer runoff for the ten hypothetical scenarios.

     For both the T  +.2°  and T + 4°C cases, the temperature increases account
for a large  fraction! of the  total reduction  in  summer  runoff.    The  next
section details, the reduction  in  summer  runoff results from a major shift in
the timing of runoff due to change in rain/snow ratios in winter and the speed
of snowmelt in  the spring.   Winter runoff shows a similar pattern.   Increases
in  temperature  alone  cause  increases  in  average  winter  runoff  due  to  a
decrease in the proportion of snow to rain and hence a decrease in the storage
of water (the snowpack) during the winter.  For the T + 2°C run with no change
in precipitation,  winter  runoff  increases 8%;  for  the T  + 4°C run  with no
change in precipitation,  winter runoff increases dramatically by 34$.

     When  precipitation   changes  are  imposed  on  the  temperature  increases,
winter  runoff   results   become  mixed—for  T  +  2°C   runs,   increases   in
precipitation cause increases in winter runoff, and decreases in precipitation
cause decreases  in winter runoff.   For the  T  +  4°C  runs,  however,  the winter
runoff changes are, for the most part, positive.  For all the runs except one,
changes  in  precipitation  lead to   increases  in  winter  runoff.    The  one
exception is the extreme—a decrease in monthly precipitation of 20%.  Even in
this case, however,  the decrease in average winter  runoff is small—only 4/6.
The percent changes in average winter runoff are plotted in Figure 3.  Table 3
summarizes these results.

     Some  of  the  changes  in  average winter  runoff are extremely  large,
particularly in the runs with increases in precipitation.  Increases in winter
precipitation of only 2Q%  lead to increases in average winter runoff of 40* to
80* for the T  + 2°C and T + 4°C runs,  respectively.   Such dramatic increases
in  runoff  raise  concerns  about  the  possibility  of  increased  flooding,
especially  in  basins  with  flood-control  systems  designed  for  different
hydrologic conditions,  or  in basins without major reservoirs.

Changes in Average Monthly Runoff

     The full  consequences for  runoff of climatic  changes can be  seen  when
average monthly  runoff  is studied.  Here  we see the  importance of looking at
temporal variations  in runoff on a scale  shorter than the annual cycle.  When
looking only at  the  average  annual figures,  the decrease in runoff from a 4°C
increase in  average temperature  is only   1% (see  Table 4).   When  individual
average monthly  changes are  evaluated,  however,  we  see that the same increase
in temperature  of 4°C causes  an increase  in average January runoff of 3956 and
a decrease in average June and July runoff of nearly 70%.  Dramatic changes in
the timing of monthly runoff  are thus hidden when only the effects on average
annual values are considered.

     For all ten hypothetical scenarios we estimated significant shifts in the
timing of  monthly runoff.   While  the increase  in  average temperature  is a
""ajor driving force  for these  shifts,  the changes in precipitation contribute
fco and amplify  the effects.  The cause of  the shift in the timing of runoff is
a decrease  in  total  winter snowfall  and  an  earlier and faster spring melting
°f  the winter  snowpack.    Even  in those  cases where  overall precipitation
decreases, the distribution of runoff over the year changes so  that spring and
summer runoff  decrease  while  runoff  during   the winter months  increases.


                                      225

-------
Table  2.   Effect of Hypothetical Temperature and Precipitation  Scenarios on
           Average Summer (JJA)  Runoff

                            (Percent Change over Base Run)
     Precipitation Change
Temperature Change

T -f 2°C

T + 4°C
                                      -73
                                                                                +20*
32
68
-22
-62
-12
-55
-1
-19
                40 -|


                20


                0
I
Oh
C
             BE
             tt
             ta

             *
-20 •


-40 ••


-60 ••


 40


 20


 0
                -20 •


                -40 •


                -60


                -80
                     T + 2«C   T + 2«C     T * 2«C     T + 2«C    T * 2«C
                               P-10%     P-20%    P*10%    P*20%

                                  T+4»C
                                   P-10%
                                         T*4«C
                                         P-20%
         T*4»C
                     •& ta.
                     Hl("' i
                     piiXj,'
                                          ii!gi^!!ii=ni,l
                                           ii!ilii
                                           !!«•
if
imjmiinnr
iteiiii:
jiOjlSKnilUt
iiSiiiiiiiiiiti
                                           MKiUilHlIB
                                           iSiiSiiSi;,:
                                           IjjjjHIjjijjji!,
T*4»C
P + 20%
      Figure  2.   Percent Change in Average Summer (June, July,  August)  Runoff
                  for  the Ten Hypothetical Scenarios
                                           226

-------
 Table 3.  Effect of  Hypothetical Temperature and Precipitation  Scenarios
           on Average Winter (DJF) Runoff

                       (Percent Change over Base Run)
      Precipitation Change
    -201
     -105
                                                                 ±20*
Temperature Change
T + 2°C
T + 4°C

-24
-4
^^^^_b_ ^_
-9 +8
+14 +34
wWMB^ri^
+25
+54
*l^~W>«
+44
+75
80  -


60  •


40  ..



20


 0
JJ,  -20 •


3  -40 ••

£  80 T
       g  604-
          40 •
          20 -
          -20-.
         -40 J-
                    T*2»C
                      iii
                     im
T+2»C
P-10%
T* J»C
P-20%
                                         T+2»C
                                         P+ 10%
                            T+4»C
                            P-10%
          T*4"C
          P-20%
           P+10%
T*2»C
P+ 20%
                                                           BpliKJiili!
                                                            IHSiSi:
                                                            ill
Figure 3.   Percent Change  in  Average Winter (December,  January, and
            February Runoff for  the Ten Hypothetical Scenarios
                                  227

-------
Table 4.   Effect  of  Hypothetical Temperature and Precipitation
            Scenarios  on Average-Monthly  and  Average-Annual  Runoff:
            A  Summary

                 (1000 Acre-feet and Percent Change over Base Run)
      Run*
JAM  FEB  MAR  APR  MAY  JUN  JUL  AUG  SEP  OCT  NOV  DEC ANNUAL
      Base  2118 2245 2660 2481 1951 1285  773   492  348  401  882 1611  17245
T+2C

T2P1

T2P2

T2P3

T2P4

T+4C

T4P1

T4P2

T4P3

T4P4

2234
.05
1879
-.11
1554
-.27
2615
.23
3001
.42
2947
.39
2520
.19
2130
.01
3398
.60
3866
.83
2370 2822 2323
.06 .06 -.06
1971 2368 1973
-.12 -.11 -.20
1605 1957 1655
-.29 -.26 -.33
2780 3277 2685
.24 .23 .08
3227 3738 3046
.44 .41 .23
2920 2806 1940
.30 .05 -.22
2474 2359 1637
.10 -.11 -.34
2047 1954 1351
-.09 -.27 -.46
3395 3270 2264
.51 .23 -.09
3888 3735 2598
.73 .40 .05
1649
-.15
1418
-.27
1203
-.38
1884
-.03
2122
.09
1078
-.45
908
-.53
748
-.62
1259
-.35
1444
-.26
1027
-.20
891
-.31
763
-.41
1165
-.09
1303
.01
556
-.57
469
-.63
387
-.70
648
-.50
742
-.42
591
-.24
514
-.33
443
-.43
667
-.14
744
-.04
262
-.66
220
-.72
181
-.77
306
-.60
352
-.54
374
-.24
328
-.33
284
-.42
420
-.15
466
-.05
163
-.67
139
-.72
116
-.76
189
-.62
215
-.56
278
-.20
246
-.29
216
-.38
311
-.11
343
-.01
150
-.57
131
-.62
113
-.68
170
-.51
189
-.46
345
-.14
299
-.26
257
-.36
394
-.02
441
.10
264
-.34
227
-.43
195
-.51
305
-.24
345
-.14
852
-.03
731
-.17
613
-.31
978
.11
1112
.26
784
-.11
670
-.24
557
-.37
902
.02
1028
.17
1828
.13
1581
-.02
1353
-.16
2097
.30
2387
.48
2113
.31
1831
.14
1574
-.02
2408
.49
2724
.69
16693
-.03
14200
-.18
11903
-.31
19271
.12
21930
.27
15984
-.07
13586
-.21
11354
-34
18513
.07
21126
.23
      Notes to Table 4.
      [Please note that  the runoff values are given in acre-feet, the standard
      unit for runoff ..in all available U.S. databases.  The conversion to
      cubic meters (nr)  is 1233 m  per acre-foot.]
           The decimal values on the lines following each run are the
           percentage change in runoff between the model run and the base run,
           or (RO-nodei -  RObase^^base^  Thus»  average-January runoff for
           the T + 2 degrees C ("T+2C") run increased by 0.05, or 5 percent,
           over the base  run ("Base").  Similarly, average-annual runoff for
           the same run decreased by 0.03, or 3 percent.  Average-January
           runoff for the T4P4 run (the last run in the table above) increased
           by 0.83, or 83 percent, over the base run, while average-annual
           runoff for this run increased by 23 percent.
      2.   Runs are coded as  followst
      Base:     Base run using historical temperature and precipitation.
      T + 2C:   Temperature increase of 2 degrees Celsius.
      T2P1:     T + 2 C} Precipitation decrease of 10 percent.
      T2P2:     T + 2 C; Precipitation decrease of 20 percent.
      T2P3*     T •»• 2 Cj Precipitation increase of 10 percent.
      T2P4:     T + 2 C} Precipitation increase of 20 percent.
      T + 4C:   Temperature increase of 4 degrees Celsius.
      T4P1:     T + 4 C; Precipitation decrease of 10 percent.
      T4P2t     T + 4 C; Precipitation decrease of 20 percent.
      T4P3t     T * 4 C} Precipitation increase of 10 percent.
      T4P4:     T ••• 4 C; Precipitation increase of 20 percent.
                                     228

-------
 Figures 4 and  5  show the average  monthly runoff, the average annual  runoff,
 and the percent  change  in  these values compared  to  the  base run for  each  of
 the ten hypothetical scenarios.

      The changes  in  the timing  of  runoff occur  primarily because  of the
 increase in  average  temperatures, which has  two  effects:    a significant
 decrease in the proportion of winter  precipitation that falls as snow and  an
 earlier and shorter spring snowmelt.   The first  effect  causes greater winter
 rainfall and  hence winter runoff, since less overall precipitation enters the
 snowpack to be  held  over until spring  snowmelt.  The  second  effect intensifies
 spring   runoff,  leading   to  additional  adverse  consequences  for both summer
 runoff  levels and soil-moisture  levels throughout  the spring and  summer.

      Changes  in   both   the  timing  and  magnitude  of  runoff  are   extremely
 important for  water  availability.   Yet changes  in runoff alone  do not  tell  us
 all there  is  to know about the vulnerability of a region to changes  in water-
 resource characteristics—changes in  other  variables must  also bs evaluated.
 Perhaps the  most important  of  these  is  the  change  in   the  soil  moisture
 available to agriculture and other  plant communities.   Soil  moisture  is one  of
 the most valuable measures  of water availability  for agricultural development
 and productivity, and  it  is  a major determinant  of  vegetative  types and
 extent.   The  next section describes in detail the changes  in soil moisture  in
 this  basin that  are expected to  occur from  the  changes  in temperature and
 precipitation described  above.

 Changes in  Average Summer and Winter Available Soil Moisture

      Average  summer  soil-moisture  values  in the  agricultural portion of the
 Sacramento  Basin,  defined as  the sum of June, July, and  August  soil  moisture,
 show  significant  and  consistent  decreases from  the  base  case for all ten
 hypothetical  scenarios.   These  decreases  range from 8%  to  44/t.   The  minimum
 decrease of Q% results  from  a  temperature  increase  of 2°C combined  with the
 maximum increase   in average  precipitation  of  20%.  The maximum decrease  in
 average summer  soil  moisture of 44# results  from a 4°C  increase in temperature
 combined with  a  20$ decrease  in average precipitation.   These results are
 summarized  in  Table  5.   Percent changes  in average  summer  soil moisture are
 plotted  in  Figure 6 for all ten  hypothetical  temperature  and precipitation
 scenarios.

     Winter soil-moisture values also  show  widespread  decreases  in  the lower
 basin—seven  of  the ten scenarios  result in  reduced  average winter  soil
moisture.   The magnitude of the  reductions  is  not nearly as  large  as the
 reductions  in  summer  soil  moisture,   but  the  winter  reductions offer  some
additional  insights  into  the   sensitivity  of  watersheds  to  changes   in
 climate.  Temperature  increases  alone  reduced winter soil  moisture  by H% and
 9%  for  2° and 4°C  increases, respectively.  These  reductions are the result  of
 increased evapotranspiration  rates.    Of  greater  interest  is the fact  that
soil-moisture   increases  were   relatively   small,    even   for  the   high-
precipitation scenarios, with an  actual  decrease  in soil  moisture  when the
 temperature increased 4°C and precipitation increased 10*.   During the winter
months,  percentage increases in precipitation have a larger effect on absolute
precipitation  than  the   same  percentage  increase  in  summer months  simply
because  overall precipitation levels are  higher.   Yet  these increases do not
manifest themselves  as  proportional increases in  winter  soil moisture.  There


                                     229

-------
            5000-


            4500 • •


            4000 • •

         &
         E  3500-
         *

         Si  3000 •>
         8

         g  2500-

         3
         1  2000 -


            1500-


            1000-


             500-
                  (A)
                   Btse Run
   Model-
              0	:	,	i	i	!	:	j	;	:	__
               JAN  FEB   MAR  APR   MAY  JUN   JUL  AUG   SEP  OCT   NOV  DEC
                                      MOUTHS

Figure 4a.   Average  Monthly  Runoff  (Model and  Base  Run) for  the
              T+2°C,  P-2Q% Scenario
             5000-


             4500
             4000- ,-
                 ''

             3500
          S  3000-^
                                         (B)
Model —>\
o
5
5
•
2SOO-
2000-
1500-
1000-
500-
0-
\ /'
L \ ,; t
\ Model 	 >! !
\ / r.
\ ! / !
k V i I f
V. ,' I \
\. 1 i 1
* . • •
•<•,'-.
^^^ -
                JAN  FEB   MAR  APR  MAY  JUN  JUL  AUG  SEP  OCT   NOV  DEC

                                       MONTHS

Figure 4b.   Average Monthly  Runoff  (Model  and  Base Run) for  the T+2°C,
              P+20% Scenario
                                      230

-------
               JAN  FEE  MAR  APR   HAY  JUN  JUL   AUG  SEP  OCT   NOV  DEC
                                    MONTHS
 Figure 5a.  Average Monthly Model  and Base  Runoff for the
              P-2Q% Scenario.
              JAN  FEB  MAR   APR  HAY  JUN  JUL   AUG  SEP  OCT  NQV   DEC
                                    MONTHS

Figure  5b.  Average Monthly Model and Base  Runoff for  the T+4°C,
             P+2Q% Scenario
                                     231

-------
Table 5.  Effect of Hypothetical Temperature and Precipitation Scenarios
          on the Average Summer (JJA) Soil Moisture (Lower Basin)

                      (Percent  Change over Base Run)
      Precipitation Change
 Temperature Change
T

T
                              -20$
-20

-38
                                                  -33
-12

-29
+20$


 -8

-26
               40 ••
I 20 •
u
2
5
*-zo -
	
« -40 •
te
£ -60 -
C
z
.3
i 40 "
s
X 20 '
s
D
* n

-20 -


-•40 -
-60 .






















T + 2»C
'•'*, * .-









T+4»




- ,"&
tar
! =.



: T
p










c i
F









+ 2»C
- 10%
^i i r
; ^t - -j








r*4»c
'-10%



: yj s
'i a ~i "^ s^.

•t "^ *


P










*
P









>2*C
-20%
i'i i- 1
^tJ|iS fttif
li flit ™







r + 4»c
-20%



i^?irr
feh* •» i,1,
r S"«a! ni
p STjito in
fe'1" 
-------
 are two principal  reasons  for this:  during  the winter months,  soils  tend to
 be near or  at saturation and  surplus moisture runs off, and  larger  precipi-
 tation events  in winter result  in more prompt storm  runoff,  which does  not
 become available to  recharge  soil  moisture.   Decreases in  precipitation  have
 the opposite effect,  which can be seen by the larger proportional decreases in
 average winter  soil-moisture  values.   Table 6 shows  the percent changes in
 average winter  soil-moisture  values  for  the ten runs  using  hypothetical
 inputs. Unlike the summer  soil-moisture results, precipitation changes are as
 effective   as  changes  in  temperatures  in  reducing  average   winter  soil
 moisture.   In this case, the lower  winter temperatures have  less of an effect
 on evapotranspiration  rates,   while  the  higher winter  precipitation has  a
 proportionally larger effect.


   Table 6.   Effect  of Hypothetical  Temperature and  Precipitation  Scenarios
             on the  Average Winter (DJF)  Soil  Moisture (Lower  Basin)
                         (Percent  Change  over  Base Run)

    Precipitation Change         -20%       -10%      0     +10%    +20%
 Temperature  Change
T + 2°C
T -c 4°C
-20
-25
-11
-16
-4
-9
+2
-3
+ 6
+ 3
Changes  in  Average Monthly Available Soil Moisture

     The  models  imply  that  monthly   soil-moisture  availability  in   the
Sacramento   Basin  using  the  hypothetical   temperature  and  precipitation
scenarios would  be reduced consistently from its base  level,, with the greatest
percentage  reductions  occurring during the  summer  months.  For seven of  the
ten  hypothetical cases, soil-moisture values  were  reduced in  every month of
the  year.    For  the other  three runs,  which  involve  increases  in monthly
precipitation, only  slight increases 4-n the soil moisture during winter months
were observed.   Table 7 shows  the average monthly  soil  moisture,  the 50-year
mean  of the annual  average soil-moisture  values,  and  the percent changes
between  these values and the values from the base run for the lower basin.

     This  section has  described the  changes  in  seasonal  and monthly  soil
moisture and runoff  that result from using a  series  of  hypothetical climate-
change scenarios  to  drive the water-balance model.   Many of these changes are
persistent  and  significant,  despite quite  variable  precipitation  patterns.
Among the most  important changes noted are major,  pervasive decreases in the
average  summer  soil moisture  and   the  volume  of  summer  runoff,   and  large
increases  in the volume of winter  runoff.    The next section  describes  the
results  of  using  temperature  and  precipitation output  from  the   eight  GCM
climate scenarios to drive the water-balance model.
GCM SCENARIOS
     Each  of  the  three  GCM  studies  produce  temperature and  precipitation
estimates  for  individual  grid points under  a doubled concentration  of atmo-
spheric  carbon dioxide.   These data  were  used  to test  the sensitivity  of
runoff and soil moisture in the study  region in the same manner as the
                                     233

-------
Table 7.   Effect of Hypothetical  Temperature and Precipitation Scenarios
            on  Average Monthly  and  Long-Term Annual Average Soil Moisture  of
            the Lower Basin (Millimeters and Percent Change over Base RunV
            Run
BASE
                       (Millimeters and Percent Change over Base Run)
                                                                              AVG
                   195  226  233  217   170   110   61   36   25   29   77  142   127
T+2C

T2P1

T2P2

T2P3

T2P4

T+4C

T4P1

T4P2

T4P3

T4P4

187
-.04
174
-.11
157
-.20
198
.01
208
.07
179
-.08
165
-.16
147
-.24
191
-.02
201
.03
220
-.03
208
-.08
192
-.15
229
.02
235
.04
214
-.05
200
-.11
183
-.19
224
-.01
231
.02
228 210
-.02 -.03
219 201
-.06 -.07
203 186
-.13 -.14
235 218
.01 .00
239 222
.03 .02
222 202
-.05 -.07
211 190
-.10 -.12
194 174
-.17 -.20
230 210
-.01 -.03
236 216
.01 -.01
158
-.07
150
-.12
137
-.19
165
-.03
171
.00
145
-.14
135
-.20
123
-.28
152
-.10
158
-.07
98
-.11
92
-.16
84
-.24
103
-.07
106
-.03
83
-.24
77
-.30
70
-.37
87
-.21
91
-.17
49
-.19
47
-.24
42
-.30
52
-.15
54
-.12
37
-.39
34
-.44
31
-.49
39
-.36
41
-.33
27
-.24
26
-.28
23
-.35
29
-.20
30
-.17
19
-.48
17
-.53
16
-.57
20
-.46
20
-.43
18
-.28
17
-.32
15
-.39
19
-.25
20
-.21
11
-.54
10
-.58
9
-.63
12
-.52
13
-.49
22
-.24
19
-.33
16
-.43
25
-.14
28
-.04
15
-.49
13
-.57
11
-.64
17
-.41
20
-.32
69
-.10
60
-.23
50
-.35
79
.02
88
.15
60
-.23
51
-.34
42
-.45
69
-.11
78
.01
131
-.08
117
-.18
101
-.29
144
.02
156
.10
121
-.15
107
-.25
91
-.36
135
-.05
146
.03
118
-.07
111
-.13
101
-.21
125
-.02
130
.02
109
-.14
101
-.20
91
-.28
115
-.09
121
-.05
                  Notes to Table 7..

                  1.   The decimal values on the lines following each run are the
                      percentage change in available soil moisture between the model run
                      and the base run, or (S\odel - SMbaae)/(SMba8e).  Thus, average-
                      January available soil moisture for the T + 2 degrees C ("T+2C"1)  run
                      decreased by 0.04, or 4 percent, over the base run ("Base").
                      Similarly, annual-average available soil moisture for the same run
                      decreased by an average of 0.07, or 7 percent.  Average-July
                      available soil moisture for the T4P4 run (the last run in the table
                      above) decreased by 0.33, or 33 percent, over the base run, while
                      annual-average available soil moisture for this run decreased by  an
                      average of 5 percent.


                  2.   Runs are coded as follows:
                  Base:     Base run using historical temperature and precipitation.
                  T + 2Ct   Temperature increase of 2 degrees Celsius.
                  T2P1:     T + 2 C} Precipitation decrease of 10 percent.
                  T2P2:     T + 2 C} Precipitation decrease of 20 percent.
                  T2P3:     T + 2 Cj Precipitation increase of 10 percent.
                  T2P4:     T + 2 C; Precipitation increase of 20 percent.
                  T •«• 4C:   Temperature increase of 4 degrees Celsius.
                  T4P1:     T + 4 Cj Precipitation decrease of 10 percent.
                  T4P2t     T + 4 C; Precipitation decrease of 20 percent.
                  T4P3:     T + 4 Cj Precipitation increase of 10 percent.
                  T4P4t     T + 4 C} Precipitation increase of 20 percent.
                                           234

-------
 hypothetical scenarios of the preceding section.  Each of the three models was
 used  to  produce  three  different scenarios:   predicted  temperature  changes
 alone;  temperature  changes  together  with the  relative (percent) change in
 precipitation;  and  temperature  changes  together  with the  absolute  change in
 precipitation.    For  reasons  described  in  Gleick  (1986b),  the  relative
 precipitation  runs  from  the MCAR model  were not  included in  the  analysis.
 Table 7 summarizes  the eight scenarios developed  from the GCM temperature and
 precipitation data.

      State-of-the-art GCMs  provide us  with  one of  our  only direct  insights
 into the behavior of global climate at hundreds of different points around the
 globe.    As  such,  precipitation  and  temperature  data  from individual  grid
 points can  be  used to develop  additional climatic scenarios to  evaluate the
 hydrologic  response  of  regional  watersheds  to  climate  changes.     These
 additional data provide a "sense" of  realism  that  cannot be matched  by even a
 wide range of hypothetical scenarios.   In addition,  as scientists continue to
 improve the  spatial resolution  and hydrologic parameterizations of GCMs,  the
 quality of  regional  detail  will  improve.   These  improvements  in regional
 output can then be used to drive water-balance evaluations of hydrologic areas
 of special interest and concern.

      All three GCMs produced precipitation and temperature  data for a  control
 (1xCOp)  scenario and  for  a  doubled C02 (2xC02) scenario.   The differences in
 model   formulations,   parameterizations,   grid   scales,   and   geographical
 resolutions  among  the three GCMs  result  in  differences in estimates  of  the
 effect of a doubled concentration  of carbon  dioxide  on  precipitation  and
 temperature.    Differences   in  the  control   runs—the  attempt  to reproduce
 existing  climate—introduce  further   variations   in  the  temperature   and
 precipitation results.

      The eight  GCM scenarios were used to drive the water-balance model.   The
 results  are  summarized  here for  two  spatial resolutions:   average seasonal
 runoff and average monthly runoff.  Significant changes in runoff patterns  are
 identified and  discussed  in  the  following  sections.

 Changes  in Average Summer  and Winter Runoff

     Significant changes  in  seasonal  runoff  that  are consistent  across  the
 different  GCMs  are observed  in each of the scenarios.  Despite differences in
 GCM  resolutions,  formulations,  and parameterizations, the  values of  summer
 runoff predicted by the  water-balance runs using GCM data  all change  in  the
 same direction  and by similar magnitudes;  winter  runoff shows similar effects
 iiT~the opposite~~direction.    Specifically, average  summer  runoff  decreases
 significantly for all  eight scenarios while average winter runoff increases in
 all  eight  scenarios.   Tables 8 and  9 summarize the  percentage changes in
 average  summer   and  average  winter runoff,  respectively, for  the eight  GCM
 scenarios.    When  only the  GCM  temperature  changes  are evaluated,  average
 summer runoff values  decrease dramatically by 40#  to 68%.   These  decreases
 persist when the precipitation changes are included,  even  under the spring and
 summer precipitation increases of  the GISS model.   All  eight GCM  scenarios
 show a major drop in summer runoff volumes, with a minimum decrease of 30% and
a maximum decrease of 68% over the historical base run.  Just as summer  runoff
decreases  in  all eight scenarios, winter runoff increases in all  eight.  The
average  winter  runoff  increases 16^-81*.  The greatest  increases occur with


                                    235

-------
Table 7.  Effect of GCM Temperature and Precipitation Scenarios  on  the
          Average Summer (JJA) Runoff


                       (Percent Change over Base Run)


 GCM1                  T Only    T and Relative P   T and Absolute P

 NCAR                    -40             n.a.                -30

 GFDL                    -50             -48                 -48

 GISS                    -68             -53                 -40
 1.   The three general circulation model data sets are: temperature
      only; temperature and relative precipitation; and temperature and
      absolute precipitation. The differences among the three runs are
      discussed in the text.

 n.a. Not included here; see Gleick (1986a, Appendix C).
Table 8.  Effect of GCM Temperature and Precipitation Scenarios on  the
          Average Winter (DJF) Runoff

                       (Percent Change over Base Run)


  GCM 1                   T Only    T and Relative P_   T and Absolute P

  NCAR                    +17             n.a.                +16

  GFDL                    +26             +34                 +33

  GISS                    +38             +81                 +66


  1.   The three general circulation model data sets are: temperature
       only; temperature and relative precipitation; and temperature and
       absolute precipitation. The differences among the three runs are
       discussed in the text.

  n.a. Not included here; see Gleick (1986a, Appendix C).
                                   236

-------
  the high  precipitation  scenarios of  the  GISS  model.   The magnitude of tho
  average summer runoff decreases  could be important to agriculture,  while the
  large  increases  in  average winter  runoff suggest  significant flooding  and
  water-management problems.   These runoff changes are plotted  in Figures  7 and
  O •

       The  consistency  of  these  changes  despite  the variations in  the  GCM
  assumptions and outputs is  the  result of two major factors:  the temperature
  increases   in  the models  are  driving significant changes  in  the  timing  of
  runoff  during  the  year,   and   although  the  precipitation   changes   make
  significant contributions  to the changes in the magnitude of runoff,  they are
  less  important in determining the timing  of that runoff than are the changes
  in  temperature.                                                             B

  Changes  in Average Monthly Runoff

      Water-balance (runs using all eight GCM  scenarios show increases  in runoff
  during  each  of  the  winter  months.    These   increases  slowly  give way to

             "                                                     *
    nP     in",    p «r  nH        "8  "* ^^ m°nthS  Wlt*    *™™™ Of
 runoff during late summer and early fall.  The GCM temperature increases alone
 produce  very  large  decreases  in  runoff during  the  summer months  and large
 increases in runoff  during January and  February.   As  examples  Figures i 9  10
 and  11  plot  the average  monthly  model runoff  prodS  for' tie  three  GCM
 temperature scenarios plotted against the average 'monthly runoff for the base
 run.   The  change  in timing  of runoff  can  be seen  clearly in  these olotf
 Although the overall change  in annual  runoff volumes for  fL  HIPP    F
 figure! is large.  Table 10 lists  the data  on avenge ™^ £^£™
 annual runoff,  and the  percentage  changes for all eight  of the GCM scenarios
 and the base and case run.                             B       une uun scenarios

      As with the hypothetical scenarios,  the changes  in  the  timing  of runoff
 in the GCM-driven  cases occur  primarily because of  the  increase in  average
 temperatures.   Higher average  temperatures cause a significant decrease in  the
 proportion  of  winter precipitation  that falls  as snow  and an  earlier  and
 shorter spring  snowmelt.  The first effect causes greater winter  rainfall  and
 runoff,  since less overall precipitation enters the snowpack to  be  held over
 until spring melt.   The second effect intensifies the magnitude of peak flows
 in spring and shortens  the overall duration  of spring runoff, which leads to
 decreases  in summer  runoff levels and depress soil-moisture levels throughout
 the  spring and  summer.

      Among  the  most  consistent  and significant  results  obtained  from this
 study are the decreases  in soil-moisture availability during critical parts of
 the  year.    The next two sections  describe  the  seasonal and  monthly soil-
 moisture changes  that result  from using  the  GCM temperature  and precipitation
 scenarios  to drive  the water-balance model  of the Sacramento Basin.   This
 section will  focus  primarily  on the consequences  of GCM- estimated changes in
 precipitation and temperature  for available  soil moisture  in the agricultural
 areas of the Sacramento  Basin.

 Changes in Average Summer and Winter Available Soil Moisture

     Water-balance  model   results   using  all   eight  GCM   scenarios  show
significant  reductions  for the base  case summer  soil-moisture values in the


                                     237

-------
      40-r
      20--
  2 -20
  BE
  QC
  U
  S -40
  ce
    -60 -•
    -80 -L
  NCAR

(a)     (c)
                              (a)
GFDL

  (b)
(c)
(a)
GISS

 (b)
                                                                         (c)
           (a)  Temperature only
           (b)  Temperature and Relative Precipitation
           (c)  Temperature and Absolute Precipitation
Figure  7.   Percent  Change in  Average Summer (June, July,  and August)
            Runoff for All Eight  GCM Scenarios
                                     238

-------
  1
  *  100
  ^  80

  fc

      60
  D
  et
      40  ••
       20  ••
        0


      -20
    NCAR

    (a)     (c)
     GPDL

(a)      (b)
                                                 (c)
(a)
GISS

  (b)
(c)
(a)  Temperature only
(b)  Temperature and Relative Precipitation
(c)  Temperature and Absolute Precipitation
Figure  8.   Percent Change In Average Winter (December, January,  and
            February)  Runoff for All  Eight GCM Scenarios
                                     239

-------
        5000 -

        4500 -

        4000-
    g
             i
        3500 ,

    £   3000-j-      -•'        \  ••
    s       i   ...--•           \    -
    W       •...•''
    S   25001                    \    \  <	 Base Run
    -j       >
    «   20007                      \_
            i         Model  	>    \   \
        1500-                        \   \_                             /
            -                            '"••   \                         /'
        1000-
        500 -:-
           JAN    FEB   MAR   APR   MAY   JUN   JUL   AUG    SEP   OCT   NOV   DEC
                                         MONTHS
Figure 9.   Average Monthly Model and  Base Runoff for  the NCAR Temperature
            Assumptions
                                       240

-------



NETERS
u
1
i
3
5



5000-
4500-
4000-
3500-
3000-
2500'
2000-
1500-

1000-
500-
n .
i
i
i
—
-!•
.*-— ~~~\ '
-x--""' \ -i-
i
••• ^- -L
\ "'• '
„.-"' \ ^^ < 	 Base Run i
\ \ }
'• y ,
\ \ / r-
Model 	 >\ \ //;
\ \ .' .,•' i
• S i . • '.
.. X \ .'.• -r-
\ \ i. :
\ t i
\. X .:•' ~^
""*
"\ ^"-v.
* . • « . t • 1 1
           JAN   FEB  MAR   APR  MAY   JUN   JUL   AUG   SEP   OCT   NOV   DEC

                                        MONTHS
Figure 10.  Average Monthly Model and Base Runoff for  the GFDL Temperature
            Assumptions
                                     241

-------


g
tf
1!

u
§
5
ml

«













?uuu
4500
4000

3500


3000
2500


2000




1500


1000


500

0
•
1
(
i
•*


.
.


-




-i


.


-


-
•
""-^
*'"-» -*
•-' \ "^^"^i
./ \ \
• s' \ \
,-"~ \ \ < 	 Base Run ,
,4
\ \ * 1
\ \ /
Model 	 >\ \ / ,
\ V 1 f
\ ^ •' /
\ \ i1 /
S 1 /
K '» ^ // -*
^ X //
\ \ *•
S*- \ *
/; 4
\ ^ / •'
VV. >v. //

x-. — _ ^^
: i 	 ; 	 i 	 i 	 i 	 i 	 i 	 1 	 1 	 ,
         JAN   FEB   MAR   APR   MAY   JUN   JUL   AUG   SEP   OCT   NOV   D€C
                                      MONTHS
Figure 11.  Average Monthly  Model  and Base Runoff for the GISS  Temperature
            Assumptions
                                      242

-------
Table  10.   Effect of GCM Temperature  and  Precipitation Scenarios on Average
              Monthly and Average Annual Runoff:   A Summary

                              (1000 Acre-feet and Percent Change OT«P Base Ron)'


                   Run2    JAHHBMARAPJRMAYJONJOLAUGSEPOCTNOV  DEC ANNUAL

                   Base   2118  2245 2660 2481 1951 1285  773  $92  348  101  882  1611  17245
                  National Center for Atmospheric Research (NCAR)  GCMt
                   T»Pa
 2518 2824 3059 2291 1428  821  440  275  217   306  842  1646  16667
   .19  .26  .15 -.08 -.27 -.36 -.43 -.44 -.38  -.24  -.05   .02   -.03
 3162 2318 3029 2278 1610  987  518  273  211   300  846  1429  16961
   .49  .03  .14 -.08 -.17 -.23 -.33 -.44 -.39  -.25  -.04  -.11   -.02
                  Geophysical Fluid Dynamics Laboratory (GFDL)  GCMt

                  T     2678 2887 2950 2181 1286  702  356  221  184  287  811  1944   16488
                          .26  .29  .11 -.12 -.34 -.45 -.54 -.55 -.47 -.28 -.08^.21   -.04
                  TjP-  3274 2847 2956 2044 1207  694  357  262  196  184  783  1876   16679
                          .55  .27  .11 -.18 -.38 -.46 -.54 -.47 -.44 -.54 -.11   .16   -.03
                  T;Pa  3261 2816 2959 2071 1205  704  362  272  212  174  772  1858   16666
                          .54  .25  .11 -.17 -.38 -.45 -.53 -.45 -.39 -.57 -.12   .15   -.03

                  Goddard Institute for Space Sciences (GISS) GCMt
                   TjPr

                   TjPa
  3100 2946 2745 1817  979  482  213  133  132  250  778 2183
   .46   .31   .03 -.27 -.50 -.62 -.72 -.73 -.62 -.38 -.12  .36
  4169 3878 4406 2499 1354  707  303  188  123  387 1263 2756
                 .01 -.31 -.45 -.61 -.62 -.65 -.03
        .97  .73  .66
      3806 3579 3910 2307 1339
                      _   .43  .71
981  333  224  135  3i|J»  i135 2549
                          .80   .59   .47 -.07 -.31 -.24 -.57 -.54 -,.61 -.14  .29  .58
15759
 -.09
22033
  .28
20642
  .20
                 Motes to Table JO..

                         note that the runoff values are given in acre-feet, the standard
                      for runoff in all available U.S.  databases.   The conversion to
                       meters (m3) is 1233 m* per acre-foot.]

                      •m.. rf«nlmal values on the lines following each run are the
                          ceoi«w        .  runoff between the model run and the base  run,
                      percentage change in runo        ^^ av.rage-January runoff for
                      °r S^temoerature-only run ("T") run Increased by 0.19, or 19
                      fch!».«t   over the base run ("Base").  Similarly, average-annual
                      percent,  over tne^      deopeaaed by 0.03, or 3 percent.  Average-

                      rUn°ff r«U?f foC the OISS Teaperature and absolute P™^;"0"
                             Jlast run in the table above) increased by 0.80,  or 80
                          J   oler the base run, while average-annual runoff  for this
                          "increased by 20 percent.
2

T:
Jjjr',
Runs are coded as follows «
     TMnnerature changes only.
                                                    .
                                        changes  and  relative precipitation changes.
                                        ohtnges  and  absolute precipitation changes.
                   (See text for details of these runs.)
                                           243

-------
lower  basin.   These reductions range  14-36%.   In  six of the eight scenarios,
average winter  soil-moisture  values  undergo modest reductions of 2-10/J, while
the  remaining  two  runs show a 3% and  4$  increase  in  soil moisture.  Table  11
and  Figure 12 present the changes in average summer soil moisture based on the
GCM  climate-change scenarios;  Table 12 presents the changes in average winter
soil moisture.

     The decreases in average summer soil moisture in the Sacramento Basin are
remarkably consistent  regardless  of which  GCM  scenario  is  used  to drive the
water-balance model.  Soil-moisture  losses  of between 20% and 40? result from
seven  of the eight  scenarios,  with  the remaining decrease of 14$ occurring  in
the GISS high-precipitation case.

     The magnitude  and  the consistency  of the average  summer soil-moisture
drying signify  a major  hydrologic impact,  especially  given that these results
are consistent with the summer soil-moisture results from the ten hypothetical
temperature  and precipitation scenarios   discussed  earlier:   all  eighteen
climate-change  scenarios yield large losses of  summer soil moisture when used
to drive the water-balance model.

Changes in Average Monthly Available Soil Moisture

     There is  a consistent  monthly  depression  of soil-moisture  availability
for  the  GCM  runs,  with  the exception  of slight increases  during  some winter
months for the  highest  precipitation scenarios  of  the GISS model.   The water-
balance model  results  using six  of the eight GCM  scenarios show  decreases  in
monthly soil moisture after March continuing through  December.   The other two
scenarios,  using the  GISS  relative  and   absolute  precipitation   data,  show
increases in soil moisture  beginning again  in  November.   Table 13 summarizes
the  average monthly  soil moisture  and the  50-year mean of the annual average
soil-moisture results for the lower basin.

RESULT HIGHLIGHTS AND DISCUSSION

     Eighteen  climate-change  scenarios  were used to  drive a water-balance
model  designed  to  evaluate  the impacts  of global  climatic changes on runoff
and soil moisture in a  major  watershed.   The scenarios included ten scenarios
with hypothetical increases  and decreases in precipitation and temperature and
eight  scenarios with changes  in precipitation and  temperature generated using
results  from  three  state-of-the-art  general  circulation  models  of  global
climate.   The results from  using these eighteen temperature and precipitation
scenarios to  drive the  water-balance model  show some  consistent and pervasive
changes in  both runoff  and  soil moisture,  despite  the fact that the scenarios
have some major difference among  them.

     The results  of the  water-balance runs show  that dramatic shifts  will
occur  in the  timing and distribution  of  both  soil moisture and runoff.   The
directions  of these shifts are independent of the level of rainfall, while the
magnitudes  of  the   soil  moisture  and  runoff changes  are  exacerbated  by
increases or  decreases  in  precipitation.   Four  particularly important  and
consistent  changes  were observed:

    •  Large   decreases  in   summer   soil-moisture  levels  for  all  eighteen
       climate-change scenarios


                                     244

-------
Table  11.   Effect of GCM Temperature and Precipitation Scenarios on  the
            Average Summer (JJA)  Soil Moisture  (Lower Basin)

                         (Percent Change over Base Run) 1
       1
   GCM

   NCAR

   GFDL

   GISS
T Only     T  and Relative P   T and

 -28              n.a.

 -33              -35

 -31              -21
                     T and Absolute P

                              -20

                              -36

                              -14
   1.    The three general circulation model  data sets are: temperature
        only; temperature and relative precipitation; and temperature and
        absolute precipitation. The differences among the three  runs are
        discussed in  the text.

   n.a. Not included  here; see Gleiok (1986a,  Appendix C).
                40
            5   20 •'
            ^   -20- '
            |  -40 +

            Z
            M  -60 J-
                      NCAR

                     (•)   (O
(•>
GFDL

  (b)
                 (c)
GISS

 (b)
                                  (c)
                   (•) Temperature only
                   (b) Temperature and Relative Precipitation
                   (c) Temperature and Absolute Precipitation
 Figure 12.  Percent Change in Average  Summer Soil Moisture (June, July,
             and  August) Over the Base  Run for All Eight  GCM Scenarios.
             (Note the consistent decreases in summer soil moisture.)
                                   245

-------
    Table  12.  Effect of GCM Temperature and Precipitation Scenarios on the
               Average Winter (DJF) Soil Moisture (Lower Basin)

                           (Percent Change over Base Run)
      GCM
          1                   T Only    T and  Relative  P    T and  Absolute P
      NCAR                     -2             n'a-                  "3

      GFDL                     -5              -*                  -3

      GISS                    -1°.              +*                  +3


      1.   The three general circulation model data sets are:  temperature
           only; temperature and relative precipitation; and temperature and
           absolute precipitation. The differences  among the three runs are
           discussed in the text.

      n.a. Not included here; see Gleick (1986a,  Appendix C).
    •  Decreases  in  summer  runoff  volumes  for  all eighteen  climate-change
       scenarios

    •  Major  shifts in the  timing of average  monthly  runoff  throughout the
       years, with spring and summer runoff shifting to winter

    •  Large  increases  in winter  runoff  volumes  for fifteen of  the eighteen
       climate-change  scenarios,   including all eight GCM  cases.   The other
       three  scenarios—all  of  which  involved  10?  or  2Q%  decreases  in
       precipitation—showed small or moderate decreases in winter runoff.

     The hydrologic changes described above will have serious implications for
many aspects of water resources, including agricultural water supply, flooding
and drought probabilities, groundwater use and  recharge,  and reservoir design
and  operation~to  name  only  a  few.     Only  by   looking  at  the  specific
characteristics of  water-resource  problems,  and their  vulnerability  to the
types of changes in runoff and  soil moisture  identified above,  can details of
future societal  impacts  be  evaluated.   Such  evaluations  must begin  now in
diverse hydrologic  basins so that  policies  for mitigating  or  preventing the
most  serious  hydrologic  impacts  of climatic  changes  can  be developed and
implemented.
                                     246

-------
Table  13.  Effect of  GCM  Temperature  and  Precipitation  Scenarios on
             Average Monthly and Long-Term  Annual Average Soil Moisture of
             the Lower  Basin

                        (Millimeters and Percent Change oyer Baae Ron)1

             Run2    JAN   FEE   MAR  APR  MAY  JUN  JUL  AUG  SEP  OCT  NOV  DEC   AVG

             BASE    195   226   233  217  170  110   61   36   25   29  77  142   127
             National  Center for Atmospheric Research (NCAR)  GCM:

             T      191  222  229  204  144   90   38   21    13   18    72   137    115
                    -.02 -.02 -.02 -.06 -.15 -.18 -.38 -.43  -.46 -.38 -.07 -.04  -.09
             T;P,    202  217  229  204  152  100   43   23    15   19    73   127    117
                    .04 -.04 -.02 -.06 -.11 -.09 -.30 -.36  -.40 -.34 -.05 -.11  -.08

             Geophysical Fluid Dynamics Laboratory (GFDL) GCM:

             T      185  218  227  207  152   82   37   20    12   17    66   129    113
                    -.05 -.03 -.03 -.05 -.11 -.26 -.39 -.46  -.50 -.40 -.15 -.09  -.11
             T}Pr    194  220  227  204  147   79   36   19    12   12    63   126    112.
                    -.01 -.03 -.03 -.06 -.13 -.28 -.41 -.47  -.51 -.59 -.18 -.11  -.12
             TjPa    198  222  228  204  145   78   36   19    12   13    64   127    112
                    .02 -.02 -.02 -.06 -.15 -.29 -.41 -.47  -.52 -.55 -.17 -.11  -.12"

             Goddard Institute for Space Sciences (G1SS) GCM:

             T      175  210  219  197  149   86   40   17     8   11    60   122    108
                    -.10 -.07 -.06 -.09 -.13 -.22 -.35 -.52  -.67 -.61 -.22 -.14  -.15
             T}P_    202  230  238  210  161   94   44   19     9   18    94   152    123
                    .03   .02  .02 -.03 -.05 -.14 -.29 -.47  -.63 -.36   .22   .07  -.03
             T;P    201  231  241  213  167  107   50   22    11   15    84   148    124
                    .03   .02  .03 -.02 -.02 -.03 -.18 -.39  -.56 -.48   .09   .04  -.02
            Notes to Table  13.

            1.    The decimal values on the lines following each run are the
                 percentage change in available soil moisture between the model run
                 and the base run, or (SM^^ - SMbase)/(SMbase).   Thus,  average-
                 June available soil moisture for the NCAR Temperature ("T") run
                 decreased by 0.18, or 18 percent, over the base run ("Base").
                 Similarly, annual-average available soil moisture  for the same run
                 decreased by an average of 0.09, or 9 percent.   Average-July
                 available soil moisture for the GISS Temperature and  absolute
                 precipitation run (T;Pa) run (the last run in the  table above)
                 decreased by 0.18, or 18 percent, over the base run,  while annual-
                 average available soil moisture for this run decreased by an
                 average of 2 percent.
           2.   Runs are coded as follows:
           T:        Temperature changes only.
           T;Pr:     Temperature changes and relative  precipitation changes.
           T;Pfl:     Temperature changes and absolute  precipitation changes.

           (See text for details of these runs.)
                                        247

-------
REFERENCES

Dickinson,  R.E.   1984.    Modeling  evapotranspiration  for  three-dimensional
     global climate  models.   In Climate  processes  and climate  sensitivity,
     eds. J.E. Hansen and T. Takahashi.   American  Geophysical  Union Monograph
     29.  Maurice Ewing,  5:  58-72.

Flaschka,  I.M.  1984.   Climatic  change and water  supply  in the  Great  Basin.
     Master's Thesis.   Department of Hydrology and Water Resources, University
     of Arizona.

Gleick, P.H. 1986a.  Methods for evaluating the regional hydrologic impacts of
     global climatic  changes Journal  of  Hydrology (accepted  for publication
     May 26, 1986).

Gleick, P.H.  1986b.   Regional  water availability and  global  climatic change:
     The  hydrologic  consequences of  increases in atmospheric C02  and  other
     trace  gases.   Ph.D. Thesis, ERG-DS  86-1.   Energy and  Resources  Group,
     University of California,  Berkeley.  688 pp.

Hansen, J., G. Russel, D. Rind, P. Stone,  A. Lacis, S.  Lebedeff, R. Rudey, and
     L. Travis.  1983.   Efficient three-dimensional global  models for climate
     studies:  Models I and II.  Monthly Weather Review. 111:4:609-62.

Hansen,  J.E., D.  Rind,   G. Russell,  P.   Stone,  I.  Fung,  R.  Ruedy, and J.
     Lerner.   1984.   Climatic  sensitivity:   Analysis  of  feedback mechanisms.
     In  Climate  processes  and  climate sensitivity,  eds.  J.E. Hansen and T.
     Takahashi.  Washington, D.C.:  American Geophysical Union.

Manabe, S., and  R. J.  Stouffer.   1980.  Sensitivity  of a global climate model
     to  an increase  of  C02 concentration  in the atmosphere.    J.  Geo.   Res^
     85:C10:5529-54.

Manabe,  S.,  and R.T. Wetherald.   1975.   The  effect of doubling the C02
     concentration on  the climate of a general circulation model.  J.  Atmos^
     Sci.  37:99-118.

Manabe, S., and  R.T.  Wetherald.  1980. On the distribution of climate change
     resulting from  an increase  in  C02-content of the atmosphere.  J.  AtmojLi.
     Sci.  37:99-118.

Manabe, S., R.T.  Wetherald, and  R.J.  Stouffer.   1981.  Summer dryness due to
     an increase of atmospheric C02 concentration.  Climatic Change  3:347-86.

Nemec,  J., and J. Schaake.   1982.   Sensitivity of water  resource systems to
     climate variation.   Hvdrological Sciences. 27:3:327-43.

Revelle,  R.R., and P.E.  Waggoner.  1983.   Effects of  a carbon  dioxide-induced
     climatic  change  on  water  supplies   in  the  western  United  States.   in
     Changing  climate.  Washington,  D.C.:    National  Academy  of  Sciences,
     National  Academy  Press.
                                       248

-------
Rind, D.,  and  S. Lebedeff.   1984.   Potential  climatic  impacts of increasing
     atmospheric C02 with emphasis on water  availability and hydrology in the
     United States.   Washington, D.C.: U.S. Environmental Protection Agency.

Schwarz, H.E.  1977.   Climate change  and  water supply:   How sensitive is the
     northeast?  In  Climate,  climatic change, and water  supply.   Washington,
     D.C.: National  Academy of Sciences.

Stockton,  C.W.,  and W.R.  Boggess.    1979.   Geohydrological  implications  of
     climate change  on water resource  development.    Fort  Belvoir,  Virginia:
     U.S. Army Coastal  Engineering Research Center.
Washington,  W.M.,  and  G.A.  Meehl.     1983.     General  circulation  model
     experiments on the climatic effects due  to  a  doubling and quadrupling of
     carbon dioxide concentration.   J.Geophys. Res. 88:C11:6600-10.
Washington, W.M.,  and  G.A. Meehl.   1984.   Seasonal cycle experiment  of the
     climate sensitivity due to a doubling of  C02  with an atmospheric general
     circulation  model   coupled   to  a  simple  mixed-layer  ocean  model.  J.
     Geophys.  Res.  89:06:9475-9503.                                         ~~
                                    249

-------
 Hydrologic Consequences of Increases
 in Trace Gases and CO2  in the Atmosphere
John R Mather and Johannes Feddema
University of Delaware
Newark, Delaware USA
INTRODUCTION

     There  has  been  much speculation concerning  the  effect of  increasing
atmospheric CCU  and other trace gases on all aspects of life on  this  planet.
Most investigators agree  that  such  increases will  result  in some  warming of
the lower layers of the atmosphere.   However,  whether  ice caps  will  melt or
grow; whether  sea  levels will  rise; whether agriculture will just  move pole-
ward uniformly;  whether there will be local adjustments in the global circu-
lation patterns; whether precipitation will change as  a result  of increased
evapotranspiration; and how all these possible  changes  will  influence socio-
economic factors, our  standard  of living,  or our quality of life are questions
subject  to endless  debate  and speculation.   Clearly, one  reason  for  our
inability to definitively  answer these questions lies in our  lack of reliable
information on the magnitude of changes  in  temperature and  precipitation to
expect as greenhouse gases increase.   However, even if reliable climatic data
were available,  uncertainties would  still exist because of the difficulty in
determining  how particular  climatic  conditions influence  such  factors  as
agricultural   markets,  human  perceptions  and  tastes, water   demands  and
supplies, or even political and economic  decisions.   We can  only continue to
work toward a  more accurate understanding  of future climatic conditions while,
at the  same time,  trying  to  translate those climatic  conditions  into more
useful  human,  economic,   physical,  or   political  responses   through  the
application of meaningful models.

     We  have  a well-tested  model  that  expresses how  atmospheric  energy
(expressed as  air temperature)  and precipitation influence the water relations
of a  place  or  area.   The  climatic  water  budget,  originally developed by
Thornthwaite in  the early  1940s and  later modified by Thornthwaite  and Mather
(1955),  has been used extensively to  provide information on  factors  such as
soil moisture  storage,  actual  evapotranspiration, water deficit,  soil water
surplus,  water runoff or streamflow, and  snow storage and-melt.   Where checks
are possible,  the  simple  water  budget  bookkeeping  procedure  developed by
                                   251

-------
Thornthwaite and Mather has been found to provide reliable data for many parts
of the world.  Annual  values  of water surplus computed from the basic data of
monthly temperature  and  precipitation approximate closely measured 'values of
streamflow.  In fact, Mather (1981) even suggested the use of the climatically
computed values of surplus as a way  to evaluate the accuracy of stream gaging
stations.   Values of computed soil moisture  storage  agree almost exactly with
values of  soil  moisture content measured  by the weighing and  drying  of soil
samples (Thornthwaite and Mather 1955).  Finally, many studies show that water
deficit or the ratio of actual evapotranspiration to potential evapotranspira-
tion is closely related to agricultural yields (Mather 1978).

     Evaluation  of  the  water   budget  bookkeeping  method  reveals  that  if
precipitation increases in an area with  no change in temperature (a surrogate
for  potential  evapotranspiration)  or with  a  decrease  in temperature,  an
increase occurs in soil moisture storage and  in stream runoff.  An increase in
precipitation accompanied by  an increase in  temperature  is  more difficult to
evaluate  because the  relative   magnitudes  of  these  changes would determine
whether  soil moisture content and  streamflow  increase or   decrease.    If
precipitation increases more than the climatic water demand,  streamflow should
increase,   while if  water demand,  as a result  of  the  atmospheric  warming,
exceeds the  increase  in precipitation,  soil  moisture  and streamflow  would
decrease.   Conversely, a  precipitation decrease accompanied  by  an increase in
climatic demand  for  water (temperature) should result in a decrease  in soil
moisture  storage and  streamflow.   Clearly,  the seasonal  patterns of  these
changes would strongly influence the actual pattern of increase or decrease in
soil moisture conditions or water surplus.

     A number of investigators  have used different  water budgeting procedures
to  evaluate  the  effect  of  predicted temperature  and  precipitation  changes
resulting  from  increases in  trace gases  and  atmospheric C02.   Most of the
predictions of climatic changes come  from  the  operation  of global circulation
models under  current  and  increased  COp conditions.   These models,  based on
different  assumptions  concerning such factors as cloud  cover,  surface rough-
ness, land-water distributions, oceanic influences, atmospheric water vapor,
and  surface-boundary layer exchanges,  provide  estimates of temperature and
precipitation for  present and  various future  scenarios.    In  this paper, we
apply the temperature and precipitation data obtained from two of these global
circulation models,  the  Goddard (GISS)  model  and the NOAA  Geophysical  Fluid
Dynamics Laboratory  (GFDL)  model,  to  the  climatic  water budget to determine
the possible  influence of predicted changes  in temperature  and precipitation
on such  factors as  soil  moisture deficit,  water surplus,  and soil  moisture
storage in twelve selected regions of the globe.  The results not only suggest
the  complex  nature of  changes  in hydrologic  factors that  will accompany an
increase in C02  and  other gases, but  also reveal some of the difficulties in
trying to draw conclusions from such modeled data.

     Any attempt to understand future hydrologic conditions  through the appli-
cation of  temperature  and precipitation data  derived from  one of the global
circulation models  is fraught  with  uncertainties  that  the user  must  fully
recognize.   First,  there  is  no reason  to  expect  that  future  climates will
merely repeat past conditions.   One cannot  necessarily  look at warm episodes
in the past period of instrumental records to model future climatic conditions
resulting  from  increased  C02.   The reasons  for the  past climatic changes are
different  and there  is every reason  to expect that the pattern of climatic
                                     252

-------
changes will also  be  different.   Second, while  investigators expect that the
predicted  increase in trace  gases  and C02  will  lead to  increases  in atmos-
pheric temperatures,  there  is some  question concerning  the  timing and magni-
tude of  the greenhouse  warming.   The  influence  of such  climatic warming on
precipitation is more in doubt since  a number  of feedback relations must also
be considered.   Third, the  available  models  do not explain present conditions
with great  accuracy;  built-in errors  might  produce even greater errors under
future  scenarios.     Fourth,  available  global  circulation  models  provide
information for a rather coarse gridwork of points.  The GISS model uses an 8°
x  10°  latitude and longitude grid  while  the  GFDL model  uses 4° x  5° grid.
These networks  cover  a  wide range  of  conditions.   Topography  can  vary from
coastal plains to mountains, while present climatic conditions might vary from
desert to  rainforest.   Evaluating  conditions at  one  spot  in  the  grid  can
provide only  rough estimates of  what  to  expect in  other parts  of  the grid
area, and point estimates might not  represent the whole grid.

BACKGROUND

     Possibly the most active workers  in  the field of modeling the hydrologic
effect of  COp  warming  have  been  Manabe  and  his  associates  at   the  NOAA
Geophysical Fluid  Dynamics  Laboratory  in  Princeton,  New  Jersey.   They have
developed their own global circulation model (the NOAA/GFDL model), which they
are continually improving and modifying, to provide a closer representation of
real world conditions.  Manabe and Wetherald (1980) used a simple version of a
global circulation model to outline  the global  pattern of soil moisture.  They
suggested that there would  be a high  latitude  region  where the rate of runoff
would increase appreciably along with  a zonal belt of decreasing soil moisture
at  slightly lower latitudes.    There would  be  regions of increased  soil
moisture  along  the east coast  of the  subtropical portion of  the continent.
The  warming of  the atmosphere  predicted by  the  model would  encourage  the
penetration of moist air into high latitudes and  result in large increases in
precipitation there.

     Manabe, Wetherald,  and Stouffer  (1981)  conducted a detailed analysis of
three different  circulation models:    the  S15,  the  G15,  and  the G21.   The
original  S15 model was an idealized section  of land and water stretching from
pole to pole.  It was 120°  wide  at  the equator  (60°  land, 60°  water)  with a
zonal wave number of the retained spectral components  of 15.   G15 had a global
computational domain and more realistic geography with  continents and oceans;
the  G21  was very  similar  to G15,  except that  it had  a  maximum zonal  wave
number of  21.  The  paper showed that  the  zonal mean  value  of  soil  moisture
reduces appreciably in summer  in  two  distinct  zones  in  middle and high lati-
tudes in  response  to the  modeled increase  in atmospheric C02.    The authors
concluded that the summer dryness resulted not  only from the  earlier ending of
the  snowmelt season,  but also  from  the earlier occurrence  of  the  spring to
summer reduction in rainfall rate.  The effect  on the  snowmelt season was more
significant in high latitudes, while the reduction  in  rainfall  rates was more
important in middle latitudes.  Results indicated  a statistically significant
increase  in both soil moisture and rate  of runoff in  high latitudes  in  all
models during all of the annual  cycle  with the  exception of the summer.

     The  authors pointed out that the G15 and  G21  models have a somewhat poor
record of  simulating  current summer  season  precipitation  (in  comparison with
the winter season precipitation).  Further,  the G21  model locates the tropical


                                      253

-------
 rainbelt over  the tropical Atlantic and  Pacific  Oceans south of the equator,
 which  does  not correspond  to  currently observed patterns.   The inability of
 the  models  to simulate  current  conditions  raises  certain  questions about
 simulating double C02 conditions as well.   The tropical rainbelt problem may
 be related  to  the abnormally high  sea  surface temperatures over the tropical
 Southern Hemisphere oceans  in the G21 model.

     Along with the enhanced poleward moisture  transport, a C02-induced summer
 dryness appears in middle  latitudes.   This results from the poleward movement
 of the subtropical dry zone by about 5° latitude during summer.  At this time,
 zonal  mean  soil moisture  is reduced  from 20%  to 60%  around  50° latitude and
 from  1051  to HQ%  around  70° latitude.   The  percentage  increase  in zonal mean
 soil moisture which is found in high latitudes  in all seasons except summer is
 of the order of 60%.   These changes were found assuming a quadrupling of C02
 concentrations rather than  the more conventional doubling of C02.

     In a recent  analysis  of this  same problem of summer soil moisture condi-
 tions  on  a  worldwide  basis,  Manabe  and  Wetherald  (Volume  1)  achieve
 essentially  the same  conclusions using a model that  includes predicted cloud
 cover  conditions.  They find  that the  increased  carbon  dioxide  conditions
 result in a  reduction in soil  moisture in summer over large regions in middle
 and high latitudes (the  North  American  Great Plains,  western Europe, northern
 Canada, and  Siberia).   There  is also  a  winter enhancement  of  soil moisture
 over  large  midcontinent  and high  latitude  land  areas.   While  the  authors
 question some  of  the  details of the climatic changes  that would accompany an
 increase in  C02,  they believe  that the basic  conclusions  of the  paper would
 remain unchanged by any imperfections in the model.

 METHODOLOGY

     Monthly temperature and precipitation data for twelve selected regions of
 the world were obtained  from data  tapes of the GISS and NOAA general circula-
 tion models.  Two different analyses were applied to each set of data.   First,
 the data of  estimated monthly  temperature and  precipitation  that  would occur
with a doubling of C02 in  the atmosphere, as well  as  the actual modeled data
of temperature and precipitation for current conditions (the "control" data),
were entered into the climatic  water budget; and values of the factors of the
water  budget were obtained from  both sets of  data.   Differences  in  such
factors as  potential  evapotranspiration, precipitation,  water deficit,  and
water surplus  were obtained by subtracting  the control  water  budget  factors
from those obtained  from the double C02  temperature  and  precipitation data.
Second,  since   the models do  not predict current  conditions  with  great
accuracy,  we felt that a more  reliable  estimate might  be obtained by applying
 information on  the changes in  monthly  temperature  and precipitation obtained
from the differences  between double C02 conditions and  control  conditions to
current station  conditions as  actually  measured.   The  Center for  Climatic
Research not only  possesses one of the most extensive collections of monthly
 temperature and precipitation  data from  stations  in  all parts  of the world,
but it has also used  those data to provide  computer-generated climatic water
budgets for  each  degree  of latitude and  longitude for all the  land  areas of
 the earth.   Information  on the actual  change in  monthly temperature from the
 control conditions to  the  double C02 conditions was applied  to the presently
 available  data   on   station   temperatures   within   each   selected  region.
 Similarly,  the percentage change in precipitation  between the modeled control


                                     254

-------
conditions  and  the  doubled C02  conditions  was determined  and  the  present
observed precipitation values were adjusted by these percentages to obtain new
precipitation values for  a  COp  doubling.    As  in the  previous case,  the
differences  between  water  budget factors  obtained from the  modified  current
data and  the actual  current data were evaluated.   Thus,  four different esti-
mates of  the effect  of doubling  C02  on  factors of  the  climatic water budget
were considered.  The four  estimates may  be summarized as follows:

     •  GISS  model:  Change  in  T,  percent  change  in  P  applied to  current
                     measured data minus  current data

     •  GISS  model:  Doubled C02  estimated  conditions  minus modeled  control
                     conditions

     •  NOAA  model:  Change  in  T,  percent  change  in  P  applied to  current
                     measured data minus  current data

     •  NOAA  model:  Doubled C02  estimated  conditions  minus modeled  control
                     conditions.

RESULTS

     Tables  1-4  indicate  the  water  budget  results for  the region  in North
America covering southeastern Texas and northern Mexico.   The area studied is
a rectangular geographical region ranging in  size from 10°-15°  of longitude
and  13°-15°  of  latitude depending on  the  model  being evaluated.  •  Data on
temperature and precipitation at  a  number of selected  grid points within the
area  were obtained  either from  our own  file  of  current data or from the
evaluation of the GISS or  NOAA  global circulation  models under control condi-
tions and double C02  conditions.

     Table  1  compares data from  our  current data  files (shown  in  the lower
portion) with data obtained by correcting  the  current data  by  the  change in
temperature and  percentage change in  precipitation  between  a  double C02 event
and  the  control situation  (the model evaluation  of a single  COp situation)
using the GISS  model (shown in  the upper portion).  The amount of water that
can be held  in  the root 2one at field capacity  is  considered to be 150 mm and
water is withdrawn from the soil  according to a linear declining availability
model.   Using  present  data,  average annual potential evapotranspiration is
found'to equal 885 mm in the area with the monthly amounts ranging from 13 mm
of potential evapotranspiration  in January to 154 mm in July.   Annual precipi-
tation totals 707  mm over  the  whole  area with  a  maximum value of  105 mm in
July and a minimum value of 22 mm in January.  Values greater than 50 mm occur
each month  from April through  October and again in December.   Soil moisture
storage  does not reach  field  capacity during  the year on the  average with
highest values  equal to  60 mm of storage in March.  No water is found in the
root zone from July  through November.  As  a  result of these dry conditions, a
water deficit of 178 mm exists  in the area with the period of deficit running
from  April  through  October.   No surplus  of water  can  occur  since  the soil
moisture storage never returns to field  capacity.

     Under  the   computations adjusted  for  the change  in  temperature  and
precipitation resulting  from  a  C02  doubling,  annual 'potential  evapotrans-
piration  increases  to 1150 mm while precipitation decreases to 692 mm.  Peak
                                     255

-------
     Table  1.  Average  Climatic Water  Budget Data  Over Texas-Mexico  Region
               Using GISS Model, 150 mm  Storage  at Field Capacity, and Linear
               Declining Availability of Soil Moisture
GISS model.
Current data modified by change in T and percent change in P.
     Month
Yearly totals:
APE
Prec
St
                                                 AE
                      1150
        692
Def    Surp
J
F
M
A
M
J
J
A
S
0
N
D
12.3
12.6
17.2
21.5
23.5
28.7
30.1
30.4
26.2
23.3
18.0
14.0
18
18
47
86
118
181
196
188
134
95
45
24
15
30
27
49
82
57
99
104
122
41
19
48
23
34
20
5
1
0
0
0
0
0
0
24
17
18
41
64
86
58
99
104
122
41
19
24
1
0
6
23
32
124
97
84
12
55
26
0
0
0
0
0
0
0
0
0
0
0
0
0
                 692
                 458
GISS model area.  Average current data.

     Month     T       APE     Prec     St
Yearly totals:
                          AE
                          Def
885
707
        707
                                                        178
                          Surp
J
F
M
A
M
J
J
A
S
0
N
D
7.7
8.8
12.5
16.9
20.7
24.6
25.9
25.8
23.0
18.1
12.9
8.2
13
16
35
64
101
138
154
145
106
66
32
16
22
28
37
50
71
71
105
94
94
52
32
14
46
59
60
48
24
2
0
0
0
0
0
37
13
16
35
63
95
92
106
94
94
52
32
37
0
0
0
1
6
46
48
51
12
13
0
14
0
0
0
0
0
0
0
0
0
0
0
0
                                      256

-------
     Table  2.  Average  Climatic Water  Budget  Data  Over  Texas-Mexico  Region
               Using NOAA Model, 150 mm  Storage at Field Capacity, and Linear
               Declining Availability of Soil Moisture
NOAA model.
Current data modified by change in T and percent change in P.

     Month       T     APE     Prec     St       AE      Def      Surp
J
F
M
A
M
J
J
A
S
0
N
D
Yearly totals:
15.7
17.9
18.6
21.8
25.3
30.6
30.9
30.4
31.8
26.5
18.4
14.9

26
37
49
81
138
193
199
186
176
132
41
22
1281
22
21
39
26
107
63
66
93
21
55
21
25
560
3
2
1
0
0
0
0
0
0
0
0
4

23
23
40
26
107
63
66
93
21
55
21
22
560
3
14
9
55
31
130
133
93
154
77
20
0
721
0
0
0
0
0
0
0
0
0
0
0
0
0

NOAA model area.  Average current data.

     Month      T      APE    Prec      St       AE      Def      Surp
J
F
M
A
M
J
J
A
S
0
N
D
10.1
11.7
14.9
18.8
22.2
25.6
25.9
25.7
23.4
19.3
13.9
10.3
18
23
44
73
111
146
152
143
107
70
33
18
24
23
22
36
62
77
86
79
89
51
34
31
19
20
9
2
0
0
0
0
0
0
1
13
18
23
32
43
64
77
86
79
89
51
33
18
0
0
12
30
47
69
67
64
18
19
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Yearly totals:         939    613               613     326
                                      257

-------
Table  3.  Average  Climatic  Water Budget  Data Over Texas-Mexico  Region Using
          GISS Model,  150  mm Storage at Field Capacity,  and  Linear Declining
          Availability of Soil Moisture
GISS model.  Doubled C02 data.

     Month      T      APE     Free
                  St
 AE
Def    Surp
J
F
M
A
M
J
J
A
S
0
N
D
Yearly totals:
10.3
11.1
14.5
20.1
21.2
24.0
25.7
26.1
25.6
20.8
14.6
11.5
•
i
18
20
41
83
101
130
152
148
128
80
35
22
957
67
91
83
88
134
125
162
145
72
37
38
63
1105
143
150
150
150
150
146
150
147
92
50
53
94

18
20
41
83
101
130
152
148
128
78
35
22
955
0
0
0
0
0
0
0
0
0
2
0
0
2
0
64
42
5
32
0
6
0
0
0
0
0
150
GISS model area.  Control run data.

     Month      T      APE     Prec
                  St
 AE
Yearly totals:
752    1197
752
Def
 0
Surp
J
F
M
A
M
J
J
A
S
0
N
D
5.7
7.3
9.8
15.5
18.4
19.9
21.5
21.5
22.3
15.6
9.4
5.7
12
17
31
65
92
103
118
112
106
59
25
12
99
88
115
90
115
155
171
133
55
47
63
66
150
150
150
150
150
150
150
150
100
88
126
150
12
17
31
65
92
103
118
112
106
59
25
12
0
0
0
0
0
0
0
0
0
0
0
0
87
71
83
25
23
52
53
21
0
0
0
30
445
                                      258

-------
Table 4.  Average Climatic  Water Budget  Data Over Texas-Mexico  Region Using
          NOAA Model,  150 mm  Storage at Field Capacity,  and Linear Declining
          Availability of Soil Moisture
NOAA model.  Doubled C02 data.

     Month      T      APE     Free      St      AE      Def      Surp
J
F
M
A
M
J
J
A
S
0
N
D
Yearly totals;
9-9
14.2
16.9
23.0
30.0
37.6
35.8
31.0
31.6
23.6
13.0
9.1
•
7
18
36
95
190
216
219
190
175
94
14
5
1260
95
103
115
57
60
29
80
132
23
102
86
70
952
150
150
150
112
7
0
0
0
0
9
80
145

7
18
36
95
165
36
80
132
23
94
14
5
706
0
0
0
0
25
179
139
58
152
0
0
0
554
83
85
79
0
0
0
0
0
0
0
0
0
246
NOAA model area.  Control run data.

     Month      T      APE     Free      St      AE      Def      Surp
J
F
M
A
M
J
J
A
S
0
N
D
4.2
7.9
13.2
19.9
26.9
32.6
30.8
26.3
23.2
16.4
8.5
4.5
3
10
34
81
162
205
199
150
105
50
12
3
103
114
64
80
35
35
104
112
97
95
139
85
150
150
150
150
29
0
0
0
0
45
150
150
3
10
34
81
156
64
104
112
97
50
12
3
0
0
0
0
6
141
95
38
7
0
0
0
100
104
30
0
0
0
0
0
0
0
22
82
Yearly totals:        1013    1063              725      287      338
                                      259

-------
summer potential  evapotranspiration increases from  154  mm to  196  mm in July
while  the  maximum of  precipitation shifts to September  with  122  mm.   Under
these  altered  conditions,  only   the  months   from  May  through  -September
experience more than 50 mm of  precipitation.   The moist season is shorter and
displaced to late summer.  With  higher potential evapotranspiration and lower
precipitation,   soil  moisture  storage  is  considerably  less with  the maximum
storage of 34 mm occurring in February.  The deficit increases markedly to 458
mm annually  from  the  current value of 178 mm, and all  months except February
and December show some deficit.  No month experiences a surplus.

     Table 2 provides the same comparison for the NOAA model.  While one might
expect  the  water  budget based  on   current  data  in  the area   to  provide
essentially the same results as  found  for current conditions in Table 1, some
differences are found because  the  actual areas evaluated by the GISS and NOAA
models differ.    In  an area with  variable conditions such  as those  in  the
Texas-Mexico region,  this can lead  to differences in average  areal tempera-
tures and precipitation.  Table 2  (lower portion) shows that current potential
evapotranspiration over the  NOAA area  equals  939 mm compared with 885 mm over
the GISS area.  Precipitation  is somewhat lower, equaling only 613 mm for the
NOAA area compared with 707  mm for the GISS area.  As a result, soil moisture
storage  is  much  lower  for  current  conditions  under the  NOAA model  and  the
deficit is greater (326 mm vs 178 mm in the GISS model).

     Applying the corrections for changes in temperature and precipitation due
to a doubling of C02 using the NOAA model, annual potential evapotranspiration
increases  to   1281  mm   while   precipitation  decreases   to  560  mm  (upper
portion).   Essentially no water is stored  in  the soil in  the  area under the
increased C02  conditions, the  deficit  increases to 721  mm and,  as before,
there  is no  surplus in any month.  The  NOAA model postulates  very dry condi-
tions for a C02 doubling with precipitation some  132 mm less than given in the
GISS model and potential evapotranspiration 131  mm greater.  Both changes work
together to result in much drier soil conditions.

     A brief glance at  Table 3 reveals one of the significant  problems of
using  the data  from the  global  models directly.  The control  portion of the
table  provides  the model  estimates of  current  conditions given by the GISS
model.  Average annual  potential evapotranspiration is 752 mm (compared with
885 mm from actual  current conditions  in Table  1) while average precipitation
has increased 1197 mm from 707 mm  in Table  1.  The GISS model predicts current
conditions that are much wetter and somewhat cooler than actually found in the
area.  All but one month in the year have precipitation values greater than 50
mm; and  soil moisture storage  is  at  field capacity  (150 mm)  in  nine of the
twelve months,  including  all summer.   Only in  the  September-November period
does  soil  moisture storage  under  these  modeled conditions drop  below field
capacity.  As a result of the high precipitation and soil moisture conditions,
no deficit occurs, while annual surplus equals 445 mm.  The GISS control model
clearly does  not  represent  current conditions  in the area.   Since starting
conditions are unrealistic,  one  cannot rely on  the  absolute value of projec-
tions which show a warming of  temperature and  an increase  in potential evapo-
transpiration from  752  mm to  957  mm.    Precipitation  decreases slightly from
1197  mm  to  1105  mm.   As  a  result, soil moisture storage  is  slightly drier
(only five months with storage at field capacity) and 2 mm of deficit occur in
October.   Surplus  decreases  to 150 mm from 445 mm.  The estimated conditions
                                      260

-------
 for a  C02  doubling are  much more moist  than  current conditions  because  the
 modeled control conditions were initially so moist.

      Table  4 shows that the NOAA model also  predicts  rather  wet conditions in
 the Texas-Mexico area  for the control period.  Average annual potential evapo-
 transpiration equals 1013 mm while precipitation equals  1063  mm.   All but  two
 months  have  values  of precipitation over 50 mm.   As a  result,  soil  moisture
 storage equals field capacity  in  six months of the year, although it is zero
 in four months.   This  rapid change  from  field capacity to no  water in  the root
 zone results in large  modeled values for both  deficit and surplus (28? mm  and
 338 mm, respectively)  while  actual current  data reveals a  deficit of  178 mm
 and no  surplus.

      The NOAA model  projects  that with a  C02  doubling, annual  potential  evapo-
 transpiration equals 1260  mm  and  precipitation drops to 952 mm.   Only three
 months  have soil moisture at field capacity.   The deficit increases to  554 mm
 from 287 nan  and the  surplus,  which still  exists,  drops from  338 mm to  246
 mm.   Since  the  NOAA model predicts  rather  moist  starting conditions as does
 the GISS model,  it provides double  COp event  conditions that  are probably more
 moist  than  they  should  be.   The  difference between  the starting  and  ending
 conditions  may,  however,  be indicative of  the  type of changes  resulting from
 an  increase  in COp.   We  have, therefore,  concentrated on the  differences  or
 changes in  the water budget  factors  between  either current or  control  condi-
 tions and  doubled COp conditions  to eliminate some of the  errors  due to  the
 inability of  the models to  represent  current  conditions.

     We evaluated water budgets for twelve selected areas {shown  in Figure 1)
 in  different  climatic  regions of  the  globe,  using  both the  GISS and NOAA
 models  and   the  two different evaluation  techniques  described above.    The
 annual  values of the water budget  factors  of prime interest  (PE,  P,  Deficit,
 Surplus) are  summarized in  Tables  5 and  6.  Table 5 provides  information from
 both  the NOAA and GISS models  based  on  the differences between average water
 budgets evaluated using  both  current temperatures and percentage changes   in
 precipitation.   The results show an  increase in potential evapotranspiration
 in  all  regions investigated,  the NOAA values ranging from Just  over 100 mm  in
 north central Siberia to  just 400 mm  in northeast  Brazil.   The GISS model
 estimates a  larger range  of increases, from 75  mm in  north central  Siberia  to
 nearly  450 mm in  northeast  Brazil.   Agreement is quite reasonable  between  the
 two models.

     Some investigators have  suggested that  the increase  in  COp in  the atmos-
 phere will result in greater changes in temperatures in high latitudes than  in
 low  latitudes.   The picture is one of significant  polar  warming.  While this
may be  true,  it does not necessarily mean that potential evapotranspiration  in
 high  latitudes will increase more  than  in  low  latitudes.    Since potential
 evapotranspiration is zero until mean monthly temperatures exceed about 0.5°C,
 increasing air temperatures that are well  below freezing will  not result   in
any  increase  in potential  evapotranspiration,  while  such temperature  changes
 in  lower latitudes  with   temperature  well  above   freezing   will   result   in
appreciable  changes  in  potential   evapotranspiration.    Because of the  cold
monthly  temperatures at  high latitudes,   increases  in temperature  due to  C02
 increases may have  no  significant  influence  on the  potential evapotranspira-
 tion in winter.  Precipitation generally  increases as a result of the increase
                                     261

-------
ro
                             Figure  1.   Annual  Change  in  the  Moisture  Index  (I  )--GISS  Model

-------
    Table 5.  Annual Water Budget Factors for Selected Regions Computed From
              NOAA and GISS Global Climate Models.  (Water budgets computed
              using current  T + AT, current P + /SAP - current T,P).

Location
North Central Siberia
South Central Canada
Upper Midwest (USA)
Pacific Northwest
Ukraine (USSR)
Southeast China
Texas and North Mexico
West Central Africa
Northeast Brazil
Southeast Australia
Southern Africa
Argentina (Pampas)
APE Prec. Deficit Surplus
NOAA GISS
119 75
138 106
255 149
171 122
153 97
143 298
342 265
347 426
380 442
248 303
299 332
191 363
NOAA GISS
55 70
208 54
-8 28
62 92
98 132
298 103
-53 -15
95 221
237 -164
-20 53
-12 66
134 291
NOAA GISS
63 4
-68 52
251 75
142 61
85 -26
14 0
395 280
253 205
155 156
267 251
311 266
57 72
NOAA GISS
0 0
2 0
-13 0
32 32
29 13
168 -195
0 0
-1 0
12 -451
0 0
0 0
0 0
Table 6.
Annual Water Budget Factors for Selected Regions Computed From
NOAA and GISS Global Climate Models  (Water budgets computed
using double carbon dioxide T + P - control T + P)

Location
North Central Siberia
South Central Canada
Upper Midwest (USA)
Pacific Northwest
Ukraine (USSR)
Southeast China
Texas and North Mexico
West Central Africa
Northeast Brazil
Southeast Australia
Southern Africa
Argentina (Pampas)
APE Prec. Deficit Surplus
NOAA GISS
82 78
146 104
252 138
151 126
174 96
113 288
247 205
168 528
198 347
360 274
269 277
185 291
NOAA GISS
131 151
298 114
-7 60
227 236
151 164
19 100
-111 -92
84 263
-56 -123
-80 110
-11 159
113 114
NOAA GISS
-2 -5
-42 0
171 5
81 7
113 -19
-2 4
267 2
84 246
254 471
412 66
244 1
78 177
NOAA GISS
48 68
109 10
-88 -74
157 117
88 49
-95 -184
-92 -295
0 -19
0 0
-27 -106
-36 -117
5 0
                                     263

-------
in COp in  the atmosphere.  Increases are found in eight of the twelve regions
according to the NOAA model and  ten  of the twelve regions on the basis of the
GISS model.   Only the Texas-Mexico  region shows a  precipitation  decrease in
both models, while  the  greatest  discrepancy  occurs in  northeastern Brazil
where the NOAA model calls for a 237-mm increase in precipitation and the GISS
model  forecasts  a  164-mm decrease  in precipitation.    The  NOAA  value  must
result partly  because the  model  locates  the intertropical  convergence  zone
south of the equator under control conditions.

     Even though  precipitation  would  generally  increase as  a result  of the
greenhouse warming, this additional water is less than the total that would be
evaporated by  the  increased potential  evapotranspiration.   As a  result, the
annual deficit would  increase  in all regions  except  south  central  Canada
(NOAA) and  the Ukraine  (GISS).    Some  of the  increases in  deficit  would be
substantial  (395  mm  in   the  Texas-Mexico  region,  311  mm  in  South  Africa),
although the  southeast  China region shows  hardly any  increase in deficit at
all.   Seven  of the twelve NOAA  regions show  either  no change in  surplus  or a
decrease, while ten of the  twelve  GISS regions also show no surplus change or
a decrease.   Two  regions show quite conflicting  results in terms of surplus.
In  southeast  China,  the  NOAA  model  shows an  increase  in surplus  (because of
the great increase in precipitation) while it shows a decrease  in surplus with
the GISS model.  Similarly, northeast Brazil also shows an increase in surplus
with  the NOAA model  and a significant decrease with  the  GISS model.   This
difference  can be  related  directly to  the  modeled values  of precipitation
which are quite different.

     Table 6 is similar  to Table 5 except that it presents data obtained from
differences in water budget factors determined from modeled control conditions
and double C02 conditions for both the NOAA and GISS models.  Potential evapo-
transpiration  increases  in all  regions  sampled.   Precipitation  increases in
all but  five regions using the  NOAA model  and all  but  two  regions using the
GISS model.   Deficit  increases  in all but  three  regions using the NOAA model
and all  but  two regions  with  the GISS  model.   Again a decrease in the deficit
is  found in  south central Canada  as in Table 5 in  the  NOAA  model and in the
Ukraine  in the GISS model.   North central Siberia experiences  small decreases
in deficit according  to  both  the NOAA  and GISS models.  The surplus decreases
or  does  not change  in  seven  of the twelve  regions with the  NOAA model and
eight of the twelve regions with the GISS model, as might be expected with the
increase of  potential evapotranspiration.   A noticeable increase in dryness
occurs in all  regions except north central Siberia  and south central Canada,
the  two  most  poleward  regions.   The  Southern  Hemisphere  regions  sampled
exhibit a strong tendency for increased dryness.

     The pattern of increased dryness found in most  regions on  an annual basis
is again found if the data for only for the three summer months are considered
(Table 7).   Decreases  in  the deficit are found  in the Siberian  and south
central  Canada  regions  according  to the  GISS and NOAA models, respectively,
while  a  marked decrease  in deficit is found  in  both the NOAA and GISS models
in west central Africa.  The Ukraine and Argentine Pampas also  show a decrease
in summer deficit using the NOAA model.

     The  Thornthwaite-Mather  water  budget  permits  the  development  of  a
moisture index  (Im),  of  the  relative  moisture or dryness  of a climate,  from
a  simple comparison of annual precipitation with potential evapotranspiration


                                     264

-------
      Table 7.  Summer Water Budget Factors for Selected Regions Computed
                 NOAA and  GISS  Global  Climate Models.  (Water budgets computed
                using current  T + AT,  current P + %L? - current T,P).

Location
North Central Siberia
South Central Canada
Upper Midwest (USA)
Pacific Northwest
Ukraine (USSR)
Southeast China
Texas and North Mexico
West Central Africa
Northeast Brazil
Southeast Australia
Southern Africa
Argentina (Pampas)
APE Prec. Deficit Surplus
NOAA GISS
38 9
20 38
159 57
85 45
41 7
34 109
138 128
78 85
116 124
105 107
120 146
91 157
NOAA GISS
29 35
121 23
-49 7
-7 29
23 63
100 2
-19 -10
104 158
-217 12
80 38
7 15
111 158
NOAA GISS
43 -11
-61 17
222 23
127 27
77 -33
0 0
157 159
-33 -33
0 0
55 69
113 130
-18 23
NOAA GISS
0 0
0 0
0 0
0 0
0 0
82 -107
0 0
-1 0
-289 -99
0 0
0 0
0 0
 (Im  =  100[{P/PE)-1]).   The data of average annual precipitation and potential
 evapotranspiration  for  each of the twelve  regions,  computed by both the NOAA
 and  GISS  models,  have been used  to determine the moisture index on the basis
 of both current data and data adjusted for modeled changes  in temperature and
 precipitation.  Since  the actual  value of the moisture index obtained in this
 way  depends on the  magnitude  of the input data, which varies greatly with the
 particular  circulation model,  it was  felt  that  only the  difference  in  the
 value  of  the  moisture  index  between current  and modeled  future conditions
 should  be considered.    This would  still  permit  evaluation  of  whether  the
 climate was  becoming  relatively  more  moist  or  dry and  it would  allow  the
 results of  the  two models  to  be  compared  even  though current  input data were
 quite different.   Figures  1 and  2  provide  information on the  relative change
 in the  moisture  index  for  each of the  twelve  regions based on  the GISS and
 NOAA models,  respectively.   In  half  of the cases,  relative  changes  in  the
 moisture index between current and modeled C02 conditions are quite small (ten
 units or less).   In seven of  the  twelve areas,  both models show the same type
 of change  in  the  moisture  index  and  in all seven of  those  cases  both models
 indicate a  shift  to drier  conditions.   In none  of the  twelve  areas  do both
 indices indicate a shift  toward more moist conditions.  Areas where one of the
 models  indicates  a  shift  to more moist   conditions include  north  central
 Siberia, south central Canada, Ukraine, southeastern  China, and west  central
 Africa.   The  shift  to  drier  conditions is most  clearly marked in  the  upper
midwest of the United States,  the Texas-Mexico area, and northeastern Brazil.
                                     265

-------
Figure 2.   Annual Change in the Moisture Index (Im)--NOAA Model

-------
     Mather (1978) investigated the  relation  between the water budget factors
of annual potential evapotranspiration and the moisture index, and the distri-
bution of natural  vegetation  in the United States  and  Canada.   He found that
within  well-defined  ranges  of  potential  evapotranspiration  and  moisture
indices, clearly  identified  natural vegetation  associations exist.   Little
overlap of vegetation types was found except in the oak-chestnut, hickory, and
pine forest regions and in certain dry semiarid vegetation regions.  Using the
values of annual potential evapotranspiration and moisture index obtained from
each of the circulation models under current and modeled C02 conditions, it is
possible to  predict  the  changes  in natural  vegetation that  might accompany
each of the climatic changes  that  is forecast by the models.  Figures 3 and 4
indicate the  nature of  the vegetation  changes for  each  of the  study areas
using data from the GISS  and  NOAA  models,  respectively.  With the NOAA model,
five of the  twelve areas experience no change  in vegetation.   In all but two
or three cases,  the  changes that do occur result from both  a  warming of the
climate and  a general  increase in  dryness.     Certain change  arrows end in
regions on  the diagram  without any vegetation indicated merely  because the
diagram  was  constructed  for  United  States  and  Canadian   vegetation  (plus
tropical rainforest)  so that vegetation conditions in other possible ranges of
potential evapotranspiration and Im were not sampled.

CONCLUSION

     The present  study  has had two  main goals:   to evaluate  the   effect of
increased atmospheric C02 on  factors  of  the water  budget in twelve  selected
regions of  the world;  and to  evaluate differences and  similarities of two
different global  models  that  have  been used  to provide  estimates of future
climatic conditions and  to  consider  the  results from the different techniques
for evaluating  the data  in order to understand better  the problems of trying
to estimate future hydrologic conditions.

     Versions of the NASA/GISS model and the NOAA/GFDL model have been used to
provide data of control  conditions as  well as double C02 conditions.  Because
the control conditions differ significantly from reality, it was felt  that the
focus of the study should be on differences rather than absolute values of the
factors of  temperature  and precipitation  provided  by  the models.   Thus, the
two  techniques  used with each  model  involved  obtaining  the  differences
between  double C0?  conditions and  control conditions; and   current  existing
conditions and current conditions as modified by the actual change in  tempera-
ture and percentage  change in precipitation found  from the  operation of each
model.

     The results  show that the models and  analysis  techniques do not provide
similar estimates  of  changes  in different parts of  the world, although there
is a general undercurrent of agreement in a majority of the regions.   Tempera-
ture and hence potential  evapotranspiration  is predicted to  increase in all
twelve  regions while precipitation  is  expected to  increase in most of the
regions.   Since  the  climatic demand  for  water is expected  to increase more
than  the  supply  of  water  by  precipitation in  most of the  regions  studied,
there  is  a  tendency  for  most regions  to show an  increase  in annual water
deficit, a  decrease  in  annual water  surplus,  and  a decrease  in summer soil
moisture storage.   Both  models  show exceptions  to  these conclusions—for
example,  in  the   Pacific  Northwest  (surplus),  Ukraine  (surplus)  and  west
central  Africa  (summer soil moisture storage).  In a  few other cases, one of


                                      267

-------
            200-
            400-
            MO-
            800-
            1000 -
            1200 -
            1400-
            1800-
            1800
                                          Oak-chMtmit
                                           -hickory
                                           -pint form*
                               •JO       20       60

                                    Moimrt indM. Im
          1 - North/central Siberia
          2 - South/central Canada
          3 - Dppec midwest (USA)
          4 - Pacific northwest
          5 - Ukraine (USSR)
          6 - Southeast China
7 - Texas and  N.  Mexico
8 - West/central  Africa
9 - Northeast  Brazil
10 - Southeast Australia
11 - Southern  Africa
12 - Argentina (Pampas)
Figure 3.   Predicted Changes  in Natural Vegetation in  Selected Regions
             as a  Result  of Increased  Carbon Dioxide—GISS Model
                                         268

-------
            1
                200-
                400-
               600-
               800-
1000-
               1200 -
               1400-
               1600 -
               1800
             1 - North/central Siberia
             ? - South/central Canada
             3 - Upper  midwest (USA)
             4 - Pacific  northwest
             5 - Ukraine  (USSR)
             6 - Southeast China
                           7 - Texas and N. Mexico
                           8 - West/central Africa
                           9 - Northeast Brazil
                           10 - Southeast Australia
                           11 - Southern Africa
                           12 - Argentina (Pampas)
Figure  4.  Predicted Changes in  Natural  Vegetation in. Selected Regions
            as  a Result  of Increased Carbon Dioxide—NOAA Model
                                     269

-------
 the  models shows a  difference  from the  foregoing  generalizations.   The most
 marked  problem areas,  where discrepancies  are greatest,  appeared to  be  in
 southeast  China and  northeast Brazil and probably result from quite different
 control or current modeled conditions.

     Consideration  of possible  vegetation  changes  that might  accompany the
 predicted  climatic  changes calls  for  changes in natural  vegetation in about
 two-thirds of  the twelve regions studied.  The changes result in part from the
 significant  warming  that  will occur  in  every  region and  in  part  from the
 general  increase in dryness  in   spite  of  the  predicted small  increases  in
 precipitation.

     The  inability   of  the models to provide  a good description  of current
 conditions  has  been  a  major  drawback  to the  study.    In   an   effort   to
 deemphasize  this  problem,   differences   rather  than  absolute  values  were
 investigated but  it  is  clear that in certain regions where small movements  in
 circulation  belts  can  result  in  large  differences  in  climatic  conditions
 (e.g., northeast  Brazil),  even  the use of differences may not produce data  of
 great reliability.   However,  the  use of the climatic water budget to evaluate
 the  combined effect  of  changes in both temperature and precipitation makes  it
 possible  to obtain  a more  rational picture of how  increases  in  COP  might
 affect  such hydrologic  factors as  soil moisture  surplus (and  hence  stream
 runoff)  water  deficit  and summer  soil  moisture  storage.   Because  of known
 relations  between water budget factors  and  natural  vegetation,  some estimate
 of  how  increased C02 will modify  the distribution of  natural  vegetation  is
 also possible.  The  picture is not necessarily bleak but it suggests a general
 increase  in  dryness  that  might lead  to  changes in vegetation  toward  a more
 drought-tolerant  type.  Marginal areas will be more greatly affected, although
 some modifications of current moisture relations and hydrologic conditions can
 be  expected  nearly   everywhere.   It is  likely  that  some of our  better agri-
 cultural areas will  experience less favorable conditions in .the future.  These
 water budget studies need  to be expanded to other areas  and the data need  to
 be  evaluated  at  particular   points  rather than as  averages over  large geo-
 graphic areas  if we  are  to  understand the  real nature  of the  changes  to  be
 expected with increased concentrations of C02 and other greenhouse gases.
REFERENCES

Manabe, S., and  R.T.  Wetherald. 1980.   On  the Distribution of Climate Change
     Resulting from an  Increase in COp  Content of  the Atmosphere, Journal of
     the Atmospheric Sciences. 37:99-118.

Manabe, S., R.T. Wetherald, and R.J. Stouffer. 1981.  Summer Dryness Due to an
     Increase of Atmospheric C02 Concentration, Climatic Change. 3:347-85.

Manabe,  S.,  and  R.T.  Wetherald.   1986.    Reduction  in  Summer Soil  Wetness
     Induced by an Increase  in  Atmospheric  Carbon Dioxide,   Science. 232:626-
     28.

Mather,  J.R.    1978.    The Climatic Water  Budget  in  Environmental Analysis.
     Lexington, MA:  Lexington Books, 239 pp.
                                     270

-------
Mather, J.R.   1981.   Using Computed Stream  Flow in Watershed Analysis, Water
     Resources Bulletin. 17(3) :474-82.

Thornthwaite, C.W., and  J.R.  Mather.   1955,   The Water Balance, Publications
     in Climatology,  Laboratory of Climatology, 8:1-104.
                                      271

-------
HEALTH

-------
The Impact of Human-Induced Climatic
Warming Upon Human Mortality: A New York
City Case Study

Laurence S. Kalkstein, Robert E. Davis
Jon A. Skindlov, Kathleen M. Valimont
Center for Climatic Research
University of Delaware
Newark, Delaware USA
ABSTRACT

     The  goal  of  this  study  is to determine if weather has an  impact  on
mortality  in New York City and to ascertain whether expected  future climatic
warming will alter the death rates  significantly.   Summer  weather appears  to
have a  significant  impact  on  New  York's  present  mortality rates,  and  a
"threshold  temperature"  of  92°F was determined,  suggesting  that mortality
increases quite rapidly when the maximum temperature exceeds this value.  Days
with high  minimum  temperatures, long  periods  with  temperatures  above  the
threshold,  and low  relative  humidities appear  to  increase  mortality  most
dramatically.  Five climatic scenarios were developed  to estimate New York's
future  weather assuming  that  warming  does  occur,  and  "acclimatized" and
"unacclimatized" mortality  rates  were  estimated  for  each  scenario.   The
unacclimatized  rates  were  computed by  using New York's  weather/mortality
algorithm  developed  from  the  historical analysis.   Acclimatized  rates  were
computed  by selecting  present-day "analog  cities" which resemble  New York's
predicted  future weather and  by  developing weather/mortality algorithms  for
them.

     Results indicated that the number  of  additional  deaths at temperatures
above the  threshold  could  increase  by  over  tenfold  if New  Yorkers  do not
become acclimatized to  the warming.  The elderly will constitute an increasing
proportion  of  these  deaths.   However,  if full acclimatization  occurs,  the
number  of  additional  deaths  above  the threshold  temperature  could be  no
different  than today.    No similar relationships  were discovered for  the
winter,  and the data suggest  that  any  changes  in  winter  weather  will  have
minimal   impact   on  New   York's   mortality   rates.      A   preliminary
precipitation/mortality  study  was undertaken, and summer days  following  a
precipitation  event had significantly lower mortality  rates than summer  days
without precipitation.   In  the winter  these results were  reversed,  and  days
with rain  (but no  snow)   had  significantly  higher  mortality  rates  than
nonprecipitation days.


                                   275

-------
 INTRODUCTION

     A procedure  has  been developed recently  to  evaluate the impact of long-
 term  climatic  warming  on  inter-regional  variations   in  human  mortality.
 Fifteen  cities around  the  country are  presently being  evaluated,  possible
 future  climatic  scenarios  are  being  developed  for each,  and  estimates of
 possible weather-related changes in mortality are being calculated.

     The objective of this report is to describe our procedure and to apply it
 to one of  our fifteen cities,  i.e., New  York City, New  York.   The impact of
 present-day weather  on  New  York's  present mortality rates  is  discussed,  and
 estimates are presented describing  the potential impact of climatic warming on
 New York's future mortality.

     Although no  previous study has attempted  to  predict the impact of future
 weather  changes on mortality,  there has  been considerable work  relating to
 present  climate/mortality relationships.   For  example,  studies  at the Centers
 for Disease  Control  have identified a  number of  factors that  may accelerate
 the onset  of  heat stroke,  including  decreases  in use  of  air conditioning,
 consumption of  fluids, and  living  in well-shaded  residences (Kilbourne et al.
 1982).  However,  other researchers have found that many causes of deaths other
 than heat stroke  increase during extreme weather (Applegate et al. 1981; Jones
 et al.  1982).  In addition, it has been shown that mortality  attributed to
 weather  varies  considerably  with  age,  sex,  and  race, although  there  is
 disagreement  among researchers  in  defining the  most  susceptible  population
 group  (Oechsli and Buechley 1970;  Bridger, Ellis,  and Taylor  1976;  Lye  and
 Kamal  1977;  Jones et al. 1982).  The impact of cold weather is less dramatic
 than hot weather,  although mortality increases have been noted  during extreme
 cold  waves  (Centers  for Disease Control 1982;  Fitzgerald  and  Jessop  1982;
 Callow, Graham, and Pfeiffer 1984).

     This  study  will  incorporate  some  approaches  used  in previous  studies
 while offering  a  new approach  to account  for  potential changes in mortality/
 weather relationships that might be attributed to acclimatization.

 PROCEDURE

     A very  detailed mortality data  base  is presently  available from  the
National Center for Health Statistics (NCHS), which contains records for every
 person who has died  in  this  country  from 1964-present  (National  Center  for
Health Statistics 1978).   The data contain information such as cause of death,
place of death, age  of  death,  date of death,  sex,  and  race.  These data were
extracted  for  the New   York Standard  Metropolitan Statistical  Area for  11
years: 1964-66, 1972-78, and 1980 (during intervening years,  a  sizable amount
of information was missing from many records).  The number of deaths for each
day were  tabulated and  divided into categories  of total deaths and  elderly
deaths (65 years  and older).   These daily death totals  were standardized to
 conform to a hypothetical "standardized city," which contains fixed population
 characteristics (Table 1).   The death  rates for New York were  adapted to  the
population characteristics of  the standardized city to  conform to  procedures
 commonly found in  the  epidemiological   literature (Mausner and Bahn  1974;
Lilienfeld  1980).   The  advantages of  this  standardization  procedure  are
 twofold.   First,  when  the  study  is   extended  beyond  New York,  inter-city
                                     276

-------
         Table 1.   Population Characteristics of the Standardized City
              Total Populat ion             3,811,000
                      Male                 1,833,000
                      Female               1,978,000
                      White                2,817,000
                      Non-White              993,000
                                Age Groups
                       0-4  Years            263,000
                       5-17 Years            759,000
                      18-24 Years            493,000
                      25-44 Years          1,122,000
                      45-64 Years            764,000
                         65 Years            411,000
comparisons will be feasible since demography  is kept constant.  Second, if a
city  has  grown rapidly  during  the study  period,  the bias  introduced  by the
increase  in deaths  that  are  due  to  population growth  is  eliminated,  and
changes  in  mortality  attributed to  environmental   factors   can  be  better
assessed.

     Apparently weather  does have some impact on daily  mortality (Figure 1).
During the  heat wave  of late July 1980 in New York,  standardized deaths rose
by  over  50% above  normal on the day  with the highest maximum temperature.
Elderly deaths  showed  similar  increases.   In  this  study,  daily  changes  in
mortality were  compared to nine  different weather  elements  that  might have
some influence on death rates (Table 2).

     Initial observations of daily standardized deaths vs.  maximum temperature
suggest that weather  has an impact only on the warmest 1056-2051  of the days;
however,   the  relationship  on   those  very  warm  days  is  impressive  (see
Figure 2).  Figures similar to Figure 2 were  developed to  compare the maximum
temperature on  the day  of the  deaths,  as well  as  one, two,  and  three days
prior to  the day  of deaths to determine  if a  time-lag exists between weather
and the mortality  response.   In the case  of New  York,  there is a one-day lag
between weather and mortality.  In addition, a "threshold  temperature," which
is  the   maximum  temperature  above  which   mortality  increases,   can  be
determined.  The threshold temperature  can be  calculated objectively by using
a sum of  squares technique (Kalkstein  1986).   The  threshold  temperature for
total  deaths  in  New  York  was  92°F;  mortality increased dramatically  at
temperatures above  this level.    This  procedure  can be repeated for winter,
where  the  threshold  temperature  represents  the  temperature  below  which
mortality increases.
                                     277

-------
                                                          -IOO
               10 11  12 13 14  15  16  17  18  19  2O 21  22 23 24
                               DATE {JULY  198O)
           	TOTAL MORTALITY  	ELDERLY MORTALITY	TEMPERATURE

      Figure  1.   Mortality During a 1980 Heat Wave in New York City


         Table 2.  Weather Variables Used in the Mortality  Study

                          Maximum Temperature
                          Minimum Temperature
                          Maximum Dewpoint
                          Minimum Dewpoint
                          Heating Degree Hours  (HDH)*
                          3AM  Visibility
                          3PM  Visibility
                          3AM  Wind  Speed
                          3PM  Wind  Speed
HDH  is  calculated  by  determining  the  total  number of  degrees  that the
temperature  is  above 90° for  the  day.   If the  temperature exceeds 90° for
2 hours on  a given day, HDH  is calculated as the sum of  the degrees above
90 for those 2 hours.
                                   278

-------
       >-  275
       ^  250

       8225
       ^  200
       £  175
      O
      on
      <
      O
      z
      to
150

125

100

 75
                     65       75       85       95

                 MAXIMUM   TEMPERATURE (°F)
            Figure 2.  Daily Summer-Season  Standardized Mortality
                       vs.  Maximum Temperature:   Mew York One Day Lag
     Once  the  threshold was established, a  multiple  regression  analysis was
performed  using  the weather  elements  described previously to  determine the
weather/mortality relationship for  days above the threshold temperature.  When
a  statistically  meaningful  relationship was determined,  an algorithm  was
developed  and  used  to  predict   the  expected  increases  in  mortality  at
temperatures above the threshold.

     The next step was an attempt to estimate changes in New York's mortality
that might  occur  with the predicted climatic warming.   In consultation with
EPA and the NASA-Goddard Institute  for Space  Sciences, investigators developed
future  weather  scenarios for  New  York  by  adding  temperature  increments  to
existing historical New  York  temperatures.   These  scenarios were created for
the period recorded by adding 1°,  2°, 4°, 5°, and 7°F to the existing weather
data.   This produced  an approximation  of what  New York's temperature regime
could be  over  the  next  100 years.    New  mortality  estimates were created, for
each of the  temperature  increments  by using  the algorithm developed from the
historical data evaluation.
                                     279

-------
     When measuring the impact of warming on future mortality,  the question of
acclimatization had to  considered.   Will New Yorkers react to  heat as they do
today,  or  will  their reaction  be  similar  to people  who presently  live in
hotter  climates?   There  is  much  disagreement  in the  literature concerning
human  acclimatization  to changing  weather.   Some  research  indicates  that
acclimatization  responses are very  rapid (Rotton  1983);  others think that it
is  a much  slower  process (Ellis 1972;  Kalkstein  and Davis  1985),  and  a few
suggest that virtually  no acclimatization  occurs  at all (Steadman, 1979).  It
is  obvious  that  the  full  range of  possibilities must  be examined  in  this
study.  First, the historical algorithm  that was developed from the previously
described multiple regression  procedure  was  applied  to the  future weather
scenarios with  the incorporated incremental  increases  in temperature.   The
mortality increases estimated  from this  procedure  imply  no  acclimatization
because an  assumption is made  that New Yorkers  will respond  to  heat  in the
future in much the same way that they do today.  Second, analog cities for New
York were established to account for full acclimatization.   For example, by
adding the temperature increment to New York's present temperature regime, its
weather will  approximate another city's present  weather  in  the U.S.   A 58F
increment added  to  New  York's present summer temperatures will yield a regime
approximating that  of Norfolk,  Virginia, today.   Since  Norfolk residents are
fully acclimatized  to this regime,  the  weather/mortality algorithm developed
for  Norfolk can  be utilized for New  York  to account for full acclimatization
when New York's temperatures rise by 5°F.

     Present-day analogs  to account for full acclimatization were selected for
New  York  for the  1°,  2°, 4°,  5°,  and  7°F increments,  and  mortality  models
similar to  the  one described for New York  were created  for them.  The analog
cities were determined  by computing for the three summer months (June,  July,
August) mean maximum  temperatures,  mean  minimum temperatures, and mean number
of days with maximum  temperatures over  90°F for over 100 cities in the United
States.  The city that best duplicated New York's regime was established as an
analogue city.   This was achieved  objectively using a  variety of statistics
for model evaluation (Willmott et al. 1985).

     Figure 3  illustrates the hypothetical  differences  expected  in mortality
with full and no acclimatization.  It is probable that the warmer analogs will
show smaller  increases in  mortality than  the original New  York model  since
residents are already acclimatized to the increased warmth.  Thus, for warming
scenarios of 7° or  more,  the  differences in predicted deaths between full and
no acclimatization situations may be  very  large (area hatched between lines 1
and  2).    In  certain  cases,  it is  possible  that  no  extra deaths will be
predicated  for full acclimatization,  as residents will  be  conditioned  to hot
weather.  For example,  in Jacksonville,  Florida,  heat waves appear to produce
no extra deaths (see Figure 4).  The relationship is so poor that it is almost
impossible  to determine a threshold temperature.
                                      280

-------
                                                    LINE 1
                                                    LINE  2
                     WARMING SCENARIO
                (degrees above the baseline)
    LINE 1:  PREDICTED DEATHS WITH NO ACCLIMATIZATION.

    LINE 2:  PREDICTED DEATHS NITH FULL ACCLIMATIZATION.

   [T77I '•  PREDICTED DEATHS WITH PARTIAL ACCLIHATIZATION,
Figure 3.  Expected Increases in Mortality  in  the Target City
          for Different Warming Scenarios
                            281

-------
    175
    150
    125
O
5  100
2   75
     50
g»
co
      0
           70       80       90       100
           MAXIMUM  TEMPERATURE (°F)
      Figure 4. Daily Summer-Season Standardized Mortality
              vs. Maximum Temperature:  Jacksonville One Day Lag
                        282

-------
RESULTS

     The  multiple  regression analysis  to  determine  those  weather elements
having  the greatest  impact  on present-day  mortality  in New  York produced a
surprisingly  strong  relationship (Table  3).   Minimum  temperature,  maximum
dewpoint,  and  heating  degree  hours  (HDH)  were  all   highly  statistically
significant and explained  almost  66%  of the variance in mortality at tempera-
tures above  the  threshold.  The  most  offending days appeared to possess high
minimum temperatures, high HDH values, and low d«wpoints, indicating that hot,
dry  conditions in Hew York appear most  conducive  to rises  in mortality.  The
results  from the  evaluation  of  the  elderly were similar,  and  the explained
variance was slightly higher.  Thus, it appears that predictive algorithms can
be  developed to  estimate mortality   in  New York at  temperatures  above  the
threshold.  These algorithms  were also used to estimate unacclimatized deaths
in New York using each of the warming  scenarios.

     Next,  the  analog  cities  were  determined,  threshold   temperatures  were
calculated, and multiple  regressions  were  developed for each (Table  4).   As
expected,  the relationships  became progressively  worse  for  the  analog cities
representing the  warmest scenarios, and  the lack of  a weather/mortality rela-
tionship for  Norfolk and Jacksonville indicated that people  in  those cities
were not  sensitive  to  even  the  warmest  temperatures because they  are fully
acclimatized to the  frequent  heat.  Thus,  there would be no expected increase
in mortality  in  New York  for the 5°  and  7°F scenarios  if  the  people become
fully acclimatized.   Note that  threshold  temperatures  were higher  for  the
warmer analog cities, supporting  the  contention that the impact  of weather on
mortality is relative on an inter-regional scale.

     The number of deaths predicted  from  the nonacclimatized New York algo-
rithm increased very rapidly with each  succeeding warming  scenario.   One of
the  reasons  for  this was  the increasing number of  days exceeding New York's
threshold temperature of 92°F for the warmer scenarios (Table 5).   At present,
the  average  monthly percentage of days exceeding this threshold is  3.3% in
June, 10/1  in July, and 3-6£  in August.   Thus, for  an  average summer  season,
only  5.7%  of  the  total  days   exceed  the  threshold  temperature.    These
percentages increase steadily as  the predicted  warming  increases,  and  for the
7°F  scenario, almost half  of the days in July  and over  one-third  of the days
in the entire summer season exceed the threshold.  Obviously the  total number
of days with  heat-related  increases  in mortality  will  also  increase if there
is no acclimatization.

     A comparison  of expected mortality increases for all age groups  with no
and  with  full acclimatization showed dramatic differences  (Table  6).    At
present in New York, the average number of additional standardized deaths that
occur on days above  the threshold temperature each month is  19 in June, 86 in
July, and  25  in   August  (the  raw,  unstandardized  totals   for  New York  are
considerably higher,  but  these figures should  be  used  with caution).   Using
the  algorithm for no acclimatization,  these figures more than doubled with a
2°F  rise in  temperature,  and increased by more  than tenfold with  a 7°F rise.
Thus, if  New Yorkers do  not  acclimatize  to  the  increasing warmth,  it  is
predicted  that  the  average  number  of  additional  standardized   deaths  will
exceed  1300  each  summer season  if the weather  warms  by 7°F  (the  raw totals
will  exceed  3200).  The  full  acclimatization results showed much different
                                      283

-------
                Table 3.   Results of the Regression Analysis
                          for Mortality Climate Relationships
                                Total Deaths

Variable
Minimum Temp. (MT)
Heating Degree Hrs. (HDH)
Maximum- Dewpoint (MD)
Intercept
Total Death Algorithm: Y =

Coef.
+2.60
+0.66
-1.48
+12.01
12.01 + 2.60

R2
.306
.593
.659

(MT) + 0.66
R2
Improv.
.306
.287
.066

(HDH) -
Level of
Signif.
.005
.005
.010

1.48 (MD)
                                Elderly Deaths
Heating Degree Hrs. (HDH)
Minimum Temp. (MT)
Maximum Dewpoint (MD)
Intercept
Elderly Death Algorithm:
Y =
  +0.51       .298      .298          .001
  +1.98       ,567      .269          .001
  -1.55       .668      .101          .010
 +20.76
20.76 + 0.51 (HDH) + 1.98 (MT)' - 1.55 (MD)
                                      284

-------
Table 4.  New York's Analog Cities, Their Threshold Temperatures,
          and the R2 of Their Regression Models
Warming Analog
Scenario Citv
1°F

2°F

4°F

5°F

7°F

Indianapolis

Philadelphia

Atlanta

Norfolk

Jacksonville

Total/ Threshold
Elderly Terno.
total
elderly
total
elderly
total
elderly
total
elderly
total
elderly
91°F
91°F
91°F
91°F
94°F
92°F
94°F
94°F
96°F
96°F
Table 5. Percentage of Days Above the
for the Six Warming Scenarios

Month
June
July
August
Season

0
3.3%
10.0*
3.6*
5.7*
Degrees
1
7.0*
12.3*
5.5*
8.3*
* of Days R2 of Reg.
Above Threshold Model
10.9
10.9
13.5
13.5
4.6
7.8
7.5
7.5
10.5
10.5
.092
.078
.210
.240
.200
.200
non-
significant
non-
significant
Threshold .Temperature
for New York
Above Present
2
8.5*
16.7*
8.8*
11.4*
4
12.1*
25.8*
21.5*
19.9*
5 7
15.2* 19.7*
31.7* 46.0*
26.4* 40.0*
24.5* 35.4*
                                285

-------
             Table 6.  Average Monthly Increase in Total Mortality
                       for the Various Warming Scenarios in New Yorka»


NO
ACCLIMATIZATION
Degrees Above
Month
June
July
August
Total

0
19(45)
86(206)
• 25(60)
130(311)

1
34(81)
110(263)
37(88)
181(432)
FULL
2
57(138)
154(368)
64(153)
276(657)
Present
4
5(12)
282(674)
170(407)
56(1354)

5
156(373)
372(890)
250(598)
778(1861)

7
253(605)
622(1488)
487(1165)
1362(3258)
ACCLIMATIZATION
Degrees Above
Month
June
July
August
Total
0
19(^5)
86(206)
25(60)
130(311)

33(79)
62(148)
29(69)
124(296)
2
32(77)
54(129)
55(132)
141(338)
Present
4
5(12)
11(26)
4(10)
20(48)

5
0
0
0
0

7
0
0
0
0
a Numbers in parentheses represent raw, unstandardized mortality estimates.
  They are calculated by multiplying the standardized values by 2.39.  The
  population of the New York metropolitan area in 2980 was 9,120,000, which is
  2.39 times the population of the standardized city (3,811,000).

b These values are not adjusted for potential future population increases.
                                      286

-------
trends.    The  seasonal  number  of  standardized  deaths  remained  virtually
constant  with   a   1°F  rise  (this  was  calculated  using  the  present-day
Indianapolis  algorithm,  which  represents New  York's analog  city for  a  1°F
rise),   and  rose   only  slightly   with  a   2°F   increase  (Philadelphia's
algorithm).    However,  for   the  warmer  scenarios,  the  acclimatized  deaths
dropped  sharply,  and  no  additional  deaths  were   predicted  at  5°  and  7°F
warming.    These results  reflect  the  present-day Norfolk  and Jacksonville
situation, where no  additional  deaths are noted at  temperatures exceeding  the
threshold.

     The  actual number  of  deaths  attributed  to  future  warming will  fall
somewhere  between   the  predicted  values  for  nonacclimatization  and  full
acclimatization, but the precise impact of future acclimatization is obviously
unknown.   We  suggest that  a lag  in acclimatization  to climatic change  is
likely,  and  factors  such  as  the  physical  composition  of  the  city  (i.e.,
building construction  designed to accommodate present weather conditions) will
delay  or prevent  full  acclimatization.   Thus, it is  improbable that New
Yorkers  will  become  as  totally  insensitive  to  hot weather  as Jacksonville
residents are  today, and the decrease  in weather-related mortality predicted
by the full acclimatization model is highly unlikely.

     Predicted  mortality  increases  for  the  elderly  show  similar  trends
(Table 7).    Very  large  increases  are  noted  with  no  acclimatization  (raw
unstandardized values  are  not provided, as they  are partially dependent upon
demographic information which is unknown, such as the proportion of population
in the  elderly category when a  7°F rise is achieved).   However,  deaths once
again decrease  to  zero with  full  acclimatization.    There is some indication
that  the  elderly  will  constitute  an  increasing   proportion  of the  total
mortality as  the-climate warms  (Table 8).  At present,  the percentage of the
standardized mortality that is attributed to the elderly at temperatures above
the  threshold  is 64£  in June,  70%  in July,  and 5^%  in August.   Using the
algorithms  for  no   acclimatization,  this  proportion  is  predicted  to  rise
significantly as  the  weather warms,  and since the deaths  are  standardized,
this does not  assume that  the elderly will constitute  a larger  proportion of
the population in the  future.

     An attempt was made to duplicate the procedure  to determine the impact of
winter weather on mortality  using the same warming  scenarios.   Winter analog
cities  were  selected, threshold  temperatures  were determined,  and  multiple
regressions were performed, but the  relationships for New York and the analog
cities were unimpressive for winter (Table 9).  Since their explained variance
was  low,  the  models  were   not  robust  enough  to  produce  any  predictive
algorithms.    Although  findings  will  probably  differ  for  other  evaluated
cities, these results  suggest that any  change  in winter weather  in the future
will have little impact on weather-related mortality in New York.

     A  final  aspect  in the New York analysis was an attempt  to determine  if
precipitation has any effect upon mortality.   No  attempt  to  estimate future
impacts of  precipitation was made,  and  the study  concentrated  on historical
relationships  only.    It appears  that  precipitation  may have  an impact  on
mortality  during both summer  and  winter  (Tabl?  10);  however,  unlike  the
temperature relationships, its  influence  does  not appear to increase  steadily
as  precipitation  amounts  increase.  Thus,  the precipitation evaluation was
                                      287

-------
Table 7.  Average Monthly Increase in Elderly Mortality
          for the Various Warming Scenarios in New York
NO ACCLIMATIZATION

Month
June
July
August
Total

0
12
60
13
85

1
23
81
21
125
Degrees
2
41
107
36
184
Above Present
4
88
199
104
391

5
116
266
161
521

7
192
447
321
960
NO ACCLIMATIZATION

Month
June
July
August
Total

0
12
60
13
85

1
19
47
24
91
Degrees
2
25
42
38
105
Above Present
4
11
22
12
45

5
0
0
0
0

7
0
0
0
0
                          288

-------
Table 8.  Percent of Total Mortality Increase Attributed
          to the Elderly for the Six Warming Scenarios

Month
June
July
August

0
6455
10%
54*

1
61%
7455
51%
Degrees
2
72?
10%
55%
Above Present
4
77?
10%
6155
5
745*
7255
6455
7
7655
7255
6655
    Table 9-   Winter  Relationships  Between Weather and
              Mortality  for  New York and Two of Its Analogs
City
New York
Nashville
(5° Analog)
Norfolk
(7° Analog)
Total/
Elderly
total
elderly
total
elderly
total
elderly
Lag Time
With Best Fit
0
0
0
0
1
1
R2 of
Regression Model
.069
.142
.132
.079
.064
.082
                           289

-------
                Table 10.  Relationships Between Precipitation
                           and Mortality in Mew Yorka
Variables
Mean Mortality for M
Each Variable5
for Each
Variable0
T- Significant
Statistic*1 Level
Sunnier
Precipitation
vs.
No Precipitation
-0.135
0.100
310
425
3.235 0.001
Winter
Precipitation
vs.
No Precipitation
Rain
vs.
No Precipitation
Snowfall
vs.
No Precipitation
3" Snow on Ground
vs.
No Snow on Ground
. 0.087
-0.079
0.192
-0.079
0.045
-0.079
0.336
-0.038
342
378
119
378
140
378
47
392
2.241 0.025

1.324 0.187€
2.489 0.013
a  All best-fit  relationships  possessed a one-day lag  between the precipita-
   tion event and the mortality response.
b  Expressed as standard deviations from the overall daily mean.
*j  Number of days in the sample.
d  T-statistic comparing the means of two samples, assuming that variances are
   not equal.
e  Not statistically significant.
                                      290

-------
limited  to  comparing mortality  rates  during  periods  of  precipitation and
nonprecipitation and determining if the difference in the mean daily mortality
rates was  statistically  significant between the two periods.  It appears that
a  one-day  lag  exists  between the precipitation episode  and  the mortality
response during all seasons,  and  that the strongest correlation between these
variables  occurred  in summer.  On  summer  days  with precipitation, mortality
averaged  .135 standard  deviation  below the  mean,  but on  those days without
precipitation, mortality averaged .100 standard deviation above  the mean.  One
possible explanation  for this relationship is  that  summer  rain may provide a
refreshing,  cooling influence which tends  to lessen discomfort and therefore,
to  lower  mortality.   Some  strong  winter relationships were also discovered,
and rain appeared  to  have a  greater  influence on mortality than snowfall.  A
significant  relationship  was  determined   between  all  precipitation and  no
precipitation  days,   but  when precipitation  was  subdivided  into  rain  and
snowfall,  only  the rain relationship proved  to be statistically significant.
Unlike the summer  findings,  mortality  was  significantly higher  on days with
rainfall,  and mean daily  mortality  rate was almost .200  standard deviation
above the  mean on those days.  Although days with snow falling appeared to
have little impact on mortality, days  with significant accumulations  on the
ground did correspond  with higher mortality  rates.   Those  days with three or
more inches  of snow on  the ground averaged  over  .330 standard deviation above
the mean.

CONCLUSION

     The objectives of  this study were to  determine the  historical relation-
ships between weather and  mortality  and to  estimate the  possible impact of
long-term climatic warming on future mortality rates in New York City.  During
the summer,  weather appears to exert  a significant  influence on mortality in
New York,  but the future  impact  is  largely dependent on whether New Yorkers
will acclimatize  to  the predicted increasing  warmth.   If  acclimatization is
slow or  is nonexistent,  thousands  of  additional  deaths may occur during each
summer season if  the mean  temperature warms to 7°F  above present  levels.
However,  changes in winter weather should have little impact on mortality.

     This  study will be  expanded to include fourteen additional cities  around
the United  States.   Analog  cities  will   be  determined,  and  inter-regional
influences of weather will  be examined.   In addition, mortality rates will be
subdivided  by  race  and additional age  categories,  and those causes  of death
that are considered to  be weather-related will be isolated  and independently
evaluated.

ACKNOWLEDGMENTS

     This  research was  supported  by  the U.S.  Environmental  Protection  Agency
under contract number 68-01-7033.   The authors thank Mr.  Dennis Tirpak,  Office
of  Policy   Analysis,  EPA,  and Dr.  Melvyn  Tockman, Associate  Professor  of
Environmental  Health  Sciences,  The   Johns  Hopkins  University,  for   their
suggestions and support.  Thanks are also extended to the National Oceanic and
Atmospheric  Administration,  for funding our  initial climate/mortality  work,
and to various scientists at the NASA-Goddard Institute  for  Space Sciences for
their interest in our project.
                                     291

-------
REFERENCES

Applegate, W.B., J.W. Runyan, Jr., L. Brasfield, M.L. Williams, C. Konigsberg,
     and  C.  Fouche.    1981.    Analysis of  the  1980  heat  wave  in  Memphis.
     Journal of the American Geriatrics Society 29:337-42.

Bridger, C.A.,  F.P.  Ellis,  and H.L. Taylor.   1976.  Mortality  in St.  Louis,
     Missouri,   during  heat  waves  in  1936,  1953,  1954,  1955  and  1966.
     Environmental Research 12:38-48.

Centers  for  Disease Control.   1982.   Exposure-related  hypothermia  deaths—
     District of  Columbia 1972-1982.   Morbidity and Mortality  Weekly  Report
     December:31-50.

Ellis, F.P.  1972.  Mortality from heat illness and heat-aggravated illness in
     the United States.  Environmental Research  5:1-58.

Fitzgerald, F.T., and C.  Jessop.   1982.   Accidental hypothermia:  A report of
     22  cases  and review of  the  literature.   Advanced  Internal Medicine 27:
     127-50.

Callow, D., T.E. Graham, and S. Pfeiffer.   1984.  Comparative thermoregulatory
     responses  to  acute cold  in  women  of Asian and European  descent.   Human
     Biology 56:19-34.

Jones, T.S., A.P. Liang,  E.M.  Kilbourne, M.R.  Griffin, P.A.  Patriarca,  S.G.G.
     Wassilak,  R.J.  Mullan,  MD; R.F. Herrick, H.D. Donnel,  Jr.,  K.  Choi, and
     S.B. Thacker.   1982:   Morbidity  and  mortality associated  with  the July
     1980 heat wave in St. Louis and Kansas City, MO.  Journal of the American
     Medical Association 247:3327-30.

Kalkstein, L.S.   1986.   Progress Report  for Contract  //NA85AA-H-AI098:   The
     Impact of Climate  Upon  Human Mortality.  National  Oceanic and Atmospheric
     Administration.

Kalkstein, L.S., and  R.E.  Davis.   1985.   The development of a weather/morta-
     lity model for  environmental impact  assessment.  Proceedings of the 7th
     Conference of Biometeorology  and Aerobiology.  334-36.

Kilbourne,  E.M.,   K.   Choi,  T.S.  Jones,   S.B.  Thacker,  and   the   Field
     Investigation Team.  1982.   Risk  factors for heatstroke:  A case-control
     study.  Journal of the American Medical Association  247:3332-36.

Lilianfeld, A.M., and  D.E.  Lilienfeld.   1980.   Foundations  of  epidemiology.
     Oxford University Press.

Lye, M., and A.  Kamal.   1977.  The effects of a  heat  wave on mortality rates
     in elderly inpatients.   The Lancet 1:529-31.

Mausner, J.S.,  and A.K.  Bahn.    1974.  Epidemiology;    An  introductory text.
     W.B. Saunders Co.
                                     292

-------
National Center  for Health  Statistics,  1978.   Standardized Micro-Data  Tape
     Transcripts.  U.S.  Department  of  Health,  Education,  and  Welfare,  DHEW
     Publication No. (PHS) 78-1213.

Oechsli, F.W.,  and R.W.  Buechley.    1970.   Excess mortality associated  with
     three Los Angeles^  September  hot spells.  Environmental Research   3:277-
     84.

Rotton, J.   1983.  Angry,  sad,  happy?  Blame the weather.  U.S.  News and World
     Report 95:52-3.

Steadman, R.G.   1979.   The assessment of sultriness:   Part I:   A temperature-
     humidity index based on human  physiology and clothing science.   Journal
     of Applied Meteorology 18:861-73.

Willmott, C.J.,  S.G.  Ackleson,  R.E.  Davis,  J.J.  Feddema,  K.M.  Klink,  D.R.
     Legates,  J.  O'Donnel,  and  C.M.   Rowe.    1985.    Statistics  for  the
     evaluation  and  comparison of  models.   Journal  of  Geophysical  Research
     90:8995-9005.
                                      293

-------