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
------- |