United States
Environmental Protection
Agency
Policy, Planning,
And Evaluation
(PM-221)
EPA-230-05-89-053
June 1989
c/EPA
The Potential Effects
Of Global Climate Change
On The United States
Appendix C
Agriculture
Volume 1
Printed on Recycled Paper
-------
THE POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE
ON THE UNITED STATES:
APPENDIX C • AGRICULTURE
Editors: Joel B. Smith and Dennis A. Tirpak
OFFICE OF POLICY, PLANNING AND EVALUATION
US. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, DC 20460
MAY 1989
-------
TABLE OF CONTENTS
Page
APPENDIX C-l: AGRICULTURE
PREFACE iii
EFFECT OF GLOBAL CLIMATE CHANGE ON AGRICULTURE
GREAT LAKES REGION 1-1
J.T. Ritchie, B.D. Baer, and T.Y. Chou
IMPACT OF CLIMATE CHANGE ON CROP YIELD IN THE SOUTHEASTERN
USA: A SIMULATION STUDY 2-1
Robert M. Peart, J.W. Jones, R. Bruce Curry, Ken Boote, and
L. Hartwell Allen, Jr.
POTENTIAL EFFECTS OF CLIMATE CHANGE ON AGRICULTURAL
PRODUCTION IN THE GREAT PLAINS: A SIMULATION STUDY 3-1
Cynthia Rosenzweig
THE ECONOMIC EFFECTS OF CLIMATE CHANGE ON VS. AGRICULTURE:
A PRELIMINARY ASSESSMENT 4-1
Richard M. Adams, J. David Glyer, and Bruce A. McCarl
CLIMATE CHANGE IMPACTS UPON AGRICULTURE AND RESOURCES:
A CASE STUDY OF CALIFORNIA 5-1
Daniel J. Dudek
EFFECTS OF PROJECT CO2-INDUCED CLIMATIC CHANGES ON IRRIGATION
WATER REQUIREMENTS m THE GREAT PLAINS STATES (TEXAS, OKLAHOMA,
KANSAS, AND NEBRASKA 6-1
Richard G. Allen and Francis N. Gichuki
APPENDIX C-2: AGRICULTURE
DIRECT (PHYSIOLOGICAL) EFFECTS OF INCREASING CO- ON CROP PLANTS
AND THEIR INTERACTIONS WITH INDIRECT (CLIMATIC) EFFECTS 7-1
Elise Rose
POTENTIAL EFFECTS OF CLIMATE CHANGE ON PLANT-PEST INTERACTIONS.. 8-1
Benjamin R. Stinner, Robin AJ. Taylor, Ronald B. Hammond,
Foster F. Purrington, and David A. McCartney
IMPACTS OF CLIMATE CHANGE ON THE TRANSPORT OF AGRICULTURAL
CHEMICALS ACROSS THE USA GREAT PLAINS AND CENTRAL PRAIRIE 9-1
Howard L. Johnson, Ellen J. Cooler, and Robert J. Sladewski
FARM-LEVEL ADJUSTMENTS BY ILLINOIS CORN PRODUCERS TO
CLIMATE CHANGE 10-1
William E. Easterling
• •
11
-------
TABLE OF CONTENTS (continued)
APPENDIX C-2: AGRICULTURE (continued)
CHANGING ANIMAL DISEASE PATTERNS INDUCED BY THE
GREENHOUSE EFFECT 11-1
Edgar Stem, Gregory A. Mertz, J. Dirck Stryker, and Monika Huppi
EFFECT OF CLIMATIC WARMING ON POPULATIONS OF THE HORN FLY,
WITH ASSOCIATED IMPACT ON WEIGHT GAIN AND MBLK PRODUCTION
IN CATTLE 12-1
E.T. Schmidtmann and JA. Miller
AGRICULTURAL POLICIES FOR CLIMATE CHANGES INDUCED BY
GREENHOUSE GASES 13-1
G. Edward Schuh
ui
-------
PREFACE
The ecological and economic implications of the greenhouse effect have been the subject of discussion within
the scientific community for the past three decades. In recent years, members of Congress have held hearings
on the greenhouse effect and have begun to examine its implications for public policy. This interest was
accentuated during a series of hearings held in June 1986 by the Subcommittee on Pollution of the Senate
Environment and Public Works Committee. Following the hearings, committee members sent a formal request
to the EPA Administrator, asking the Agency to undertake two studies on climate change due to the greenhouse
effect.
One of the studies we are requesting should examine the potential health and environmental
effects of climate change. This study should include, but not be limited to, the potential impacts
on agriculture, forests, wetlands, human health, rivers, lakes, and estuaries, as well as other
ecosystems and societal impacts. This study should be designed to include original analyses, to
identify and fill in where important research gaps exist, and to solicit the opinions of
knowledgeable people throughout the country through a process of public hearings and
meetings.
To meet this request, EPA produced the report entitled The Potential Effects of Global Climate Change on the
United States. For that report, EPA commissioned fifty-five studies by academic and government scientists on
the potential effects of global climate change. Each study was reviewed by at least two peer reviewers. The
Effects Report summarizes the results of all of those studies. The complete results of each study are contained
in Appendices A through J.
Appendix Subject
A Water Resources
B Sea Level Rise
C Agriculture
D Forests
E Aquatic Resources
F Air Quality
G Health
H Infrastructure
I Variability
J Policy
GOAL
The goal of the Effects Report was to try to give a sense of the possible direction of changes from a global
wanning as well as a sense of the magnitude. Specifically, we examined the following issues:
o sensitivities of systems to changes in climate (since we cannot predict regional climate change, we
can only identify sensitivities to changes in climate factors)
o the range of effects under different wanning scenarios
o regional differences among effects
o interactions among effects on a regional level
iv
-------
o national effects
o uncertainties
o policy implications
o research needs
The four regions chosen for the studies were California, the Great Lakes, the Southeast, and the Great Plains.
Many studies focused on impacts in a single region, while others examined potential impacts on a national scale.
SCENARIOS USED FOR THE EFFECTS REPORT STUDIES
The Effects Report studies used several scenarios to examine the sensitivities of various systems to changes in
climate. The scenarios used are plausible sets of circumstances although none of them should be considered to
be predictions of regional climate change. The most common scenario used was the doubled CO2 scenario
(2XCO2), which examined the effects of climate under a doubling of atmospheric carbon dioxide concentrations.
This doubling is estimated to raise average global temperatures by 1.5 to 4-5°C by the latter half of the 21st
century. Transient scenarios, which estimate how climate may change over time in response to a steady increase
in greenhouse gases, were also used. In addition, analog scenarios of past warm periods, such as the 1930s, were
used.
The scenarios combined average monthly climate change estimates for regional grid boxes from General
Circulation Models (GCMs) with 1951-80 climate observations from sites in the respective grid boxes. GCMs
are dynamic models that simulate the physical processes of the atmosphere and oceans to estimate global climate
under different conditions, such as increasing concentrations of greenhouse gases (e.g., 2XCO2).
The scenarios and GCMs used in the studies have certain limitations. The scenarios used for the studies assume
that temporal and spatial variability do not change from current conditions. The first of two major limitations
related to the GCMs is their low spatial resolution. GCMs use rather large grid boxes where climate is averaged
for the whole grid box, while in fact climate may be quite variable within a grid box. The second limitation is
the simplified way that GCMs treat physical factors such as clouds, oceans, albedo, and land surface hydrology.
Because of these limitations, GCMs often disagree with each other on estimates of regional climate change (as
well as the magnitude of global changes) and should not be considered to be predictions.
To obtain a range of scenarios, EPA asked the researchers to use output from the following GCMs:
o Goddard Institute for Space Studies (GISS)
o Geophysical Fluid Dynamics Laboratory (GFDL)
o Oregon State University (OSU)
Figure 1 shows the temperature change from current climate to a climate with a doubling of CO- levels, as
modeled by the three GCMs. The figure includes the GCM estimates for the four regions. Precipitation
changes are shown in Figure 2. Note the disagreement in the GCM estimates concerning the direction of
change of regional and seasonal precipitation and the agreement concerning increasing temperatures.
Two transient scenarios from the GISS model were also used, and the average decadal temperature changes
are shown in Figure 3.
-------
FIGURE 1. TEMPERATURE SCENARIOS
GCM Estimated Change in Temperature from 1xCO2 to 2xCO2
8
6
WINTER
Great Southeast Great California United
Lakes Plains States*
Great Southeast Great California United
Lakes Plains States'
Great Southeast Great California United
Lakes Plains States*
GIS'3
GFDL
OSU
* Lovter 43 St?tes
-------
FIGURE 2. PRECIPITATION SCENARIOS
GCM Estimated Change in Precipitation from 1xCOa to 2xCO2
Great Southeast
Lakes
Great California United
Plains States'
l.U
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
.nfi .
i
i
1
-i JL 1
I
v\
il
riNTER
HJ
f\
Great Soul1 least Great California United
Lakes Plains States*
1.0
0.8-
0.6-
0.4-
0.2-
0.0
-0.2-
-0.4-
-0.6
No
Change
I
SUMMER
Great Southea;! Great Caltfomia United
Lakes Plains States*
GISS
GFDL
OSU
* Lower 48 States
-------
4
3.5
0 3
g 2.5
< 2
UJ
Q.
UJ
1.5
1
0.5
0
4
3.5
3
O
ut 2.5
oc
I 2
flC
UJ - .
o. 1.5
UJ
1
0.5
0
3.72
2.99
2.47
1.72
1.36
0.70
0.88
0.30
1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s
TRANSIENT SCENARIO A
1.26
1.02
0.59
0.18
0.35
V//,
1980s 1990s 2000s 2010s
TRANSIENT SCENARIO B
2020s
FIGURE 3.
GISS TRANSIENTS "A" AND "B" AVERAGE
TEMPERATURE CHANGE FOR LOWER 48 STATES
GRID POINTS.
VUl
-------
EPA specified that researchers were to use three doubled CO, scenarios, two transient scenarios, and an analog
scenario in their studies. Many researchers, however, did nothave sufficient time or resources to use all of the
scenarios. EPA asked the researchers to run the scenarios in the following order, going as far through the list
as time and resources allowed:
1. GISS doubled CO2
2. GFDL doubled CO2
3. GISS transient A
4. OSU doubled CO2
5. Analog (1930 to 1939)
6. GISS transient B
ABOUT THESE APPENDICES
The studies contained in these appendices appear in the form that the researchers submitted them to EPA.
These reports do not necessarily reflect the official position of the ILS. Environmental Protection Agency.
Mention of trade names does not constitute an endorsement.
IX
-------
EFFECT OF GLOBAL CLIMATE CHANGE ON AGRICULTURE GREAT LAKES REGION
by
J.T. Ritchie
B.D. Baer
T.Y. Chou
Department of Crop and Soil Sciences
Plant and Soil Sciences Building
Michigan State University
East Lansing, MI 48824-1325
Contract No. CR-814601-01-0
-------
CONTENTS
FINDINGS 1-1
CHAPTER 1: INTRODUCTION 1-2
DESCRIPTION OF THE ECOLOGICAL SYSTEM 1-2
RECENT LITERATURE 1-2
ORGANIZATION OF THIS REPORT 1-3
CHAPTER 2: METHODOLOGY * 1-4
THE EFFECTS MODELS 1-4
Development of the Model 1-4
Limitations Inherent in the Models 1-5
THE WEATHER SCENARIOS 1-6
The Scenarios Used 1-6
Limitations of the Weather Scenarios 1-6
SIMULATIONS WITH THE MODEL 1-6
CHAPTER 3: RESULTS 1-9
DIRECT EFFECTS NOT CONSIDERED 1-9
Yield 1-9
Irrigation Water Demand 1-9
DIRECT CO, EFFECTS CONSIDERED 1-10
FIGURES AND TABLES 1-10
SUMMARY GRAPH 1-17
LIMITATIONS 1-17
CHAPTER 4: INTERPRETATION OF THE RESULTS 1-23
CHAPTER 5: IMPLICATION OF RESULTS 1-24
ENVIRONMENTAL IMPLICATIONS 1-24
REFERENCES 1-25
APPENDIX A: PROPERTIES OF SOILS USED 1-26
APPENDIX B: TABLES OF RUN RESULTS 1-30
11
-------
Ritchie
FINDINGS1
Using weather scenarios created from two climate models, Goddard Institute for Space Studies (GISS)
and Geophysical Fluid Dynamics Laboratory (GFDL), and baseline observed weather data sets, corn and
soybean production was simulated for normal and changed climates using the CERES-Maize model and the
SOYGRO soybean model. The primary analysis was completed without including the direct effects of CO,
on photosynthesis and transpiration. Thus, changes in temperature and rainfall were the principal causes of
yield changes. Further analysis included the direct effects of CO2-
For the humid Great Lakes region, changes in temperature had the greatest effect on model-predicted
crop yields. In most cases, an increase in temperature caused a decrease in the duration of crop life cycle.
The more extreme GFDL weather change caused a decrease in yield ranging from 3% to 50% for irrigated
corn with the decreases being greater for the more southern stations. The soybean model predicted less yield
decreases than the corn model under the irrigated GFDL conditions, the decrease ranging from zero to a
maximum of 30%. The maximum decreases occurred in the southernmost locations. In the most northern
latitudes where the warmer conditions provided a longer frost-free growing season, the increased
temperatures had a beneficial effect on simulated yields. Because the GISS model generated smaller
temperature increases, the effects on yield were less extreme but followed the same general pattern as the
GFDL model.
The rainfed crop yield under GFDL weather was reduced quite substantially when compared to
irrigated crops at most sites. With a few exceptions, when using GISS weather with a slight increase of rain
during the growing period, yield decreases were relatively small. The amount of irrigation water required for
optimum yields was closely related to the amount of rainfall in the growing season. Water requirements
increased an average of about 90% under GFDL conditions when compared to the baseline weather, and
decreased an average of about 30% under GISS conditions.
At Fort Wayne a longer season corn cultivar adapted to a more southern climate could compensate for
some of the lost yield due to climate change. Full compensation could not be obtained, however, because the
grain-filling duration is not substantially different between corn cultivars.
The direct effects of CO2 were studied by running versions of the maize and soybean models with
modified photosynthesis and transpiration calculations. The direct effect of COy as approximated in the
modified crop models, increased yields when compared to weather effects alone for both crops at all
locations. In some situations the direct effects overcame the predicted weather-related yield losses.
'Although the information in this report has been dunded wholly or partly by the US. Environmental
Protection Agency under contract no. CR-814601-01-0, it does not necessarily reflect the Agency's views, and
no official endorsement should be inferred from it.
word maize refers to the crop Zeamavs L. which, in the United States, is usually called corn.
1-1
-------
Ritchie
CHAPTER 1
INTRODUCTION
DESCRIPTION OF THE ECOLOGICAL SYSTEM
Our part of the Global Climate Change study concentrated on corn and soybean production in the
Great Lakes region, using weather from 18 sites in 10 states (the eight states bordering the Great Lakes, plus
Iowa and Missouri). This region extends to the northern limit of the major growing region for corn and
soybeans in North America. The long, cold winters throughout most of the region place a constraint on crop
productivity because of the short growing season. However, com production is high in the region.
Variability of yields is also high owing to the variation in precipitation and temperature in the area
The variation in the water-holding capacity of the soils in this region also contributes to the variability
in yield.
RECENT LITERATURE
Simulation models of crop production and yield have been used to examine potential effects of CO2
enrichment on crop production since at least 1970. Duncan and Barfield (1970) showed that while CO, was
limiting photosynthesis at certain layers in a corn canopy under ambient CO2 conditions, it was nonlimiting at
concentrations of 600 parts per million.
In a U.S. Department of Commerce report (1975), yield simulation models were used extensively to
evaluate the impact of climate change on the biosphere. The focus of that study was to determine how
increased stratospheric flight would affect weather and crop production. The principal thrust of the report
dealt with lowered temperatures, although some simulations of crop yield were done for elevated
temperatures. In the same report, results of corn yield simulations (Benci et al., 1975) indicated that elevated
temperatures would decrease the length of the growing season and move the com belt northward. This led
to the conclusion that U.S. corn production would increase with a small (2°C) temperature increase because
a larger section of the northern states (Minnesota, Wisconsin, and Michigan) would have greater production
possibilities. Results of the soybean yield simulation (Curry and Baker, 1975) indicated that a 2°C increase
in temperature would decrease yield by 4% and 13% in Ohio and Indiana, respectively, and cause no change
in Iowa yields. These projections were believed to be the combined result of a decreased growing season and
an increased photosynthetic rate by the plant.
Crop models were used in Ontario, Canada, to study how crop yields would be affected by climate
change as predicted by the GFDL and GISS models for doubled CO, (Smit, 1987). Their corn yield
simulation using both scenarios indicated that corn could be produced in northern Ontario where the season
at present is too short, and that yields in southern Ontario would be decreased by 10% to 35% for all but
poorly drained soils. This latter finding primarily resulted from a shortening of the growing season due to an
approximate 1.5°C increase in the average temperature as predicted by both global climate models.
Van Keulen et al. (1981) used dynamic crop growth simulation models to evaluate net assimilation and
transpiration and their ratios for C3 (wheat) and C4 (com) crops exposed to increased (430 ppm) CO2.
Their results demonstrated that stomatal behavior was the key factor in determining plant response to
increased CO2 under nonlimiting water and nutrient conditions when the length of the season was the same.
Other studies (World Meteorological Organization (WMO), 1984; Carter et al., 1984) evaluated the
potential climate effects of increased CO, on crop production, exclusive of physiological effects. A meeting of
experts organized by the WMO and the international Meteorological Institute in Stockholm concluded that
1-2
-------
Ritchie
studies with mechanistic crop models coupled with climate change scenarios are an appropriate first step
approximation in studying the impacts of increasing CO2 on crop growth and yield (WMO, 1984). Carter et
at. (1984) tested the sensitivity of crop models to daily and monthly time resolutions and found that monthly
climatic variables are adequate for limited crop modeling studies. They reported differences between short-
term and long-term responses to climate change.
Few studies have considered both climatic and physiological effects simultaneously. Baker et al. (1985)
adapted a detailed crop-climate model to investigate die interactive effects of CO,, leaf area index (LAI),
and the environment on midday crop water-use and water-use efficiency. The results showed that increased
CO2 in conjunction with increased LAI can offset the lowered transpiration caused by increased stomatal
resistance. Stewart (1986) used results from a doubled CO, experiment to run a generalized crop growth
model for Saskatchewan spring wheat with a 15% increase in photosynthetic capacity. Predicted climate
changes caused a reduction in wheat yield even with the increase in photosynthetic capacity.
ORGANIZATION OF THIS REPORT
In this report we discuss the CERES-Maize and SOYGRO crop models, and how they were used with
the global warming weather scenarios. The strengths and limitations of using these models with the
predicted weather are examined. Eighteen sites in the Great Lakes region were chosen to represent a cross-
section of the region for simulation of yields under the baseline weather and two CMC-modeled weather
conditions with doubled CO2. We also simulated irrigated yields in order to evaluate the impact of
temperature changes alone on simulated yields. We briefly evaluated how changing genetic types can help
compensate for some yield loss. In order to determine how direct effects of CO2 increase may influence
crop yields, separate runs were made to compare them with the more direct effects of temperature and
rainfall alone on simulated yields. Appendices of model run output summaries and soil properties are
included at the end of the report.
1-3
-------
Ritchie
CHAPTER 2
METHODOLOGY
THE EFFECTS MODELS
Development of the Model
The CERES-Maize and SOYGRO simulation models are designed to predict the growth components
and yield of different corn and soybean varieties for all cropping seasons and all types of environments where
the crops are generally grown. The models are designed primarily to predict:
1. Phenological development or duration of growth stages as influenced by plant genetics, weather,
and soil factors;
2. Apical development as related to morphogenesis of vegetative and reproductive structures;
3. Extension growth of leaves and stems and senescence of leaves;
4. Biomass production and partitioning;
5. Root system dynamics; and
6. The effect of soil-water deficit and nitrogen deficiency on the photosynthesis and photosynthate
partitioning in the plant system.
The models simulate the values of the predicted variables over a sampling interval of one day, that is,
these values are a sequence of numbers spaced at 24-hour intervals. The models are programmed in
FORTRAN 77 and set up to run interactively on any IBM-compatible microcomputer with at least 256
kilobytes of random access memory (RAM). In a Compaq microcomputer with 640 kilobytes RAM,
simulation time of one cropping season for the CERES-Maize model with N fertilizer as nonlimiting (i.e.,
nitrogen subroutines are shut off) takes about 15 to 20 seconds.
The input variables necessary to run the crop models can be divided into three categories: (1) the
exogenous input variables, which are uncontrollable and may be stochastic in nature; (2) the controllable
input variables, which are deterministic in nature; and (3) the system parameters, which are the coefficients
in the analytical equations describing the model.
The exogenous input variables are the daily solar radiation (MJ m"2 day*1), maximum and minimum air
temperature (°C); and rainfall (mm day*1).
The controllable input variables are the beginning day of the simulation; day of the year for sowing;
plant population (plants m*2); row spacing (m); depth of sowing (cm); day of the year and amount of
irrigation (mm) (amount of irrigation can be automatically calculated by the program); and nitrogen
fertilization status. For application of the model to climate change, nitrogen fertilizer is assumed to be
nonlimiting.
The natural system parameters are the latitude of the production area, the soil parameters, and the
initial conditions of the soil profile: soil albedo; upper limit of stage 1 soil evaporation (mm); soil-water
drainage constant; and USDA Soil Conservation Service curve number to calculate runoff (CN2). There are
also parameters for each soil layer: the lower limit of plant extractable soil water (volume fraction); drained
upper limit soil water content (volume fraction); saturated water content (volume fraction); weighting factor
for new root growth distribution; bulk density; and initial soil water content (volume fraction).
1-4
-------
MATERIAL 5HLONGS TO:
US EPA TOXICS UPRAHY
401 M ST SW / Ti.;-7?3
WASHINGTON, DC 20460 Ritchie
(202) 260-3944
The system design parameters are the genetic coefficients of the variety. For CERES-Maize these
coefficients are PI (thermal time required from emergence to end of juvenile stage); P2 (rate of photo-
induction; in degree-days per hour); P5 (thermal time required for grain filling); G2 (potential kernel
number); and G3 (maximum daily rate of kernel fill, in mg per kernel). SOYGRO has similar types of
coefficients to describe genetic variation in plant growth and development. These are described in more
detail in another section of this report by Peart et al.
The simulation models were developed from experimental data and expert opinion (where
experimental data are not available) from many locations over the past eight years. Because they are models
with simplified functions representing mechanistic responses, they are not calibrated for any particular
location, environment, or soil type.
The CERES-Maize model is documented in the book, CERES-Maize: A Simulation Model of Maize
Growth and Development, edited by Jones and Kiniry (1986), and is available from the Texas A&M Press.
The SOYGRO model is documented in a technical publication by Wilkerson et al. (1983). Users' guides are
available from the primary developers of the models (Ritchie-Maize3; Jones-SOYGRO ).
However, since the published documentations were developed, both models have been modified and in
the past three years, both models have been subjected to international testing.
Limitations Inherent in the Models
The limitations and/or assumptions of the crop models are (1) weeds, diseases, and insect pests are
controlled to the extent that they have no economic effect; (2) except for nitrogen, all nutrients required for
plant growth are nonlimiting; (3) there are no highly problematic soil conditions, such as high salinity and
acidity, heavy compaction, or trace element deficiencies; (4) there are no catastrophic weather events such as
hailstorm, tornado, flood, excessive rain, and typhoons; (5) the model, except as amended for a part of this
study, does not consider the direct effects of CO2 on photosynthesis and transpiration.
This latter limitation could cause the model to produce a lower estimate of yield for higher than
normal CO2- The photosynthetic rate at any light level would be greater than the value calculated unless
increased canopy resistance offsets the expected photosynthetic increase. The transpiration rates may be
lower than calculated as a result of increased stomatal resistance at elevated CO2 levels.
The models attempt to account for the supply of biomass (photosynthesis) to support organ growth
and respiration and as well as the demand for the biomass as determined by the potential growth rate of
plant organs. The potential rate of expansion of leaves, stem, and grain is influenced by temperature. An
increased supply of assimilate caused by high CO, may not cause a more rapid rate of aboveground organ
growth, possibly canceling some of the benefit of increased photosynthesis.
Partitioning of assimilates into various growing organs follows a qualitative (Brouwer, 1965; Whisler et
al., 1986) pattern of priorities. The primary principles are that (1) if the soil-supplied materials (water and
nutrients) are nonlimiting, and the atmospheric-supplied energy and materials (light and CO2) are in limited
supply, the plant top parts have priority; and (2) if the atmospheric sources are nonlimiting but the soil-
supplied materials are limited, the root system has priority for the assimilates. In general, the elevated CO2
makes photosynthesis less limiting. The plant tops may be growing at an optimum rate and the extra
assimilate supply may be partitioned to the root system, increasing its size and, thus, may have little influence
on aboveground biomass growth rates, unless plants can develop tillers or new branches as a sink for the
•^Dr. Joe T. Ritchie, Homer Nowlin Chair, Michigan State University, Department of Crop and Soil
Science, East Lansing, MI 48824-1325.
4Dr. Jim Jones, Department of Agricultural Engineering, University of Florida, Gainesville, FL 32611.
1-5
-------
Ritchie
added assimilate supply. Since modern corn does not tiller or branch to any significant extent, there may be
little net effect on growth rates of aboveground biomass caused by CO, increases. However, soybeans can
branch more heavily if additional assimilate is available. Therefore, direct effects of CO2 could cause a
greater growth rate in soybean than in corn.
THE WEATHER SCENARIOS
The Scenarios Used
From baseline weather data (1951-1980) provided by National Center for Atmospheric Research,
Boulder, Colorado, we used 18 weather stations in the 10 states of our study area. These stations (with the
soils and cultivars used for the run of the models) are shown in Table LA. The weather data from NCAR
contained the maximum and minimum air.temperatures and precipitation; solar radiation was generated
from the temperatures and precipitation using equations from the weather generator program WGEN
(Richardson, 1985).
For this study we used the predicted average temperature, precipitation, and solar radiation from the
GISS and GFDL doubled CO2 steady-state scenarios. We multiplied the baseline weather data by the
percent changes between the model values for current averages in a grid cell to predicted future averages in
the same cells. Table IB shows the average changes from the baseline weather of all the GISS grid boxes
that overlapped our study area. Table 1C shows the average climate change for three strips of GFDL grid
cells that were in our study area
Limitations of the Weather Scenarios
The crop models are sensitive to the dates of occurrence of precipitation as well as to the total amount
of the rainfall. Since the scenarios simply multiply the daily baseline rainfall amount by the percent change
in rainfall amount, the scenarios did not simulate change in frequency of rain occurrence. A month with
fewer days of heavier rain might have more evapotranspiration and cause more plant stress than a month
with an identical amount of rainfall over a greater number of days. Any rainfall frequency change could
impact on frequency and duration of plant stress.
The changed maximum and minimum temperatures were calculated by multiplying the baseline
temperatures by the change in average temperatures predicted by the climate models. Thus the change in
average temperature assumes little change in daily variations in temperature. The temperature functions for
plant development are linear between 8>C and 34°C. If the daily variations are incorrect from the GCM
model assumptions, there would be little change in the model outcomes as long as the temperatures are
within this range. Simulated yields would be more uncertain with the greater the duration within a day that
the temperature is above 34°C. This condition would occur in a few days of July and August for the GFDL
scenario in the southern regions of the Great Lakes area.
SIMULATIONS WITH THE MODEL
We generated a list of typical soils to be simulated. From that list we chose a soil that is most
representative of the region for each of the 18 sites. A variety for each crop was chosen based on genetic
parameters of maturity type and photoperiod that were suited for the current climate, but not necessarily the
new climate. Runs of both crops were then made with the baseline weather, GISS doubled CO2 weather,
and the GFDL doubled CO2 weather. Both rainfed and irrigated conditions were simulated. Irrigated yields
were simulated for two reasons. One was to help assess the influence of temperature change alone on yields
and the other was to provide information on how valuable irrigation would be as a management alternative
for the changed climate. Yield, irrigation water demand, and season length for each year's simulation were
used to do our analysis.
1-6
-------
Ritchie
Table 1A. Weather Stations Used
Station
Soil Type
Cultivar
Soybean Corn
Duluth,MN
Saint Cloud, MN
Des Moines, IA
Springfield, MO
Saint Louis, MO
Green Bay, WI
Madison, WI
Peoria, IL
Fort Wayne, IN
Indianapolis, IN
Flint, MI
Muskegon Co., MI
Cleveland, OH
Columbus, OH
Albany, NY
Buffalo, NY
Pittsburgh PA
Williamsport PA
Medium Sandy Loam
Medium Silty Loam
Deep Sandy Loam
Medium Sandy Loam
Medium Silty Loam
Medium Silty Clay
Medium Sandy Loam
Deep Sandy Loam
Deep Silty Loam
Shallow Silty Loam
Medium Silty Loam
Medium Sandy Loam
Deep Silty Clay
Deep Sandy Loam
Deep Silty Loam
Deep Silty Loam
Medium Silty Clay
Medium Silty Clay
EVANS
EVANS
MGAL2
MG-02
MG-02
EVANS
EVANS
MG-02
MG-01
MG-04
MG-01
EVANS
MG-01
MG-02
MG-01
MG-01
MG-01
MG-01
EDO
EDO
PIO 3183
PIO 3147
PIO 3147
EDO
DEKAB XL71
PIO 3720
PIO 3720
PIO 3183
DEKALBXL45
DEKALBXL45
PIO 3720
PIO 3183
DEKALBXL45
DEKALBXL45
PIO 3720
DEKALBXL45
Table IB: Average GISS Climate Change for the Entire Region
Temperature
•C
Precipitation
mm/month
March-May
June-August
Sept-Nov
4.5
3.5
43
5.5
3.4
-14.4
Table 1C: Average GFDL Climate Change for the East-West Sections of the Region
Latitude
May
June
July
Aug
44°-49°
Temp Precip
°C mm
3.6 8.9
9.4 -45.0
9.4 -30.8
8.1 -13.5
40°-44°
Temp Precip
°C mm
3.6 9.4
7.0 -9.2
8.0 -39.6
4.3 0.8
36°-40°
Temp Precip
°C mm
2.4
6.4
7.4
4.2
-6.6
-10.7
7.7
1-7
-------
Ritchie
At one site we selected a variety of corn that would be better suited to the changed Fort Wayne
climate. Plant breeders will continue to develop new adapted varieties as the climate changes. The variety
chosen for simulation was one that is presently a longer season type adapted to Missouri and southern
Indiana.
Relatively simple modifications were made to the CERES-Maize and the SOYGRO models to take
into account the direct effects of CO2 on photosynthesis and transpiration in maize and soybean plants.
These modifications were designed to simulate the increase in stomatal resistance as it affects photosynthesis
and transpiration. The modifications were also done to determine the influence of increased CO2 in the
efficiency of biomass production. These modified models were run with the two weather scenarios for all
sites.
A more technical description of the modifications made for the direct effect of CO2 can be found in
the Southeastern agriculture section of this report by Peart et al. These modified models were run with the
two weather scenarios for all sites.
1-8
-------
Ritchie
CHAPTERS
RESULTS
DIRECT EFFECTS NOT CONSIDERED
The response of irrigated com yield to temperature change (the irrigated crops should not be affected
by precipitation changes) was negative at all latitudes simulated except for the 46°-48° degree zone
represented by Duluth (Tables 3 and 6). With the GISS model the mean yields decreased an average of
11%, but the percentage change ranged from -3% at Green Bay to -28% at Springfield. At Duluth, the GISS
scenarios increased yields by 86% (Table 6). With the more drastic increase in temperature predicted by the
GFDL scenario, irrigated corn yields decreased an average of 43% except for Duluth, which had a 36%
increase predicted (Table 3). These irrigated yields represent the best that presently adapted varieties can
do.
Rainfed corn yields followed the trend of decreases in irrigated yields, with decreases in most latitude
zones except the most northern ones (Tables 2 and 5). The GISS simulation decreases averaged about 16%
for most locations (Table S) with the increase at Duluth being about 49%. For the GFDL runs the rainfed
corn yield decreases averaged about 50%, with no increase at any latitude (Table 2). This reduction was
primarily associated with lower precipitation, although higher temperatures contributed to shorter growth
duration.
Soybean yields under irrigation changed much less than the irrigated corn yields. With the GISS runs,
on average there was little change at all locations except Duluth, which had a 181% increase (Table 12). The
GFDL runs for irrigated soybeans had yield decreases averaging about 14% except for Duluth, which has a
175% increase, and Buffalo with a 3% increase (Table 9).
Rainfed soybean yields with the GISS were reduced (by an average of 13%) for two- thirds of all
locations, while the other locations had an increase of yield ranging from 0.1% for Des Moines to 118% for
Duluth (Table 11). The GFDL weather reduced rainfed yields by about 55%, with Duluth increasing 6%
(Table 8).
Irrigation Water Demand
Water demand represents the amount of irrigation required to produce the irrigated yields. This
assumes 100% efficiency of application and availability of water at the desired time. Both of these conditions
are usually not possible to achieve because of unevenness of water applied by an irrigation system and an
inability to deliver the water at the exact time of need.
The baseline irrigation amounts for corn varied from between 46 mm at Buffalo to 288 mm at
Indianapolis (Tables 4 and 7). The GISS weather resulted in quite a mixture of changes in water required at
different locations, but on average there was little significant change (4.4%). This probably results from a
mixture of changing the growing season precipitation along with increasing temperatures and also decreasing
the duration of the total growing season. Water-holding capacity of the soil chosen for each location also has
an influence on irrigation water requirements.
For the GFDL scenario, irrigation water requirements were much higher owing to the large decrease
in rainfall in that model. Percentage increases for irrigation averaged about 50%, but the range of percent
change was very large, going from -7.4% in Indianapolis to +174% in Duluth (Table 4).
1-9
-------
Ritchie
For soybeans, the baseline irrigation requirement was somewhat less variable than for corn, with mean
values ranging from 84 mm at Duluth to 288 mm at Indianapolis (Tables 10 and 13). The GISS scenario
primarily caused an increase in water demand in the range of 10-40%, but at Springfield there was a -4%
change in demand for irrigation water (Table 13). The GFDL weather also caused large increases in
soybean irrigation demand, averaging about 90% increase (Table 10).
We are confident that the model's predicted direction of irrigation demand change is accurate. The
primary concern with regard to both water demand and yield is the possible direct effect of CO2 on
photosynthesis and transpiration as discussed in Chapter 2. In the section below we show the results of the
models modified to account for the direct effects of CO2. Also, we expect that new varieties will be found to
partially offset reduced yields. This point is discussed further in Chapter 4.
DIRECT CO2 EFFECTS CONSIDERED
The direct effect of CO2, in combination with the weather of the GISS and GFDL scenarios, lessens
the impact on yield caused by change of the weather alone. The direct effect had a larger beneficial
influence on soybean yields than it did on corn yields. Decreases in irrigated corn yield were somewhat less
with the direct effects than with the effects of weather alone (Tables 3 and 6). The range of irrigated corn
yield changes was approximately +5% to -48%, except in Duluth where there was an increase of 100% due
to the longer season. The rainfed corn usually yielded higher under the direct effect because of a better
water supply resulting from lowered transpiration. The rainfed yield changes ranged around -40% to +160%
(Tables 2 and 5). In general the northern sites had more positive yield changes than sites in the southern
region of the Great Lakes.
For most sites, rainfed and irrigated soybean in the GISS scenario (Table 11 and 12) and irrigated
soybeans in the GFDL scenarios (Table 9), the yield actually increased over the baseline yields. Yield
increases ranged from 12% to 465% with the greater increases being at the northern sites where the season
length is critical (Tables 9,11, and 12). For rainfed soybean in the GFDL scenarios, the yield increase
caused by the direct effects was not enough to overcome the yield decrease caused by the extreme weather in
about half the cases. The change in yields ranged from +163% to -84% with no clear geographic trend,
probably because of the strong influence soil type has on rainfed yield (Table 8).
The direct effect of CO2 decreased the irrigation water demand for corn. The greater decreases in
water demand were in the southern region (Tables 4 and 7) where as much as 172 mm less water would be
required for the soil and weather used for Indianapolis. The soybean GISS runs showed a decrease in
irrigation need when compared to weather effects alone, but most sites still had an increased demand when
compared to baseline. The direct effects of CO2 with the GISS weather changed soybean irrigation demand
from the baseline by values ranging from -13% to +32%, compared to the weather effects alone, which
changed demand in a range from -4% to +42% (Table 13). Under GFDL, the direct CO2 effects caused a
slight increase in demand in the north and a slight decrease in demand in the south when compared to
weather effects alone (Table 10). However, in all cases there was an increased demand (from 40% to 200%)
from the baseline weather.
FIGURES AND TABLES
Because the yield and irrigation water requirements vary both in space and time, we decided to provide
graphic information on the temporal variations in some important aspects of the corn study for a central
Great Lakes location: Fort Wayne, Indiana. This type of information helps to visualize the response more
easily than is possible in the summary tables.
The relatively small year-to-year differences in irrigated corn yield (Fig. 1) are mostly caused by
variations in growing season temperature. The yields produced by GISS weather change vary with the
1-10
-------
Ritchie
baseline yields in most years, while the yields with the GFDL weather change are considerably lower, have
less temporal variation, and do not follow the pattern of the baseline. This results from a shorter growing
season due to the approximate 8°C temperature increase predicted by the GFDL model.
The year-to-year variations in rainfed corn yield (Figure 2) are much greater than the irrigated yield
variations. Yields for GISS weather yields are somewhat more stable with almost the same mean as the
baseline yields. Yields from GFDL are also more stable than the baseline, but have a considerably reduced
mean value. This decrease is the result of both shorter seasons and less rainfall.
The influence of the length of the growing season on irrigated corn yield at Fort Wayne is shown in
Figure 3. Each datum point is a one-year simulation result for the three weather scenarios. This information
clearly demonstrates idea that a temperature increase causes decreases in yield by decreasing the growing
season in regions where the baseline temperature is high enough to provide a full season corn crop. The
season length varies considerably between weather patterns, with the baseline season averaging 138 days, the
GISS 116 days, and the GFDL 92 days.
The CERES-Maize and SOYGRO models calculate a soil-water deficit factor every day in order to
modify the calculated potential growth and transpiration rates for water deficit conditions. Equations in the
soil-water submodel calculate the potential transpiration rate and the potential root absorption rate. The
soil-water deficit factor is zero when the potential absorption rate equals or exceeds the potential
transpiration rate. When the potential absorption is less than the potential transpiration rate, the soil-water
deficit factor increases as the absorption rate decreases. The deficit conditions are averaged for various
stages of plant development to help interpret the extent to which water deficits influence yield. The water
deficit factor is calculated so that a value of zero means no deficit during the season and a value of one
means very large plant-water deficit. For the rainfed corn simulation at Fort Wayne, the largest soil water
deficit conditions are for the GFDL weather with its lower summer rainfall (Figure 4 and Table 1C). GISS
and baseline weather have rather similar average deficit factors, with minimum values at or near zero and
maximum values at about 0.2 and 03, respectively. This result is in agreement with observations that rainfed
corn yields are quite good in much of the Great Lakes region because of the reliable supply of rainfall during
the growing season.
The year-to-year variations in irrigation water demand were so great that it seemed most appropriate
to express the results in cumulative irrigation amounts (Figure 5) for corn at Fort Wayne. The GISS weather
pattern, on average, required less irrigation than the baseline or the GFDL. Average yearly irrigation
amounts were about 67, 87, and 120 mm for the GISS, baseline, and GFDL scenarios, respectively. The
drier years with greatest demands would require about 175% of the averages.
Ml
-------
Ritchie
^o
£ 6-1
O 4-|
2-
.
/ \/\ /\/\ A A
\ X-, / " v v/ v ^^-
1951 1956 1961 1966 1971
Year
V
— Baseline
GISS 2xCO.
•- GFDL 2xC02
1976 1981
Figure 1. Annual variation in irrigated corn yield over 30 years comparing baseline weather with GISS and
GFDL weather.
1-12
-------
Ritchie
1951
— Baseline
GISS
- GFDL 2xCO,
1981
Figure 2. Annual variation in rainfed yield over 30 years comparing baseline weather with GISS and GFDL
weather.
1-13
-------
Ritchie
I
z
g
•?
-------
Ritchie
16-
14-
12-
o>
-------
Ritchie
-- GFDL 2xCO,
1956 1961 1966
Year
1971 1976 1981
Figure 5. Water demand for irrigated corn accumulated over 30 years for corn under baseline, GISS, and
GFDL weather.
1-16
-------
Ritchie
As an estimation of a possible way to compensate for the lower yields caused by shorter seasons,
longer season corn cultivar (B73 * M017) was compared with the adapted one chosen for the present study
(PIO 3720) at Fort Wayne. For comparison, Figures 6 and 7 show the temporal variation in the irrigated
and rainfed corn yields of PIO 3720 for baseline and GFDL weather. For this comparison, we chose GFDL
because of the two climate models GFDL has the more extreme temperature differences from the baseline.
Note that although the compensation due to cultivar change does not bring the yield to the baseline, this
result provides evidence that already existing cultivars can be introduced into a region that will help overcome
some of the detrimental effects of higher temperature on corn yields. The primary reason why the
compensation was not better is that the grain filling period has less genetic variation available than the
vegetative growth period. However, as climate changes in the future, plant breeders may be able to find
sources of breeding material with longer grain fill duration that can be incorporated into high yielding corn
for better yields at higher temperatures.
SUMMARY GRAPH
The most significant information from this study are yields of com and soybean under the different
climate scenarios (Figures 8 and 9). The irrigated yields are presented because they provide information on
how temperature change alone affects yields. The latitude of the sites studied had some influence on
baseline yields and on yield response to climate change. Baseline yields of com are a maximum in the 44°-
46°. latitude zone, although those in the 40°-44° zones are almost as good. These zones include the present
corn belt. With the GISS temperature increase of about 4°C during the growing season, the maximum
irrigated yields shift farther north, being maximum at 46°-48°. The GFDL temperature increase of about 6°
to 8°C in the region lowered average yields compared to the other scenarios, and also shifted the maximum
yielding area to latitudes farther north.
Irrigated soybean yields generally tend to be higher at the lower latitudes for both climate change
scenarios and for the baseline. This is likely the result of a greater dependence of the soybean's season
length on photoperiod than is the case with corn. The 4°C GISS temperature increase greatly improved the
higher latitude yields over the baseline yields by increasing the otherwise unusually short and uncertain
growing season length. The GFDL temperature increases reduced yields by about 20% at all latitudes except
the 46"-48° latitude region.
LIMITATIONS
The primary limitations on the results from this study center on four issues:
1. Direct effects of increased CO2 on photosynthesis and biomass production rates: Higher
temperatures increase plant respiration, but higher CO2 increases gross photosynthesis. Research
to date indicates that the net biomass production rates will be greater for most crop plants with
higher CO2- Since the CERES-Maize model primarily produced yield reductions of corn due to
shorter seasons, the bias in this limitation would cause a low estimate of biomass production in
corn without greatly affecting yields. Grain filling rates at temperatures above 18°C are constant,
yet the temperature influences duration of grain fill in the range from 8°C to 34°C.
1-17
-------
Ritchie
o
o
- PIO 3720, Boseline
B73«M017. 2xCO,
-- PIO 3720. 2xCO,
1951
1956
1961
1966
Year
1971
1976
1981
Figure 6. Yield of irrigated com (variety PIO 3720) grown under baseline and GFDL weather conditions
and variety B73*MO17 grown under GFDL weather conditions.
1-18
-------
Ritchie
1951
PIO 3720. Baseline
B73«MO17. 2xCO,
PIO 3720. 2xCO,
1981
Figure 7. Yield of dryland corn (variety PIO 3720) grown under baseline and GFDL weather conditions and
variety B73*MO17 grown under GFDL weather conditions.
1-19
-------
Ritchie
0)
14-,
12-
10-
8-
4-
2-
0
GFDL 2xC02
GISS 2xC02
Baseline
46-48 44-46
42-44
Latitude
40-42 38-40
Figure 8. Comparison of average irrigated corn yield for five latitude zones in the Great Lakes region with
baseline, GISS, and GFDL weather.
1-20
-------
Ritchie
3.5-1
3.0-
2.0-1
2
1.0-
0.5-
0.0
GFDL 2xC02
GISS 2xC02
Baseline
46-48 44-46
42-44
Latitude
40-42 38-40
Figure 9. Comparison of average irrigated soybean yield for five latitude zones in the Great Lakes region
with baseline, GISS, and GFDL weather.
1-21
-------
Ritchie
2. Direct effects of increased CO. on transpiration: In rainfed crop production, the possible direct
effect of CO2 causing a reduction in transpiration would result in some yield improvement over
the present modeled values as demonstrated in the direct effect results. However, this would
occur only in years and locations where a significant soil water deficit reduced yields in the
baseline analysis.
3. Effects of changing climate on pests: The assumption that pests do not affect yields is always a
limitation in yield estimation. Increased temperatures will create different environments for
diseases, insects, and weeds such that difficult, new pest problems may emerge. Any associated
humidity changes will also affect pests, especially leaf diseases. Higher temperatures without
higher vapor pressures would lower the relative humidity, thus decreasing some disease
susceptibility.
4. Technical limitations in the model assumptions and relationships exclusive of pest problems: The
limitations inherent in the models were discussed in an earlier section of this document and will
not be repeated here. However, three limitations that could influence the results of this study
need further discussion.
A. Biomass production rates are calculated using the concept of a constant light conversion
efficiency. This efficiency is expressed in grams biomass per megajule of intercepted
photosynthetic radiation. Temperature extremes and soil-water deficits decrease the
efficiency. This simplified analysis of biomass production assumes that respiration is
proportional to the gross photosynthesis. Since respiration increases at higher temperatures,
the bias in the modeled biomass without the direct affects of CO2 would be toward lower
production than predicted.
B. Partitioning of the biomass is one of the greatest uncertainties in model descriptions of a
complex biophysical system. Partitioning is dynamic and has feedback systems that are
difficult to measure and model. The CERES-Maize model calculates a potential rate of
growth of aboveground organs at various stages of growth. This potential rate is
independent of biomass production rate and is usually dependent on temperature. It is only
when the biomass production rate cannot support the potential organ growth rates that top
growth is reduced. Otherwise, biomass not needed for top growth produces extra root
growth. The balancing of this dynamic partitioning of biomass between growing organs on a
plant is probably the major source of error in simulation models.
C. Determination of the number of seed or kernels that will fill is a major determinant of yield.
This number is usually determined during a relatively short time when plants change from
growing leaves to growing grains. In CERES-Maize, grain number is calculated as a
function of the rate of biomass production during pollination and cob growth. This is about
20 days in the Great Lakes region with baseline temperatures and would be somewhat
shorter with increased temperatures.
1-22
-------
Ritchie
CHAPTER 4
INTERPRETATION OF THE RESULTS
In our analysis of climate change in the Great Lakes region, it seems clear that the increase in
temperature is the major cause of yield reduction of both corn and soybeans. The smaller temperature
increase of the GISS model caused smaller yield reductions compared to those calculated using the GFDL
model The primary exception to this trend was the northernmost location, Duluth, which has a short and
variable frost-free growing season for corn and soybeans under the baseline conditions. With temperature
increases, northern areas can have longer, more stable seasons, thus providing an improved environment for
crop production.
The changes in precipitation for the two climate scenarios studied had a relatively minor influence on
the results of the study as is evident from the rainfed yields for both corn and soybeans. The small increase
in precipitation during most of the growing season for the GISS model actually decreased the duration of
water deficit periods as compared to the baseline weather. The major reductions in precipitation during June
and July for the GFDL weather increased water deficit durations when compared to the baseline weather.
However, because of the considerably shorter plant life cycle caused by higher temperatures, the water deficit
influence was of secondary importance.
The water deficit problem caused by the GFDL weather would be more important on soils with low
water holding capacity in the root zone such as sandy soils, shallow soils, or soils with high water tables in the
spring that lower during the summer.
If higher temperatures reduce general crop production capability in much of the U.S., the Great Lakes
region should become more critical for crop production because of water availability for supplemental
irrigation. Crops grown in humid regions have lower irrigation water requirements than crops in arid regions
that presently depend heavily on diminishing water supplies for crop production. If the humid regions are to
provide a stable supply of food, supplemental irrigation will be needed in many soils to produce the yield
levels that were simulated for this report. In drier years in the Great Lakes region water needs are approxi-
mately 25 cm for baseline corn and soybean irrigation, but the average need is less than 10 cm. However,
the irrigation systems and water supply must be designed for the more severe drought conditions.
We have attempted to demonstrate that a part of the yield reduction in corn related to shorter seasons
can be overcome with selection of new cultivars that have a longer growing season. There is a limit,
however, to the degree of genetic material that is available for selection of cultivars that completely
compensate for weather changes. When the limit of season length change for cultivars of a crop is reached
and there is still more growing season left, then double cropping can be practiced to increase annual
production. This is a common practice in the southeastern U.S. where soybeans often follow a winter wheat
crop or where two corn crops are grown in one season.
The CERES-Maize and SOYGRO models are sensitive to temperature as it affects plant development
and to the soil water balance as it affects plant stress. Gradual changes in these two variables should be
properly accounted for in the models. The direct effects of CO2 on photosynthesis and transpiration were
unaccounted for in versions of the models developed from field data in recent literature.
However, the direct effects as incorporated in the model usually provided some compensation for yield
losses caused by shortened seasons. It would be rather certain that for much of the season when leaf area
index is relatively low, there would be less water used by the crop and thus there would be a reduction in
plant stress and irrigation water requirement.
1-23
-------
Ritchie
CHAPTERS
IMPLICATION OF RESULTS
ENVIRONMENTAL IMPLICATIONS
With temperature changes predicted by the GISS and GFDL models used, there would be major
changes in the best type of crop to grow, especially in the northernmost latitudes of the eastern VS. and the
central latitudes around 38°. The northern states would become much more productive for annual crops like
corn and soybeans because of the lengthening of the frost-free period. However, many of the glacial till soils
in the northern latitudes are not as productive as the corn belt soils. Thus, large increases in crop production
would require irrigation to supplement rainfall and create a greater need for chemical fertilizers.
The careless use of both irrigation and fertilizers in humid regions on sandy soils creates a definite
environmental hazard to the region's groundwater. It would be necessary to carefully manage water and
fertilizer in such a way that chemical concentrations of water-soluble nutrients, like nitrogen and potash,
would be kept to a minimum to prevent leaching, yet be kept high enough to support good plant growth.
This goal can be achieved by frequently applying the fertilizers in small quantities to "spoon-feed" the plants
as they need the nutrients rather than by applying everything at once and expecting the soil to retain it until
the plants use it. Most major commercial farmers currently apply fertilizer once before planting to save time
and money. Improved soil and water management policy would need to deal with this problem if more of
the fragile lands along the eastern UJS.-Canada border are to be cultivated.
The increased length of the growing season in the northern United States could increase the demand
for farmland in this region. Because of the presence of forests and wetlands in the northern VS., care will
have to be taken to choose land most suitable for cropland. Environmental impact studies, with a view
toward ground and surface water contamination problems, will need to be conducted before starting drainage
projects and deforestation to create new cropland.
In the southern Great Lakes region, the frost-free seasons would be lengthened considerably and the
crop growth cycles would be shortened. To cope with this, new varieties would be needed to lengthen the
growing cycle. However, there is a limit to how much this would achieve within a crop species. One
alternative would be to plant crops such as perennials that can be grown as annuals, e.g. cotton, a crop that
continues its growing cycle through the whole season. Another alternative would be to grow two crops within
one season in the southern Great Lakes region. Soils there are usually quite good
The water supply for the irrigation needed to stabilize production in the humid regions would come
from both surface and subsurface sources. Detailed studies would be needed to evaluate the impact of
irrigation on streamflow and water table levels. The main problem will be to have a stable water supply
when the region is quite dry during droughts. Storage sites and use of water from the Great Lakes would
also need to be considered.
1-24
-------
Ritchie
REFERENCES
Baker, J.T., Allen, L.H. Jr., and Beladi, S.E. "Simulations of interactions of climate, CO2 and leaf area on
crop water-use efficiency." Agronomy Abstracts, pp.10,1985.
Benci, J.F., E.CA. Runge, R.F. Dale, W.G. Duncan, R.B. Curry, and LA. Schaal. "Effects of hypothetical
climatic changes on production and yield of corn." CLAP Monograph 5. Part 2. Climatic Effects. NTIS PB-
247-726, U.S. Department of Commerce, Department of Transportation, Washington, D.C. 1975. 36 pp.
Brouwer, R. "Root growth of cereals and grasses." In: F.L. Milthorpe and J.D. Ivins (ed.) The Growth of
Cereals and Grasses. Bubberworths, London. 1965. pp. 153-166.
Carter, T.R., Konjin, N.T., Watts, R.G. "The role of agroclimatic models in climate impact analysis."
International Institute for Applied Systems Analysis. Working Paper 84-98. 2361 Laxenburg, Austria, 1984.
pp.26.
Curry, R.B., and C.H. Baker. "Climatic change as it affects soybean growth and development." CLAP
Monograph 5. Part 2. Climatic Effects. NTIS PB-247-726, VS. Department of Commerce, Department of
Transportation, Washington, D.C. 1975.16 pp.
Duncan, W.G. and Barfield, B J. "Predicting effects of CO, enrichment with simulation models and a digital
computer." Trans, of the ASAE, 13:246-248,1970.
Jones, CA. and J.R. Kinky (ed.). CERES-Maize: A simulation of maize growth and development. Texas
A&M Press, College Station, Texas, 1986.194 pp.
Richardson, C.W. "Weather simulation for crop management models." Trans, of ASAE, 28:5,1602-1606,1985.
Smit, B. "Implications for climate change for agriculture in Ontario." Summary of land evaluation group
reports. Climate Change Digest. Atmospheric Environment Science, Environment Canada. 1987.
Stewart, R.B. "Climatic change - Implications for the prairies." Paper presented at the Royal Society for
Canada Symposium, Winnipeg, Manitoba 1986.
UJS, Department of Commerce. "Impacts of climatic change on the biosphere." CLAP Monograph 5. Part 2.
Climatic Effects. NTIS PB-247 726. Washington, D.C. 1975.
van Keulen, H., van Laar, H.H., Louwerse, W., and Goudriaan, J. "Physiological aspects of increased CO2
concentration." Experientia, 36:786-792.1981.
Whisler, F.D., B. Acock, D.N. Baker, R.E. Fry, H.F. Hodges, J.R. Lambert, H.E. Lemmon, J.M McKinion,
and V.R. Reedy. "Crop simulation models in agronomic systems." Adv. in Agron. 40:141-208.1986.
Wilkerson, G.G., J.W. Jones, KJ. Boote, KJ. Ingram, and J.W. Mishoe. "Modeling soybean growth for crop
management." Trans, of the ASAE, 26:1, 63-73,1983.
WMO. "Report of the WMO/UNEP/ICSU-SCOPE expert meeting on the reliability of crop-climate models
for assessing the impacts of climatic change and variability." WCP-90,1984. 31 pp.
1-25
-------
Ritchie
APPENDIX A: PROPERTIES OF SOILS USED
For each soil there are various lines of information. Line one contains:
Bare soil albedo, unitless.
Upper limit of stage 1 soil evaporation, mm.
Soil water drainage constant, fraction drained per day.
SCS curve number used to calculate daily runoff, unitless.
For the remaining lines there is information about each of the n soil layers:
Thickness of soil layer, cm.
Lower limit of plant-extractable soil water of soil layer, cm3/cm3.
Drained upper limit soil water content for soil layer, cm3/cm3.
Saturated water content for soil layer, cm /cm3.
Initial soil water content for soil layer cm3/cm3.
Weighting factor for soil layer to determine new root growth distribution, unitless.
Moist bulk density of soil in soil layer, g/cm3.
Organic carbon concentration in soil layer, %.
MEDIUM SILTY CLAY
.11 6.00 JO 87.00
10. .215 361 .416 361 1.000 135 1.74
15. 216 361 .415 361 .819 136 1.66
20. .218 361 .414 361 .607 136 1.45
25. .221 361 .412 361 .407 137 1.12
30. .225 360 .409 360 .247 137 .73
30. .228 360 .407 360 .135 138 37
1-26
-------
Ritchie
DEEP SILTY CLAY
.11 6.00 JO 85.00
10. 215 361 .416 361 1.000 135 1.74
15. .216 361 .415 361 .819 136 1.66
25. .218 361 .414 361 .607 136 1.45
30. .221 361 .412 361 368 137 1.09
30. .225 360 .409 360 .202 138 .65
30. .229 360 .407 .360 .111 138 .29
30. .231 360 .405 360 .061 139 .09
30. .231 360 .405 360 .033 139 .01
SHALLOW SILTY LOAM
.12 6.00 .20 81.00
10. .0% .245 .415 .245 1.000 136 1.16
15. .097 .245 .415 .245 .819 136 1.10
15. .098 .245 .414 .245 .607 136 .97
20. .100 .245 .413 .245 .449 136 .77
MEDIUM SILTY LOAM
.12 6.00 30 79.00
10. .0% .245 .415 .245 1.000 136 1.16
15. .097 245 .415 245 .819 136 1.10
20. .098 .245 .414 245 .607 136 .97
25. .100 245 .413 .245 .407 136 .75
30. .103 .245 .411 .245 247 137 .49
30. .105 .245 .409 .245 .135 137 .24
DEEP SILTY LOAM
.12 6.00 .40 77.00
10. .0% .245 .415 245 1.000 136 1.16
15. .097 .245 .415 .245 .819 136 1.10
25. .098 245 .414 .245 .607 136 .97
30. .100 .245 .412 .245 368 136 .72
30. .103 .245 .411 .245 .202 137 .43
30. .105 .245 .409 .245 .111 137 20
30. .107 .245 .408 245 .061 138 .06
30. .107 .245 .408 .245 .033 138 .01
1-27
-------
Ritchie
SHALLOW SANDY LOAM
.13 6.00 .40 74.00
10. .082 .211 342 .211 1.000 1.58 .70
15. .083 .211 342 .211 .819 1.58 .66
15. .083 .211 341 .211 .607 1.59 58
20. .085 .211 340 .211 .449 159 .46
MEDIUM SANDY LOAM
.13 6.00 50 70.00
10. .082 .211 342 .211 1.000 158 .70
15. .083 .211 342 .211 .819 158 .66
20. .083 .211 341 .211 .607 159 58
25. .085 .211 340 .211 .407 159 .45
30. .086 .211 339 .211 .247 159 .29
30. .088 .211 338 .211 .135 1.60 .15
DEEP SANDY LOAM
.13 6.00 50 68.00
10. .082 .211 342 .211 1.000 158 .70
15. .083 .211 342 .211 .819 158 .66
25. .083 .211 341 .211 .607 159 58
30. .085 211 340 211 368 159 .43
30. .087 211 339 .211 .202 159 .26
30. .088 211 338 211 .111 1.60 .12
30. .089 211 337 211 .061 1.60 .04
30. .089 211 337 211 .033 1.60 .01
30. .089 211 337 211 .018 1.60 .00
SHALLOW SAND
.15 4.00 .40 75.00
10. .030 .106 319 .106 1.000 1.66 .29
15. .031 .106 319 .106 .819 1.66 .28
15. .031 .106 318 .106 .607 1.66 .24
20. .031 .106 318 .106 .449 1.66 .19
1-28
-------
Ritchie
MEDIUM SAND
.15 4.00 - JO 70.00
10. .030 .106 319 .106 1.000 1.66 .29
15. .031 .106 319 .106 .819 1.66 .28
20. .031 .106 318 .106 .607 1.66 .24
25. .031 .106 318 .106 .407 1.66 .19
30. .032 .106 317 .106 .247 1.66 .12
30. .033 .106 317 .106 .135 1.66 .06
DEEP SAND
.15 4.00 .60 65.00
10. .030 .106 319 .106 1.000 1.66 .29
15. .031 .106 319 .106 .819 1.66 .28
25. .031 .106 318 .106 .607 1.66 .24
30. .032 .106 318 .106 368 1.66 .18
30. .032 .106 317 .106 .202 1.66 .11
30. .033 .106 317 .106 .111 1.66 .05
30. .033 .106 316 .106 .061 1.66 .01
30. .033 .106 316 .106 .033 1.66 .00
30. .033 .106 316 .106 .018 1.66 .00
1-29
-------
Ritchie
APPENDIX B: TABLES OF RUN RESULTS
For both crops for each of the two scenarios there are three tables comparing results of the new
weather with the baseline weather at each of the stations. These three tables compare non-irrigated yields,
irrigated yields, and water demand for the irrigated crops. The stations are broken up into five groups based
on latitude:
Group 1 46°-48°
Group 2 44°-46°
Group 3 42°-44°
Group 4 40°-42°
GroupS 38°-40°
For each station there are 10 columns of information (5 each for weather effects and weather effects
with direct effects of CO2) about either crop yield or water demand:
Mean of variable (yield or water demand) for 30 years under baseline weather.
Mean of variable under new weather conditions (GISS or GFDL).
Mean of the difference between present and new weather.
Percent change of variable from baseline weather to the new weather.
Percent of uncertainty of the change from present to new weather. Uncertainty is calculated to be
the quotient of the standard deviation of the difference of the means divided by the mean of the
baseline value. This allows a relative comparison between the variability of yield and irrigation
demand differences due to climate changes.
1-30
-------
Table 2: Great Lakes Corn Yield (Kg/ha) -• Baseline and GFDL 2xC02 -- Rainfed
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, NN
II. St. Cloud, NN
Green Bay, WI
III. Madison, WI
Flint, HI
Nuskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Noines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Willfamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
5804.5
3769.2
5986.1
8651.1
7006.6
2940.9
11054.8
12068.5
7429.3
7751.8
11357.7
10943.2
9037.2
8939.6
6069.1
6399.3
3595.7
9358.7
Mean
2xC02
4049.5
945.2
2375.1
3744.4
3678.1
845.9
5470.0
6380.1
2427.4
3518.7
5555.7
5457.9
4094.6
4006.1
4805.0
4270.3
2135.4
5043.0
Mean
Oiff
-1754.9
-2824.0
-3611.0
-4906.7
-3328.5
-2095.0
-5584.8
-5688.4
-5001.9
-4233.2
-5802.1
-5485.4
-4942.6
-4933.5
-1264.1
-2129.1
-1460.3
-4315.7
Mean
Xchange
-30.2
-74.9
-60.3
-56.7
-47.5
-71.2
-50.5
-47.1
-67.3
-54.6
-51.1
-50.1
-54.7
-55.2
-20.8
-33.3
-40.6
-46.1
Uncer
Xchange
13.2
13.1
10.6
7.0
9.4
15.5
4.0
2.8
8.4
7.9
4.3
5.3
7.1
6.8
12.6
11.0
16.2
3.7
Mean
Baseline
5804.5
3769.2
5986.1
8651.1
7006.6
2940.9
11054.8
12068.5
7429.3
7751.8
11357.7
10943.2
9037.2
8939.6
6069.1
6399.3
3595.7
9358.7
Mean
2xC02
9652.7
3487.3
6598.2
7220.9
7993.7
3079.8
10307.6
8502.3
6027.8
5060.1
7312.6
7987.9
7886.7
7500.9
7164.1
7347.1
4853.3
5753.2
Mean
Diff
3848.3
-281.9
612.1
-1430.2
987.2
138.9
-747.2
-3566.2
-1401.5
-2691.7
-4045.1
-2955.3
-1150.5
-1438.6
1095.1
947.8
1257.6
•3605.5
Mean
Xchange
66.3
-7.5
10.2
-16.5
14.1
4.7
-6.8
-29.5
-18.9
-34.7
-35.6
-27.0
-12.7
-16.1
18.0
14.8
35.0
-38.5
Uncer
Xchange
13.6
16.7
11.6
7.0
9.4
18.0
4.2
2.7
9.0
7.6
4.3
4.9
6.7
6.4
12.0
10.8
18.4
3.8
-------
Table 3: Great Lakes Corn Yield (Kg/ha) -- Baseline and 6FDL 2xC02 -- Irrigated
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Ouluth, MN
II. St. Cloud. NN
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Will import, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
7046.0
13424.8
12601.7
12734.9
12462.5
12509.5
11912.7
12402.8
12709.5
9793.8
12979.0
12912.0
12956.9
11771.2
11632.6
11304.5
9705.6
9608.1
Mean
2xC02
9592.7
7441.2
7094.4
6942.3
7737.4
7941.9
6800.8
7756.0
6196.2
4634.6
6694.1
7328.6
7156.8
7056.7
6923.2
6932.0
6044.8
5278.5
Mean
Oiff
2546.7
-5983.6
-5507.3
-5792.6
-4725.1
-4567.6
-5112.0
-4646.7
-6513.3
-5159.2
-6284.9
-5583.4
-5800.1
-4714.5
-4709.4
-4372.5
-3660.7
-4329.5
Mean
Xchange
36.1
-44.6
-43.7
-45.5
-37.9
-36.5
-42.9
-37.5
-51.2
-52.7
-48.4
-43.2
-44.8
-40.1
-40.5
-38.7
-37.7
-45.1
Oncer
Xchange
11.5
2.9
3.5
3.0
3.5
2.6
2.8
2.8
2.9
3.3
2.9
2.7
2.6
3.1
3.7
4.2
3.4
3.4
Mean
Baseline
7046.0
13424.8
12601.7
12734.9
12462.5
12509.5
11912.7
12402.8
12709.5
9793.8
12979.0
12912.0
12956.9
11771.2
11632.6
11304.5
9705.6
9608.1
Mean
2xC02
10562.3
8255.2
7820.9
7692.5
8567.3
9141.7
7500.9
8502.2
7107.0
5089.0
7310.9
7991.4
7828.4
7706.3
7538.3
7608.7
6590.1
5724.9
Mean
Oiff
3516.3
-5169.6
-4780.8
-5042.4
-3895.1
-3367.8
-4411.8
-3900.6
-5602.6
-4704.8
-5668.1
-4920.6
-5128.4
-4064.9
-4094.4
-3695.8
-3115.5
-3883.2
Mean
Xchange
49.9
-38.5
-37.9
-39.6
-31.3
-26.9
-37.0
-31.4
-44.1
-48.0
-43.7
-38.1
-39.6
-34.5
-35.2
-32.7
-32.1
-40.4
Uncer
Xchange
11.7
2.9
3.6
3.2
3.7
2.7
2.8
2.8
3.2
3.4
3.0
2.8
2.7
3.2
3.9
4.4
3.5
3.5
-------
Table 4: Great Lakes Corn Irrigation Demand (ma/year) — Baseline and GFDL 2xC02
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, MN
II. St. Cloud, MN
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Will import, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Coluntous, OH
Mean
Baseline
73.6
238.7
193.0
132.8
156.7
253.2
59.2
45.7
163.5
131.2
82.9
80.4
144.8
111.8
179.5
180.4
287.9
61.8
Mean
2xC02
201.6
281.2
234.5
169.3
188.9
266.7
122.7
110.9
214.2
176.1
118.4
135.5
190.4
184.5
166.0
187.1
266.7
125.3
Mean
Diff
127.9
42.5
41.5
36.5
32.2
13.5
63.5
65.2
50.6
44.9
35.5
55.1
45.6
72.7
-13.5
6.7
-21.2
63.5
Mean
Xchange
173.7
17.8
21.5
27.5
20.6
5.3
107.2
142.8
31.0
34.2
42.8
68.6
31.5
65.1
-7.5
3.7
-7.4
102.9
Uncer
Xchange
20.1
5.8
8.5
12.1
11.1
5.9
22.9
26.7
9.1
14.4
19.6
20.9
12.1
17.8
10.7
9.4
5.4
24.4
Mean
Baseline
73.6
238.7
193.0
132.8
156.7
253.2
59.2
45.7
163.5
131.2
82.9
80.4
144.8
111.8
179.5
180.4
287.9
61.8
Mean
2xC02
54.3
135.5
78.7
47.3
48.3
133.8
.0
5.9
80.5
38.8
6.7
14.3
40.6
44.8
55.8
56.6
116.0
10.1
Mean
Diff
-19.4
-103.2
-114.4
-85.5
-108.4
-119.4
-59.2
-39.8
-83.1
-92.4
-76.2
-66.0
-104.2
-67.0
-123.7
-123.7
-171.9
-51.6
Mean
Xchange
-26.3
-43.2
-59.2
-64.4
-69.2
-47.2
-100.0
-87.1
-50.8
-70.4
-91.9
-82.2
-71.9
-59.9
-68.9
-68.6
-59.7
-83.6
Uncer
Xchange
17.1
5.5
7.9
10.4
9.9
5.2
17.0
19.5
8.4
13.0
15.4
16.2
10.9
ll.1
9.0
8.2
5.1
18.8
-------
Table 5: Great Lakes Corn Yield (Kg/ha) -- Baseline and 6ISS 2xC02 -- Rainfed
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I.
Duluth, MN
II. St. Cloud. MN
Green Bay. WI
III. Madison, WI
Flint, MI d.
Muskegon Co.. MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Will import, PA
V. Springfield, NO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
5804.5
3769.2
5986.1
8651.1
7006.6
2940.9
11054.8
12068.5
7429.3
7751.8
11357.7
10943.2
9037.2
8939.6
6069.1
6399.3
3595.7
9358.7
Mean
2XC02
8635.6
3943.3
5547.6
7588.3
5790.0
2522.9
8874.0
8948.1
6517.4
6549.7
10141.1
8133.0
7087.9
7245.1
6674.0
6302.5
3630.4
6998.1
Mean
Diff
2831.1
174.1
-438.4
-1062.8
-1216.6
-418.0
-2180.8
-3120.4
-911.9
-1202.1
-1216.7
-2810.3
-1949.3
-1694.4
604.9
-96.9
34.7
-2360.6
Mean
Xchange
48.8
4.6
-7.3
-12.3
-17.4
-14.2
-19.7
-25.9
-12.3
-15.5
-10.7
-25.7
-21.6
-19.0
10.0
-1.5
1.0
-25.2
Uncer
Xchange
15.9
17.2
12.9
8.5
11.3
18.4
5.5
3.5
10.0
8.4
4.3
6.5
9.1
8.0
13.4
12.3
18.2
3.8
Mean
Baseline
5804.5
3769.2
5986.1
8651.1
7006.6
2940.9
11054.8
12068.5
7429.3
7751.8
11357.7
10943.2
9037.2
8939.6
6069.1
6399.3
3595.7
9358.7
Mean
2xC02
14241.9
10066.6
13075.1
11998.8
11096.1
7785.9
11613.1
11191.9
11238.5
8371.8
12074.5
11583.3
11331.0
10401.4
8923.1
9261 .4
7316.2
7925.0
Mean
Diff
8437.4
6297.4
7089.0
3347.7
4089.6
4845.0
558.4
-876.6
3809.2
620.0
716.7
640.0
2293.8
1461.8
2854.1
2862.0
3720.5
-1433.7
Mean
Xchange
145.4
167.1
118.4
38.7
58.4
164.7
5.1
-7.3
51.3
8.0
6.3
5.8
25.4
16.4
47.0
44.7
103.5
-15.3
Uncer
Xchange
12.1
20.2
10.6
7.6
10.1
26.9
4.6
3.0
9.5
7.5
4.4
5.1
7.2
6.8
12.0
10.9
18.6
3.6
-------
Table 6: Great Lakes Corn Yield (Kg/ha) -- Baseline and GISS 2xCO2 -- Irrigated
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, NN
II. St. Cloud, HN
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
7046.0
13424.8
12601.7
12734.9
12462.5
12509.5
11912.7
12402.8
12709.5
9793.8
12979.0
12912.0
12956.9
11771.2
11632.6
11304.5
9705.6
9608.1
Mean
2xC02
13116.1
11633.4
12167.7
11025.4
10775.2
11350.2
10659.9
10231.1
10517.9
7685.1
11015.0
10403.7
10467.7
9395.9
8398.2
8586.3
7806.2
7244.7
Mean
Diff
6070.1
-1791.4
-434.0
-1709.5
-1687.3
-1159.4
-1252.9
-2171.7
-2191.6
-2108.7
-1963.9
-2508.4
-2489.2
-2375.3
-3234.5
-2718.2
-1899.4
-2363.4
Mean
Xchange
86.2
-13.3
-3.4
-13.4
6 -13.5
-9.3
-10.5
-17.5
-17.2
-21.5
-15.1
-19.4
-19.2
-20.2
-27.8
-24.0
-19.6
-24.6
Uncer
Xchange
10.8
2.4
3.2
3.1 '
3.6
2.7
3.4
3.0
3.1
3.2
3.0
2.9
3.0
3.4
3.9
4.3
3.1
3.2
Mean
Baseline
7046.0
13424.8
12601.7
12734.9
12462.5
12509.5
11912.7
12402.8
12709.5
9793.8
12979.0
12912.0
12956.9
11771.2
11632.6
11304.5
9705.6
9608.1
Mean
2xC02
14308.5
12777.7
13286.4
12070.9
11867.9
12651.1
11627.2
11191.9
11833.5
8405.4
12074.5
11589.9
11383.9
10307.6
9220.8
9412.2
8504.8
7925.0
Mean
Oiff
7262.5
-647.1
684.7
-664.0
-594.6
141.6
-285.5
-1210.9
-876.0
-1388.4
-904.5
-1322.1
-1572.9
-1463.7
-2411.8
-1892.3
-1200.8
-1683.1
Mean
Xchange
103.1
-4.8
5.4
-5.2
-4.8
1.1
-2.4
-9.8
-6.9
-14.2
-7.0
-10.2
-12.1
-12.4
-20.7
•16.7
-12.4
-17.5
Uncer
Xchange
10.8
2.5
3.3
3.3
3.8
2.9
3.7
3.2
3.3
3.4
3.1
3.0
3.2
3.6
4.1
4.5
3.3
3.4
-------
Table 7: Great Lakes Corn Irrigation Demand (mm/year) -- Baseline and Giss 2xC02
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, MM
II. St. Cloud. MN
Green Bay, WI
HI. Madison, WI
Flint, MI
Nuskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Noines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williansport, PA
V. Springfield. MO
St. Louis, MO
Indianapolis, IN
Colunbus, OH
Mean
Baseline
73.6
238.7
193.0
132.8
156.7
253.2
59.2
45.7
163.5
131.2
82.9
80.4
144.8
111.8
179.5
180.4
287.9
61.8
Mean
2xC02
133.7
217.5
200.0
124.3
159.2
244.8
93.2
66.9
145.7
114.2
65.9
93.2
145.7
115.1
122.8
143.1
253.1
59.2
Mean
Oiff
60.1
-21.2
6.9
-8.5
2.5
-8.4
34.0
21.2
-17.8
-17.0
-17.0
12.8
.8
3.3
-56.7
-37.3
-34.7
-2.5
Mean
Xchange
81.6
-8.9
3.6
-6.4
1.6
-3.3
57.3
46.4
-10.9
-12.9
-20.5
15.9
.6
3.0
-31.6
-20.7
-12.1
-4.1
Uncer
Xchange
22.7
6.5
8.6
13.0
12.3
6.2
25.7
27.6
10.4
15.3
19.7
21.5
13.1
19.9
10.8
10.0
5.5
23.9
Mean
Baseline
73.6
238.7
193.0
132.8
156.7
253.2
59.2
45.7
163.5
131.2
82.9
80.4
144.8
111.8
179.5
180.4
287.9
61.8
Mean
2xC02
11.8
75.4
31.3
18.5
31.3
94.8
4.2
.8
32.2
12.6
.0
5.9
17.8
16.9
35.5
34.7
97.4
.8
Mean
Diff
-61.9
-163.3
-161.7
-114.3
-125.3
-158.4
-55.0
-44.8
-131.4
-118.6
-82.9
-74.5
-127.1
-94.8
-144.0
-145.6
-190.5
-60.9
Mean
Xchange
-84.0
-68.4
-83.8
-86.0
-80.0
-62.6
-92.9
-98.2
-80.3
-90.4
-100.0
-92.7
-87.7
-84.8
-80.2
-80.7
-66.2
-98.7
Uncer
Xchange
14.4
5.4
7.4
10.0
9.8
5.7
17.7
18.7
8.3
12.0
15.1
15.4
10.5
15.0
8.4
7.7
5.2
17.7
-------
Table 8: Great Lake Soybean Yield (Kg/ha) •• Baseline and GFDL 2xC02 - Rainfed
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, MN
II. St. Cloud, MN
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland. OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
705.0
1818.0
1481.3
1569.3
1846.7
1336.3
2365.7
2316.3
1942.0
1772.0
2398.3
2314.3
2311.3
2173.3
1805.7
2079.0
1147.7
2215.7
Mean
2xC02
748.3
103.7
593.7
690.0
939.7
520.0
993.7
1192.7
675.0
622.7
1061.3
1035.3
1022.3
1014.0
1281.0
1180.7
634.3
924.0
Mean
Diff
43.3
-1714.3
-887.7
-879.3
-907.0
-816.3
-1372.0
-1123.7
-1267.0
-1149.3
-1337.0
-1279.0
-1289.0
-1159.3
-524.7
-898.3
-513.3
-1291.7
Mean
Xchange
6.1
-94.3
-59.9
-56.0
-49.1
-61.1
-58.0
-48.5
-65.2
-64.9
-55.7
-55.3
-55.8
-53.3
-29.1
-43.2
-44.7
-58.3
Uncer
Xchange
19.6
13.2
8.6
11.1
9.5
9.1
6.7
7.3
8.5
8.6
6.6
7.9
8.9
9.1
12.8
10.7
17.5
8.5
Mean
Baseline
705.0
1818.0
1481.3
1569.3
1846.7
1336.3
2365.7
2316.3
1942.0
1772.0
2398.3
2314.3
2311.3
2173.3
1805.7
2079.0
1147.7
2215.7
Mean
2xC02
1858.3
298.0
1351.3
1641.3
1934.3
1289.0
2158.0
2700.7
1647.3
1495.7
2343.0
2360.3
2256.7
2051 .3
2540.3
2335.0
1220.3
2123.0
Mean
Diff
1153.3
-1520.0
-130.0
72.0
87.7
-47.3
-207.7
384.3
-294.7
-276.3
-55.3
46.0
-54.7
-122.0
734.7
256.0
72.7
-92.7
Mean
Xchange
163.6
-83.6
-8.8
4.6
4.7
-3.5
-8.8
16.6
-15.2
-15.6
-2.3
2.0
-2.4
-5.6
40.7
12.3
6.3
-4.2
Uncer
Xchange
30.6
13.4
12.2
14.9
12.6
12.0
8.4
9.1
11.3
10.9
8.3
10.2
11.0
11.5
17.0
14.0
20.1
10.7
-------
Table 9: Great Lakes Soybean Yield (Kg/ha) -- Baseline and GFDL 2xC02 -- Irrigated
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, MN
II. St. Cloud, MN
Green Bay, WI
III. Madison, WI
Flint, MI
Nuskegon Co., Ml
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Noines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, NO
St. Louis, NO
Indianapolis, IN
Colunbus, OH
Nean
Baseline
877.3
2625.0
2619.7
2552.0
2930.7
2522.0
3179.3
3011.3
3205.3
3385.7
3261.0
3051.3
2973.0
2866.3
3340.0
3526.3
3084.3
3079.3
Hean
2xC02
2408.3
1926.7
2370.7
2165.3
2750.0
2299.3
2919.7
3094.0
2505.0
2350.7
2739.3
2865.7
2848.3
2737.0
2952.0
2945.7
2265.3
2811.7
Nean
Diff
1531.0
-698.3
-249.0
-386.7
•180.7
-222.7
-259.7
82.7
-700.3
-1035.0
-521.7
-185.7
-124.7
-129.3
-388.0
-580.7
-819.0
-267.7
Nean
Xchange
174.5
-26.6
-9.5
-15.2
-6.2
-8.8
-8.2
2.7
-21.8
-30.6
-16.0
-6.1
-4.2
-4.5
-11.6
-16.5
-26.6
-8.7
Uneer
Xchange
18.8
6.4
5.7
7.8
4.8
4.9
2.0
4.5
3.2
3.2
2.9
4.6
5.8
4.5
4.0
3.6
4.2
5.7
Nean
Baseline
877.3
2625.0
2619.7
2552.0
2930.7
2522.0
3179.3
3011.3
3205.3
3385.7
3261.0
3051.3
2973.0
2866.3
3340.0
3526.3
3084.3
3079.3
Nean
2xC02
4957.3
4395.7
4442.3
4398.3
4641.3
4597.7
4554.0
4903.3
4462.3
4118.3
4537.0
4824.7
4887.3
4525.3
4957.7
4753.0
3218.3
4680.0
Nean
Diff
4080.0
1770.7
1822.7
1846.3
1710.7
2075.7
1374.7
1892.0
*
1257.0
732.7
1276.0
1773.3
1914.3
1659.0
1617.7
1226.7
134.0
1600.7
Nean
Xchange
465.0
67.5
69.6
72.3
58.4
82.3
43.2
62.8
39.2
21.6
39.1
58.1
64.4
57.9
48.4
34.8
4.3
52.0
Uncer
Xchange
17.5
5.7
7.5
8.6
6.4
7.1
2.5
4.5
3.3
4.2
3.0
4.6
5.7
4.6
4.4
4.6
5.8
-------
Table 10: Great Lakes Soybean Irrigation Demand (mm/year) -- Baseline and GFDL 2xC02
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, MN
II. St. Cloud, NN
Green Bay, WI
III. Madison, WI
Flint, Ml
Muskegon Co., MI
Albany, NY
Buffalo, NY.
IV. Peoria, IL
Des Noines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
83.9
202.5
193.7
126.2
169.5
157.4
150.3
134.6
166.2
206.8
155.0
147.4
139.2
135.4
193.4
204.8
287.6
127.7
Mean
2xC02
244.4
439.4
405.0
247.3
303.1
263.1
330.9
288.4
333.0
395.0
305.2
347.3
329.7
361.1
279.3
336.5
414.1
309.4
Mean
Diff
160.5
236.9
211.3
121.2
133.6
105.7
180.6
153.8
166.8
188.2
150.2
199.9
190.5
225.7
85.8
131.7
126.5
181.7
Mean
Xchange
191.4
117.0
109.1
96.0
78.8
67.2
120.2
114.2
100.4
91.0
96.9
135.6
136.8
166.6
44.4
64.3
44.0
142.3
Uncer
Xchange
20.2
14.4
12.8
18.2
15.7
11.3
15.9
14.7
12.3
10.8
13.7
16.8
16.5
25.1
15.0
15.1
12.7
16.9
Mean
Baseline
83.9
202.5
193.7
126.2
169.5
157.4
150.3
134.6
166.2
206.8
155.0
147.4
139.2
135.4
193.4
204.8
287.6
127.7
Mean
2xC02
257.2
482.2
405.6
258.8
280.7
258.5
299.6
268.2
319.6
376.1
288.7
317.5
313.2
332.8
270.0
321.8
400.8
289.2
Mean
Diff
173.3
279.7
211.9
132.6
111.2
101.1
149.3
133.6
153.4
169.2
133.7
170.1
174.0
197.4
76.6
117.0
113.3
161.5
Mean
Xchange
206.7
138.1
109.4
105.1
65.6
64.3
99.3
99.2
92.3
81.8
86.2
115.4
125.0
145.8
39.6
57.1
39.4
126.5
Uncer
Xchange
21.7
12.2
15.1
17.7
15.4
11.3
15.5
14.7
12.7
10.8
13.3
17.4
17.6
25.5
15.9
15.1
12.5
17.6
-------
Table 11: Great Lake Soybean Yield (Kg/ha) -• Baseline and GISS 2xC02 -- Rainfed
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth. MN
II. St. Cloud, NN
Green Bay, WI
III. Madison. WI
Flint, NI
Nuskegon Co., NI
Albany, NY •
Buffalo, NY
IV. Peoria, IL
Des Hoines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamport, PA
V. Springfield, HO
St. Louis, NO
Indianapolis, IN
Columbus, OH
Nean
Baseline
705.0
1818.0
1481.3
1569.3
1846.7
1336.3
2365.7
2316.3
1942.0
1772.0
2398.3
2314.3
2311.3
2173.3
1805.7
2079.0
1147.7
2215.7
Nean
2xC02
1542.0
873.3
1644.7
1599.0
.0
1241.3
1882.3
2208.0
1890.3
1774.0
2414.3
2075.3
2171.0
1993.0
2110.7
2201.0
901.0
2065.7
Nean
Oiff
837.0
-944.7
163.3
29.7
-1846.7
-95.0
-483.3
-108.3
-51.7
2.0
16.0
-239.0
-140.3
-180.3
305.0
122.0
-246.7
-150.0
Nean
Xchange
118.7
-52.0
11.0
1.9
-100.0
-7.1
-20.4
-4.7
-2.7
.1
.7
-10.3
-6.1
-8.3
16.9
5.9
-21.5
-6.8
Uncer
Xchange
26.4
15.3
11.1
14.0
7.6
12.9
8.4
8.6
10.5
10.6
7.5
9.3
10.4
10.5
14.7
12.9
18.0
9.6
Nean
Baseline
705.0
1818.0
1481.3
1569.3
1846.7
1336.3
2365.7
2316.3
1942.0
1772.0
2398.3
2314.3
2311.3
2173.3
1805.7
2079.0
1147.7
2215.7
Nean
2xC02
3454.3
2032.0
3492.3
3529.7
3460.7
2871.7
3765.3
4363.7
3843.7
3731.3
4442.0
4194.7
4193.0
3717.3
3765.3
3860.0
1543.7
4156.0
Nean
Diff
2749.3
214.0
2011.0
1960.3
1614.0
1535.3
1399.7
2047.3
1901.7
1959.3
2043.7
1880.3
1881.7
1544.0
1959.7
1781.0
396.0
1940.3
Nean
Xchange
390.0
11.8
135.8
124.9
87.4
114.9
59.2
88.4
97.9
110.6
85.2
81.2
81.4
71.0
108.5
85.7
34.5
87.6
Uncer
Xchange
32.9
19.0
13.9
16.1
14.5
20.1
9.7
9.2
13.3
1
7.9
10.4
11.7
12.3
18.8
16.5
21.3
10.2
-------
Table 12: Great Lakes Soybean Yield (Kg/ha) -- Baseline and GISS 2xC02 -- Irrigated
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth, HN
II. St. Cloud, MM
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
877.3
2625.0
2619.7
2552.0
2930.7
2522.0
3179.3
3011.3
3205.3
3385.7
3261.0
3051.3
2973.0
2866.3
3340.0
3526.3
3084.3
3079.3
Mean
2xC02
2466.0
3002.0
3023.0
2698.0
3219.3
2610.3
3266.0
3412.3
3255.0
3297.7
3270.3
3254.7
3265.3
2982.0
3344.7
3451.3
2704.3
3279.7
Mean
Diff
1588.7
377.0
403.3
146.0
288.7
88.3
86.7
401.0
49.7
-88.0
9.3
203.3
292.3
115.7
4.7
-75.0
-380.0
200.3
Mean
Xchange
181.1
14.4
15.4
5.7
9.9
3.5
2.7
13.3
1.5
-2.6
.3
6.7
9.8
4.0
.1
-2.1
-12.3
6.5
Uncer
Xchange
21.1
6.0
4.9
8.1
3.3
5.3
1.8
4.4
2.5
2.5
3.0
4.7
5.6
4.5
3.4
2.8
4.1
5.6
Mean
Baseline
877.3
2625.0
2619.7
2552.0
2930.7
2522.0
3179.3
3011.3
3205.3
3385.7
3261.0
3051.3
2973.0
2866.3
3340.0
3526.3
3084.3
3079.3
Mean
2xC02
4477.7
5184.0
5073.0
5001.7
5228.3
4800.3
5099.7
5349.3
5317.3
5431.7
5217.0
5269.3
5263.7
4916.7
5361.7
5330.0
3818.7
5222.7
Mean
Diff
3600.3
2559.0
2453.3
2449.7
2297.7
2278.3
1920.3
2338.0
2112.0
2046.0
1956.0
2218.0
2290.7
2050.3
2021.7
1803.7
734.3
2143.3
Mean
Xchange
410.4
97.5
93.7
96.0
78.4
90.3
60.4
77.6
65.9
60.4
60.0
72.7
77.0
71.5
60.5
51.1
23.8
69.6
Uncer
Xchange
25.6
5.2
4.8
6.1
3.4
7.4
2.0
4.4
2.2
2.2
2.7
4.3
5.6
1.8
3.3
3.1
4.5
5.7
-------
Table 13: Great Lakes Soybean Irrigation Demand (mm/year) -- Baseline and GISS 2xC02
Weather Effects Alone
Weather and Direct C02 Effects
Group Weather Station
I. Duluth. MN
II. St. Cloud, MN
Green Bay, WI
III. Madison, WI
Flint, MI
Muskegon Co., MI
Albany, NY
Buffalo, NY
IV. Peoria, IL
Des Moines, IA
Fort Wayne, IN
Cleveland, OH
Pittsburgh, PA
Williamsport, PA
V. Springfield, MO
St. Louis, MO
Indianapolis, IN
Columbus, OH
Mean
Baseline
83.9
202.5
193.7
126.2
169.5
157.4
150.3
134.6
166.2
206.8
155.0
147.4
139.2
135.4
193.4
204.8
287.6
127.7
Mean
2xC02
113.7
216.2
218.1
139.6
211.9
167.8
213.8
183.5
183.5
211.9
166.3
206.0
188.4
182.9
185.2
212.4
365.6
162.1
Mean
Diff
29.8
13.7
24.4
13.5
42.3
10.4
63.5
48.9
17.3
5.1
11.3
58.6
49.1
47.5
-8.2
7.6
78.1
34.4
Mean
Xchange
35.5
6.8
12.6
10.7
25.0
6.6
42.2
36.4
10.4
2.5
7.3
39.8
35.3
35.0
-4.3
3.7
27.1
26.9
Uncer
Xchange
19.7
11.2
11.3
16.0
14.4
10.8
15.1
13.3
11.6
9.3
13.1
16.2
16.6
23.0
14.7
14.1
12.0
14.8
Mean
Baseline
83.9
202.5
193.7
126.2
169.5
157.4
150.3
134.6
166.2
206.8
155.0
147.4
139.2
135.4
193.4
204.8
287.6
127.7
Mean
2xC02
111.1
199.1
195.1
137.5
183.4
156.0
184.7
161.1
155.0
193.4
139.2
175.3
159.9
162.0
167.8
192.0
347.5
139.1
Mean
Diff
27.2
-3.4
1.4
11.4
13.9
-1.4
34.4
26.5
-11.2
-13.4
-15.9
27.9
20.6
26.5
-25.6
-12.9
59.9
11.4
Mean
Xchange
32.4
-1.7
.7
9.0
8.2
-.9
22.9
19.7
-6.8
-6.5
-10.2
19.0
14.8
19.6
-13.2
-6.3
20.8
9.0
Uncer
Xchange
18.9
11.0
11.2
14.7
14.0
10.9
15.5
13.4
11.7
8.8
12.0
16.1
16.7
22.5
14.5
13.8
11.7
14.6
-------
IMPACT OF CLIMATE CHANGE ON CROP YIELD IN THE SOUTHEASTERN U.SA.:
A SIMULATION STUDY
by
Robert M. Peart
J. W. Jones
R. Bruce Curry
Ken Boote
L. Hartwell Allen, Jr.
Institute of Food and Agricultural Sciences
University of Florida
Gainesville, FL 32611
Contract No. CR814600
-------
CONTENTS
Page
FINDINGS 2-1
CHAPTER 1: INTRODUCTION 2-3
DESCRIPTION OF THE ECOLOGICAL SYSTEM 2-3
LITERATURE REVIEW 2-3
ORGANIZATION OF THIS REPORT 2-6
Chapter 2: METHODOLOGY 2-7
THE EFFECTS MODEL 2-7
Crop Model Description 2-7
CERES Maize and SOYGRO Common Features 2-7
Sensitivity of SOYGRO to Changes in Weather Conditions 2-8
Direct Effects of COj Enrichment 2-13
Limitations Inherent in the Models 2-18
THE WEATHER SCENARIOS 2-21
The Scenarios Used 2-21
Issues Resulting from the Scenarios 2-21
Limitations of the Weather Scenarios 2-21
CROP MANAGEMENT 2-24
SIMULATION EXPERIMENTS 2-26
CHAPTER 3: RESULTS 2-27
RELATIVE CONTRIBUTIONS BY EACH CLIMATE VARIABLE 2-27
SUMMARY OF YIELD RESULTS FOR ALL LOCATIONS 2-27
Annual Variability 2-35
Soybean Simulations for Climate Effect Only 2-35
Maize Simulations for Climate Effect Only 2-40
Soybean Simulations for Combined Climate and Direct Effects 2-40
Maize Simulations for Combined Climate and Direct Effects 2-44
WATER-USE RESULTS 2-44
Soybean Water-Use Results for Combined Climate and Direct Effects 2-44
Maize Water-Use Results for Combined Climate and Direct Effects 2-44
Irrigation Requirements - Soybeans 2-44
CHAPTER 4: IMPLICATIONS OF RESULTS 2-50
ENVIRONMENTAL IMPLICATIONS 2-50
SOCIOECONOMIC IMPLICATIONS 2-50
POLICY IMPLICATIONS 2-50
REFERENCES 2-51
11
-------
Peart
FINDINGS1
Simulations of soybean and com (maize) growth for the southeastern U.SA. were run for 30 years of
weather data, 1951-80, for 19 locations, with and without supplemental irrigation, and then with weather data sets
adjusted for climatic changes predicted by the GISS and the GFDL General Circulation Models for a doubling
of the carbon dioxide concentration. Then the crop models SOYGRO and CERES-Maize were modified to
account for the increase in photosynthesis and decrease in transpiration due to increased carbon dioxide.
In general, the climatic variable effects alone caused decreases in yields in the range of 25% for rainfed
soybeans under the GISS scenario, but the GFDL weather dropped rainfed soybean yields 73%. For rainfed
corn (maize), GISS weather lowered yields only 8%, but the GFDL weather cut yields by 65%.
Adding the effect of the carbon dioxide enrichment changed these yield results significantly for soybeans, but
not for com (maize). For the GISS weather scenarios and rainfed soybeans, yields for doubled carbon dioxide
at 6 of the 19 locations had yield increases of over 24%; 5 of the 19 had decreases of 10 to 17%; and the average
change over all locations was an increase of 9%. The GFDL weather scenario was in general more drastic
(higher temperatures and lower rainfall at critical reproductive-growth periods), and the average yield reduction
was 55%. Corn (maize) results were roughly the same as for the simulations with climate effects only.
Irrigation mitigated the weather effects somewhat, and soybeans yielded about 18% less than the base
weather under both doubled carbon dioxide scenarios when only climate effects were included, but they yielded
about 14% more when the combined climatic and direct effects were in the model. Irrigated corn (maize)
yielded about 20% less under all combinations.
Irrigation demand was based on the accumulated irrigation water applied during each simulation run. For
the base weather case, the averages were:
rainfall - 508 mm,
evapotranspiration - 590 mm,
irrigation requirement - 224 mm.
For runs made using the SOYGRO model incorporating the combined climate effect and the direct CO2 effects
on the plant, the GISS scenario increased the potential irrigation demand by 33%, while the GFDL scenario
increased the demand by 133%.
Water-use efficiency was studied at several locations, and results, compared with base weather results,
ranged from very significant reductions in efficiency to some cases of little change. Water-use efficiency was
directly correlated with yield.
An obvious, but striking, conclusion is the great difference in results from use of the two different weather
scenarios. It is very important to the future of southeastern U.SA. agriculture to have acceptable estimates of
these future weather trends. This cannot be solved in a short-term project, but should be carried on for a
number of years, so that actual data can be used to validate the accuracy of the forecasts.
A second important conclusion is the value of physiologically based crop simulation as a method to study this
type of problem. Simulation can handle changing environmental variables, even when these variables have
different and sometimes opposite effects on yield. One prevalent view is that higher temperatures of the changed
'Although the information in this report has been funded wholly by the U.S. Environmental Protection
Agency under Contract No. CR814600, it does not necessarily reflect the Agency's views, and no official
endorsement should be inferred from it.
2-1
-------
Peart
climate will drastically reduce yields, while another view points to the increased photosynthesis under higher
concentrations of carbon dioxide. Neither simplified view can make accurate forecasts of effects on actual yields.
Third, to perhaps oversimplify our results, corn yields in the southeastern US A. would be reduced by the
weather changes associated with doubling carbon dioxide, either moderately or drastically, depending on the
weather scenario. Soybean yields would be affected little or cut in half, depending on the weather scenario.
Both statements are based on the more common rainfed, nonirrigated situation.
Fourth, under either scenario, irrigation water demand would be greatly increased for two reasons: (1) crops
currently irrigated would require much more water, the amount depending on the scenario; and (2) more
acreage would be irrigated as weather changes cause more frequent crop failures.
A fifth conclusion should be stressed, and that is the annual variability of weather and its amplifying effect
on crop yields. For example, the GISS scenario at Memphis, 1951-60, had yearly yields both higher and lower
than the base weather. Therefore, with only modest changes in average yield, some drastic year-to-year
variations will occur and will have important economic impact on the areas affected (see Adams et al. chapter
in this report).
2-2
-------
Peart
CHAPTER 1
INTRODUCTION
DESCRIPTION OF THE ECOLOGICAL SYSTEM
This study covers soybean and com production in the southeastern U.SA., and it is based on simulations
by the SOYGRO V5.41 and the CERES-Maize models using actual weather data for 1951-1980 and the same
data except for modification for the effect of doubled carbon dioxide on the weather.
This southeastern region studied is shown in Figure 1 and includes Louisiana, Arkansas, Kentucky, Virginia,
Florida, Alabama, Mississippi, Tennessee, Georgia, North Carolina, and South Carolina. For summarizing
results, the region was subdivided into Delta, Uplands, and Coastal Plains areas, where the locations share some
soil and climate similarities.
The southeastern region includes the practical southern limit for corn and soybean production in the U. S.
High temperatures and variability of precipitation produce significant stress on these crops in a normal year in
the Southeast. Soils are quite variable, and many have a low water-holding capacity, further amplifying the
potential crop stress problem. Table 1 lists the sites, the soil types used in the simulation, and the soybean
varieties used in SOYGRO.
For soybeans in particular, there is a distinct relationship between the effect of temperature and rainfall at
the time of reproductive fruit growth. Soybean varieties have been adapted to most areas of the region.
Irrigation is not in widespread use for at least two reasons, lack of a reliable water source and a low benefit/cost
ratio. On the other hand, excess water may have a destructive effect on yield, particularly during plant
establishment time and during harvest.
LITERATURE REVIEW
There has been no agreement as to the combined effect of carbon dioxide changes on crop production,
especially on the interaction of direct effects and indirect effects. The presence of carbon dioxide in the
atmosphere could cause changes in important weather variables such as temperature, solar radiation and
precipitation. The plant also responds directly to increased carbon dioxide with increased rates of photosynthesis
and somewhat reduced transpiration. Waggoner (1983) predicted that a warmer and drier climate would reduce
crop production, but he added that this does not take into account increased photosynthesis rates due to the
increase in carbon dioxide.
Soybean growth and development simulators reported in the literature include: SOYMOD, developed at
the Ohio Agricultural Research and Development Center/The Ohio State University (Meyer et al. 1979);
SOYGRO, developed at the University of Florida (Wilkerson et al. 1985); and GLYCYM, developed by USDA
(Acock et al., 1983). All are physiologically based simulators. SOYMOD and GLYSYM have a more detailed
physiological structure and operate on an hourly time step. SOYGRO operates on a daily time step, is more
user-friendly, and has been much more widely validated and used than the other two. More details about
SOYGRO and reasons for chosing it for this study are given in a later section in this report.
Dynamic crop growth simulation was used by van Keulen et al. (1981) to show that stomatal behavior was
the key factor in determining plant response to increased carbon dioxide under nonlimiting water and nutrient
conditions. Stewart (1986) ran a generalized crop growth model for Saskatchewan spring wheat with a 15%
increase in photosynthetic capacity, using results from a doubled carbon dioxide experiment, and he found that
with the predicted climate changes, a net reduction in wheat yield was indicated, even with the increase in
photosynthetic capacity.
2-3
-------
Peart
Table 1. List of Study Sties Soil Types and Varieties
SITE CITY
BIAL BIRMINGHAM, AL
MBAL MOBILE, AL
MGAL MONTGOMERY,AL
LRAR LTTTLE ROCK, AR
TLFL TALLAHASSEE, FL
ATGA ATLANTA, GA
MOGA MACON, GA
LUKY LOUISVILLE, KY
BRLA BATON ROUGE, LA
SHIA SHREVEPORT, IA
MEMS MERIDIAN, MS
CHNC CHARLOTTE, NC
RANG RALEIGH, NC
CLSC COLUMBIA, SC
MPTN MEMPHIS, TN
NSTN NASHVILLE, TN
LEVA LYNCHBURG, VA
NOVA NORPORK, VA
WINC WILMINGTON, NC
SOIL TYPE
DEEP SILT LM
MED SANDY LM
MED SANDY LM
MED SILT LM
ORGB SNDY LM
DEEP SILT LM
MED SANDY LM
MED SILT LM
SHL SILTY CLY
MED SILT LM
MED SILT LM
MED SILT LM
MED SANDY LM
MED SANDY LM
MED SILT LM
MED SILT LM
MED SILT LM
MED SILT LM
MED SANDY LM
VARIETY
FORREST
BRAGG
BRAGG
FORREST
BRAGG
FORREST
BRAGG
ESSEX
TRACY
FORREST
FORREST
FORREST
TRACY
BRAGG
FORREST
ESSEX
ESSEX
ESSEX
TRACY
*RUNS FOR MAIZE MADE WITH SAME SOIL TYPES AS FOR SOYBEANS.
FOR MAIZE ONLY ONE VARIETY USED - MCCURDY 84AA.
2-4
-------
Peart
Figure 1. Location of study sites.
2-5
-------
Peart
Bisbal (1987) found in Florida an increase in soybean yield at doubled carbon dioxide levels in controlled
chambers with other environmental conditions ideal. Allen et al. (1987) predicted that soybean yields would
increase about 32% with a doubling of carbon dioxide, which is in general agreement with KimbalTs (1983)
conclusions of an increase of about 33% ±6%.
The two general circulation models that were used to provide the 2xCO2 climate senarios were developed
by (1) Goddard Institute for Space Studies (GISS) (Hansen et al., 1988); and (2) the Geophysical Fluid
Dynamics Laboratory (GFDL) (Manabe et al., 1987). The characteristics of these two types of GCMs have been
analyzed by Schlesinger and Mitchell (1985).
ORGANIZATION OF THIS REPORT
The remainder of this report will describe the methodology used to obtain data on yields for soybeans and
corn for base weather (1951-1980) and for the climate modified by doubled carbon dioxide in the atmosphere.
Results are presented for 30 years of weather data for 19 southeastern U.S. locations in 11 states for current or
base levels of carbon dioxide, both with and without irrigation, and also for two sets of weather data based on
doubled carbon dioxide (2xCO2). Results are given for the climate effects only, and also for the climate effects
plus the carbon dioxide enrichment effect on photosynthesis and on transpiration.
Some interpretation is made of these results. Environmental and socioeconomic implications are suggested,
and policy recommendations are made.
2-6
-------
Peart
CHAPTER 2
METHODOLOGY
THE EFFECTS MODEL
The effects of climate change on soybean and corn yields in the southeastern UJSA. were studied using
existing models for these two crops. The SOYGRO V5.41 and CERES-Maize models were chosen to simulate
soybean and corn, respectively, for several seasons. First, these models have been available and documented for
several years. Second, these models respond to the major climate variables of solar radiation, temperature, and
precipitation and include the effects of soil characteristics on water availability for crop growth. Third, they
have been validated for a range of soil and climate conditions in the UJSA. and other countries, and are being
used by scientists at various research institutions. Fourth, these models were developed with compatible data
structures so that the same soil and climate data bases could easily be used with both crop models. Finally, both
these models have user-oriented interfaces to facilitate their use in studies such as this. These factors were
important for this study, which attempted to provide credible estimates of the impact in a very short time period
of 6 months.
The impact of climate change on crop growth and yield includes the direct effect of CO- changes on the
crop as well as indirect effects of this and other atmospheric gases on climate variables such as rainfall and
temperature. This study was divided into two parts. First, the original SOYGRO and CERES-Maize models
were used without modification to estimate the effects of temperature, solar radiation, and rainfall changes
suggested by two General Circulation Models (GISS and GFDL) on crop yield. The second part of the study
involved the development of equations to change crop photosynthesis and evapotranspiration processes in both
SOYGRO and CERES-Maize due to increased CQy and the use of these modified models to estimate the
combined direct and indirect effects of CO2 on crop growth and yield. These two models were not originally
developed to account for CO2 effects at normal ambient concentrations.
Next, a description is given of the original SOYGRO and CERES-Maize models with emphasis on how
climate variables affect yield estimates. Then, the modifications of the models to include direct effects of CO2
on photosynthesis and transpiration are presented.
Crop Model Description
CERES Maize and SOYGRO Common Features: The SOYGRO and CERES-Maize models were designed
to simulate crop growth and yield under a range of soil and climate conditions where these crops are normally
grown. Both models predict the phonological development or duration of vegetative and reproductive growth
stages as affected by variety, weather, and soils. Photosynthesis and the production and partitioning of biomass
into leaves, stems, roots, and fruit are estimated daily depending on these crop, soil, and weather factors. A
component model of the soil-root system integrates the effects of rainfall, root growth dynamics, and
climate-induced evapotranspiration to predict day to day water availability to the plants and the resulting
development of water stress. Water stress causes reductions in canopy development, photosynthesis, partitioning
of biomass, and senescence or abortion of plant material, depending on the timing and severity of stresses. The
models were developed primarily using field data from experiments over a range of locations and time.
The SOYGRO crop growth model was first described by Wilkerson et al. (1983) and subsequent
modifications by Wilkerson et al. (1985) and Jones et al. (1988a). Version 5.41 includes the basic carbon and
nitrogen balances described by Wilkerson et al. (1983), but major modifications to the original model were made
for describing phonological development, soil-water balance, and effects of temperature, radiation, and daylength
on process rates. The soil water model in version 5.41 was adopted from Ritchie (1985) and is the same as that
used in the CERES-Maize model. The phenology model used in SOYGRO V5.41 is described in detail by Jones
et al. (1988b).
2-7
-------
Peart
The inputs to the models include the natural system inputs, management inputs, crop and genetic input
variables used in the model equations, and initial conditions.
The original CERES-Maize model was described by Jones and Kiniry (1986). Modifications in the
input-output structure and user-interface of the original version were made to conform to the standard crop
model inputs and outputs developed by IBSNAT (1986). The SOYGRO V5.41 model also used this input-output
structure, which made it possible to simulate both crops using the same natural system (weather and soil) inputs,
and the outputs could be analyzed by the same procedures.
The natural system inputs consist of weather and soil data and the site latitude. Weather is considered an
uncontrollable input consisting of daily solar radiation (MJ/m2-day), maximum and minimum air temperatures
(°C), and rainfall (mm/day). Daylength is computed from the day of year and latitude. The soil parameters
include the soil albedo, a soil water drainage rate constant, upper limit of stage 1 soil evaporation (mm), a runoff
curve number, and characteristics describing each layer in a one-dimensional profile. For each layer, the layer
depth, lower limit of plant-extractable water, drained upper limit of soil water, and saturated water content (in
volume fractions) are inputs. In addition, a root growth weighting factor is input for each layer for use in
distributing new root growth as the season progresses. By changing these natural system inputs, the crop models
can simulate growth and yield for existing conditions where these inputs are available or for hypothetical
conditions where these inputs are estimated to represent those conditions, e.g., estimated future climate changes.
The management inputs are the beginning day of the simulation; planting day; plant density (plants/m2); row
spacing (m); depth of planting (cm); whether irrigated automatically, according to a particular experiment or
schedule, or not at all; and, if irrigated according to a particular schedule, the dates and amounts of water
applied. Nitrogen was assumed to be nonlimiting for all simulations, although the CERES-Maize model has
the capability to vary this management variable as well. In SOYGRO, nitrogen supply to the plant is through
the nitrogen fixation process, which occurs in proportion to carbohydrate availability. Initial condition inputs are
values of soil water in each zone and initial plant weight at emergence.
The crop and genetic inputs are different for the soybean and corn models. In SOYGRO, a file of crop
coefficients supplies all the values for basic growth processes such as photosynthesis, respiration requirements
for synthesis of different plant parts, effects of temperature on development, photosynthesis, and seed growth
rates, the effect of solar radiation on photosynthesis, the effect of leaf nitrogen content on photosynthesis, and
the effects of phenological stage on partitioning under ideal, nonstressed conditions. These inputs remain the
same for all soybean cultivars. This file remained unchanged in all the climate change simulations. A second
input file in SOYGRO contains parameters for a wide range of cultivars adapted to latitudes ranging from about
8°N to 44°N latitude. These coefficients are inputs to those processes that vary significantly among soybean
cultivars, and can be divided into those related to reproductive development and those related to biomass
accumulation and partitioning. The reproductive development coefficients are (1) the sensitivity of the cultivar
to photoperiod and (2) the thermal or photothermal time thresholds required for each stage to occur. Jones et
al. (19885) described the development model and the coefficients. The photoperiod sensitivity changes
significantly among cultivars and is the major determinant of the range of latitudes for which each cultivar is
adapted. The biomass and partitioning parameters describe the differences among cultivars using maximum seed
and shell growth rate (mg/day), leaf size, and maximum rate of flower and pod addition.
The CERES-Maize model has a similar input file for describing each cultivar's sensitivity to photoperiod,
duration of stages, and biomass coefficients, although they are different parameters. These are described by
Jones and Kiniry (1986).
Sensitivity of SOYGRO to Changes in Weather Conditions. In order to understand the results from crop
simulation studies in which several climate variables are changed, it is useful to study the effects of each variable
acting alone on the overall growth and yield of a crop. A sensitivity analysis was conducted by Boote et al.
(1988) to study the effects of temperature, solar radiation, and photoperiod on soybean growth and yield using
SOYGRO V5.41. Since temperature and radiation are two of the weather variables projected to change by the
General Circulation Models, a summary of the results of this study is presented here.
2-8
-------
Peart
Daily development and biomass accumulation rates are the results of various processes in the model. Each
of these processes may be affected by temperature in different ways. For example, vegetative node development
rate increases with temperature up to 28 to 30°C (Hesketh et al., 1973), whereas reproductive development rate
is optimal for temperatures of 21 to 28°C (Parker and Borthwick, 1943). Table 2 lists the sources of data used
to develop the relationships between process rates and temperature, solar radiation, and daylength in SOYGRO.
Table 2. Processes Affected by Temperature, Radiation, and Daylength in SOYGRO V5.41
1. Temperature
a. Photosynthesis (Hofstra and Hesketh, 1975)
b. Maintenance Respiration (McCree, 1974)
c. Vegetative Node Development (Hesketh et al., 1973)
d. Leaf area growth (Thomas and Raper, 1978)
e. Duration of Reproductive Stages (Parker and Borthwick, 1943)
f. Pod and Seed Addition Rates (Thomas and Raper, 1981)
g. Seed Growth Rates (Egli and Wardlaw, 1980)
h. Evapotranspiration (Priestly and Taylor, 1972)
2. Solar Radiation
a Evapotranspiration (Priestly and Taylor, 1972)
b. Photosynthesis (Ingram et al., 1981)
3. Daylength
a. Duration of Reproductive Stages (Thomas and Raper, 1976)
b. Pod and Seed Addition Rates (Fisher, 1963)
c. Partitioning of Carbon to Fruit (Cure et al., 1982)
In the sensitivity analysis, all weather variables were held constant for a baseline run and then each variable
was varied one at a time. The Bragg cultivar was used, no water stress was allowed, and temperature was set
to 28°C. Photon flux density was set to 35 moles/m2-day. Two cases were simulated to evaluate temperature
effects. Daylength was fixed at 12 h and temperature varied. In the second case, daylength was held to 14 h
for 50 days then switched to 12 h for the remainder of the season. In the first case, an incomplete canopy
occurred in some years, but in the second case, the 50-day period at 14 h ensured that a full canopy would
develop at all temperatures.
The integrated results of temperature changes are shown in Figure 2a-e. Days to flower decreased as
temperature increased to about 30°C, after which days to flower started increasing. The number of nodes on
the plant increased from 7 to over 11 for the same temperature range (Figure 2b). The effects of increasing
temperature on duration of growth stages depended on temperature. For example, increasing temperature
between 15 and 25°C had a major effect on time to flower, whereas increasing from 25 to 35°C had very little
effect.
In order to evaluate independent temperature effects on seed growth and yield without the effect of
temperature on vegetative growth and canopy size, simulations were run with a 14-h daylength for the first 50
days to ensure a full canopy.
Seed yield increased rapidly as temperature increased from 15 to 20°C and was optimal at about 24°C
(Figure 2c). Decreases in seed yield between 24 and 34°C can be attributed to a combination of shortened seed
2-9
-------
Peart
£T
of
UJ
C/)
i
I 10
100
90
80
70
60
50
40
30-
20-
-e s-
i
20
30
4O
TEMPERATURE,*C
Figure 2a. Simulated days to flower for soybean at various constant air temperatures.
12
"
co IM
UJ
o
i \OA
9 -
8 -
7 •
10
20 30
TEMPERATURE, *C
4O
Figure 2b. Simulated relationship between maximum V stage (main stem nodes) of soybeans at various
constant air temperatures.
2-10
-------
Peart
o»
E
Q
_)
UJ
Q
UJ
UJ
CO
10
TEMPERATURE, "C
Figure 2c. Simulated seed yield of soybeans as a function of various season-long, constant air
temperatures.
0.6
0.5-
x
uj 0.4
£ 0.3^
UJ
cc
S 0.2-
O.I
-e—B-
10
20
30
40
TEMPERATURE, C
Figure 2d. Simulated harvest index of soybeans (seed yield divided by total above
ground biomass) as a function of various season-long air temperatures.
2-11
-------
Peart
en
lJ
M
CO
Q
UJ
LJ
CO
0.20
0.18 -
0.16 -
0.14 -
0.12 -
0.10 -
0.08 -
0.06 -
0.04 -
0.02 -
0
10
i i i
20 30
TEMPERATURE, CC
40
Figure 2e. Simulated effect of season-long constant temperatures on soybean seed size.
2-12
-------
Peart
fill duration and reduced seed growth rate. The rapid decrease in yield above 34°C is attributed to the modeled
effects of temperature on photosynthesis. Harvest index (seed weight/total above-ground biomass) was shown
by Baker et al. (1988) to decrease from 0.52 to 0.46 as temperature was increased from 23 to 33° C for soybeans
when averaged over 330 and 660 ypm CO- concentration. Simulated results (Figure 2d) show an increase in
harvest index, HI, as temperature increased from 14 to 20°C, no change between 20 and 28°C, then a decrease
for temperatures above 28°C. The increases in seed yield and harvest index as temperature increased between
14 and 20° were due in large part to the function derived from data of Thomas and Raper (1981), which causes
decreased pod and seed setting in this range of temperatures. Average seed size also decreased with
temperature (Figure 2e) in the simulated results, which is consistent with literature (e.g., Baker et al., 1988).
Figure 3 shows the effect of solar radiation (expressed as moles/m2-day) on total biomass and seed yield.
There was a maximum yield at about 45 to 50 moles/m2-day.
These results of SOYGRO sensitivity analyses showed one major effect relevant to the climate changes
predicted by the GCMs. Increases in temperature on the order of 4 to 6°C could cause significant increases in
soybean yield if the baseline weather was cool, such as in the northern latitudes, and significant decreases in yield
if the base line temperatures were already in the range of 25-30°C. At constant day and night temperature of
30°C, yield was roughly 20% lower than that at 24°C, and at 36°C, it was 35-40% lower. These results were
obtained without taking into account the effect of increases in temperature on crop water use and the possibility
for increased water stress and additional yield reductions. They were also obtained, however, using air
temperature as the determinant of growth processes in the model. Increases in air temperature can cause
increases in atmospheric vapor pressure deficit, which should increase plant canopy transpiration rates. The
increase in transpiration rates will cause foliage-to-air temperature difference to increase (Allen, 1986; Idso,
1987). However, the actual foliage temperature will actually always increase as air temperature increases.
Furthermore, elevated CO2 will cause partial stomatal closure and induce a higher foliage temperature also.
Direct Effects of CCX. Enrichment. The climate change scenarios predicted by GCMs were for conditions
expected under doubled atmospheric CO2 concentrations. The effects of temperature, solar radiation, and
precipitation on crop growth processes were discussed in the previous section. CERES-Maize and SOYGRO
were first used to simulate changes in crop yields across the southeastern UJS. caused by changes in these climate
variables alone. However, since CO2 also directly affects plant growth processes, final assessment of the impact
of doubling CO2 concentration can only be made by considering both climate and direct effects on plants.
When plants are exposed to increases in CO, concentration, there are both immediate and long-term effects
on crop growth, all other conditions being equal (Acock and Allen, 1985; Allen, 1986; Cure and Acock, 1986).
One immediate effect is that leaf stomatal resistance increases and plants tend to lose less water for the same
environmental conditions (Rogers et al., 1983; Valle et al., 1985a,). However, the anticipated decrease in water
use by plants in high CO2 is moderated by increases in leaf temperatures caused by partial stomatal closing,
which increases the vapor pressure of water inside the leaves. The increase in atmospheric CO2 concentration
also results in higher gradients for diffusion of CO2 into plant leaves and increases photosynthetic rates. This
increase in photosynthesis occurs even though there is partial closure of stomata, and the magnitude of the
increase varies with the plant's pathway for fixing carbon (C. vs. C4). In the longer term, increased
photosynthesis may increase biomass growth rates significantly ana development rates slightly, and modify other
anatomical features of the plant, such as starch storage and specific leaf weight (Bisbal, 1987; Allen et al., 1988).
There is also an interaction between these longer-term changes in canopy growth and development and crop
water use. Plant canopies under elevated CO2 produce leaf area faster, capture more of the incoming energy,
and thus tend to transpire at a higher rate than would be expected if one only considered the short-term stomatal
resistance effect on lowering canopy water use.
2-13
-------
Peart
C7»
E
o"
_i
UJ
o:
o
CO
CO
g
CD
6 -
5 -
4 -
3 -
2 -
I -
BIOMASS
SEED
YIELD
20 40
PAR, MOLES/DAY
60
Figure 3. Simulated effect of total daily accumulated photosynthetically active solar radiation (PAR) on total
biomass and seed yield in soybeans.
2-14
-------
Peart
In the CERES and SOYGRO models, increased growth over the long run is accounted for in the original
models. However, the direct effects of CO, on photosynthesis and evapotranspiration were not included in these
models. Our objective here was to develop a method that will give a first approximation of the changes in
canopy photosynthesis and the transpiration rates by plants exposed to elevated CO2 in comparison with those
same plants if exposed to normal atmospheric CO, levels under the same climate conditions. Our approach was
to compute ratios of daily photosynthesis and evapotranspiration rates of a canopy exposed to elevated CO2 to
those rates of the same canopy if exposed to ambient CO, conditions.
i) Photosynthesis. Increases in photosynthesis rates by plants under elevated CO, concentrations are
well documented (Rogers et al., 1983; Cure, 1985; Cure and Acock, 1986). Although process-level models that
describe changes in photosynthesis by plants under various light and CO, conditions exist, the crop models used
in this study did not include the direct effects of CO2 on photosynthesis rates. Therefore, modifications were
made in both SOYGRO and CERES-Maize models to cause increases in photosynthesis under the double CO2
scenarios.
Various sources of literature were reviewed to obtain the best estimates of increases in canopy
photosynthesis rates for both soybean and corn. Soybean has a C-3 carbon fixation pathway and thus is more
responsive to increases in atmospheric CO, than is corn. Table 3 shows values of percentage increase in soybean
photosynthesis for double CO2 reported by various authors. The values reported by Allen et al. (1987) for
mid-day, high-light conditions were about 50% increase in canopy photosynthesis for double CO2. This increase
would not, however, represent the daily total photosynthesis increase of a canopy because of the similarity of the
photosynthesis values under low-light conditions and the fact that light levels go through cycles during the
daytime. This was demonstrated by simulating hourly values of tomato (a C-3 plant) canopy photosynthesis using
the model derived by Acock et al. (1978) fit to tomato canopy photosynthesis data taken by J. W. Jones and E.
Dayan in Gainesville, FL. These values were summed over each day to obtain daily canopy rates. The percent
increase in the daily canopy photosynthesis rates of tomato under double CO, conditions (+21%) was less than
the increase of instantaneous canopy photosynthesis under high light (+31%) (Jones, J.W. and E. Dayan,
unpublished). In other words, the increase in daily integrated photosynthesis was 33% less than the increase in
instantaneous mid-day photosynthesis rate. This relationship between relative increases in
instantaneous mid-day canopy rates and daily total canopy rates varied with total daily radiation. Reducing the
relative increase in instantaneous soybean canopy photosynthesis under high light by 33% results in a relative
daily response. Therefore, we felt justified in reducing the relative response from a 50% increase (instantaneous)
to a 35% increase (daily canopy) in photosynthesis under double CO2 conditions. This value is also consistent
with the short-term net assimilation rate increase in soybean in contrast to the instantaneous increase in leaf
carbon exchange of 78% by soybean reported by Cure (1985). Using this value for Gainesville weather data,
seed yield increases of 40% occurred, which are similar but about 5% higher than the increases in yield reported
by Allen et al. (1987) for soybean under double CO2 conditions.
A similar evaluation of corn canopy photosynthesis was conducted and a value of +15% was selected. This
value was also reduced by 33% to +10%, which is more representative of daily integral canopy values and
similar to the 9.0% increase in short-term net assimilation rate for corn reviewed by Cure and Acock (1986).
In the crop models, factors of 135 and 1.10 were multiplied by photosynthesis rates computed under current
CO2 concentrations for soybean and corn, respectively. It was assumed for this first approximation that the
relative increase in photosynthesis was independent of other factors. The effects of water stress, temperature,
and leaf area were handled exactly as before in the two models.
ii) Evapotranspiration. Estimates of crop water use in the crop models are based on a simplified
energy balance approach that depends on net radiation (RN) and temperature using the Priestly-Taylor (1972)
method. A schematic of the current method for computing crop water use (soil evaporation and plant
transpiration) is given in Figure 4. Both models use this method, which was developed and implemented
originally by J.T. Ritchie (1985).
2-15
-------
Peart
TEMP
RAD
EOS
Figure 4. Variable Dependency Diagram showing how EP, ES, and ET are computed in the CERES and
SOYGRO models. Directions of arrows show functional dependence, i.e., EOS « f(EO, LAI).
2-16
-------
Peart
Table 3. Response of Soybean Photosynthesis and Growth to Doubled Atmospheric CO2
Concentration From Several Selected Literature Sources
SOURCE
1. Idso et al. (1987)
2. Baker et al. (1988)
3. Acock et al. (1985)
4. Elwell et al. (1987)
5. Elwell et al. (1987)
6. Allen et al. (1987)
7. Allen et al. (1987)
8. Acock (per. comm.)
RELATIVE
INCREASE COMMENTS
130 Based on crop growth rate.
130 Based on final crop yield.
1.65-1.80 Based on measurements ofgross
(Gross leaf photosynthesis).
1.75 Sensitivity analysis of SOYMOD
(Gross leaf photosynthesis).
1.60 Sensitivity analysis of SOYMOD
(Final yield).
1.50 Based on midday measurements of
total canopy photosynthesis.
130 Based on yield summary of 6
locations.
130 Growth chamber measurement,
total shoot growth.
In this method, a potential evapotranspiration rate (EO) is first computed for the entire canopy (soil plus
plants), using the Priestly-Taylor equation. Potential soil evaporation (EOS) is then computed by partitioning
EO based on energy captured by the plants - a function of leaf area index (LAI). Actual soil evaporation (ES)
depends on the time from the last soil wetting and on availability of water in the top soil zones. Then, potential
plant transpiration (EOF) is computed by partitioning EO as a function of leaf area index. Existing root length
density, RLD(Z), and water content, W(Z), distributions in the soil are used to estimate the maximum supply
of water to the plant. When supply is greater than EOP, then actual transpiration (EP) is set equal to EOP.
When supply is less than EOP, then stomata closure would have occurred some time during the day because of
water stress, and EP would be set equal to supply from the root system and would be less than EOP. Finally,
actual ET is the sum of ES and EP, and is always less than or equal to EO.
The Penman-Monteith equation, which uses net radiation (R ) as an input, was used as the basis for a ratio
to modify evapotranspiration under elevated CO2 conditions (taken from France and Thornley, 1986):
5 R
AE
a) ' Pa> Sa
(1)
+ id +
where AE is evapotranspiration rate in energy units, s is the slope of the saturated vapor pressure vs. temperature
curve, 7 is the psychometric constant, Rn is net radiation, c_ is specific heat of the air at constant pressure, p is
air density, (ps(T.) - p.) is vapor pressure deficit of the air; ga is the boundary layer conductance between the
canopy and the bulk air, and gg is the canopy conductance to water vapor.
If this equation is applied twice to the same canopy and same environment except for CO, concentration,
then the only variable that changes in equation 1 is gc, the canopy resistance to vapor transport. In other words,
IL, c-, p, T_, p(TA p., wind speed 00, leaf area, plant height, and g^would be the same for both cases.
Therefore, when we take a ratio of AE* (under elevated CO2 levels) to AE, we obtain:
2-17
-------
Peart
AEC s + i (l + ga/gj
RATIO = — (2)
AE s + 7
where g^0 is the canopy conductance to water vapor under elevated CO2 conditions. Aerodynamic methods for
computing gg(= l/Ra) based on Thorn (1972), as implemented by Jagtap (1987), are used.
The canopy resistance, RC, is computed by assuming all leaves act as parallel resistances, and
Rc = (rL + rb)/LAI (3)
where rL is the leaf stomatal resistance, s/m, and LAI is leaf area index, and r^ is the leaf boundary layer
resistance. Leaf resistances r, are computed as functions of CO2 concentrations Tor r, for corn and soybean
using the equations developedby Rogers et al. (1983). Then, gg and gcc (canopy conductance under elevated
CO2) can be computed:
EC = VRg (4)
8CC =
The computation of RATIO requires temperature, wind speed, LAI, and CO2, but is independent of solar
radiation. LAI is obtained directly from the CERES and SOYGRO models.
Figure 5 shows a schematic of the modifications required to adjust EOF in CERES and SOYGRO. It is
assumed that the potential plant transpiration (EOF) is changed under elevated CO2 conditions due to increased
stomatal closure and changes in the partitioning of energy captured by the canopy. This also assumes that the
overall evapotranspiration of a full canopy is directly affected by CO2 through its effect on EOF. This is shown
in the diagram (Figure 5) by the computation of RATIO (using Equation 2 and the calculated values for g_, gp,
and g°c which depend on LAI, wind speed, temperature and CO2). Once RATIO is computed, EOPC is
computed and EP, EOS, ES, and ET are computed as before, but using EOF0 instead of EOF. This procedure
will result in a lower transpiration rate for higher CO2 levels, and a lower ET on a daily basis, but may or may
not change seasonal ET by the same proportion because of the altered LAI growth under elevated CO2
conditions.
In Table 4, this procedure was compared with changes in ET under elevated CO, conditions from more
detailed simulations using the SPAM model (Allen, 1986). The responses of RATIO to CO2 and to temperature
were similar to those reported by Allen for a full canopy (LAI=4) and wind speed of 3.6 m/s. Under very low
LAI values, the RATIO for double CO2 approached the RATIO of leaf stomatal resistances, which was about
0.68 for soybean. For higher LAI values, the RATIO approached 0.98.
Limitations Inherent in the Models. The results obtained in this study were based on two crop simulation
models and are thus subject to the assumptions and limitations of these models. The models were developed
under a range of soil and climate conditions and tested over others. However, neither of the models has been
tested under the conditions suggested by the GCMs. The models do account for changes in solar radiation,
temperature, and precipitation.
2-18
-------
Peart
CERES,
SOYGRO
CROP
MODELS
EOS
Figure 5. Variable Dependency Diagram showing how potential transpiration (EOF) is
modified under elevated CO2 levels in SOYGRO and CERES crop models.
2-19
-------
Peart
However, most of the data used to derive the relationships affecting processes in the models were for
temperatures below 35°C. In many cases, the two GCMs projected temperatures above 35 and even 40°C.
Relationships of processes to temperatures in this range are extrapolated. In SOYGRO, much of the yield
decline occurred because of lower seed growth rates, pod setting, and photosynthesis above 35°. In
CERES-Maize, much of the yield decrease in this temperature range was due to shorter seed filling periods.
The models have been tested over a wide range of rainfall and irrigation conditions, but do not account for
flooding, which could occur during time periods when increases in precipitation occur. The models also assume
that soil nutrients and micronutrients are not limiting and that there are no major soil problems such as acidity,
high compaction, or salinity. The models also assume that pests (insects, diseases, weeds) are controlled and
pose no limitation to crop growth and yield. Therefore, the results of the models should be used as an indicator
of the relative effects of climate change on yield. Absolute yields harvested by farmers over an area may be
lower than simulated yields because of the occurrences of some of these limitations to crop growth over time and
space.
The first approximation of the direct effects of COL were included as an estimate of combined effects of
climate and plant response to changes in CO2. Modifications to the models were made to compute
photosynthesis and evapotranspiration rates under increased CQ>2 but other parameters, such as rate of leaf
appearance, appears to be affected also for soybean (Baker et al., 1988) and other crops such as tomatoes (E.
Dayan, J.W. Jones, unpublished data, Gainesville, FL). Interactions between photosynthesis, temperature, and
solar radiation occur in the models because these variables affect the same growth processes. An overall climate
change may have higher CO2, tending to increase photosynthesis, and higher temperature, which decreases
photosynthesis. The combined effect could be higher or lower photosynthesis rates.
However, the methods used to implement changes in photosynthesis and transpiration in these models need
to be improved for future studies. The ratio method for computing changes in ET due to stomatal closure under
elevated CO>2 conditions mimicked the results of more detailed models. However, changes in atmospheric vapor
pressure were not included in the current model, and changes in plant temperature due to stomatal closure
under elevated CO2 conditions were not computed or used to modify plant growth processes. It is not clear at
this time just how much difference those limitations would cause. Allen (1986) showed that increases in CG>2
to 800 vpm could cause plant temperatures to be 2-3°C higher than if they were under 330 vpm for the same
climate conditions.
Table 4. Comparison of the Ratio of Soybean Canopy ET Under Elevated COL Concentrations Using the
Derived RATIO in This Paper and the More Detailed Simulations by Allen (1986) Using the SPAM
Model (Leaf Area Index = 4.0; Average Wind Speed = 3.6 m/s; Leaf Boundary Layer Resistance
of 10 s/m.)
CO2 Temp = 31°C Temp = 18°C
RATIO!/ LHA2/ RATIO!/ LHA2/
450
600
800
.977
.950
.916
0.970
0.938
0.903
.966
.926
.878
0.960
0.938
0.897
I/Ratio of plant transpiration under elevated CO2 to that under ambient CO2 conditions computed using the
ratio of Penman-Monteith equations.
2/From Allen, L.H., Jr. 1986.
2-20
-------
Peart
Similarly, changes in photosynthesis rates under doubled CO, were modeled by increasing gross
photosynthesis rates by 35% for soybean and 10% for corn on a daily, canopy basis. Auxiliary simulations with
a canopy photosynthesis model were used to demonstrate the adequacy of these values relative to published
canopy rates under high-light, mid-day conditions. However, this increase for both crops depends on light
intensities, and changes in cloudiness over time or under the proposed climate scenarios could significantly
reduce the beneficial effects of CO2 on canopy photosynthesis rates because of the shapes of plots of
photosynthesis vs. light intensity under various CO2 concentrations.
The models do account for increased growth rates resulting from CO, increases. For soybean, the data
reported by Baker et al. (1988) provide a basis for comparing the realism ofihe model to simulate physiological
responses under combinations of CO2 and temperature. The time constraints of this study did not allow for a
detailed comparison, but general comparisons between SOYGRO results and those data were good. For
example, simulated increases in LAI under doubled CO2 were similar to those reported by Baker et al. (1988),
and decreases in harvest index and seed size were similar. Seed yield increases in the Baker et al. (1988) study
were about 45, 24, and 15% for the low (26/19), medium (31/24), and high (36/29) temperature treatments,
respectively. Under ambient temperatures in Gainesville, increases in seed yield were about 40% with the
modifications under well-watered conditions.
THE WEATHER SCENARIOS
The Scenarios Used
In this study we used three weather scenarios:
a) Standard weather data for 30 years (1951/80) for 19 locations in 11 states in the southeastern U.SA.
b) Standard weather modified by the ratios provided by the GISS GCM model for two grid points near
Charlotte, NC, and Memphis, TN. Using rectangles to delineate the applicable range of each grid point, we
identified which locations were in the respective rectangles for the Charlotte and Memphis grid points and
applied the ratios to the solar radiation, temperature, and precipitation data for these locations. No
interpolations were made between grid points.
c) Standard weather was modified by ratios provided by the GFDL GCM model for eight grid points, near
St. Louis, MO, Greenville, MS, New Orleans, LA, Huntingdon, WV, Augusta, GA, Gainesville, FL, Washington,
DC, and a grid point in the Atlantic Ocean off the Virginia coast. The same system of rectangles centered on
grid points was used to apply specified ratios to data for sites within the rectangles. No interpolations were
made between these grid points. Ratios for temperature, solar radiation, and precipitation were used in both
cases. The standard weather data was provided by NCDC, Asheville, NC, by way of NCAR, Boulder, CO, with
the help of Roy Jenne. This standard weather data included only temperature and precipitation. Solar radiation
was generated by a synthesis program, WGEN, developed by Richardson (Richardson and Wright, 1984,
Richardson, 1985) and modified by Hodges et al. (1985).
Table 5 gives the location data as well as the appropriate GISS and GFDL grid points. Figures 6 and 7
provide the location of the grid points for the two weather scenarios for doubled carbon dioxide.
Issues Resulting from the Scenarios
Limitations of the Weather Scenarios
i) Precipitation variability. When considering rainfall and crop growth, a serious question arises related to
changes in rainfall such as suggested in the scenarios used in this study. Will the change in rainfall be reflected
in more/fewer rainfall events or the same number of events with higher/lower amounts in each event? The
2-21
-------
Peart
Table 5. List of Study Sites With Related GISS and GFDL Grid Box Mid-Points
SITE
CITY
LAT
LNG
GISS
BIAL BIRMINGHAM, AL
MBAL MOBILE, AL
MGAL MONTGOMERY, AL
LRAR LITTLE ROCK, AR
TLFL TALLAHASSEE, FL
ATGA ATLANTA, GA
MCGA MACON, GA
LUKY LOUISVILLE, KY
BRLA BATON ROUGE, LA
SHLA SHREVEPORT, LA
MEMS MERIDIAN, MS
CHNC CHARLOTTE, NC
RANG RALEIGH, NC
CLSC COLUMBIA, SC
MPTN MEMPHIS, TN
NSTN NASHVILLE, TN
LBVA LYNCHBURG, VA
NOVA NORFORK, VA
WING WILLMINGTON, NC
33.34
30.41
32.18
34.44
30.23
33.39
32.42
38.11
30.32
32.28
32.20
35.10
35.52
33.57
35.03
36.07
37.20
36.54
34.16
86.45
88.15
86.24
92.14
84.22
94.26
83.39
85.44
91.09
93.49
88.45
80.5
78.47
81.07
89.59
86.41
79.12
76.12
77.55
MEMPHIS
MEMPHIS
MEMPHIS
MEMPHIS
CHARLOTTE
CHARLOTTE
CHARLOTTE
MEMPHIS
MEMPHIS
MEMPHIS
MEMPHIS
CHARLOTTE
CHARLOTTE
CHARLOTTE
MEMPHIS
MEMPHIS
CHARLOTTE
CHARLOTTE
CHARLOTTE
GFDL
GREENVILLE, MS
NEW ORLEANS, LA
AUGUSTA, GA
GREENVILLE, MS
GAINESVILLE, FL
AUGUSTA, GA
AUGUSTA, GA
ST. LOUIS, MO
NEW ORLEANS, LA
GREENVILLE, MS
GREENVILLE, MS
AUGUSTA, GA
AUGUSTA, GA
AUGUSTA, GA
GREENVILLE, MS
ST. LOUIS, MO
HUNTINGTON, WV
WASHINGTON, DC
ATLANTIC OCEAN
2-22
-------
Peart
Figure 6. GISS grid boxes and mid points (near Memphis, TN, and Charlotte, NC) and weather data locations
used in this study.
Figure 7. GFDL grid boxes and mid points (near St. Louis, MO; Huntington, WV; Washington, DC; Greenville,
MS; Augusta, GA; New Orleans, LA; Gainesville, FL; and in the Atlantic Ocean off the Virginia
coast).
2-23
-------
Peart
answer to this question could completely change the results of this study in terms of rainfed yields if it suggested
that the change was in terms of events rather than amounts. The assumption we made for runs reported was
that the amount of rainfall was changed but the number of events was not.
The importance of using daily rainfall data as we did in this study is emphasized in a 4-year study of
turfgrass water requirements at Fort Lauderdale, FL. Allen et al. (1978) showed that on a whole-year basis,
annual rainfall exceeded annual evapotranspiration by 198 to 561 mm. However, on a monthly water-budget
basis, irrigation requirements computed for the whole year ranged from 142 to 356 mm. Furthermore, irrigation
requirements for daily water budgets were much higher, 569 to 589 mm, for very shallow-rooted turfgrass on
droughty soil with available water of 20 mm in the top 300 mm of soil. For a soil holding; 71 mm of avadable
water in the rooting zone, the calculated annual irrigation requirements ranged from 213 to 427 mm. Therefore,
crops are likely to have more serious water stress periods under actual daily conditions than would appear when
the rainfall is averaged across each month.
ii) Variability of the precipitation ratios. The variability of the precipitation ratios from month to month for
both GISS and GFDL is much greater than the variability of the temperature and solar radiation ratios. This
variability is of particular concern when dealing with crops such as soybeans and corn where timing of
morphologic development related to soil moisture stress greatly affects final yield. This large variability of
precipitation makes the meaningfulness of the results open to question.
In Table 6, for example, note the variations in monthly Columbia, SC, precipitation for July, August, and
September under the three scenarios:
Base (actual 30-vr. ave.) GISS GFDL
July 136 mm 192 mm 88 mm
August 140 mm 173 mm 60 mm
September 107 mm 86 mm 87 mm
August is a critical month for reproductive fruit growth.
iii) Representative plant temperature. Based on the information provided by NCAR, the air temperature
is calculated by the GCMs at some elevation considerably above the crop. The ratio of this temperature with
and without double CO2 when applied to the 1.5 m air temperature from the historic data base may not
represent the actual air temperature above the plant canopy. A second assumption is that the air temperature
at 1.5 m (such as measured and reported to NCDC) is the plant temperature in the models. Since temperature
has major influences on most crop growth processes, a major limitation is the assumption that the historic air
temperature as modified by the GCM scenarios is the plant temperature. Plant temperatures are often different
from air temperatures and are influenced by vapor pressure deficit of the air. Thus, errors in computing air
temperature at 1.5 m and errors in assuming that plant temperature is the same as air temperature may result
in errors in the predictions of the impact of climate and CO2 enrichment effects. The magnitude of these errors
is not known at present, but future work should investigate this limitation.
iv) The monthly time step of the ratios provided by the GLM scenarios may be too large for accurate results
from the plant models because a month is a very long time when calculating the timing of the physiological
events in the plant life cycle, such as flowering, fruit set, and maturity.
v) Sites for which data were provided do not correspond very well with major growing areas in the region.
CROP MANAGEMENT
No attempt was made to evaluate the effects of crop management on the results of the study of doubled
CO2 effects on crop yield due to the time and resources constraints of the project.
2-24
-------
Peart
Table 6. Summary of Weather Data, 30-year Averages (1951-80) of Three Sites in the DeUaOVIemphis),
Uplands (Charlotte), and Coastal Plains (Columbia) (PREC. - mm, TMAX and TMIN = c;
COLUMBIA, SC
MONTH
Jan.
Feb.
Mar.
Apr.
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
MEAN
AVG.
PREC.
111.25
101.35
131.06
91.19
97.28
112.52
135.89
U0.21
107.44
65.28
63.75
88.90
103.84
YR. SOLAR
STO
383.60
STD
TMAX
13.45
15.28
19.47
24.99
28.78
31.76
33.27
32.77
29.74
24.72
19.53
14.86
24.05
RADIATION
GISS
392.87
TMIN
0.67
1.47
5.51
10.26
15.04
18.93
21.17
20.80
17.74
10.16
4.77
1.49
10.67
GFDL
387.28
1
PREC.
120.93
126.07
162.13
107.42
110.80
147.63
192.01
172.60
85.63
63.32
47.56
81.52
118.14
GISS
THAX
15.75
17.42
23.48
27.05
31.65
34.29
35.23
34.14
33.02
28.08
21.89
16.64
26.55
THIN
2.97
3.60
9.52
12.32
17.91
21.47
23.14
22.17
21.02
13.53
7.13
3.27
13.17
I
PREC.
85.55
106.82
150.33
104.77
80.55
49.62
88.46
59.73
86.81
44.00
68.73
114.77
86.68
GFDL
TMAX
15.57
17.60
22.20
28.36
31.27
35.85
38.19
35.71
33.80
27.71
22.98
18.24
27.29
TMIN
2.79
3.78
8.24
13.63
17.53
23.03
26.09
23.74
21.80
13.15
8.22
4.87
13.91
MEMPHIS, TN
MONTH
Jan.
Feb.
Mar.
Apr.
Hay
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
MEAN
PREC.
117.09
109.98
138.18
146.56
128.52
90.93
102.36
95.00
91.95
60.20
105.92
123.1?
109.16
STD
TMAX
9.05
11.65
16.33
22.70
27.21
31.35
33.07
32.36
29.04
23.60
16.33
11.26
22.00
1
TMIN
-0.59
1.14
5.49
11.24
16.06
20.48
22.56
21.58
17.83
10.73
5.06
1.?n
11.07
PREC.
78.34
185.32
85.67
116.95
183.28
110.12
153.85
83.79
100.13
36.54
90.03
90.30
109.53
GISS
THAX
12.00
14.64
20.66
26.24
29.51
33.89
35.44
34.92
33.68
26.77
21.73
15.13
25.38
I
THIN
2.37
4.13
9.82
14.79
18.36
23.02
24.93
24.13
22.46
13.91
10.47
5.15
14.46
PRC.
91.10
123.51
198.97
84.86
104.10
61.47
43.20
75.33
180.31
42.44
106.24
196.00
108.96
GFDL
TMAX
11.31
13.59
19.04
25.68
30.46
33.89
36.03
34.33
32.32
26.40
19.57
14.78
24.78
TMIN
1.67
3.07
8.20
14.23
19.31
23.02
25.52
' 23.54
21.11
13.53
8.31
4 79
13.86
AVG. YR. SOLAR RADIATION
STO GISS GFOL
383.90 390.11 385.76
CHARLOTTE. NC
MONTH
Jan.
Feb.
Mar.
Apr.
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
MEAN
AVC.
PREC.
96.5
96.8
122.7
83.1
92.5
90.7
99.1
95.3
91.2
69.1
72.4
86.4
91.3
STD
TMAX
10.26
12.12
16.56
22.45
26.41
29.76
31.47
31.12
27.83
22.21
16.61
11.58
21.53
TMIN
-0.71
0.03
3.92
9.12
14.01
18.11
20.33
20.05
16.78
9.76
4.29
0.44
9.68
| PREC.
104.92
120.39
151.76
97.84
105.31
118.97
139.97
117.25
72.68
67.02
54.00
79.19
102.44
GISS
TMAX
12.53
14.22
20.52
24.49
29.25
32.28
33.43
32.48
31.09
25.54
18.95
13.33
24.01
TMIN |
1.55
2.13
7.88
11.16
16.86
20.63
22.29
21.42
20.04
13.09
6.63
2.20
12.16
PREC.
74.22
102.00
140.72
95.43
76.55
39.99
64.49
40.58
73.68
46.57
78.04
111.49
78.65
GFDL
TMAX
12.35
14.40
19.26
25.79
28.87
33.82
36.36
34.04
31.86
25.17
20.03
14.91
24.74
TMIN
1.38
2.31
6.62
12.46
16.48
22.17
25.22
22.98
20.81
12.72
7.71
3.78
12.89
YRLY. SOLAR RADIATION
STD GISS GFDL
389.20 398.61 392.93
2-25
-------
Peart
A review of the results would suggest that adaptation to changing climate by changing cultivars could possibly
compensate for some of the negative responses to climate change. The extent of this is unknown and needs
further investigation. For both corn and soybean, varieties are available and the models could be used to make
an evaluation of this management practice. Earlier planting probably would be adopted by farmers as weather
becomes warmer in the early spring, and the impact of this practice could be evaluated using the models.
Double cropping is a management practice that could be applied in areas where the changing climate
provided a longer growing season. The limitation would be the availability of adequate soil moisture at planting
time for the second crop as well as for the growth and development of the second crop. This practice could also
be evaluated for both corn and soybeans using the modeling.
SIMULATION EXPERIMENTS
Extension specialists or researchers in each of the 11 southeastern states were contacted to get current
recommended varieties, soil types, planting dates, and area planted in soybeans, as well as the same information
for corn. Table 1 presents the locations of the 19 weather sites, soil types, and cultivars used.
A set of correction parameters from the GISS and the GFDL data was applied to each weather site. Two
grid points from GISS and eight grid points from GFDL data were used, and the parameters applied for each
month to the appropriate weather site data without geographic interpolation. WGEN was used to generate solar
radiation data for each of the 19 weather sites using the temperature and precipitation data from the weather
data provided by NCAR.
Based on information obtained from personal contacts in each of the southeastern states, we chose a soil
type from a generic soil base list that most closely characterized a typical field soil in the vicinity of each of the
19 sites. Note that this generic soil list was the same overall list as used by Ritchie of Michigan State University
for the study of corn and soybeans in the Great Lakes region and by Rosenzweig for wheat in the Great Plains
region. Runs of 30-year simulations for particular locations were made and the growing season dynamics plotted.
Runs were made for rainfed and irrigated management. For the irrigated management, when the available
moisture dropped to 40% in the upper 0.5 m of the profile, water was applied to fill the soil profile to Held
capacity. An application efficiency of 75% was used.
As one looks over the results for the many runs of SOYGRO for various years and locations and weather
scenarios, it is obvious that it would be valuable to be able to compare this data with data from runs in which
a weather scenario was used that assumed a doubling of the atmospheric CO, level, but without the
accompanying climate change suggested by the GISS and GFDL scenarios. This study would be a top priority
for further work, if and when additional resources are available.
2-26
-------
Peart
CHAPTERS
RESULTS
RELATIVE CONTRIBUTIONS BY EACH CLIMATE VARIABLE
For a detailed study of the individual effects of the various weather factors, simulations were done for the
1951 weather data for Memphis, TN, and for the same data with the GISS doubled CO2 factors applied. The
SOYGRO model that responds to climate effects and direct effects of CO2 was used. Precipitation, temperature,
solar radiation, and carbon dioxide levels were changed individually between the standard or base data and the
GISS 2xCO2 data to produce results for each of the above factors alone.
Figures 8 and 9 show the changes in precipitation, and maximum daily temperature over the season,
indicating about a 20% increase in rainfall over the season and 4-6°C increase in temperature. Solar radiation
changed very little (not shown). The simulated crop growing season length changed very little for any of the
cases. Leaf area index (LAI) increased significantly under elevated CO, conditions (Figure 10), but much less
under any of the other cases. The 33% increase in LAI under double CO2 conditions is similar to the data
reported by Baker et al. (1988) for soybean.
Seed yield was affected by temperature and precipitation (Figure 11). A water stress in the selected year
occurred during the middle of the seed filling period. Increased rainfall for case P reduced the magnitude of
the stress and caused about a 24% yield increase. Temperature increases of 4-6° during the season resulted in
a 68% decrease in yield. This large decrease is due to the direct effect of high temperature on growth processes
in SOYGRO and to increased evapotranspiration demand, which increased the length and severity of the
drought. The direct effect of CO2 acting alone caused about a 58% increase in seed yield, due to both increases
in photosynthesis and reduced water stress (lower seasonal evapotranspiration (Figure 12).
Summaries of biomass, seed yield, and seasonal evapotranspiration are presented in Table 7. In general,
temperature increases proposed by the GISS model caused considerable decreases in biomass and seed yield.
Although increased precipitation at this site and double CO, increased yield when acting alone, their combined
effect was not enough to overcome the temperature-induced yield loss, and a net loss of 11.5% occurred. It was
interesting to note that seasonal ET decreased by 2% under direct CO2 effects alone, but when all effects were
included, ET increased by almost 11% primarily due to increased temperatures. Results from the GFDL
scenarios would probably be more severe because of the lower amounts of precipitation simulated by that model.
SUMMARY OF YIELD RESULTS FOR ALL LOCATIONS
Simulations were run for all 19 locations, rainfed and irrigated, three weather scenarios (base, GISS 2xCO,
and GFDL 2xCO2) with SOYGRO and CERES-Maize models with climate effects alone. The same set of
conditions was used for runs of SOYGRO modified for the combined climatic and direct effects. For CERES-
Maize, four locations were run with the model modified for combined climatic and direct effects. These results
were obtained for some 315 sets of 30-year runs, each requiring 1/2 to 2 hours of run time on microcomputers.
Table 8 is an example of the summary spreadsheets that were obtained for each combination. These
spreadsheets calculated means, standard deviations, percentage differences, uncertainties, and yields in metric
and English units. Spreadsheet input data was taken from output data of the simulations. Length of the time
from planting to crop maturity, total evapotranspiration during that period, rainfall, total irrigation water applied,
and seed weight are shown for each year's run.
2-27
-------
Peart
£
E
900
750 -
- 600-
450-
300 -
150 -
o
1951 MEMPHIS, TN
GISS, 2XC03,
MAY 14
JUN II
JUL09
AUG06
SEP 03
OCTOI
Figure 8. Cumulative rainfall as a function of time of year, based on SOYGRO runs for both GISS and
STANDARD weather sets for Memphis, TN. 1951. The version of SOYGRO used had been
modified to include combined climatic and the direct effects of 2xCO2 on photosynthesis and
transpiration.
2-28
-------
Peart
O
UJ
or
i
a:
UJ
o_
x
<
50
45 -
40-
35 -
30
25 -
20
1951 MEMPHIS, TN
MAY 14
JUN II
JUL09
AUG06
SEP 03
OCTOI
Figure 9. Maximum daily air temperatures as a function of time of year for the same conditions as Figure 8.
2-29
-------
Peart
12.0
1951 MEMPHIS, TN
MAY 14
JUN
JUL09
AUG06 SEP 03 OCT 01
Figure 10. Simulated leaf area index as a function of time of year for the case where direct effects of 2xCO2
(GISS) alone are included for the 1951 year of weather data from Memphis, TN. The standard
curve represents the effect of normal ambient CO2.
2-30
-------
Peart
4000-
3200-
o>
JC.
UJ
240O-
1600-
g 800 -I
CO
/B «^
/ B
AMBIENT C0
MAYI4
JUNII
JUL09
AUG 06
SEP 03 OCTOI
Figure 11. Simulated seed yield vs time of year using the 1951 Memphis, TN weather data except the standard
weather was modified by the GISS climate model parameters for precipitation alone (P), solar
radiation alone, (S) temperature alone (T), and combined (S+P+T).
2-31
-------
Peart
4800
4000 H
S. 3200
^.
o>
Q- 2400 -
LJ
>- 1600
o
LJ
£ 800 H
0
1951 MEMPHIS, TN
MAY 14
STANDARD
JUN
JUL09
AUG06
SEP 03
OCTOI
Figure 12. Simulated seed yield versus time of year comparing the results for the combined climate and direct
effects of 2xCO2 (GISS) with those of the standard weather data
2-32
-------
Peart
Table 7. Sensitivity of Simulated Yields to Each Climate Change Factor Alone,
Based on the GISS Climate Change Scenario and Double C02, and to
Combinations for the Memphis, TN Site Using 1951 Weather Data
Biomass Seed Yield Evapotranspiration
Factor kg/ha kg/ha mm
Standard Run 6690 2520 617
Percentage Change
Solar Radiation (S) -0.6 -0.8 -0.2
Precipitation (P) +15.5 +23.8 +5.0
Temperature (T) -44.1 -68.3 +4.1
Carbon Dioxide (C) +53.9 +57.9 -2.0
S+P+T -29.4 -41.3 +12.3
S+P+T+C +7.8 -11.5 +10.9
2-33
-------
Peart
Table 8. SOYGRO, Modified for Direct C02, GISS Weather, RAINFED
University of Florida, EPA Global Climate Change Project, 1988
MP.TN SOYGRO- 2 Planting Date: 135
2C02 RAINFED GISS STD RAINFED
Yr Length GET
51 125 684
52 126 491
53 125 390
54 130 414
55 125 658
56 125 543
57 123 698
58 125 652
59 125 706
60 126 661
61 127 486
62 125 644
63 126 593
64 125 629
65 125 518
66 126 464
67 124 612
68 125 530
69 126 484
70 127 651
71 126 619
72 125 585
73 127 587
74 124 647
75 125 513
76 125 599
77 129 531
78 127 671
79 126 613
80 134 520
SUMMARY,
IR Rain SdWtco | Length GET IR Rain SdWt
0 647 223
0 328 29
0 398 38
0 252 2
0 664 219
0 417 90
0 857 417
0 626 312
0 703 441
0 580 134
0 305 107
0 589 215
0 531 295
0 524 292
0 624 68
0 408 90
0 533 473
0 512 84
0 312 90
0 553 293
0 498 340
0 556 192
0 537 266
0 881 324
0 430 127
0 558 177
0 463 101
0 614 329
0 621 349
0 634 20
Univ. of Florida
MP.TN SOYGRO
2C02 RAINFED GISS
Length GET
Meanl26.0 579.8
SDV 2.02 83.3
SDVS .37 15.2
Yield Analysis
Means :
Base Yld, muA:
2C02 Yld, muB:
Diff. (muC):
Diff.sigma muC:
% Diff ,muC/muA:
sigma muC/muA,
Uncertnty : + , -
IR Rain SdWtco
.0 538.5 204.6
.0 143.6 131.7
.0 26.2 24.0
125 620 0 523 252
125 449 0 274 47
124 357 0 277 36
126 385 0 218 14
125 593 0 517 257
125 489 0 330 76
124 633 0 706 388
125 577 0 526 335
125 610 0 560 431
125 588 0 526 160
128 428 0 254 74
123 558 0 502 225
126 528 0 416 277
125 558 0 437 332
124 483 0 519 97
127 443 0 346 84
127 561 0 433 414
126 486 0 419 130
125 442 0 292 86
125 573 0 443 306
125 566 0 438 334
125 532 0 440 189
124 526 0 447 319
126 598 0 700 316
124 466 0 343 126
127 532 0 443 168
123 455 0 367 134
123 571 0 544 355
125 536 0 534 348
126 468 0 504 79
EPA Project, 1988
Planting Date:
STD RAINFED
Length GET IR Rain SdWt
125.1 520.4 .0 442.6 213.0
1.19 69.9 .0 117.0 124.6
.22 12.8 .0 21.4 22.7
Summary Report: Definitions:
g/m2 Bu/acre Length (days) -Growing Season
213 31.7 GET (mm)- Evapotranspiration
204.6 30.4 IR (mm) -Cumulative Irrigation
-8.4 -1.2 Rain (mm) - Cumulative Rain
Diff
-29
-18
2
-12
-38
14
29
-23
10
-26
33
-10
18
-40
-29
6
59
-46
4
-13
6
3
-53
8
1
9
-33
-26
1
-59
Diff
-8.4
26.5
33 4.9 SdWtco (g/sq.m)-Yield, GISS, 2C02
-3.9% SdWt (g/sq.m)- Base Yield
15.5% R. M. Peart, Agr. Eng. Dept. 4/28/88
2-34
-------
Peart
Uncertainty percentages are shown in some tables, and they were calculated as follows:
. . *w c "c
The uncertainty of % change = — +_ x 100
where nc = difference between mean of standard yield and the mean of yield as a result of modified climate.
/ia = mean of standard yield
02 •* »
- * *
where CT/ia and *Vb are the standard deviations of the appropriate yields.
In this discussion, we will focus more on rainfed than irrigated cases, since the great majority of these crops
are not irrigated in the Southeast. We discuss the results first with the original SOYGRO and CERES-MAIZE
models, which are sensitive to weather variables, but assume current concentrations of CO2 in the ambient air.
Table 9 shows the rainfed and irrigated results for all locations. Figures 13 and 14 show the 30-year average
rainfed and irrigated soybean results for all locations and for both the climate alone (GISS-N and GFDL-N)
scenarios and the climate and direct CO, effects (GISS-D and GFDL-D) scenarios. The combined (-D)
scenarios account for carbon dioxide enrichment, or the direct effects.
Annual Variability. Most results shown in this report are 30-year averages for actual 1951-1980 weather at
a given location or for weather data modified by the GISS or GFDL weather models. The following figure was
included to emphasize the year to year variability at a sample site. A shorter period of 10 years (1951-60) was
chosen for these examples. Figure 15 shows 10 years of yearly results at Memphis, TN, 1951-60, and the
variability is striking. Three of the years would be considered crop failures, when value of the yield hardly
covered harvesting costs, while three other years produced outstanding yields. Figure 15 results include the
modified SOYGRO results, which account for carbon dioxide enrichment. GISS results show slightly higher than
base yields for only 2 of the 10 years at Memphis, and GFDL results are lower in all years, and drastically lower
in most of the 10 years. Variability between years is even greater with the GISS weather than the base weather.
Since the GFDL yields were so much lower, variability was less.
Soybean Simulations for Climate Effect Only. For climate effect alone the discussion in this section will refer
only to the models of climate alone (-N). The results, therefore, are only influenced by the change in weather
variables and not by any direct influence of CO2 on photosynthesis and transpiration.
In the Delta area, where normal yields are lower, the climate predicted by carbon dioxide doubling (2xCO2)
cuts yields almost in half for GISS and reduces the yields to near crop failure for the GFDL case. For
commercial and economic reference, 1800 kg/ha is a little less than 30 bushels per acre, a typical yield, and
soybean prices have been the $6-$8/bushel range. Thus the Delta GFDL-N results would mean a gross income
of about $30- $40 per acre, not enough to pay land capital costs.
The Coastal Plains locations had better base yields and reductions of about 20% due to the GISS weather
and 70% due to the GFDL weather. The GFDL results were more variable in their effect among locations, with
reductions of 78% to 90% for Columbia, Macon, Meridian, Montgomery, and Norfolk, which means 30 years
of mostly crop failures. The other three locations, Mobile, Tallahassee, and Wilmington, all coastal areas, were
reduced less.
2-35
-------
Peart
Table 9. SOYGRO Results, Kg/ha, Southeastern USA, 30-yr Average
N - Doubled C02 , weather effects only.
D - Doubled CO,, weather and direct effects.
SOYGRO Rainfed Yields, Kg/ha, 30-Yr. Ave.
Location | Climate Alone,N | Comb. Climate & Direct,D
DELTA:
COASTAL PL:
UPLANDS :
BASE
1701
2698
2724
GISS-N
982
2067
2313
GFDL-N
309
817
878
GISS-D
1451
2978
3313
GFDL-D
572
1358
1393
AVERAGE: 2495 1929 733 2777 1205
SOYGRO, Irrigated Yields, Kg/ha, 30-yr Average
DELTA: 3948 2814 3120 3931 4425
COASTAL PL: 3770 3144 3097 4265 4381
UPLANDS: 3852 3370 3070 4686 4388
AVERAGE: 3837 3158 3092 4347 4390
DELTA - Baton Rouge, LA, Little Rock, AR, Memphis, TN, Shreveport, LA.
COASTAL PL - Columbia, SC, Mobile, AL, Macon, GA, Meridian, MS,
Montgomery, AL, Norfolk, VA, Tallahassee, FL, Wilmington, NC.
UPLANDS - Atlanta, GA, Birmingham, AL, Charlotte, NC, Lynchburg, VA,
Louisville, KY, Nashville, TN, Raleigh, NC.
2-36
-------
Peart
UPLANDS
cm-u tm-« tnvt
i: m-iinct. -i •- IIBCI
COASAL PLAIN
MU tUHl OIM OU-I OM
. -I-.IIBO
wt
-------
Peart
UPLANDS
CUM cm-i cm-i
Kind. Hi: m-imct. -I: IIKC1
COASTAL PLAIN
tm-i on-*
-------
Peart
M
>-
SO¥GRO, DIR, EFFECTS, HEHPHIS VIELDS
RAINFED, i§51-60
1DWJ-
4000-
A W W
3500-
innn
3080-
2500-
2000-
1500-
1000-
500-
Q-
7
1
I
1
/
/
\
\
\
\
\ l/hm
r
J\k" rn.
/
1
1
1
-\
\
\
\
\
\
I
/
1
\
\
n
7
/
/
/
/
/
/
\
\
\
\
\
\
\
\
\
\
\
\
L
/i^i
1
/
/
/
/
/
/
/
\
\
\
\
\
\
\
\
1
-
/
/
/
/
\
\
\
\
\
\
\
\
\
\
\
1
/
/
/
\
\
\
\
51 52 53 54 55 56 57 58 59 60
YEAR
Q BASE HEATHER QGISS, 2«C02 0GFDL-D, 2«C02
Figure 15. Typical annual yield variations, 1951-60.
2-39
-------
Peart
Most of the Upland locations had more moderate reductions for the GISS weather, but drastic reductions
for GFDL. Lynchburg, VA, had a peculiar tiny increase with GISS, probably due to more rainfall at some times
and more moderate temperatures.
In summary, the rainfed climate-effect-only SOYGRO model used with the GISS weather scenario shows
serious yield reductions, from 20% to almost 50% for 13 of the 19 locations, and less serious reductions for the
other 6 locations. Used with the GFDL weather scenario, the model showed very serious yield reductions of not
less than 37% and averaging 72.7% for all locations.
The irrigated results are shown in Tables 10 and 11, and naturally show less drastic effects, since the
irrigation cancels any lowered rainfall effects of the doubled carbon dioxide weather scenarios. However, the
average effects show similar reductions of 17.5% for the GISS scenario and 19.2% for GFDL. Comparing the
GFDL results for rainfed and irrigated implies that rainfall reductions were a major factor in the GFDL model
results for climate effects alone.
Maize Simulations for Climate Effect Only. Table 12 summarizes the Maize model results, nonmodified
for direct effects. Rainfed GISS reductions from the base runs are much less drastic at Baton Rouge, Little
Rock, Montgomery, and Shreveport than for soybeans. These four locations account for most of the change in
the percent difference between soybeans and maize for rainfed GISS results. The lower sensitivity to temperature
effects in this range probably account for the difference between soybeans and maize at the four locations.
GFDL rainfed results were drastically reduced, averaging 64.6%.
Irrigated maize showed results similar to soybeans, with average reductions of 18.2% for GISS and 27.6%
for GFDL doubled carbon dioxide scenarios compared to the base weather. These reductions were more
uniform among locations than was the case for soybeans.
Soybean Simulations for Combined Climate and Direct Effects. As described earlier in this report,
SOYGRO was modified with a set of constants to increase the photosynthetic rate and increase the storaatal
resistance to transpiration for the doubled carbon dioxide. This effect is often called "carbon dioxide enrichment,"
and is carried on in some commercial greenhouses to increase the rate of growth of plants. This effect has led
some writers to assume that doubling the carbon dioxide in the atmosphere would increase crop production, but
the answer is not that simple.
The simulation process, where all the effects of the different environmental conditions, rainfall, solar
radiation, temperature, carbon dioxide concentration, and others are integrated on a daily basis, is necessary to
estimate the overall effects of such a change. These results are based on quickly made modifications to the
models, and more detailed work on the models and in growth chambers are needed, but they give the best
estimates we can now make of the results of the doubled carbon dioxide scenarios.
Table 11 gives, by areas of the southeastern U.SA., yield results for the modified SOYGRO under rainfed
conditions for the GISS and GFDL weather scenarios. Under the GISS conditions, 7 of the 19 locations had
yield reductions, and the average change was an increase of 9.1%. Under the GFDL conditions, all but two
locations were drastically reduced in yields, with an average reduction of 54.6%. For the GISS data, the Delta
area had reductions of about 15%, the Coastal Plains increased in yield by about 10%, and the Upland area had
an increase of about 20%. For the GFDL weather, all areas suffered about equally.
2-40
-------
Peart
Table 10. SOYGRO Simulation, Univ. of Florida, 30-yr Averages
Non-Modified for C02 Effect on P-Syn., 3-18-88
G1SS | GDFL
Yld, Kg/ha Diff..Uncrty, | Kg/ha Diff, Uncrty,
Location Base 2*C02 Pet. + ,-,% | 2*C02 Pet. +,-,%
NON- IRRIGATED (RAINFED)
AT.GA
BI.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB.AL
MC.GA
ME, MS
MG.AL
MP.TN
NO.VA
NS.TN
RA.NC
SH.LA
TL.FL
WI.NC
AVE.:
Location
AT.GA
BI.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB.AL
MC.GA
ME, MS
MG.AL
MP.TN
NO.VA
NS.TN
RA.NC
SH.LA
TL.FL
WI.NC
3618
3477
1775
2408
2603
2529
1641
2448
3450
1890
2374
2051
2132
2999
2226
2360
1258
3147
3067
2497
Yld,
Base
4270
3961
3618
4048
3948
3712
4008
3685
3753
3934
3961
4001
4048
3806
3887
3396
4116
3369
3389
3235
2394
915
2239
2085
2562
982
1903
2286
1426
1500
1244
1325
2771
1567
2293
706
2508
2717
1929
1
Kg/ha
2*C02
3800
3013
2710
3578
3537
3658
2878
3194
2804
3396
2932
3067
2952
3457
3093
3255
2717
2892
3067
-10.6%
-31.0%
-48.4%
-7.1%
-20.0%
1.5%
-40.3%
-22.1%
-33.8%
-24.5%
-36.9%
-39.3%
-37.9%
-7.7%
-29.6%
-2.8%
-43.5%
-20.3%
-11.4%
-24.5%
GISS
5.0%
4.7%
12.1%
12.7%
11.5%
11.1%
15.7%
11.2%
5.9%
12.4%
11.9%
14.5%
13.3%
7.7%
12.4%
10.8%
15.8%
4.4%
5.4%
10.4%
1
Diff. .Uncrty, |
Pet.
-10.9%
-23.9%
-25.1%
-11.5%
-10.5%
-1.3%
-28.1%
-13.3%
-25.4%
-13.8%
-26.1%
-23.3%
-27.1%
-9.2%
-20.5%
-4.1%
-34.1%
-14.2%
-9.6%
+,-,% 1
IRRIGATED
1.6%
1.8%
2.9%
1.5%
2.6%
1.9%
2.3%
2.2%
3.0%
2.8%
2.3%
2.5%
2.2%
1.6%
1.9%
2.3%
2.7%
3.3%
2.4%
807
1614
437
188
565
646
276
1533
1957
161
518
330
377
276
1063
296
148
1553
1177
733
-77.7%
-53.5%
-75.2%
-92.1%
-78.2%
-74.3%
-83.0%
-37.4%
-43.3%.
-91.4%
-78.1%
-83.8%
-82.5%
-90.8%
-52.2%
-87.6%
-88.2%
-50.7%
-61.6%
-72.7%
4.4%
5.1%
11.7%
9.6%
9.6%
8.8%
13.2%
10.6%
6.7%
9.6%
10.2%
11.6%
11.0%
5.9%
11.0%
7.9%
13.3%
5.0%
6.5%
9.0%
GFDL
Kg/ha
2*C02
3120
3188
3208
3020
3362
3060
3114
3120
3376
3208
3046
3181
3181
2885
2919
3067
2979
2730
2986
Diff, Uncrty,
Pet.
-26.9%
-19.5%
-11.3%
-25.4%
-14.9%
-17.6%
-22.4%
-15.3%
-10.1%
-18.6%
-23.2%
-20.5%
-21.3%
-24.2%
-25.0%
-9.7%
-27.6%
-19.0%
-12.0%
+ ,-,%
2.2%
1.8%
2.8%
1.8%
2.8%
1.9%
2.1%
2.2%
2.6%
2.9%
2.2%
2.6%
2.0%
1.8%
-2.0%
2.4%
2.6%
3.4%
2.4%
AVE.
3837
3158 -17.5% 2.3%
3092 -19.2% 2.1%
2-41
-------
Peart
Table 11. EPA Climate Change Project, 6/6/88,
SOYGRO, Modified for Direct CO, Effects, 30-yr. ave. Results
Loc. Base Wthr GISS-D Diff. Uncrty, GFDL-D Diff. Uncrty,
kg/ha kg/ha % % kg/ha % %
RAINFED SOYGRO
AT.GA
BI.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB.AL
MC.GA
MP.TN
ME, MS
MG.AL
NS,TN
NO.VA
RA.NC
SH.LA
TL.FL
WI.NC
Ave.
Std. Dev
3618
3470
1775
2408
2603
2529
1641
2441
3450
1890
2132
2374
2051
2226
2999
2361
1251
3147
3067
2496
639
4587
3470
1480
3181
3060
3692
1439
2663
3174
. 2105
1849
2179
1843
2320
4008
3275
1036
3632
3820
2780
969
26.
.
-16.
32.
17.
46.
-12.
9.
-8.
11.
-13.
-8.
-10.
4.
33.
38.
-17.
15.
24.
9.
8%
0%
7%
1%
6%
0%
3%
1%
0%
4%
2%
2%
2%
2%
6%
7%
2%
4%
6%
1%
6.
5.
13.
14.
13.
13.
18.
13.
6.
14.
15.
13.
16.
14.
8.
12.
18.
5.
6.
0
6
6
8
6
0
0
4
7
0
1
7
6
6
7
6
5
3
2
1197
2482
935
282
982
1130
464
2340
3174
343
652
854
659
1762
437
558
235
2616
1802
1206
871
-66
-28
-47
-88
-62
-55
-71
-4
-8
-81
-69
-64
-67
-20
-85
-76
-81
-16
-41
-54
.9%
.5%
.3%
.3%
.3%
.3%
.7%
.1%
.0%
.9%
.4%
.0%
.9%
.8%
.4%
.4%
.2%
.9%
.2%
.6%
5
6
14
9
11
10
14
12
7
10
11
11
12
12
6
8
13
6
8
.1
.4
.2
.8
.2
.4
.0
.7
.8
.0
.8
.2
.8
.9
.3
.7
.6
.0
.2
IRRIGATED
Location
AT.GA
BI.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB.AL
MC.GA
MP.TN
ME, MS
MG.AL
NS.TN
NO.VA
RA.NC
SH,LA
TL.FL
WI.NC
AVE:
STD. DEV
Base
kg/ha
4270
3961
3618
4049
3948
3712
4008
3685
3753
3934
4042
3961
4001
3887
3806
3396
4116
3369
3389
3837
248
GISS-D
kg/ha
5219
4203
3719
4956
4734
5077
4042
4519
3685
4580
4116
4049
4075
4358
4829
4468
3847
3995
4176
4350
442
Diff. Uncrty, :
%
22
6
2
22
19
36
22
-1
16
1
2
1
12
26
31
-6
18
23
13
.2%
.1%
.8%
.4%
.9%
.8%
.8%
.6%
.8%
.4%
.8%
.2%
.8%
.1%
.9%
.6%
.5%
.6%
.2%
.7%
%
2
2
3
1
3
2
2
2
3
3
2
2
3
2
1
2
3
4
3
2
.1
.2
.7
.8
.3
.3
.9
.6
.8
.6
.9
.8
.2
.4
.9
.9
.5
.3
.2
.9%
GFDL-D
kg/ha
4513
4519
4439
4311
4687
4365
4439
4459
4492
4459
4526
4358
4513
4284
4176
4264
4297
4223
4143
4393
137
Diff. Uncrty,
%
5
14
22
6
18
17
10
21
19
13
12
10
12
10
9
25
4
25
22
14
.7%
.1%
.7%
.5%
.7%
.6%
.7%
.0%
.7%
.3%
.0%
.0%
.8%
.2%
.7%
.5%
.4%
.3%
.2%
.9%
%
2
2
3
2
3
2
2
2
3
3
2
2
3
2
2
2
3
4
3
2
.5
.1
.5
.0
.5
.4
.7
.7
.3
.9
.5
.6
.2
.4
.5
.8
.1
.2
.0
.9%
2-42
-------
Peart
Table 12. CERES-MAIZE Simulation, 30-yr Averages, Univ. of Florida
Non-Modified for C02 Effect on P-Syn., 3-18-88
BASE : GISS : GDFL
RAINFED Yld, : Yld, Diff..Uncrty, : Yld, Diff, Uncrty,
Location Kg/ha Kg/ha Pet. + .-,% Kg/ha Pet. +.-,%
AT.GA
BI.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB,AL
MC.GA
ME, MS
MG.AL
MP.TN
NO.VA
NS.TN
RA.NC
SH.LA
TL.FL
WI.NC
AVE. :
IRRIGATED
Location
AT.GA
B1.AL
BR.LA
CH.NC
CL.SC
LB.VA
LR.AR
LU.KY
MB.AL
MC.GA
ME, MS
MG.AL
MP.TN
NO.VA
NS.TN
RA.NC
SH.LA
TL.FL
WI.NC
13236
12609
3206
8906
8718
10196
6593
9706
10390
8459
8358
7962
8149
10390
8610
9597
5281
12033
9022
BASE
Yld,
Kg/ha
15210
13351
14627
14591
13777
15880
14115
15758
14122
13582
13532
13697
13892
14915
14202
14461
14396
14007
11874
9951
2940
8855
7602
9864
6999
8877
8135
8264
7040
7803
7544
9561
8142
9561
6326
10304
8313
.
i Yld,
Kg/ha
12761
10866
12307
12674
11211
13560
11384
12544
11507
11024
10794
10858
11327
11644
12141
12271
11183
11175
-10.3%
-21.1%
-8.4%
-.5%
-12.8%
-3.3%
+6.2%
-8.6%
-21.7%
-2.4%
-15.8%
-2.0%
-7.4%
-7.9%
-5.5%
-.4%
+19.7%
-14.4%
-7.9%
5.2%
4.3%
24.2%
10.6%
11.5%
11.1%
14.0%
11.5%
8.1%
9.5%
10.2%
9.5%
13.0%
8.8%
11.1%
9.7%
18.8%
5.9%
10.9%
Diff. .Uncrty,
Pet.
-16.1%
-18.6%
-15.9%
-13.1%
-18.6%
-14.6%
-19.3%
-20.4%
-18.5%
-18.8%
-20.2%
-20.7%
-18.5%
-21.9%
-14.5%
-15.1%
-22.3%
-20.2%
+ ,-,%
3.5%
3.7%
3.1%
2.6%
3.4%
4.7%
2.9%
3.9%
3.5%
3.1%
3.4%
2.6%
3.3%
3.6%
3.3%
3.9%
2.2%
3.2%
5714
6622
1578
2133
2356
3062
2313
6845
5548
1491
2543
1477
2990
3278
5181
2313
2147
6939
3585
1
: Yld,
Kg/ha
10239
10714
11312
10376
9115
11067
10938
11010
10613
8790
10779
8762
11082
9670
10642
10275
10938
10440
-56.8%
-47.5%
-50.7%
-76.1%
-73.0%
-70.0%
-64.9%
-29.5%
-46.6%
-82.4%
-69.6%
-81.4%
-63.4%
-68.4%
-39.9%
-75.9%
-59.4%
-42.3%
-64.6%
GDFL
4.7%
4.7%
21.5%
8.7%
10.3%
10.9%
11.1%
8.3%
8.2%
9.8%
8.0%
11.7%
8.5%
10.9%
8.3%
14.3%
6.4%
9.8%
Diff, Uncrty,
Pet.
-32.7%
-19.8%
-22.7%
-28.9%
-33.8%
-30.3%
-22.5%
-30.1%
-24.9%
-35.2%
-20.3%
-36.1%
-20.2%
-35.1%
-25.1%
-29.0%
-24.0%
-25.5%
+ ,-,%
3.2%
3.9%
3.2%
2.3%
3.1%
4.5%
3.0%
3.5%
3.5%
3.1%
3.4%
2.6%
3.3%
3.2%
3.4%
3.8%
2.2%
3.1%
AVE: 14340 11735 -18.2% 3.3% 10376 -27.6% 3.2%
2-43
-------
Peart
For irrigated soybeans, under GISS, about half the locations had significant increases in yield, while the rest
had insignificant changes compared to the base case. The average increase in yield for all locations was 13.7%.
For the GFDL weather, results were about the same, with an average increase of 14.9%.
Maize Simulations for Combined Climate and Direct Effects. Time constraints prevented running Maize
for all locations, but rainfed results from four locations, Charlotte, NC, Macon, GA, Meridian, MS, and
Memphis, TN, are shown in Table 13. Modification of MAIZE for carbon dioxide enrichment did not have as
great an effect on yields because the physiological effects are less, and this is seen in Figure 15, which shows
results for modified and nonmodified MAIZE for both GISS and GFDL weather scenarios. Briefly, under GISS
weather, insignificant yield changes occurred, but under GFDL weather, drastic yield reductions occurred in the
range of 75%.
WATER-USE RESULTS
Soybean Water-Use Results for Combined Climate and Direct Effects. Since the modification for carbon
dioxide enrichment affected transpiration of water from the plant leaves to the air, we are interested in how the
combined effect and climate-alone models compared in water use efficiencies. Figure 17 shows these results for
three rainfed locations for SOYGRO. We defined water-use efficiency as the seed yield divided by the total
evapotranspiration during the growing season. Study of the SOYGRO yield results in Figure 16 along with the
water-use efficiencies shown in Figure 17 show a very close relationship of water-use efficiency to yield. Allen
et al. (1985) and Jones et al. (1985a) pointed out that changes in WUE were strongly related to changes in
photosynthetic rates and only weakly related to changes in transpiration rates under various CO2 treatments at
constant temperatures. However, Jones et al. (1985b) showed that temperature increases from 28°C to 35°C
with other factors constant reduced soybean WUE about 26%. Allen et al. (1985) and Jones et al. (1985a) also
showed that increasing crop leaf area index from 33 to 6.0 decreased WUE about 17%. Evapotranspiration is
a function of air temperature and relative humidity, crop leaf area, soil water potential, and leaf water potential
among other factors, so the weather, as well as the crop, has a strong effect on it.
Comparing water-use efficiencies for the combined-effect SOYGRO to those of the climate-alone model,
the direct effect significantly increases water-use efficiency. This is because it increases yields somewhat and
reduces evapotranspiration somewhat. The GFDL scenario reduces yields drastically, and therefore also reduces
water-use efficiency in the same way.
Maize Water-Use Results for Combined Climate and Direct Effects. Figures 18 and 19 show yields and
water-use efficiencies, respectively, for MAIZE and the several scenarios. Since yields change less between the
modified and nonmodified model, water-use efficiencies are not affected much by the carbon dioxide direct
effect. The GFDL scenario shows drastic yield and water-use efficiency reductions, whereas the GISS weather
shows little change from the base weather.
Irrigation Requirements - Soybeans. As a basis for an estimate of the potential increase in the demand on
water resources in the southeastern region, a summary of average rainfall, evapotranspiration, and irrigation
amount was extracted from the simulation runs. The numbers represent averages of 19 locations x 30 years of
weather data. For the base (standard) weather case, the average rainfall was 508 mm, average ET 590 mm,
average irrigation requirement 224 mm. For runs made using the climate scenarios incorporating the combined
climatic effect and the direct CO2 effects on the plant, the GISS scenario would increase the potential irrigation
demand on the average 33%, while the GFDL scenario would increase the demand by 133%.
2-44
-------
Peart
Table 13. CERES Maize, Combined Climate and Direct Effects on Yields
Charlotte, Hacon, Meridian and Memphis
Loc. BASE GISS-D Diff, % Uncrty, GFDL-D Diff., Uncrty,
kg/ha kg/ha % % kg/ha % %
RAINFED
CH.NC 9020 8995 -.3% 10.7% 2172 -75.9% 8.8%
MC.GA 8580 8386 -2.3% 9.6% 1488 -82.7% 8.3%
ME,MS 8455 7156 -15.4% 10.4% 2599 -69.3% 9.9%
MP.TN 8254 8009 -3.0% 13.1% 2636 -68.1% 10.7%
IRRIGATED
CH.NC 14845 12811 -13.7% 2.5% 10558 -28.9% 2.3%
MC.GA 13344 10834 -18.8% 3.1% 8624 -35.4% 3.1%
ME,MS 13778 10991 -20.2% 3.5% 10953 -20.5% 3.5%
MP.TN 14242 11543 -19.0% 3.3% 11317 -20.5% 3.2%
2-45
-------
Peart
SOYGRG,CHARLOIIE,COLUHBIMEHPHIS,mU)
-D: DIR.EFF,, -N: NO
CL HP
WEATHER SCENARIO
BASE QGISS-D 0GISS-N 0GFDL-D 0GFDL-N
Figure 16. Rainfed yields for soybeans for Charlotte, NC, Columbia, SC, and Memphis, TN based runs with
SOYGRO modified to include both climatic and direct CO2 effects.
2-46
-------
Peart
^
UJ
OJ
UJ
h-
CE
SOVGRO-CHARt,.COiUH,.HEMS, HATER USE
RfllNFED, -i/r'DIR.EFF,, -H = IWWR,Eff
CHARLOTTE
COLUHBIA
HEHPHIS
QGISS-D SGISS-N H.GFDL-D HGFDL-N
Figure 17. Water used for soybeans for the same three locations and data used in Figure 16.
2-47
-------
Peart
CERES HAIZE, RAINFED 4 YIELDS
-D : DIRECT EFFECTS, -H - NO DIR, EFF.
16808 -1
7000-
6000-
5000-
HC ME
LOCATION
0BASE QGISS-D 0GISS-M BSCFDL-D 0GFDL-N
Figure 18. Rainfed yields for maize for CH-Charlotte, NC; MC-Macon, GA; ME-Meridan, MS; and MP-
Memphis, TN.
2-48
-------
28 n
Peart
, RAINFED, 30-VR, AUE,
-D = DIRECT EFFECTS, -N = NO DIR, EFF,
15 H
1H
o
_J
UJ
M
>-
V
z£3
|V/N>
m
II
II
?§
II
^1
HC HE
LOCATION
HGISS-D HGISS-N
Figure 19. Water efficiency for maize.
2-49
-------
Peart
CHAPTER 4
IMPLICATIONS OF RESULTS
ENVIRONMENTAL IMPLICATIONS
Two important environmental implications are noticeable with only a cursory look at results. The first is the
greatly increased demand for irrigation likely to be caused by these changes over time. This demand has
implications both for water resource availability and water quality. The second is the likely increase in soil
erosion caused by the increased rainfall during some periods.
Even in the most drastic case shown by the GFDL weather data, the crop failure results with rainfed
soybeans can be converted to increased yields with the addition of irrigation. With these changes occurring slowly
over many years, farmers are likely to slowly increase irrigation, as, indeed, they are currently doing in the
Southeast, as insurance against the crop failures. Gradually, the irrigation water demand for previously installed
systems would increase, and concurrently, more systems would be installed.
SOCIOECONOMIC IMPLICATIONS
The changes shown in this report would have very drastic and important socioeconomic implications.
Soybeans could be dropped as an economic crop over wide areas of the Southeast, including at least the southern
part of the important Mississippi delta area Corn would be affected less, but it is less important in the region
than is the soybean. These suggested implications are based on the assumption of current prices and costs and
assume no shifting of cultivars or changes in other management practices. Further research is required to
evaluate the effects of changes in management as climate changes occur.
Other socioeconomic implications include (1) competition for water resources between the potential
increased agricultural use and nonagricultural demands, (2) increased fertilizer use could increase if other high-
value, nonlegume crops replaced soybeans, (3) effect of changing growing seasons on the supply of inputs such
as fertilizers, labor, etc. Many of these implications will become defined clearly based on the work of Dr. Rich
Adams, Washington State University, in a companion project.
POLICY IMPLICATIONS
For agricultural production program policies, implications depend on effects on other important crops of the
region, notably cotton and peanuts. A combination of increased soybean irrigation by some farmers and reduced
acreage by others might maintain overall total production while seriously reducing income of those forced by the
weather to drop soybean production. If peanut yield were adversely affected, the production control and price
support program might be unnecessary. Similarly, cotton might also be a candidate for deregulation if the new
weather caused cotton yield reductions.
Irrigation water demand seems very likely to increase substantially over time, and state and regional water
policies will need modification.
2-50
-------
Peart
REFERENCES
Acock, B. and L. H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations, pp. 53-97. In:
B. R. Strain and J. D. Cure, (eds.) Direct Effects of Carbon Dioxide on Vegetation. DOE/ER-0238, US. Dept.
of Energy, Carbon Dioxide Research Division, Washington, DC.
Acock, B., D. A. Charles-Edwards, D. J. Fritter, D. W. Hand, L. J. Ludwig, J. Warren-Wilson, and A. C. Withers.
1978. The contribution of leaves from different levels within a tomato canopy to canopy photosynthesis: An
experimental examination of two canopy models. J. Exp. Bot. 29:815-827.
Acock, B., V. R. Reddy, H. F. Hodges, D. N. Baker, and J. M. McKinion. 1985. Photosynthetic Response of
Soybean Canopies to Full-Season Carbon Dioxide Enrichment. Agronomy Journal 77:942-947.
Acock, B., V. R. Reddy, F. D. Whistler, D. N. Baker, J. M. McKinion, H. F. Hodges, and K. J. Boote. 1983.
Response of Vegetation to Carbon Dioxide, Series Number 002, the soybean crop simulator GLYCIM: Model
Documentation 1982. Crop Simulation Research Unit, USDA, ARS and Dept of Agronomy, Mississippi State
University. Joint Program of the US. Dept. of Energy, Carbon Dioxide Research Division, and US. Dept. of
Agriculture, Agricultural Research Service, Washington, DC. 316 pp.
Adams, R. M., et al., 1988. (Chapter in this report) The economic effects of climate change on US. agriculture.
Allen, L. H., Jr. 1986. Plant responses to rising CO,. Proceedings, 79th Annual Meeting of the Air Pollution
Control Association. Minneapolis, MN. #86-93. 33 pp.
Allen, L. H., Jr. 1979. Potentials for Carbon Dioxide Enrichment. In: Modification of the Aerial Environment
of Crops, B J. Barfield and J.F. Gerber (eds.), Monograph No. 2, American Society of Agricultural Engineers,
pp. 500-519. St. Joseph, Michigan.
Allen, L. H., Jr., K. J. Boote, J. W. Jones, P. H. Jones, R. R. Valle, B. Acock, H. H. Rogers, and R. C. Dahlman.
1987. Response of Vegetation To Rising Carbon Dioxide: Photosynthesis, Biomass and Seed Yield of Soybean.
Global Biogeochemical Cycles, 1:1-14.
Allen, L. H., Jr., P. Jones, and J. W. Jones. 1985. Rising Atmospheric CO2 and Evapotranspiration. National
Conference on Advances in Evapotranspiration, American Society of Agricultural Engineers, December 16-17,
1985. Chicago, Illinois, pp. 13-27
Allen L. H., Jr., J. S. Rogers, and E. H. Stewart. 1978. Evapotranspiration as a benchmark for turfgrass
irrigation. Proc. Annual Florida Turf-grass Management Conference. 26:85-97.
Baker, J. T., L. H. Allen, Jr., K. J. Boote, P. Jones, and J. W. Jones. 1989. Response of soybean to air
temperature and CO2 concentration. Crop Sci. (accepted).
Bisbal, Evelin C. 1987. Effects of subambient and superambient carbon dioxide levels on growth, development
and total nonstructural carbohydrate of soybean. M. S. Thesis, Agronomy Dept., Univ. of Florida, Gainesville,
FL.
*
Boote, K. J., J. W. Jones, and G. Hoogenboom. 1988. Sensitivity Analysis of SOYGRO to environmental factors.
Proceedings of Crop Simulation Workshop. March 1-3, University of Florida, Gainesville, FL. p 21.
Cure, J. D. 1985. Carbon Dioxide Doubling Responses: A Crop Survey. In B. R. Strain and J. D. Cure (ed.),
Direct Effects of Increasing Carbon Dioxide on Vegetation. US. Department of Energy, Carbon Dioxide
Research Division, DOE/ER-0238, Washington, DC. pp. 99-116.
2-51
-------
Peart
Cure, J. D. and B. Acock. 1986. Crop responses to carbon dioxide doubling: A literature survey. Agr. and
Forestry Metior. 38:127-145.
Cure, J. D., R. P. Patterson, C. D. Raper, Jr., and W. A. Jackson. 1982. Assimilate distribution in soybeans as
affected by photoperiod during seed development. Crop Sci. 22:1245-1250.
Egli, D. B. and I. F. Wardlaw. 1980. Temperature response of seed growth characteristics of soybeans. Agron.
J. 72:560-564.
Elwell, D. L., R. B. Curry, and M. E. Keener. 1987. Determination of Potential Yield-limiting Factors of
Soybeans Using SOYMOD/OARDC. Agricultural Systems 24:221-241
Fisher, J. E. 1963. The effects of short days on fruit set as distinct from flower formation in soybeans. J. Bot.
41:871-873.
France, J. and J. H. M. Thornby. 1984. Mathematical Models in Agriculture. Butterworths. Boston.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lads, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
three-dimensional global models for climate studies: Models I and II. Mon. Wea. Rev. Ill: 609-662.
Hansen, J., I. Fung, A. Lacis, S. Lebedeff, D. Rind, R. Ruedy, G. Russell, and P. Stone. 1988. Global climate
changes as forecast by the GISS 3-d model. In Press. J. Geophvs. Res..
Hesketh, J. D., D. L. Myhre, and C. R. Willey. 1973. Temperature control of time intervals between vegetative
and reproductive events in soybeans. Crop Sci. 13:250-254.
Hively, William. 1988. Science Observer. Global Change. American Scientist, 76:127-130.
Hofstra, G., and J. D. Hesketh. 1975. The effects of temperature and CO2 enrichment on photosynthesis in
soybean. In: R. Marcelle (ed.) Environmental and Biological Control of Photosynthesis. W. Junk. The Hague,
Netherlands, pp. 71-80.
Hodges, T., V. French, and S. LeDuc. 1985. Yield Model Development: Estimating Solar Radiation for Plant
Simulation Models. AgRISTARS YM-15-00403.
IBSNAT. 1986. Technical Report 5, Decision Support System for Agrotechnology Transfer (DSSAT).
Documentation for IBSNAT Crop Model Input and Output Files, Version 1.0. Dept. Agronomy and Soil Sci.,
College of Trop. Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Idso, S. B. 1982. Non-water-stressed baselines: A key to measuring and interpreting plant water stress.
Agricultural Meteorol. 27:59.
Idso, S. B., B. A. Kimball, M. G. Anderson, and J. R. Mauney. 1987. Effects of Atmospheric CO2 Enrichment
on Plant Growth: The Interactive Role of Air Temperature. Agriculture, Ecosystems and Environment (In
Press).
Ingram, K. T., D. C. Herzog, K. J. Boote, J. W. Jones, and C. S. Barfield. 1981. Effects of defoliating pests on
soybean canopy CO2 exchange and reproductive growth. Crop Sci. 21:961-968.
Jagtap, S. S. 1986. Theoretical development, analysis, and experimental validation of an evapotranspiration
model for developing crops. Ph.D. Dissertation. Agric. Engineering Department, University of Florida,
Gainesville, PL. 239pp.
Jones, C. A. and J. R. Kiniry (eds.). 1986. CERES-Maize: A simulation model of maize growth and
development. Texas A&M University Press. College Station, TX.
2-52
-------
Peart
Jones, J. W., K. J. Boote, S. S. Jagtap, G. Hoogenboom, and G. G. Wilkerson. 1988a SOYGRO V5.41: Soybean
crop growth simulation model. User's Guide. Florida Agr. Exp. Sta. Journal No. 8304, IFAS. University of
Florida. Gainesville. S3 pp.
Jones, J. W., K. J. Boote, S. S. Jagtap, and J. W. Mishoe. 1988b. Soybean Development. Chapter 5 In Modeling
Plant and Soil Systems. R. J. Hanks and J. T. Ritchie (eds.). American Society of Agronomy, Madison, WI.
(In Press).
Jones, P., L. H. Allen, Jr., J. W. Jones, K. J. Boote, and W. J. Campbell. 1984. Soybean Canopy Growth,
Photosynthesis, and Transpiration Responses to Whole-Season Carbon Dioxide Enrichment. Agronomy Journal,
76:633-637.
Jones, P., L. H. Allen, Jr., and J. W. Jones. 1985. Responses of Soybean Canopy Photosynthesis and
Transpiration to Whole-Day Temperature Changes in Different CO, Environments. Reprinted from Agronomy
Journal, 77:242-249. March-April.
Jones, P., J. W. Jones, and L. H. Allen, Jr. 1985. Seasonal Carbon and Water Balances of Soybeans Growth
Under Stress Treatments in Sunlit Chambers. American Society of Agricultural Engineers, 28(6):2021-2028.
Jones, P., J. W. Jones, and L. H. Allen, Jr. 1985. Carbon Dioxide Effects on Photosynthesis and Transpiration
During Vegetative Growth in Soybeans. Reprinted from Soil and Crop Science Society of Florida Proceedings,
44:129-134.
Jones, P., J. W. Jones, and L. H. Allen, Jr. 1985. Seasonal Carbon and Water Balances of Soybeans Growth
Under Stress Treatments in Sunlit Chambers. American Society of Agricultural Engineers, 28(6):2021-2028.
Jones, P., L. H. Allen, Jr., J. W. Jones, and R. Valle. 1985. Photosynthesis and Transpiration Responses of
Soybean Canopies. Agronomy Journal, 77:119-126, January-February.
Kimball, B. A. 1983. Carbon dioxide and agricultural yield: An assemblage and analysis of 430 prior
observations. Agron. J. 75:779-788.
Manabe, S. and R. T. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of the Atmospheric Sciences, 44:1211-1235.
McCree, K. J. 1974. Equation for the rate of dark respiration of white clover and grain sorghum as functions
of dry weight, photosynthetic rate, and temperature. Crop Sci. 14:509-514.
Meyer, G. E., R. B. Curry, J. G. Streeter, and H. J. Mederski. 1979. SOYMOD/OARDC: Dynamic simulator
of soybean growth, development, and seed/yield. OARDC Research Bulletin 1113, Ohio Agricultural Research
and Development Center, Wooster, Ohio. 36 pp.
Parker, M. W. and H. A. Borthwick. 1943. Influence of temperature on photoperiodic reactions in leaf blades
of Biloxi soybeans. Bot. Gaz. 104:612-619.
Priestly, C. H. B. and R. J. Taylor. 1972. On the assessment of surface heat and evaporation using large scale
parameters. Monthly Weather Review 100:81-92.
Richardson, C. W. 1985. Weather simulation for crop management models. Transactions of the ASAE, 28:1602-
1606.
Richardson, C. W. and D. A. Wright. 1984. WGEN: A model for generating daily weather variables. VS.
Department of Agriculture, Agricultural Research Service, ARS-8,83 p.
2-53
-------
Peart
Ritchie, J. T. 1985. A user-oriented model of the soil water balance in wheat, p. 293-305. In,: Wheat Growth
and Modeling, E. Fry and T. K. Atkin (eds.). Plenum Publishing Corporation, NATOASI Series.
Rogers, H. H., G. E. Bingham, J. D. Cure, J. M. Smith, and K. A. Surano. 1983. Responses of selected plant
species to elevated carbon dioxide in the field. J. Environ. Oual. 12:569.
Schlesinger, M. E., and J. F. B. Mitchell. 1985. Model projections of the equilibrium climatic response to
increased carbon dioxide, pp. 81-147. In: M. C. MacCracken and F. M. Luthur, (eds.) Projecting the Climatic
Effects of Increasing Carbon Dioxide. DOE/ER-0237, VS. Dept. of Energy, Carbon Dioxide Research Division,
Washington, DC.
Stewart, R. B. 1986. Climatic change - implications for the prairies. Paper presented at the Royal Society for
Canada Symposium, June 2-4, Winnipeg, Manitoba.
Stewart, R. B. 1986. Climatic Change: Implications for the Prairies. Transactions of the Royal Society of
Canada, Series V, Volume I.
Thomas, J. F. and C. D. Raper, Jr. 1976. Photoperiodic control of seed filling for soybeans. Crop Sci. 16:667-
672.
Thomas, J. F. and C. D. Raper, Jr. 1978. Effect of day and night temperatures during floral induction on
morphology of soybeans. Agron. J. 70:893-898.
Thomas, J. F. and C. D. Raper, Jr. 1981. Day and night temperature influence on carpel initiation and growth
in soybeans. Bot. Gaz. 142:183-187.
Valle, R., J. W. Mishoe, W. J. Campbell, J. W. Jones, and L. H. Allen, Jr. 1985a. Phojtosynthetic responses of
"Bragg" soybean leaves adapted to different CO2 environments. Crop Sci. 25:333-339.
Valle, R., J. W. Mishoe, J. W. Jones, and L. H. Allen, Jr. 1985b. Transpiration rate and water use efficiency
of soybean leaves adapted to different CO2 environments. Crop Sci. 25:477-482.
van Keulen, et al. 1981. Physiological aspects of increased CO2 concentration. Experientia 36: 786-792.
Waggoner, Paul E. 1983. Agriculture and a climate changed by more carbon dioxide. Chap. 6, in Changing
Climate. A report of the carbon dioxide assessment committee. National Academy Press, Washington, DC.
Wilkerson, G. G., J. W. Jones, K. J. Boote, K. T. Ingram, and J. W. Mishoe. 1983a. Modeling soybean growth
for management. Trans. ASAE 26:63-73.
Wilkerson, G. G., J. W. Jones, K. J. Boote, and J. W. Mishoe. 1985. SOYGRO V5.0: Soybean Crop Growth
and Yield Model. Technical Documentation. University of Florida, Gainesville, FL 32611. 253 pp.
2-54
-------
POTENTIAL EFFECTS OF CLIMATE CHANGE ON
AGRICULTURAL PRODUCTION IN THE GREAT PLAINS
A SIMULATION STUDY
by
Cynthia Rosenzweig
Department of Geography
Columbia University
NASA/Goddard Space Flight Center
Institute for Space Studies
New York, NY 10025
IAG No. DW80932629-01-1
-------
CONTENTS
Page
ACKNOWLEDGMENTS ui
FINDINGS 3-1
CHAPTER I: INTRODUCTION 3-2
Description of the Agricultural System 3-2
Literature Review 3-3
Organization of This Report 3-4
CHAPTER 2: METHODS 3-5
Crop Models 3-5
Modifications of the CERES Models for CO2 Enrichment 3-6
Limitations Resulting from the Crop Models 3-7
Climate Change Scenarios 3-9
Limitations Resulting from the Climate Scenarios 3-14
Climate Data 3-14
Soils 3-14
Management Variables 3-14
Simulations 3-15
CHAPTER 3: RESULTS AND DISCUSSION 3-18
Climate Change Alone 3-18
Combined Climatic and Direct Effects of CO2 3-23
Interpretation of Results 3-29
CHAPTER 4: IMPLICATIONS OF RESULTS 3-31
Environmental Implications 3-31
Socioeconomic Implications 3-31
Further Research 3-31
REFERENCES 3-33
APPENDICES 3-35
-------
ACKNOWLEDGMENTS
I thank Drs. J.T. Ritchie and J.W. Jones and their colleagues at Michigan State University and the
University of Florida, Institute of Food and Agricultural Sciences, for their collaboration in this work, Dr. David
Rind for helpful advice throughout the project, Rich Goldberg for the programming, and Christopher Shashkin
for word-processing and graphics. I am also grateful to Dr. Timothy Carter for his careful reading of the
manuscript and to several anonymous reviewers for useful suggestions.
111
-------
Rosenzweig
FINDINGS1
This study is a first step in linking models of climate change to models of crop growth. The results should
be regarded as indications of the sensitivity of wheat and corn production in the central and southern Great
Plains to projected climatic changes, rather than as predictions. The uncertainties in the study lie primarily in
the following areas:
The global climate models were not designed for regional studies and their results are often on too large
a spatial scale to project effects on crop production realistically. The crop models, while among the best now
available for large-area studies, are semi-empirical, and may not provide accurate estimates for the extreme
climatic conditions implied in the scenarios. The physiological (or "direct") effects of CO, on crop growth and
water use are only approximated in the crop models. Accurate prediction of these effects awaits further,
specially designed experiments and continuing model development. The range of possible alternative cropping
strategies was not fully explored, such as substitution of different crop species and double-cropping.
The following specific results are found in the modeling study:
1. Projected climate changes cause simulated wheat and corn yields to decrease in the southern and central
Great Plains. Decreases in modeled grain yields are caused primarily by increases in temperature which
shorten the duration of crop life cycle, thus curtailing the production of harvestable biomass.
2. When the direct effects of increased concentrations of CO2 on crop growth and water use are combined with
the effects of climate change in simulations, an enhancement of modeled crop yields compensated for the
negative effects of climate change in some cases, but not in others. The more severe the climate change
scenario, the less compensation the physiological effects of CO2 provide, especially in dryland simulations.
3. In climate change simulations, the amount of water needed for automatic irrigation increases in areas where
precipitation decreases and irrigated yields are higher and less variable year-to-year compared to dryland
yields. These results suggest a potential for increased demand for irrigation in the region.
4. Adjusting the planting dates of wheat and com does not significantly ameliorate the effects of one climate
change scenario on modeled yields. Changing wheat cultivars to ones with lower vernalization requirements,
lower photoperiod sensitivity, and longer grain-filling periods, in addition to delaying planting dates,
overcomes yield decreases at some sites, but not at others.
'Although the research in this report has been funded wholly or partly by the U.S. Environmental
Protection Agency under IAG #DW80932629-01-1, it does not necessarily reflect the Agency's views, and no
official endorsement should be inferred from it
3-1
-------
Rosenzweig
CHAPTER I
INTRODUCTION
The objective of this study is to characterize the direction, magnitude, and uncertainty of potential climate
change-induced alterations in wheat and corn yield in the central and southern Great Plains region. The climate
change is that predicted to arise as a result of increasing carbon dioxide (CO2) and other radiatively active trace
gases in the earth's atmosphere. Climate change simulation experiments are designed for both dryland and
irrigated conditions in order to estimate relative changes in crop yield, evapotranspiration, water applied for
irrigation, and duration of crop growth. Potential production management adjustments to climate change, such
as farmer responses to changes in length of growing season and climate regime, are included in some model runs
as shifts in planting date and substitution of more climatically appropriate cultivars.
The effects of increasing atmospheric CO, on photosynthesis and transpiration are also approximated in
some model runs, based on results from published reports of controlled environment experiments. The
simulations involving climate change effects alone provide an "extreme case" scenario, while the simulations with
combined climatic and physiological effects of CO2 represent a more moderate impact.
Description of the Agricultural System
There are nearly 100,000 farms in Nebraska, Kansas, Oklahoma and Texas, occupying over 111 million acres.
Farmers in these states grow a third of the nation's wheat and a seventh of the nation's corn, primarily on deep
prairie soils. The importance of this abundant grain crop production to both the UJS. and international grain
supply justifies an analysis of the potential effects of climate change on agriculture in these states.
National attention has also recently focused on the water resources in the region, especially on overuse of
the Ogallala Aquifer (High Plains Associates, 1982). An evaluation of how changes in demand for water for
irrigation could exacerbate or alleviate water scarcity in the areas fed by this and other aquifers will be useful
to federal, state, and local decision-makers responsible for acceptable water supply and quality.
Although irrigation is important in certain areas, agriculture in the Great Plains region is primarily dryland
farming, i.e., without irrigation. This causes agriculture to be vulnerable to climatic stresses, particularly to
recurring drought episodes, such as the severe droughts of the 1930s (Worster, 1979; Hurt, 1981). In the Dust
Bowl period, crop failure and economic depression led to farm abandonment and to migration away from the
region.
If global climate change brings increased frequency of high temperature extremes and droughts in the Great
Plains, regional agriculture could again be negatively affected. Farm acreage and field size have both expanded
recently in the region. Despite the adoption of conservation tillage techniques, drought-resistant cultivars, and
risk management programs, some analysts argue that the region remains particularly vulnerable to climate-
induced reductions in crop yields and may be one of the first UJS. agricultural regions to exhibit impacts of
climate change (Warrick, 1984). Some global climate models project pronounced reductions in soil moisture in
mid-continental areas in summer, a prediction which implies potentially severe impacts on dryland farming and
increased demand for irrigation in the Great Plains.
Literature Review
There have been several systematic modeling studies of climate change impacts on agriculture in the Great
Plains. Warrick (1984), in a historical approach, analyzed the vulnerability of the region to a possible recurrence
of the 1930s drought by running a dryland crop yield statistical model tuned to 1975 technology with 1934 and
1936 temperature and precipitation conditions. He found the recurrence of 1930s conditions in the region would
result in wheat yield reductions of over 50%. Others have used an agroclimatic zone approach (Rosenzweig,
3-2
-------
Rosenzweig
1985) or a basic parametric crop yield model (Terjung et al., 1984; Liverman et al., 1986) to study potential
changes in crop location, yield, evapotranspiration, and irrigation requirements with climate change estimates
alone.
Terjung et al. (1984) used a crop water demand and yield model to investigate irrigated corn production
sensitivity to differing temperature, precipitation, and solar radiation fluctuations. They found that in the central
Great Plains, evapotranspiration and total water applied for irrigation were very sensitive to climate variations.
Liverman et al. (1986) continued this modeling and found that the lowest irrigated yields occurred under cloudy,
hot, and very dry climate scenarios. Under dryland cropping, minimum yields occurred under sunny-hot and
sunny-warm scenarios with very dry conditions.
Using an agroclimatic approach, Rosenzweig (1985) found that lack of cold winter temperatures in the
southern Great Plains may necessitate a change from winter to spring wheat cultivars with climate change
projected for a doubling of CO,. Changes in temperature, precipitation, and solar radiation were considered.
The study found that decreasea water availability may also increase demand for irrigation.
Few studies simultaneously consider both the climatic and physiological effects of increased COL.
Experiments in field chambers and controlled environments have shown that increased atmospheric CC^
concentration increases photosynthesis and yield and improves water-use efficiency (Kimball, 1983; Acock and
Allen, 1985; Cure, 1985). Kimball (1983) estimated an increase in crop yields due to a doubling of carbon
dioxide of about 33% +/- 6%. However, the relative effects of climate changes, particularly the increased
temperatures predicted by global climate models (GCMs), and physiological changes on crop production in the
field are still very much in question.
Robertson et al. (1987) estimated the impact of climate change on yields and erosion using the Erosion
Productivity Impact Calculator (EPIC) in Bell County, Texas, with increased energy/biomass conversion
efficiencies to estimate the effects of doubling current levels of carbon dioxide on plant growth and yields.
Results showed that modeled wheat yields in Texas decreased and modeled corn yields increased only marginally
owing to moisture stress.
The CERES-Wheat model has been used to estimate yield changes with combined CO2 and climate effects
for the southern Great Plains (Rosenzweig, 1987). In this study, the direct effects of elevated CO2 (increased
photosynthesis and improved water use) compensated for the negative effects of climate change (temperature,
precipitation, and solar radiation changes) in years with adequate rainfall, but did not reduce crop failures in dry
years. The CERES-Maize model has been used to project an increase in corn production in Illinois with the
physiological effects of COy but not the climate effects (Decker and Achutuni, 1987).
Potential effects of predicted climate change (Williams et al., 1988) and combined climatic and physiological
CO, effects (Stewart, 1986) on spring wheat production in Saskatchewan, Canada, have been estimated. The
results of the Williams et al. (1988) study, which used both historic and climate change scenarios, agroclimatic
indices, and a crop growth model, suggest that a shift to a warmer long-term climate, even if precipitation
increases, would reduce spring wheat yields, decrease wind erosion potential, enhance average potential biomass
productivity, and increase frequency and severity of droughts. Stewart (1986), with similar methodology, found
that spring wheat production in Saskatchewan would fall with climate change scenarios, both with and without
the direct effects of COy and that any decrease in precipitation from current levels would significantly reduce
yields and production.
3-3
-------
Rosenzweig
Organization of This Report
The remainder of this report consists of three sections. The next section describes the methods and
limitations of the analysis. The following section describes and interprets the results of the set of crop modeling
experiments with two GCM-derived scenarios of climate change induced by a doubling of CO, concentration.
The environmental and socioeconomic implications of these results are set forth in the final section, along with
some policy considerations and suggestions for future research. Tabulated site and soil characteristics, statistical
methods, and model results are included as appendices.
3-4
-------
Rosenzweig
CHAPTER 2
METHODS
Crop Models
Potential changes in crop production in the Great Plains were modeled with CERES-Wheat (Ritchie and
Otter, 1985) and CERES-Maize* (Jones and Kiniry, 1986). The CERES models were chosen because they
simulate crop responses to the major factors which affect crop yields, i.e., climate, soils, and management, and
because they have been widely validated. Management practices which may be varied in the models include
cultivar, planting date, plant population, row spacing, and sowing depth. The presence of these variables permits
experiments that simulate management adjustments by farmers to climate change.
The CERES models were developed with experimental data from many locations over a period of time.
They have been validated over a wide range of environments (Otter-Nacke et al., 1986) and are not specific to
any particular location or soil type. Thus they are suitable for use in a study in which the baseline (present-
day) climate ranges from semi-tropical conditions in southern Texas to mid-continental conditions in Nebraska.
The validation of the CERES crop models over different environments also serves to enhance predictive
capability for the climate change scenarios.
The CERES models were designed to predict the growth and yields of wheat and corn varieties in different
types of environments where the crops are generally grown. The models employ simplified functions to predict
crop growth and yield as influenced by plant genetics, weather (daily solar radiation, maximum and minimum
temperatures, and precipitation), soil, and management factors. Modeled processes include phenological
development, i.e., duration of growth stages, growth of vegetative and reproductive plant parts, extension growth
of leaves and stems, senescence (aging) of leaves, biomass production and partitioning among plant parts, and
root system dynamics. The CERES models also simulate the effects of soil-water deficit and nitrogen deficiency
on photosynthesis and pathways of carbohydrate movement in the plant. The nitrogen portions of the models
were not used in this study; thus nitrogen fertilizer is assumed to be nonlimiting.
Input variables are the daily solar radiation (MJ m"2 day'1), maximum and minimum air temperatures (°C),
and precipitation (mm day ). The user specifies the beginning day of the simulation, plant population (plants
m ), row spacing (m), depth of sowing (cm), and irrigation regime. Also needed are the latitude of the
production area, soil characteristics and initial conditions of the soil profile, and genetic coefficients of the crop
variety.
The soil characteristics are soil albedo, upper limit of Stage 1 soil evaporation (mm), soil-water drainage
constant, and the USDA Soil Conservation Service curve number, which is used to calculate runoff. For each
soil layer, parameters describe the lower limit of plant-extractable soil water (volume fraction), the drained
upper limit water content (volume fraction), the saturated water content (volume fraction), a weighting factor
for new root growth distribution, the bulk density, and the initial soil water content (volume fraction).
The genetic coefficients for CERES-Wheat relate to photoperiod sensitivity, duration of grain filling,
conversion of biomass to grain number and grain filling, vernalization, stem size, tillering habit, and cold
hardiness. For CERES-Maize, the genetic coefficients are the thermal time required from emergence to end
of juvenile stage, rate of photo-induction (degree-days per hour), thermal time required for grain filling, potential
kernel number, and maximum daily rate of kernel fill (mg per kernel).
CERES-Maize simulates the growth and development of Zea mays L., known in the United States as
corn.
3-5
-------
Rosenzweig
Modifications of the CERES Models for CO, Enrichment
A method was developed to give an approximation of the changes in photosynthesis and evapotranspiration
caused by a doubling of CO2 from 330 to 660 ppm (Peart et al., 1988). The approach was to compute ratios of
daily photosynthesis and evapotranspiration rates for a canopy exposed to elevated CO2 to those rates of the
same canopy exposed to ambient CO, conditions. These ratios were then applied to the model-calculated rates
of photosynthesis and evapotranspiration for current CO2 concentrations.
Photosynthesis. Experimental results were reviewed to obtain estimates of increases in canopy
photosynthesis for both corn and wheat. Plants with a C3 carbon fixation pathway are more responsive to
increases in atmospheric CO2 than plants with C4 pathways (Acock and Allen, 1985). Peart et al. (1988) con-
ducted an evaluation of corn (C4) canopy photosynthesis and selected a value of 15% for the increase in
instantaneous rate. This value was reduced by 33% to 10% for the daily integrated increase in canopy
photosynthesis, accounting for lower light intensities in the morning and evening. This value is consistent with
values of plant light-use efficiency at normal and high CO, concentrations given by Charles-Edwards (1982).
Wheat has a C3 carbon fixation pathway and appears to lie between soybean (35% as used by Peart et al., 1988)
and corn in its photosynthetic response to increases in atmospheric CO2 (Cure, 1985).
Therefore, a value of 25% was chosen for increase in daily canopy photosynthesis for wheat under doubled
CO2 conditions.
In the crop models, the photosynthesis rates for current CO2 concentrations for wheat and corn were thus
multiplied by factors of 1.25 and 1.10, respectively, to simulate conditions in the doubled CO2 environment. It
is assumed that the relative increase in photosynthesis was independent of other factors, so that no interactions
with water stress, temperature, and leaf area were included, other than those already in the models. Changes
in respiration were not taken into account.
Evapotranspiration. Increased CO2 concentration increases leaf stomatal resistance, resulting in lowered
transpiration rates per unit leaf area (Acock and Allen, 1985). However, this decrease in water use is offset by
the photosynthetically enhanced production of greater leaf area; thus the total canopy transpiration rate under
elevated CO, is higher than is accounted for by increased stomatal resistance alone.
In the CERES models, potential plant transpiration was changed under elevated CO2 conditions due to
increased stomatal closure and changes in the partitioning of energy captured by the canopy with increased leaf
area index (LAI). The Penman-Monteith equation was used to develop a ratio of transpiration under elevated
CO2 conditions to that under ambient conditions (Peart et al., 1988):
„ s Rn * cp f (Ps^a) ' Pa> Ea _ m
AE = ~ ~^^~~ ^ "~^~^~~ ~~~ '^^~~~^^~ \1)
s + 7(1 +
where AE is evapotranspiration rate in energy units, s is the slope of the saturated
vapor pressure-temperature curve, -7 is the psychometric constant, Rn is net radia-
tion, c is specific heat of the air at constant pressure, p is the density of air, (ps(TJ - pj
is the vapor pressure deficit of the air, ga is the boundary layer conductance between the canopy and the bulk
air, and g_ is the canopy conductance to water vapor.
To derive this method, Peart et al. applied the Penman-Monteith equation to the same canopy and
environment, except for differing CO2 concentrations. The only variable which changes in the two cases is the
canopy conductance to vapor transport. Therefore, a ratio of evapotranspiration rates under elevated and
ambient CO2 concentrations is obtained:
3-6
-------
Rosenzweig
AEC s + 7 (l+g./gj
RATIO = --- -2-=£ - (2)
AE s +
where gjj is the canopy conductance to water vapor under elevated CO2 conditions.
The canopy resistance is computed by
Rc "
where rL is the leaf stomatal resistance (s m"1), LAI is the leaf area index, and rb is the leaf boundary layer
resistance.
Then, the canopy conductances for ambient and elevated CO2 (g,. and g,.) are computed by:
(4)
To calculate the evapotranspiration ratio in CERES-WheaL stomatal resistance values for well-watered (0.48
and 0.63 s cm'1) and drought-stressed (0.78 and 0.75 s cm'1) winter wheat under 330 and 660 ppm from
Chaudhuri et al. (1986) were used for irrigated and dryland runs, respectively. These experimental results show
that elevated CO2 increased stomatal resistance of well-watered wheat plants, while a slight decrease in stomatal
resistance was observed under drought conditions. For corn, leaf resistance was computed as a function of CO2
concentration using the equation developed by Rogers et al. (1983).
Temperature, windspeed, LAI and CO, concentration are needed to calculate RATIO. Average daily
temperature is computed from maximum and minimum temperatures which are inputs to the CERES models;
windspeed is set at 2.0 m s . LAI is specified directly as calculated in models. The ratio procedure results in
a lower transpiration rate for higher CO2 levels on a daily basis, but may or may not change seasonal
evapotranspiration by the same proportion because of the increased LAI in increased CO2 conditions.
Differences in canopy temperature, canopy height, and leaf vapor pressure with increased CO2 are not taken into
account.
Limitations Resulting from the Crop Models
The CERES models contain many simple, empirically derived relationships. For example, the use of a
thermal time scale, (i.e., growing degree days) may not accurately represent the behavior of different crop
cultivars in different environments, and the absence of vapor pressure deficit in the evaporation equations means
that important advection effects are not considered. The photosynthesis equations do not explicitly include
maintenance respiration, which responds quite differently to temperature than photosynthesis. Some of the
assumptions of the modeling study are listed in Table 1.
Also, the relationships in the models may or may not hold under differing climatic conditions. Most of the
data used to derive the relationships in the crop models were obtained with temperatures below 35°C, whereas
the projected temperatures for doubled CO2 are often above 35 or even 40°C during the growing period. While
the models do simulate temperature effects on photosynthesis, leaf extension, vernalization, and winterkill, they
do not include a temperature effect on pollination and its viability. Technology and climatic tolerances of crop
cultivars are held constant, even though both are likely to adapt to changing climate.
3-7
-------
Rosenzweig
Table 1. Assumptions and Limitations of Crop Models
1. Thermal time scale for phenologjcal stages.
2. Absence of vapor pressure deficit in evaporation.
3. Lack of maintenance respiration in carbohydrate production.
4. Behavior of crops in high temperature not well defined.
5. Lack of temperature effect on pollination and pollen viability.
6. Beneficial physiological effects of CO2 may be overestimated.
7. Higher leaf temperatures due to increased CO2 not modeled.
8. Interactions of increased CO2 with high-temperature and water stress
not simulated.
9. Yields not limited by pests and lack of nutrients.
10. Technology and climatic tolerances of cultivars held constant.
3-8
-------
Rosenzweig
The modifications of the models for the direct effects of CO, engender uncertainties as well. In particular,
they do not consider the interactions of increased CO, with higher temperatures or water stress, which could
result in either higher or lower photosynthesis rates (Rose, 1988). Nor were changes in plant temperature due
to stomatal closure under high CO2 explicitly modeled. The Penman-Monteith model is not appropriate for
incomplete canopies.
Some other assumptions of the modeling study are that all nutrients are nonlimiting; weeds, diseases, and
insect pests are controlled; there are no problem soil conditions such as high salinity, acidity, or heavy
compaction; and there are no catastrophic weather events such as hail, tornadoes, floods, high winds, or heavy
storms. All these assumptions tend to bias simulated yields upwards. The CERES models do not simulate
windspeed, thereby ignoring changes in evapotranspiration driven by changes in windspeed. Technology and
climatic tolerances of crop cultivars are held constant, even though both are likely to adapt to changing climate.
The direct effects of CO2 in the crop modeling study may be overestimated for two reasons. First,
experimental results from controlled environments, used to derive the crop model simulation of increasing CO2
effects, may not replicate variable, windy, and pest-infested (e.g., weeds, insects, and diseases) field conditions.
Second, because other radiatively active trace gases besides CO2, such as methane (CH4), are also increasing,
the equivalent warming of a doubled CO2 climate will occur before actual doubling of atmospheric CO2. A level
of 660 ppm CO, concentration was assumed for the crop modeling experiments, while the CO, concentration
in 2060, when the equivalent warming of doubled CO, occurs in the GISS GCM transient run, is estimated to
be 555 ppm (Hansen et al., 1988).
Climate Change Scenarios
CERES-Wheat and CERES-Maize were run at 14 locations in Nebraska, Kansas, Oklahoma, and Texas with
baseline observed climate (1951-1980) and climate change scenarios developed on the basis of estimates from
the global climate models of the Goddard Institute for Space Studies (GISS) and Geophysical Fluid Dynamics
Laboratory (GFDL). Availability of daily climate data from 1951 to 1980 and geographical distribution
determined choice of locations. The climate models produce climate change results in distinct latitude by
longitude grids. Climate stations and the GISS and GFDL gridboxes in the central and southern Great Plains
are shown in Figure 1.
The climate change scenarios were developed from average monthly changes in temperature, precipitation,
and solar radiation calculated for each GCM gridbox for current and doubled CO, conditions (Smith and
Tirpak, 1988). Observed daily climate variables were multiplied by monthly ratios of climate variables from the
GCM doubled CO2 runs over the variables simulated for current conditions from the appropriate gridbox. No
interpolations were made between gridboxes. Seasonally and annually averaged temperature and precipitation
changes from the climate change scenarios are shown in Table 2 and Figure 2 for the study area.
The magnitudes of climate changes from the GFDL scenario and the climate of the 1930s drought in
Nebraska and Kansas are compared in Figure 3. While the climate change scenario decreases in precipitation
are about the same as those during the most severe drought years (1934 and 1936) in the area, the climate
change scenario temperatures are about 3°C higher than the Dust Bowl temperatures.
3-9
-------
Rosenzweig
31.30
Z3.48
Climolt
Study
6IS1
44.44
40.00
35.55
31.
26.66
22.22
95 90
SCOTT^BLUFF
• NORTH NORPDL
CLIMATE STATIONS
hmote Chonqc Sfudy
6FDL
Figure 1. Climate stations and GCM gridboxes used in crop modeling study; a) GISS GCM, b) GFDL GCM.
3-10
-------
Rosenzweig
Table 2. Temperature (°C) and Precipitation (mm/month) Changes* in the Great Plains, GISS and GFDL
Climate Change Scenarios
Latitude 39.13-46.96*N 31.30-39.13'N 23.48-31.30*N
Temp Precip Temp Precip Temp Precip
D,J,F
M,A,M
J,J,A
S,0,N
Annual
5.8
4.8
3.8
5.2
14.9
5.1
20.7
0.6
-6.0
5.1
4.9
4.3
4.1
5.0
4.6
-5.4
-18.9
-6.6
0.0
-7.7
4.7
3.7
4.5
4.1
4.3
-16.5
9.0
-10.2
-12.0
-7.4
GFDL
Latitude 40.00-44.44*N 35.55-40.OO'N 31.11-35.55*N 26.66-31.ll'N
Temp Precip Temp Precip Temp Precip Temp Precip
D,J,F
M,A,M
J,J,A
S,0,N
Annual
5.0
4.8
7.7
5.5
5.8
4.2
7.7
-28.2
1.9
-3.6
5.1
5.2
5.9
4.8
5.3
3.9
9.0
-30.1
3.6
-3.4
4.8
5.1
3.4
4.6
4.5
3.6
-7.8
11.8
-9.7
-0.5
4.2
4.5
3.2
4.6
4.1
-7.4
-7.9
66.4
3.2
13.6
*Values from study sites are averaged latitudinally-
3-11
-------
Rosenzweig
a)
WINTER
SPRING
SUMMER
FALL
ANNUAL
GISS
GFDL
b)
P
R
E
C
I
P
I
T
A
T
I
O
N
C
H
A
N
G
E
M
M
/
D
A
Y
0.2
-O.H
-0.2 H
-0.3
I I WINTER
SPRING
SUMMER
FALL
ANNUAL
GISS
GFDL
Figure 2. Average change in (a) temperature and (b) precipitation over Great Plains study sites for GISS and
GFDL climate change scenarios.
3-12
-------
Rosenzweig
a)
SPRING
SUMMER
1934 - 1936
QFDL
b)
SPRING
SUMMER
1934 » 1936
GFDL
Figure 3 Comparisons of observed climate (mean of 1934 and 1936) and GFDL dimate change scenario at
study sites in Nebraska and Kansas; (a) temperature, and (b) precipitation.
3-13
-------
Rosenzweig
The use of observed climate as a baseline is advantageous because it can be related directly to observed
crop yields in the recent past. Changes occurring under scenario conditions can be assessed relative to both
actual climate and to simulated, yet validated yields. This make the study results more relevant to current
agricultural production.
Limitations Resulting from the Climate Scenarios
Current climate models oversimplify certain aspects of the climate system, such as ocean, cloud, and land-
surface processes. In particular, precipitation and the hydrological cycle are often poorly simulated by GCMs.
The GCMs were not specifically designed for regional studies, and regional representation of current climate is
often inaccurate. Thus, the use of GCM-generated scenarios for the regional case studies for the U.S. EPA
Report to Congress must be approached with caution. In addition, since atmospheric trace gases are increasing
gradually without a predicted artificial plateau at 600 ppm CO~ the 2xCO2 climate change scenario represents
an unrealistic step change to a different climate equilibrium that will probably never be realized.
This work uses predicted changes in mean climate variables from the GCMs and does not consider
alterations in interannual climate variability. For example, the number of days of precipitation remains the
same in the baseline and climate change scenarios, while the amount of precipitation on each of those days is
adjusted by the GCM ratio. The frequency of extremes such as maximum temperatures changes in the climate
change scenarios, but the patterns of the extreme episodes are determined by the observed climate. The lack
of changes in the patterns of extreme events is particularly important, because runs of climate extremes (e.g.,
prolonged hot spells during grain filling and drought) can decrease crop productivity (Mearns et al., 1984). For
dryland crops, yields may change considerably depending on whether a change in precipitation is caused by more
or fewer events or by higher or lower precipitation per event.
Climate Data
Observed daily maximum and minimum temperatures and daily total precipitation from the National Climate
Data Center, Asheville, NC, were provided by Dr. Roy Jenne of the National Center for Atmospheric Research,
with interpolation of missing data by Dr. Amos Eddy, of the Oklahoma Climatological Survey. Observed daily
solar radiation, another CERES input, is lacking in consistent length of record, sites, and calibration.
Therefore, daily solar radiation was simulated for each site according to the method of Richardson and Wright
(1984) as modified by Hodges et al. (1985). In this method, daily solar radiation is estimated based on
correlations between departures of observed daily solar radiation from long-term daily means and departures of
daily maximum and minimum temperatures from long-term daily means stratified according to wet and dry days.
The correlations at sites for which long-term daily means are available have been computed (Richardson and
Wright (1984); these were interpolated to estimate daily solar radiation for the study sites.
Soils
The CERES models were run for three agricultural soils at each study site. The three soils were chosen
from the description of the Major Land Resource Area of each location to represent low, medium, and high
productive capacity (Appendix 1) (USDA, 1981). Soil characteristics for these representative soils were specifi-
ed by twelve generic soil types by Drs. Joe T. Ritchie and J. W. Jones (Appendix 2).
Management Variables
For CERES-Wheat, cultivar, plant population, row spacing, sowing depth, and planting date windows
(periods when planting normally occurs) were specified for each location according to information on current
practices provided by local county extension agents. For CERES-Maize, cultivars were specified according to
Jones and Kiniry (1986), and other variables were specified as suggested by county agents for Nebraska sites.
3-14
-------
Rosenzweig
Simulations
CERES-Wheat and CERES-Maize were run for 30 years of baseline climate under dryland and irrigated
conditions at the study sites. In the irrigated simulations, the soil moisture profile was automatically filled to the
drained upper limit when the soil water in the top meter of soil was less than 80% of that amount; the efficiency
of irrigation was assumed to be 100%, that is, all water applied was available for crop use. The irrigated simula-
tions are unrealistic since few farmers fully irrigate wheat, but they were done in order to study relative changes
in applied irrigation water and the stability of yields under irrigated conditions.
Reported and modeled baseline yields for wheat and corn are shown in Tables 3 and 4. County wheat
yields reported from 1985 are generally lower than yields simulated at the study sites; modeled dryland corn
yields are both above and below reported yields, depending on location. Modeled irrigated yields of both corn
and wheat are consistently higher than observed yields because of the automatic filling of the model soil water
profile.
The crop models were run again with the GISS and GFDL climate change scenarios; percent change and
standard deviations of percent change (see Appendix 3) were calculated for differences in crop yields,
evapotranspiration, and water applied for automatic irrigation. Changes in maturity date were also computed.
Another set of simulations was executed with the crop models modified for the direct effects of CO,, using the
GISS and GFDL scenarios at all study sites.
Adjustment experiments were carried out with the CERES models and the GISS climate change scenario.
These simulation experiments show how farmers might adjust management variables to mitigate the negative
effects or to take advantage of possible beneficial effects of the projected climate changes. In one adjustment
experiment, planting date was shifted by the average number of days that the first frost in the fall changed (this
was later at all locations in the GISS climate change scenario). Infestations of the Hessian fly (Phvtophaga
destructor), which damages wheat sown too early in the fall in some parts of the Great Plains, were not
considered. In another adjustment experiment, planting dates of CERES-Maize were advanced between 20
and 30 days, according to earlier last spring frosts in the GISS climate change scenario.
Another adjustment farmers may make to climate change is to plant cultivars adapted to the new climate
regime. To test the effect of such an adjustment on modeled crop yields, new cultivars chosen on the basis of
vernalization requirement and photoperiod sensitivity were used in CERES-Wheat. Since winter wheat cultivars
with high vernalization requirements need cold temperatures to induce reproductive growth (Evans et al., 1975),
warmer temperatures in the winter in the GISS climate change scenario allow shifts to cultivars with
intermediate or no vernalization requirements (i.e., spring-type wheat cultivars). Wheat cultivars also vary in
photoperiod sensitivity, i.e., need for long hours of daylight to flower (Evans et al., 1975). The warmer
temperatures of the GISS 2XCO, climate change scenario hasten green-up in the spring during a period with
short days. Thus, cultivars with less photoperiod sensitivity are required.
Simulations were done using a new cultivar with either an intermediate vernalization requirement or no
vernalization requirement at all, in addition to the changed planting date. Cultivars were also selected with low
sensitivity to photoperiod in order to avoid negative photoperiod effects caused by delayed flowering during short
days.
3-15
-------
Rosenzweig
Table 3. Observed and Modeled Baseline Wheat Yields
CERES**
State County Dryland* Drvland
SD
CERES**
Irria.* Irricr.
SD
kg/ha kg/ha kg/ha kg/ha kg/ha kg/ha
Nebraska
Douglas
Hall
Lincoln
Madison
Scotts Bluff
Kansas
Ford
Sherman
Sedgwick
Oklahoma
Oklahoma
Tulsa
Texas
Potter
Cameron
2 898
2 359
1 820
2 426
2 763
1 530
3 417
1 961
1 833
1 611
1 483
0
4 325
3 593
2 465
3 300
1 634
717
732
811
682
618
1 794 792
1 381 649
3 885 1 039
3 553
6 273
1 186
2 379
865
705
657
423
2 965
3 100
3 774
2 763
3 774
5 000
5 035
4 982
4 828
5 282
3 188 5 926
3 868 6 043
2 648 5 704
3 659 4 767
- *** 6 579
3 727
0
6 431
2 583
470
369
649
395
370
427
370
434
366
551
405
355
*Observed 1985.
**Mean of modeled yields on 3 representative generic soils,
1951-1980.
***Not available
Sources: 67th Annual Report and Farm Facts. Kansas State Board of
Agriculture; 1986 Nebraska Agricultural Statistics.
Nebraska Dept. of Agriculture; Oklahoma Agricultural
Statistics 1985. Oklahoma Agricultural Statistics
Service; 1985 Texas Field Crop Statistics, Texas Dept. of
Agriculture.
3-16
-------
Table 4. Observed and Modeled Baseline Maize Yields
Rosenzweig
State County Dryland*
CERES**
Dryland
SD
Irriq.
CERES**
Irriq. SD
kg/ha kg/ha kg/ha kg/ha kg/ha kg/ha
Nebraska
Hall 7 607 5 573
Madison 5 030 7 064
Lincoln 3 898 3 194
Douglas 6 790 8 643
Scotts Bluff 2 200 1 194
Kansas
Ford
Sherman
Sedgwick
Oklahoma
Tulsa -** 10 333
Oklahoma - 8 244
Texas
Bexar 4 583 4 096
McLennon 4 885 8 515
9 915 2 635
3 646 1 588
3 772 6 891
1 621
2 248
1 691
1 121
1 206
1 583
1 434
2 138
1 580
1 711
1 235
2 122
8 676
8 928
8 613
8 865
7 859
0
8 443
9 242
5 639
0
9 730 664
11 802 682
9 281 1 080
9 572 652
11 289 1 468
11 178
10 238
10 515
12 064
12 265
6 396
12 412
753
634
736
961
1 302
550
640
*Observed 1985
**Mean of yields on 3 representative generic soils, 1951-1980
***Not available
Sources: 1985 Kansas Farm Facts. Kansas State Board of
Agriculture; 1986 Nebraska Agricultural Statistics.
Nebraska Dept. of Agriculture; Oklahoma Agricultural
Statistics 1985. Oklahoma Agricultural Statistics
Service; 1985 Texas Field Crop Statistics. Texas Dept. of
Agriculture.
3-17
-------
Rosenzweig
CHAPTERS
RESULTS AND DISCUSSION
Climate Change Alone
Wheat. Under the GISS climate change scenario without the physiological effects of CO~ modeled dryland
wheat yields decrease in every location, with larger decreases toward lower latitudes (Table 5a). Yield decreases
range from 10 to 55%; mean decrease was about 30%, assuming equal area at each site. The yield decreases
are driven primarily by the increased temperatures in the climate change scenario, which cause the duration of
crop growth stages, particularly grain fill, to be shortened. Shortening of the grain filling period reduces the
amount of carbohydrates available for grain formation and harvestable yield. Maturity dates of wheat occur, on
average, about three weeks earlier in the GISS climate change scenario (see Appendix 4).
Total dryland crop evapotranspiration (ET) also decreases at every site (Table 5a). Although the higher
temperatures of the climate change scenario cause daily rate of ET to increase, total crop ET decreases due
to the significant shortening of the crop growing season.
Results with the GFDL climate change scenario are similar to those with the GISS scenario. Dryland
CERES wheat yields decrease everywhere, with reductions ranging from 12 to 55%; mean decrease is about
33%, again assuming equal area at each site (Table 6a). Large decreases occur at both higher and lower
latitudes. Total crop also is reduced everywhere, again due to shortening of the crop growing season. Maturity
dates of modeled dryland wheat advance by up to four weeks in the GFDL scenario (see Appendix 5).
In the automatic irrigation simulations with both GISS and GFDL scenarios, modeled wheat yields and
crop ET generally decrease, but not as much as in the dryland cases (Tables 5b and fib). Standard deviations
of the percent changes are also lower. Dryland and irrigation yields for the entire 30 years of simulation are
shown in Figure 4 for Amarillo, Texas. The high temperatures of climate change scenarios have a negative
effect on crop growth, even when adequate water is Table S.available, by shortening crop duration: maturity
dates occur about three weeks earlier in the irrigated as well as the dryland simulations for both GISS and
GFDL climate change scenarios (Appendices 4 and 5).
Even though total crop ET generally decreases, water applied for irrigation at most study sites either
remains the same or increases with the modeled climate change (Tables 5b and fib). This is because water
applied for irrigation depends on modeled soil moisture which in turn depends on precipitation as well as
evaporation. Increases in water applications range from about 10 to 50%. With the GISS climate change
scenario, water applied for irrigation remains almost the same in the northern gridbox where precipitation
increases, and increases at most sites in the central and southern gridboxes where precipitation decreases (see
Table 2). With the GFDL scenario, significant increases in irrigation occur at half of the study sites, especially
in the northern gridboxes where precipitation decreases greatly during the growing season (see Table 2).
Planting winter wheat too early in the fall can decrease yields because of excessive growth before the onset
of cold weather. If global warming extends the period between last frost in the spring and first frost in the fall,
farmers may adjust planting date of winter wheat accordingly. When planting date windows are delayed in
dryland and irrigated CERES-Wheat simulations with the GISS 2xCO, climate change scenario, yields improve
over those for the original planting date under this scenario in only a few cases (Table 7). This shows that the
modeled yield decreases in the climate change scenario are not caused primarily by too early fall planting.
Results of the adjustment experiment with cultivars better adapted to the changed climate show that a
change in cultivar brings wheat yields back up to or improves on baseline levels at two-thirds of the dryland
sites (Table 7). In the irrigated runs, yields equal to or higher than baseline yields occur at more than half of
3-18
-------
Rosenzweig
Table 5. CERES-Wheat Percent Changes in Yield, Evapotranspiration, and Water Applied for Irrigation With
GISS Climate Change Scenario; a) Dryland and b) Irrigated
a)
DRYLAND CERES-WHEAT GISS 2xC02 EXPERIMENT
b)
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
IRRIGATED
YIELD
%A
-3.2
a -3.7
f 6.1 *
-11.6 *
s 6.5 *
0.8
-6.9 *
-10.7 *
-21.5 *
-19.6 *
-18.3 *
-48.3 *
YIELD
%A
-10.5
-17.1 *
-25.9 *
-13.9 *
-27.7 *
-10.5
-38.7 *
-33.6 *
-33.9 *
-45.4 *
-55.4 *
-45.3 *
YIELD
SD %A
6.0
6.0
9.6
4.3
8.4
16.1
11.7
6.9
3.3
5.9
12.3
4.1
CERES-WHEAT GISS
YIELD
SD %A
2.4
2.0
1.8
2.4
2.8
1.9
2.0
2.0
2.0
2.1
1.8
3.3
ET
SD %A
-9.9
-8.4
-8.7
-10.5
-3.0
-8.3 *
-0.3 *
-2.5 *
2.7
3.1 *
3.8 *
-11.2 *
ET
%A
-12.7
-13.2
-17.7
-11.2
-14.2
-15.4
-16.7
-14.3
-5.8
-11.6
-11.8
-19.0
ET
SD %A
* 2.6
* 2.6
* 2.8
* 2.3
* 2.8
* 4.1
* 3.6
* 3.4
* 1.6
* 2.4
* 4.3
* 3.0
2XC02 EXPERIMENT
ET
SD %A
1.2
1.5
1.3
1.3
1.4
1.2
1.2
1.2
1.4
1.2
1.4
2.3
IRRIG
%A
-2.5
2.0
3.2
-6.8
9.4 *
-0.8
16.5 *
20.0 *
49.2 *
31.8 *
17.9 *
-4.1
IRRIG
SD %A
3.9
4.2
3.4
5.2
3.7
3.3
2.6
5.1
8.1
5.2
3.3
4.1
*Greater than 2x SD %' change. 3-19
-------
Rosenzweig
Table 6. CERES-Wheat Percent Changes in Yield, Evapotranspiration, and Water Applied for Irrigation With
GFDL Climate Change Scenario; a) Dryland and b) Irrigated
a)
DRYLAND CERES-WHEAT GFDL 2xC02 EXPERIMENT
YIELD YIELD ET
b)
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
IRRIGATED
YIELD
%
-30.1
-26.0
-45.3
-20.1
-40.4
-46.9
-12.3
-18.8
-20.4
-26.9
-55.1
-40.4
SD
* 5.
* 5.
* 9.
* 4.
* 8.
* 11.
13.
* 6.
* 3.
* 6.
* 13.
* 3.
%
6
9
1
1
3
7
2
8
3
2
0
9
CERES-WHEAT GFDL
YIELD
% SD %
-9.9 *
-9.2 *
-2.2
-16.3 *
0.2
-14.8 *
-15.6 *
-17.9 *
-19.4 *
-20.8 *
-17.3 *
-42.7 *
2
1
1
2
2
3
1
2
2
2
.3
.9
.9
.3
.8
.1
.9
.0
.4
.1
1.9
3.1
ET
SD %
-2.2
-1.0
3.3
-3.8
4.8
1.8
-1.4
-4.0
-11.6
-1.5
-0.9
*
*
*
*
*
-8.0 *
%
-19.2
-15.7
-17.7
-11.8
-19.2
-23.5
-10.1
-7.7
-4.3
-8.8
-27.5
-13.3
ET
SD
* 2
* 2
* 2
* 2
* 3
* 3
* 4
* 3
* 1
* 2
* 4
* 2
%
.7
.8
.9
.4
.0
.7
.0
.4
.4
.5
.2
.9
2xCO2 EXPERIMENT
ET
SD %
1.4
1.6
1.3
1.3
1.5
1.2
1.3
1.3
1.4
1.2
1.6
2.1
IRRIG
%
14.6
13.6
21.2
2.8
22.8
17.4
5.9
-2.3
-2.3
12.4
13.2
-3.4
*
*
*
*
*
*
*
IRRIG
SD %
4.2
4.3
3.6
5.2
3.9
3.2
3.4
5.3
7.4
5.2
3.4
4.1
^Greater than 2x SD % change,
3-20
-------
Rosenzweig
Table 7. CERES-Wheat Yield GISS 2xCO, Adjustment Experient for Planting Date and Combined Planting
Date and Change in Cultivar; a) Dryland and b) Irrigated
a)
DRYLAND CERES-WHEAT YIELD GISS 2XCO2 ADJUSTMENT EXPERIMENT
b)
CC % CC SD % CC+PD % CC+PD SD
?*te
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
CHANGE
-10.5
-17.1 *
-25.9 *
-13.9 *
-27.7 *
-10.5
-38.7 *
-33.6 *
-33.9 *
-45.4 *
-55.4 *
-45.3 *
CHANGE
6.0
6.0
9.6
4.3
8.4
16.1
11.7
6.9
3.3
5.9
12.3
4.1
IRRIGATED CERES-WHEAT
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
CC %
CHANGE
-3.2
-3.7
6.1 *
-11.6 *
6.5 *
0.8
-6.9 *
-10.7 *
-21.5 *
-19.6 *
-18.3 *
-48.3 *
CC SD %
CHANGE
2.4
2.0
1.8
2.4
2.8
1.9
2.0
2.0
2.0
2.1
1.8
3.3
CHANGE % CHANGE
-33.4 *
-38.7 *
-15.3
-11.8 *
-7.1
-2.0
-37.8 *
-28.6 *
-32.4 *
-45.9 *
-50.8 *
-48.4 *
GISS 2XCO2
3.1
3.2
9.4
4.2
8.6
16.6
11.6
6.7
3.2
5.9
12.3
3.9
CC+PD+C CC+PD+C
% CHANGE SD%
-13.8 *
-19.4 *
13.6
6.7
27.2 *
20.9
25.4
9.1
18.8 *
2.1
152.7 *
-30.2 *
CHANGE
3.7
4.0
10.6
5.5
9.8
19.0
14.4
6.7
4.0
7.6
20.8
4.4
ADJUSTMENT EXPERIMENT
CC+PD % CC+PD SD
CHANGE %
25.9 *
42.1 *
1.5
-13.7 *
2.7
-0.6
-8.7 *
-12.8 *
-21.8 *
-19.6 *
-19.3 *
-50.4 *
CHANGE
4.5
9.5
1.8
2.4
2.9
1.9
2.0
2.0
2.0
2.1
1.8
3.2
CC+PD+C CC+PD+C
% CHANGE SD%
56.6 *
257.2 *
10.4 *
5.6
19.7 *
18.5 *
-18.3 *
-8.6 *
31.3 *
34.5 *
-30.0 *
-33.6 *
CHANGE
4.8
11.3
3.0
3.9
3.5
3.3
2.2
2.4
2.5
2.7
1.8
3.8
'Greater than 2 x SD % change.
CC = Climate change alone
CC+PD = Climate change plus change in planting date
CC+PD+C = Climate change plus change in planting date plus change
in cultivar
3-21
-------
Rosenzweig
Dryland
1956
1961 1966
YEAR
1971
1976
Base Run
GISS: 2xCO2
GFDL: 2xC02
Automatic Irrigation
1976
Base Run
GISS: 2xCO2
GFDL: 2xC02
Figure 4. CERES-Wheat yields for Amarillo, Texas with GISS and GFDL Climate Change Scenarios; a)
Dryland and b) Irrigated
3-22
-------
Rosenzweig
the locations. At two sites, Amarillo (dryland) and Grand Island (irrigated), the change in cultivar results in
very large increases in yields, although this may be caused by poorly specified cultivars in the baseline simulation.
Corn. Modeled corn yields are less negatively affected by the GISS climate change scenario than modeled
wheat yields. While dryland CERES-Maize yields do decrease everywhere, the decreases are significant in only
7 out of 14 locations (Table 8a). Yield decreases range from 4 to 43%; the mean decrease is 17%, assuming
equal area Corn yield decreases are somewhat lower at lower latitudes, perhaps due to the use of cultivars
already adapted to high temperatures at the southern study sites in the baseline simulation. Total crop ET
decreases, but not significantly, with the GISS climate change scenario, implying that the increased daily
evapotranspiration is approximately offset by the shortened season (Table 8a). Maturity dates of corn advance
significantly (11 to 30 days) in the dryland case.
Dryland corn yield decreases are very large in the hotter and drier GFDL scenario, particularly at higher
latitudes (Table 9a). Decreases range from 9 to 90%; mean decrease is about 50%. The yield decreases,
especially those at higher latitudes, are caused by the combined effects of high temperatures shortening the
grain filling period and increased moisture stress. The GFDL scenario has pronounced reductions of about 30
mm per month in summer precipitation (see Table 2) in the two northern gridboxes of the study area, which
occur during critical growth stages of corn, i.e., flowering and grain filling (Doorenbos and Kassam, 1979). Total
crop ET decreases everywhere and maturity dates are advanced by an average of three weeks in the GFDL
scenario.
Irrigated corn yields decrease significantly at all locations in both the GISS and GFDL scenarios (Tables
8b and 9b). Even with irrigation, yields decreases from 9 to 21% occur in the GISS scenario and from 13 to
37% in the GFDL scenario. Compared to the less severe GISS case, where water applied for irrigation increases
significantly at only about half the study sites, the more severe GFDL climate change scenario causes water
applied for irrigation to increase everywhere, in one location by over 100%. Maturity dates advance by about two
and one-half and three weeks in the GISS and GFDL scenarios respectively.
To simulate farmer adjustment to a longer growing season, the planting date window in CERES-Maize
was set earlier according to changes in last spring frost in the GISS scenario. This results in some amelioration
in dryland yield decreases, but declines are still large (up to 32%) in most locations
(Table 10).
Combined Climatic and Direct Effects of CO2
Wheat. The direct effects of CO, are able to mitigate the decreased wheat yields in the dryland case in
some but not all locations, in both the uISS and GFDL scenarios. Yield values for latitudinal bands are shown
in Figure 5 for the GISS and GFDL climate change scenarios with and without the direct effects of CO^. Sites
at more southern latitudes show less compensation by the direct effects. With the GISS dryland scenario, 6 of
the 12 locations have yield reductions; with the GFDL dryland scenario, 7 of the 12 locations have yield
reductions. When automatic irrigation is applied, wheat yields improve over the baseline in almost all locations
with combined climatic and direct effects of CO2 in both the GISS and GFDL scenarios (Figure Sb). However,
yields still decrease in all scenarios at the southernmost study sites.
Corn. Dryland corn yields increase under the less severe GISS climate scenario when combined with direct
CO, effects, but decrease significantly in half of the locations with the more severe GFDL scenario (Figure 6a).
An interesting result occurs with the irrigated corn runs in that yields decreased compared to baseline irrigated
corn yields almost everywhere, despite the positive effects of increased photosynthesis and stomatal resistance
(Figure 6b). As simulated in CERES-Maize, this is caused by the high temperature advancement of
development stages, particularly grain filling, which cause yield decreases despite increased photosynthate and
improved water use attributable to the direct effects of CO2- The lower photosynthetic response to CO2 of corn
(10% increase) as compared to wheat (25%) also contributes to this result.
3-23
-------
Rosenzweig
i
Table 8. CERES-Maize Percent Changes in Yield, Evapotranspiration, and Water Applied for Irrigation with
GISS Climate Change Scenarios; a) Dryland and b) Irrigated
a)
CERES-MAIZE YIELD GISS 2xCO2 EXPERIMENT
Site
Dryland
YIELD % YIELD SD%
CHANGE CHANGE
ET % ET SD
CHANGE % CHANGE
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha-
North Platte
KANSAS
Good land
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. city
TEXAS
Amarillo
Waco
San Antonio
Brownsville
b)
YIELD %
Site CHANGE
NEBRASKA
Norfolk -20.2 *
Grand Island -18.9 *
Scotts Bluff -12.3 *
Omaha -22.6 *
North Platte -12.5 *
KANSAS
Good land -18.7 *
Dodge City -21.4 *
Wichita -17.5 *
OKLAHOMA
Tulsa -8.7 *
Okla. City -10.5 *
TEXAS
Amarillo -17.1 *
Waco -11.5 *
San Antonio -18.8 *
Brownsville -19.4 *
-18.6
-19.8
-33.2
-24.5
-4.0
-26.7
-42.9
-26.6
-10.5
-12.6
-12.5
-5.4
-3.5
-7.1
* 7.7
* 7.2
22.3
* 3.1
12.8
20.6
* 14.7
* 7.4
* 3.7
* 5.3
16.4
5.9
7.4
11.5
Irrigated
YIELD SD% ET %
CHANGE
1.6
1.6
2.6
1.7
2.6
1.5
1.8
2.0
1.8
2.3
1.3
1.6
2.0
3.1
CHANGE %
0.6
1.7
3.9 *
1.6
1.6
7.8 *
12.8 *
9.4 *
5.0 *
8.0 *
11.6 *
-1.9
-2.4
-5.4 *
-3.2
-3.1
-1.6
-3.6 *
-3.3
-1.3
-5.5
-4.5
-2.5
-3.0
-1.2
-4.4
-4.1
-2.2
ET SD
CHANGE %
1.1
1.0
1.9
1.2
1.5
1.1
1.3
1.9
1.3
1.9
1.2
1.3
1.6
2.3
-2.6
2.8
5.3
1.4
3.8
5.2
4.8
2.8
1.7
2.1
5.3
2.3
3.6
5.2
IRRIG
CHANGE
12.0
7.1
6.9
14.2
7.6
14.9
28.1
29.4
15.7
20.5
28.1
4.0
-4.7
-3.6
IRHIG
SD% CHANGE
* 4.1
3.6
3.7
* 4.8
4.5
* 3.3
* 4.6
* 6.7
* 5.0
* 5.2
* 3.7
4.1
4.7
3.6
* Greater than 2x SD % change
3-24
-------
Rosenzweig
Table 9. CERES-Maize Percent Changes in Yield, Evapotranspiration, and Water Applied for Irrigation With
GFDL Climate Change Scenario; a) Dryland and b) Irrigated
a)
CERES-MAIZE GFDL 2xCO2 EXPERIMENT
Drylane
YIELD YIELD SD
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
okla. city
TEXAS
Amarillo
Waco
San Antonio
Brownsville
b)
YIELD %
Site CHANGE
NEBRASKA
Norfolk -35.0
Grand Island -33.3
Scotts Bluff -22.7
Omaha -37.3
North Platte -26.4
KANSAS
Goodland -32.4
Dodge City -25.3
Wichita -21.0
OKLAHOMA
Tulsa -13.3
Okla. City -10.9
TEXAS
Amarillo -24.5
Waco -16.4
San Antonio -13.3
Brownsville -22.8
% CHANGE %
-76.0 *
-74.5 *
-83.4 *
-63.4 *
-66.8 *
-90.1 *
-66.7 *
-42.2 *
-23.5 *
-8.2
-38.1 *
-18.2 *
-12.2
8.8
Irrigated
YIELD SD% ET %
*
*
*
*
*
*
*
*
*
*
*
*
*
*
CHANGE CHANGE
1.7 26.1
1.9 30.6
2.7 36.4
1.7 32.4
2.6 23.2
1.6 28.3
1.8 20.9
2.1 16.4
2.0 13.2
2.3 3.7
1.5 23.6
1.7 -0.6
2.1 -1.6
3.1 -1.6
CHANGE %
6.4
5.9
19.6
3.2
10.9
17.7
13.7
7.2
3.5
5.5
16.0
6.0
7.7
10.6
ET SD
% CHANGE
* 1.6
* 1.6
* 2.1
* 1.7
* 1.9
* 1.6
* 1.6
* 2.1
* 1.5
* 1.8
* 1.5
1.3
1.5
2.3
ET
CHANGE
-24.3
-21.1
-25.6
-14.3
-28.2
-38.8 *
-15.8 *
-7.7 *
-4.4 *
-3.4
-14.7 *
-7.4 *
-8.3 *
-0.8
,
IRRIG
% CHANGE
87.3
78.9
73.0
107.2
64.8
68.4
47.9
51.9
40.7
10.7
48.8
8.1
7.8
6.7
ET SD
% CHANGE
2.7
2.8
4.9
1.8
3.9
4.6
4.5
2.8
1.6
2.1
5.4
2.5
3.4
4.8
IRRIG
SD% CHANG
* 4.7
* 4.6
* 4.1
* 5.8
* 5.3
* 3.6
* 4.8
* 6.6
* 5.1
* 5.1
* 4.3
4.2
4.8
3.6
* Greater than 2x SD % change
3-25
-------
Rosenzweig
Table 10. Dryland CERES-Maize Yield GISS 2xCO2 Adjustment Experiment for Planting Date
DRYLAND
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
CC %
CHANGE
IN YIELD
-18.6 *
-19.8 *
-33.2
-24.5 *
-4.0
CC
SD %
CHANGE
7.7
7.2
22.3
3.1
12.8
CC+PD %
CHANGE IN
YIELD
-15.2
-11.0
-28.0
-18.9 *
-4.2
SD %
CHANGE
7.8
7.4
22.5
3.5
12.8
KANSAS
Goodland
Dodge City
Wichita
•26.7
•42.9 *
•26.6 *
20.6
14.7
7.4
•30.8
•32.0 *
•20.0 *
20.2
15.6
7.4
OKLAHOMA
Tulsa
Okla. City
TEXAS.
Waco
San Antonio
•10.5 *
•12.6 *
-12.5
-5.4
3.7
5.3
16.4
5.9
-7
0
4 *
1
-4.5
-4.7
3.6
5.7
5.9
7.8
*Greater than 2 x SD % change.
CC = Climate change alone
CC-t-PD = Climate change plus change in planting date
3-26
-------
CERES-WHEAT YIELDS
DRYLAND
Rosenzweig
38-40 N
36-38 N
LATITUDE
34-36 N
I BASE
3 GISS dlGISS'DE
IGFDL
CERES-WHEAT YIELDS
IRRIGATED
<34 N
40-42 N
38-40 N
36-38 N
LATITUDE
34-36 N
I BASE
GISS EZlGISS'DE
IGFDL
<34 N
IGFDL*DE
Figure 5. CERES-Wheat yields with GISS and GFDL climate change scenarios with and without the direct
effects of CO2: a) dryland and b) irrigated.
3-27
-------
Rosenzweig
T
h
o
u
8
a
n
d
8
CERES-MAIZE YIELDS
DRYLAND
38-40 N
36-38 N
LATITUDE
34-36 N
CERES-MAIZE YIELDS
IRRIGATED
40-42 N 38-40 N 36-38 N
LATITUDE
34-36 N
I BASE
iGISS CZUGISS+DE
IGFDL
<34 N
• BASE I
Si GISS
CDGISS-DE
i^GFDL. I
llGFDL-DE
<34 N
Figure 6. CERES-Maize yields with GISS and GFDL climate change scenarios with and without the direct
effects of CO2: a) dryland and b) irrigated.
3-28
-------
Rosenzweig
Interpretation of Results
It should again be noted that many uncertainties are embedded in this study due to the assumptions and
simplifications of the crop and climate models. Therefore, these results should be interpreted as the potential
sensitivity of the modeled crops in the region to the range of climate change scenarios designed for the study.
Climate Change Alone. The results of the crop modeling studies with climate change scenarios alone show
that increase in temperature is the major cause of modeled yield reductions in both wheat and corn in the
southern and central Great Plains. In most cases, increases in temperature during the growing season cause a
decrease in the duration of the crop life cycle, particularly the grain filling period. In general, the smaller
temperature increases of the GISS scenario cause smaller yield reductions compared to the larger reductions
occurring in the GFDL scenario.
Wheat yields decrease more at lower latitudes where wheat is close to its boundary of adaptation. These
changes imply shifts in wheat production more northward in the region. Modeled yields of corn, a C4 crop more
adapted to high temperatures, show an opposite effect in that larger decreases occur at higher latitudes. If
climate changes occur at the level projected by the GISS scenario, corn cultivars already adapted to high
temperatures may be available for growth in the region.
Changes in precipitation have a relatively minor effect on modeled wheat yields in the Great Plains, although
they influence the amount of water applied in the irrigation simulations. Changes in precipitation do affect
modeled corn yields in some locations. The largest yield reductions occur in corn with the GFDL climate
scenario, which has very high temperatures and pronounced summer dryness to which corn is susceptible.
Combined Climate and Direct Effects of CO2. The direct effects of CO2, as modeled in this study,
compensate for or even ameliorate, the climate change impacts in many locations, but not in all. Particularly
with the hotter and drier GFDL scenario, modeled yield decreases of both corn and wheat are significant in the
dryland case. Irrigated corn is more negatively affected than wheat in the combined scenarios because of the
lower photosynthetic response of corn to CO, Because yield increases caused by higher levels of CO2 do not
overcome the negative effects of the predicted climate change in every location, even with the possibly
overestimated direct effects as incorporated in the crop models, the results of this study suggest that the
expectation that the direct CO, effects will compensate for climate change may be overly optimistic.
Irrigation. Irrigation partially mitigates the effects of climate change on both modeled corn and wheat
yields in the climate change scenarios. Modeled irrigated yields are also more stable under climate change
conditions. Modeled annual corn yields for dryland and irrigated corn in Grand Island, Nebraska, are shown
in Figure 7. These results suggest that increased irrigation may be needed to fully counteract the negative effects
of climate change in the Great Plains, and that regional demand for irrigation water would be likely to rise in
response to the predicted changes. This would occur for two reasons: first, crops currently irrigated would
require more water where precipitation decreases; and second, more acreage would be irrigated as high
temperatures increase the variability of crop yields. Increased irrigation would thus be needed to ensure
acceptable and stable yield levels.
Adjustments. Adjusting planting date of wheat to later in the fall does not significantly ameliorate the
effects of the GISS climate change scenario on CERES-Wheat yields. Earlier corn planting dates slightly reduce
yield decreases. Changing to more climatically adapted wheat cultivars with lower vernalization
requirements and lower photoperiod sensitivity, in addition to delaying planting dates, overcomes yield decreases
at some sites, but not in others. Thus, there appear to be some cultivars available for adaptation to the
projected climate change, but these adaptations may not be efficacious at all locations. This suggests that
development of heat- and drought-tolerant cultivars should be included in plant-breeding objectives for the
region.
3-29
-------
Rosenzweig
Dryland
c
o
R
N
Y
I
E
L
D
K
G
/
H
A
1976
Base Run GlSS: 2xCO2
GFDL: 2xC02
Automatic Irrigation
c
o
R
N
Y
I
E
L
D
K
G
/
H
A
ioooh
0000
1951
1976
Base Run
GlSS: 2xCO2
GFDL: 2xC02
Figure 7. CERES-Maize yields for Grand Island, Nebraska with climate change scenarios alone; (a) dryland and
(b) irrigated.
3-30
-------
Rosenzweig
CHAPTER 4
IMPLICATIONS OF RESULTS
Given the projections of a virtually unidirectional warming trend driven by increasing concentrations of
atmospheric trace gases and the potential for increased drought stress caused by higher temperatures and/or
insufficient precipitation, agriculture as it is practiced now in the Great Plains may become more difficult to
sustain in the future. If climate change occurs as predicted, agriculture may become more environmentally
damaging and economically marginal in the region.
Environmental Implications
The primary environmental implication of the results is a probable increase in demand for irrigation in the
region, particularly with more severe climate changes. Heightened demand for irrigation could place stress on
the already depleted Ogalalla Aquifer and other water resources in the region. Many of the problems associated
with intense groundwater use (for example, water depletion and soil erosion) could be exacerbated by global
warming. However, availability of and competition for water supplies also may change with climate change, and
defining the extent to which irrigation can provide an economic buffer against climate change requires further
study.
Due to the potential for increased demand for water for irrigation, primary consideration should be given
to sound environmental programs for managing water resources in terms of both quantity and quality.
Competition for water resources between the potential increased agricultural use and nonagricultural demands
may increase. This would imply that state and regional water policies would need modification in light of such
potentialities.
Socioeconomic Implications
Yield decreases in dryland fanning and need for increased irrigation may cause adverse economic
consequences to farmers and rural communities in the region. Crop production may shift to the north, causing
changes in production centers, markets, transportation, and storage. This is because the industry of agriculture
is composed of many people besides farmers, such as farm equipment manufacturers, fertilizer and seed
suppliers, and rural bankers. If climate change is extreme, climate change could cause dislocation of rural
communities in the Great Plains through farm abandonment.
Further Research
Models of crop growth are needed which are both physiologically detailed and validated for wide areas.
This is important both for simulation of agricultural yields under current climate and those projected for
conditions of climate change and increased CO2. In particular, crop models should include photosynthetic and
transpiration processes responsive to concentration of CO2. The method used in this study for simulating
evapotranspiration does not include atmospheric vapor pressure or potential changes in plant temperatures due
to stomatal closure. These should also be modeled more explicitly. Responses of all crop processes to
combined high CO2 levels and high temperatures need to be better understood.
The GISS GCM has been run with gradually increasing trace gases, and use of this "transient" climate
change scenario will allow analysis of a more realistic trajectory of crop response to gradual wanning. Running
the crop models with observed climate for the Great Plains from the drought of the 1930s would provide a
comparison of the crop responses to predicted future climate change with crop responses for an extreme climate
event of the recent past It would also help validate the crop models for climate change scenarios.
Extension of the study area to the Northern Great Plains, including North and South Dakota, would give
a fuller picture of potential response of wheat to climate change over its entire range of production in the UJS.
3-31
-------
Rosenzweig
Finally, further adjustment experiments with both wheat and corn under several climate change scenarios
are also desirable, both with and without the physiological effects of COy as are calculations of changes in
agroclimatic indices such as thermal heat sums and drought indices for the region.
3-32
-------
Rosenzweig
REFERENCES
Acock, B. and Allen, L.H. Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In Strain, B.R.
and Cure, J.D. (eds.), Direct effects of increasing carbon dioxide on vegetation. US Department of Energy,
DOE/ER-0238. Washington, DC.
Charles-Edwards, DA. 1982. Physiological Determinants of Crop Growth. Academic Press. New York. 161 pp.
Chaudhuri, U.N., R.B. Burnett, E.T. Kanemasu, and M.B. Kirkham. 1986. Effect of elevated levels of CCX. on
winter wheat under two moisture regimes. U.S. Department of Energy, Response of Vegetation to Carbon
Dioxide #029.
Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In Strain, B.R. and Cure, J.D. (eds.), Direct
effects of increasing carbon dioxide on vegetation. US Department of Energy, DOE/ER-0238. Washington, DC.
Decker, W.L. and R. Achutuni. 1987. A review of national and international activities on modeling the effects
of increased COu concentrations on the simulation of regional crop production. Progress Report to Carbon
Dioxide Research Division, U.S. Department of Energy, University of Missouri-Columbia.
Doorenbos, J. and A.H. Kassam. 1979. Yield Response to Water. FAO Irrigation and Drainage Paper 33. Food
and Agriculture Organization of the United Nations. Rome. 193 pp.
Evans, L.T., I.F. Wardlaw, and RA. Fischer. 1975. Wheat. In L.T. Evans (ed.) Crop Physiology. Cambridge
University Press. Cambridge, pp. 101-149.
Hansen, J., I. Fung, A. Laos, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
changes as forecast by the GISS 3-D model. Journal of Geophysical Research (in press).
High Plains Associates. 1982. Six-State High Plains Oeallala Aquifer Regional Resources Study. Austin, TX.
Hodges, T., French, V., and LeDuc, S. 1985. Yield Model Development: Estimating Solar Radiation for Plant
Stimulation Models. AgRISTARS YM-15-00403.
Hurt, R.D. 1981. The Dust Bowl. Nelson-Hall. Chicago.
Jones, CA. and Kiniry, J.R. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development.
Texas A&M Press College Station.
Kimball, B A. 1983. Carbon dioxide and agricultural yield: an assemblage and analysis of 430 prior observations.
Agronomy Journal 75:779-788
Liverman, D.M., WJi. Terjung, J.T. Hayes, and L.O. Mearns. 1986. Climatic change and grain com yields in
the North American Great Plains. Climatic Change 9:327-347.
Mearns, L.O., Katz, R.W., and Schneider, S.H. 1984. Extreme high temperature events: Changes hi their
problems with changes in mean temperature. Journal of Climate and Applied Meteorology 23:1601-1613.
Otter-Nacke, S., Goodwin, D.C. and Ritchie, J.T. 1986. Testing and validating the CERES-Wheat model in
diverse environments. AgRISTARS YM-15-00407 JSC 20244.
3-33
-------
Rosenzweig
Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, LH. Allen, Jr. 1988. Impact of Climate Change on Crop Yield
in the Southeastern USA. In Smith, J.B. and DA. Tirpak (eds.), The Potential Effects of Global Climate
Change on the United States. US. Environmental Protection Agency, Report to Cngress. Washington, DC.
Richardson, C.W., and Wright, DA. 1984. WGEN: A Model for Generating Daily Weather Variables. US
Department of Agriculture, Agricultural Research Service, ARS-8, 83p.
Ritchie, J.T. and Otter, S. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
model. In Willis, W.O. (ed.). ARS Wheat Yield Project. USDA-ARS. ARS - 38. pp. 159-175.
Robertson, T., V.W. Benson, J.R. Williams, CA. Jones, and J.R. Kiniry. 1987. Impacts of climate change on
yields and erosion for selected crops in the southern United States. In M. Meo (ed). Proceedings of the
Symposium on Climate Chance in the Southern United States: Future Impacts and Present Policy Issues. Science
and Public Policy Program. University of Oklahoma Norman, OK.
Rogers, H.H., G.E. Bingham, J.D. Cure, J.M. Smith, and KA. Surano. 1983. Responses of selected plant species
to elevated carbon dioxide in the field. Journal of Environmental Ouai. 12:569.
Rosenzweig, C. 1985. Potential CO2-induced climate effects on North American wheat-producing regions.
Climatic Change 7:367-389.
Rosenzweig^ C. 1987. Climate change impact on wheat: The case of the High Plains. In M. Meo (ed.).
Proceedings of the Symposium on Climate Change in the Southern United States: Future Impacts and Present
Policy Issues. Science and Public Policy Program. University of Oklahoma. Norman, OK.
Smith, J. and DA. Tirpak (eds.). 1988. Report to Congress on the Potential Effects of Global Climate Change
on the United States. Chapter III. Study Methods. US. Environmental Protection Agency. Washington, DC.
Stewart, R.B. 1986. Climatic change: Implications for the Prairies. Trans. Roy. Soc. Can/Series V/I:67-%.
Terjung, W.H., D.M. Liverman, and J.T. Hayes. 1984. Climatic change and water requirements for grain corn
in the North American Great Plains. Climatic Change 6:193-220.
US. Department of Agriculture. 1981. Land Resource Regions and Major Land Resource Areas of the United
States. Soil Conservation Service. Agricultural Handbook #2%.
US. Department of Agriculture. 1982. Basic Statistics. 1977 National Resource Inventory. SB-686.
US Department of Agriculture. 1985. Agricultural Statistics. US Government Printing Office. Washington, DC.
Warrick, R A. 1984. The possible impacts on wheat production of a recurrence of the 1930s drought in the US.
Great Plains. Climatic Change 6:5-26.
Williams, G.D.V., RA. Fautley, K.H. Jones, R.B. Stewart, and E.E. Wheaton. 1988. Estimating effects of
climatic change on agriculture in Saskatchewan, Canada. In M.L. Parry et al. (eds.). The Impact of Climatic
Variations on Agriculture. Vol. 1: Assessments in Cool Temperate and Cold Regions. IIASA, UNEP. Kluwer
Academic Publishers. Dordrecht, The Netherlands.
Worster, D. 1979. Dust Bowl: The Southern Great Plains in the 1930s. Oxford University Press. New York.
3-34
-------
APPENDIX 1. STUDY SITES
Rosenzweig
A. Climate station, tape ID#, latitude and longitude, county, Land
Resource Region (LRR), Major Land Resource Area (MLRA), generic
soil types* and soil ID numbers for sites used in Great Plains
study.
1. Grand Island, NE 14935
H. deep silt loam, #6
M. deep sandy loam, #9
L. med. silt loam, #5
2. Norfolk, NE 14941
H. deep silty clay, #3
M. deep silt loam, #6
L. med. silt loam, #5
3. N. Platte, NE 24023
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sand,
40.58N 98.19W Hall
H 71
41.59N 97.26W Madison M 102B
41.08N 100.41W Lincoln G 65, H 72
4. Omaha, NE 14942
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sandy loam, #9
41.18N 95.54W Douglas M 106,107
5. Scotts Bluff, NE 24028 41.52N 103.36W
H. med. silty clay, #2
M. deep silt loam, #6
L. deep sandy loam, #9
6. Dodge City, KS 13985 37.46N 99.58W
H. shallow silty clay, #1
M. med. silt loam, #5
L. deep sand, #12
7- Goodland, KS 23065 39.22N 101.42W
H. shallow silty clay, #1
M. deep silt loam, #6
L. deep sandy loam, #9
S. Bluff G 67
Ford
H 73
Sherman H 72
* H = Generic soil with highest drained upper limit of plant
extractable water of agricultural soils present in production
area/or MLRA
M = Generic soil with medium water-holding capacity of
agricultural soils present at site in production area/or
MLRA.
L = Generic soil with lowest water-holding capacity of
agricultural soils present in production area/or MLRA.
3-35
-------
Rosenzweig
35.24N 97.36W
36.12N 95.54W
32.26N 99.41W
8. Wichita, KS 3928
H. deep silty clay, #3
M. deep silt loam, #6
L. med. silt loam, #5
9. Okla. City, OK 13967
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sand, #12
10. Tulsa, OK 13968
H. deep silty clay, #3
M. deep silt loam, #6
L. med. silt loam, #5
11. Abilene, TX 13962
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sandy loam, #9
12. Amarillo, TX 23047
H. deep silty clay, #3
M. deep sandy loam, #9
L. med. silt loam, #5
13. Brownsville, TX 12919
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sandy loam, #9
14. El Paso, TX 23044 31.48N 106. 24W
H. deep silty clay, #3
M. deep sandy loam, #9
L. deep sand,
37.39N 97.25W Sedgwick H 75,80A
35.14N 101.42W
25.54N 97.26W
15. Midland, TX 23023 31.56N 102. 12W
H. deep silty clay, #3
M. deep silty loam, #9
L. med. silt loam, #5
16. San Antonio, TX 12921 29.32N 98.28W
H. deep sandy loam, #9
M. med. silt loam, #5
L. med. sandy loam, #8
Oklahoma H 80A
Tulsa M 112
Taylor H 78
Potter H 77
Cameron I 83D
El Paso D 42
Midland H 77, I 81
Bexar I 81, J 86,87
17. Waco, TX 13959
H. deep silty clay, #3
M. deep silt loam, #6
L. deep sandy loam, #9
31.37N 97.13W McLennon J 85,86
3-36
-------
Rosenzweig
B. Land Resource Regions (LRR) and Major Land Resource Areas
(MLRA) of sites in Great Plains study.
D Western Range and Irrigated Region
42 Southern Desertic Basins, Plains, and Mountains
G Western Great Plains Range and Irrigated Region
65 Nebraska Sand Hills
67 Central High Plains
H Central Great Plains Winter Wheat and Range Region
71 Central Nebraska Loess Hills
72 Central High Tableland
73 Rolling Plains and Breaks
75 Central Loess Plains
77 Southern High Plains
78 Central Rolling Red Plains
80A Central Rolling Red Prairies
I Southwest Plateaus and Plains Range and Cotton Region
81 Edwards Plateau
83D Lower Rio Grande Valley
J Southwestern Prairies Cotton and Forage Region
85 Grand Prairie
86 Texas Blackland Prairie
87 Texas Claypan Area
M Central Feed Grains and Livestock Region
102B Loess Uplands and Till Plains
106 Nebraska and Kansas Loess-Drift Hills
107 Iowa and Missouri Deep Loess Hills
112 Cherokee Prairies
Source: USDA, Soil Conservation Services, 1981: Land Resource
Regions and Major Land Resource Areas of the United States.
Agriculture Hand-book 296.
3-37
-------
Rosenzweig
APPENDIX 2. GENERIC SOIL TYPES FOR USE IN CLIMATE CHANGE STUDY
Format line #1 of soil
IDUMSL IX,12
PEDON 1X,A12
TAXON 1X,A60
# assigned to a soil type
SCS pedon number
Soil classification
Format
Format
1
.11
10.
15.
15.
20.
-1.
2
.11
10.
15.
20.
25.
30.
30.
-1.
line #2
SALB
U
SWCON
CN2
TAV
AMP
DMOD
SWCON1
SWCON2
SWCON3
RWUMX
PHFAC3
line #3
DLAYR(L)
LL(L)
DUL(L)
SAT(L)
SW(L)
WR(L)
BD(L)
OC(L)
NH4(L)
N03(L)
PH(L)
6.00
.513
.513
.514
.516
6.00
.513
.513
.514
.516
.518
.520
F6.2
IX, F5,
IX, F6,
IX, F6,
lx,F5,
IX, F5.
1X,F3.
IX, E9.
IX, F6,
IX, F5.
IX, F5.
IX, F4.
F6.0
IX, F6.
lx,F6.
IX, F6.
IX, F6.
lx,F6.
IX, F5.
IX, F5.
IX, F4.
IX, F4.
IX, F4.
SHALL
.10 8
.680
.679
.679
.677
MEDIU
.20 8
.680
.679
.679
.677
.676
.674
Bare soil albedo
.2 Upper limit stage 1 evaporation, mm
.2 Soil H2O drainage constant fraction drained per day
,2 SCS curve # used to calculate daily runoff
1 Annual average ambient soil temperature, °C
1 Annual average amplitude in mean monthly soil temp., oc
1 Soil mineralization factor (Default = 1)
2 Coefficient in steady state solution (Default=.00267)
1 Coefficient in steady state solution (Default=58)
2 Coefficient in steady state solution (Default=6.68)
2 Maximum daily root water uptake (Default=0.03)
2 Variable to reduce apparent photosynthesis (Default=l)
Thickness of soil layer L, cm
.3 Lower limit of plant-extractable H2O cm**3/cm**3
3 Drain upper limit soil H2O content for layer L
3 Saturated H2O content for layer L cm**3/cm**3
3 Default soil H2O for layer L cm**3/cm**3
3 Weighting factor for soil depth L to determine root
growth distribution, no units
2 Moist bulk density of soil in layer L g/c**3
2 Organic carbon concentration in layer L %
1 Default ammonium in layer L, mg elememtal N/kg soil
1 Default nitrate in layer L, mg elememtal N/kg soil
1 Default pH in layer L in 1:1 soil: water slurry
SHALLOW SILTY CLAY
89.00 6.9 13.9 1.0
.730 .680 1.000
.729 .679 .819
.729 .679 .607
.727 .677 .449
87.00 6.9 13.9 1.0
.730 .680 1.000
.729 .679 .819
.729 .679 .607
.727 .677 .407
.726 .676 .247
.724 .674 .135
.27E-02
1.35
1.36
1.36
1.36
1.74
1.66
1.45
1.16
.27E-02
1.35
1.36
1.36
1.37
1.37
1.38
1.74
1.66
1.45
1.12
.73
.37
58.0
2.5
2.4
2.2
2.1
58.0
2.5
2.4
2.2
2.0
1.8
1.5
6.
3.3
3.2
3.0
2.7
6.
3.3
3.2
3.0
2.7
2.3
1.9
68
6.5
6.5
6.5
6.5
68
6.5
6.5
6.5
6.5
6.5
6.5
03 1
03 1,
3-38
-------
Rosenzweig
3
.11
10.
15.
25.
30.
30.
30.
30.
30.
-1.
4
.12
10.
15.
15.
20.
-1.
5
.12
10.
15.
20.
25.
30.
30.
-1.
6
.12
10.
15.
25.
30.
30.
30.
30.
30.
-1.
7
.13
10.
15.
15.
20.
-1.
8
.13
10.
15.
20.
25.
30.
30.
-1
DEEP SILTY CLAY
6.00
.513
.513
.514
.516
.519
.521
.522
.522
•
6.00
.106
.106
.107
.108
6.00
.106
.106
.107
.108
.110
.111
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
.
•
•
•
•
•
30 ' 85.
680
679
679
677
675
674
673
673
SHALLOW
20 81.
262
262
262
261
MEDIUM
30 79.
262
262
262
261
260
259
00
730
729
729
727
725
724
723
723
SILT
00
312
312
312
311
SILT
00
312
312
312
311
310
309
6.9 13
.680
.679
.679
.677
.675
.674
.673
.673
LOAM
6.9 13
.262
.262
.262
.261
LOAM
6.9 13
.262
.262
.262
.261
.260
.259
.9 1.0
1.000
.819
.607
.368
.202
.111
.061
.033
.9 1.0
1.000
.819
.607
.449
.9 1.0
1.000
.819
.607
.407
.247
.135
.27E-02
1.35
1.36
1.36
1.37
1.38
1.38
1.39
1.39
1.74
1.66
1.45
1.09
.65
.29
.09
.01
.27E-02
1.37
1.37
1.37
1.38
1.16
1.10
.97
.77
.27E-02
1.37
1.37
1.37
1.38
1.38
1.39
1.16
1.10
.97
.75
.49
.24
58.0
2.5
2.4
2.2
2.0
1.7
1.4
1.1
.8
58.0
2.5
2.4
2.2
2.1
58.0
2.5
2.4
2.2
2.0
1.8
1.5
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.6 6.5
2.2 6.5
1.8 6.5
1.3 6.5
.9 6.5
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.7 6.5
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.7 6.5
2.3 6.5
1.9 6.5
1.00
1.00
1.00
DEEP SILT LOAM
6.00
.106
.106
.107
.108
.110
.111
.112
.112
•
•
•
•
9
•
•
•
•
40 77.
262
262
262
261
260
259
258
258
00
312
312
312
311
310
309
308
308
6.9 13
.262
.262
.262
.261
.260
.259
.258
.258
.9 1.0
1.000
.819
.607
.368
.202
.111
.061
.033
.27E-02
1.37
1.37
1.37
1.38
1.38
1.39
1.39
1.39
1.16
1.10
.97
.72
.43
.20
.06
.01
58.0
2.5
2.4
2.2
2.0
1.7
1.4
1.1
.8
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.6 6.5
2.2 6.5
1.8 6.5
1.3 6.5
.9 6.5
1.00
SHALLOW SANDY LOAM
6.00
.086
.086
.086
.087
6.00
.086
.086
.086
.087
.088
.089
•
•
•
•
•
•
*
•
•
•
•
•
40 74.
220
220
220
219
MEDIUM
50 70.
220
220
220
219
219
00
400
400
400
400
6.9 13.9 1.0
.220
.220
.220
.219
1.000
.819
.607
.449
.27E-02
1.61
1.61
1.61
1.61
.70
.66
.58
.46
58.0
2.5
2.4
2.2
2.1
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.7 6.5
1.00
SANDY LOAM
00
400
400
400
400
400
218 .400
6.9 13.9 1.0
.220
.220
.220
.219
.219
.218
1.000
.819
.607
.407
.247
.135
.27E-02
1.61
1.61
1.61
1.61
1.62
1.62
.70
.66
.58
.45
.29
.15
58.0
2.5
2.4
2.2
2.0
1.8
1.5
6.68 .03
3.3 6.5
3.2 6.5
3.0 6.5
2.7 6.5
2.3 6.5
1.9 6.5
1.00
3-39
-------
Rosenzweig
.13 6.00 .50 68.00 6.9 13.9 1.0 .27E-02 58.0 6.68 .03 1,
10. .086
15. .086
25. .086
30. .087
30. .088
30. .089
30. .089
30. .089
30. .089
-1.
10
.15 4.00 .40 75.00 6.9 13.9 1.0 .27E-02 58.0 6.68 .03 1.
10. .032
15. .032
15. .032
20. .032
-1.
11
.15 4.00 .50 70.00 6.9 13.9 1.0 .27E-02 58.0 6.68 .03 1
10. .032
15. .032
20. .032
25. .032
30. .033
-1.
12
.15 4.00 .60 65.00 6.9 13.9 1.0 .27E-02 58.0 6.68 .03 1.
10. .032
15. .032
25. .032
30. .032
30. .033
30. .033
30. .033
30. .033
-1.
DEEP SANDY
.50
.220
.220
.220
.219
.218
.218
.218
.217
.217
68.00
.400
.400
.400
.400
.400
.400
.400
.400
.400
LOAM
6.9 13.9 1.0
.220
.220
.220
.219
.218
.218
.218
.217
.217
1.000
.819
.607
.368
.202
.111
.061
.033
.018
1
1
1
1
1
1
1
1
1
.27E-02
.61
.61
.61
.61
.62
.62
.62
.62
.62
.70
.66
.58
.43
.26
.12
.04
.01
.00
58.0
2.5
2.4
2.2
2.0
1.7
1.4
1.1
.8
.5
3
3
3
2
2
1
1
6.
.3
.2
.0
.6
.2
.8
.3
.9
.5
68
6
6
6
6
6
' 6
6
6
6
.5
.5
.5
.5
.5
.5
.5
.5
.5
SHALLOW SAND
.40
.107
.107
.107
.107
75.00
.370
.370
.370
.370
6.9 13.9 1.0
.107
.107
.107
.107
1.000
.819
.607
.449
1
1
1
1
.27E-02
.66
.66
.66
.66
.29
.28
.24
.19
58.0
2.5
2.4
2.2
2.1
3
3
3
2
6.
.3
.2
.0
.7
68
6
6
6
6
.5
.5
.5
.5
MEDIUM SAND
.50
.107
.107
.107
.107
.106
DEEP
.60
.107
.107
.107
.107
.106
.106
.106
.106
70.00
.370
.370
.370
.370
.370
SAND
65.00
.370
.370
.370
.370
.370
.370
.370
.370
6.9 13
.107
.107
.107
.107
.106
6.9 13
.9 1.0
1.000
.819
.607
.407
.247
.9 1.0
.107 1.000
.107
.107
.107
.106
.106
.106
.106
.819
.607
.368
.202
.111
.061
.033
.27E-02
1,
1,
.66
.66
1.66
1.66
1.66
.29
.28
.24
.19
.12
.27E-02
1.
1.
1.
1.
1.
1.
1.
1.
66
66
66
66
66
66
66
66
.29
.28
.24
.18
.11
.05
.01
.00
58.0
2.5
2.4
2.2
2.0
1.8
58.0
2.5
2.4
2.2
2.0
1.7
1.4
1.1
.8
6.68
3
3
3
2
2
.3
.2
.0
.7
.3
6
6
6
6
6
.5
.5
.5
.5
.5
6.68
3
3
3
2
2
1
1,
(
.3
.2
6
6
.5
.5
.0 6.5
.6
.2
.8
.3
.9
6.5
6.5
6.5
6.5
6.5
3-40
-------
APPENDIX 3.
PERCENT CHANGE STATISTICS
Rosenzweig
1. Individual run
BASELINE
YIELD
1
2xC02
YIELD
YIELD
DIFFERENCE
2xC02-BASELINE
Observed mean
nb
fib
nc
Observed standard
deviation
CTa
Standard deviation
of observed mean
- M«
Observed standard
deviation of mean
yield difference
fie
Ma
Mean and uncertainty
of percent change
MC ±
x 100
2. Summary of 3 soils at one site
Summary mean and
uncertainty of
change over all
soils; L, M, H =
low, medium, and
high production
capacity
MC
X 100
3-41
-------
Rosenzweig
APPENDIX 4. CERES-WHEAT MATURITY DATE GISS 2xCO2 EXPERIMENT
a)
b)
DRYLAND
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Aiaarillo
Brownsville
MEAN
DIFF
-25 *
-25 *
-30 *
-24 *
-27 *
-27 *
-20 *
-20 *
-8 *
-17 *
-18 *
-11 *
SD
MEAN
DIFF
0.9
1.0
0.9
0.9
0.9
0.9
0.9
0.8
0.9
0.9
0.9
1.6
%A
-13.3 *
-14.0 *
-16.0 *
-13.7 *
-14.7 *
-15.2 *
-12.3 *
-12.3 *
-10.5 *
-10.9 *
-11.6 *
-13.7 *
SD %A
0.5
0.5
0.5
0.5
0.5
0.5
0.6
0.5
0.6
0.6
0.6
2.2
IRRIGATED
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
MEAN
DIFF
-24 *
-25 *
-29 *
-24 *
-27 *
-28 *
-21 *
-20 *
-8 *
-16 *
-17 *
-11 *
SD
MEAN
DIFF
0.9
0.9
0.9
0.9
0.9
0.9
1.0
0.8
0.9
0.9
1.0
1.4
%A
-13.1 *
-13.9 *
-15.8 *
-13.7 *
-14.6 *
-15.5 *
-12.4 *
-12.2 *
-10.4 *
-10.8 *
-11.3 *
-13.8 *
SD %A
0.5
0.5
0.5
0.5
0.5
0.5
0.6
0.5
0.6
0.6
0.6
2.0
*Greater than 2x SD % change.
3-42
-------
APPENDIX 5. CERES-WHEAT MATURITY DATE GFDL 2xCO, EXPERIMENT
Rosenzweig
a)
DRYLAND
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
MEAN
DIFF
-22
-22
-29
-21
-24
-27 *
-22 *
-21 *
-27 *
-18 *
-23 *
-15 *
SD
MEAN
DIFF
0.9
0.9
0.9
0.8
0.8
0.8
0.9
0.8
0.9
0.9
0.9
1.6
%A
-11.8
-12.1
-15.3
-11.8
-13.0
-15.0 *
-13.4 *
-13.1 *
-18.3 *
-12.1 *
-14.7 *
-18.5 *
SD %A
0.5
0.5
0.5
0.5
0.4
0.5
0.6
0.5
0.6
0.6
0.6
2.2
b)
IRRIGATED
Site
NEBRASKA
Norfolk
Grand Island
Scotts Bluff
Omaha
North Platte
KANSAS
Goodland
Dodge City
Wichita
OKLAHOMA
Tulsa
Okla. City
TEXAS
Amarillo
Brownsville
MEAN
DIFF
-21
-21
-28
-21
-24
-28 *
°22 *
-21 *
-27 *
-18 *
-23 *
-13 *
SD
MEAN
DIFF
0.9
0.9
0.8
0.8
0.8
0.8
1.0
0.8
0.9
0.9
1.0
1.5
%A
-11.5
-12.0
-15.1
-11.7
-12.9
-15.5 *
-13.3 *
-13.1 *
-18.3 *
-11.9 *
-14.9 *
-16.8 *
SD %A
0.5
0.5
0.4
0.5
0.4
0.5
0.6
0.5
0.6
0.6
0.6
2.1
*Greater than 2x SD % chanqe.
3-43
-------
THE ECONOMIC EFFECTS OF CLIMATE CHANGE ON UJS. AGRICULTURE:
A PRELIMINARY ASSESSMENT
by
Richard M. Adams
J. David Glyer
Department of Agricultural and Resource Economics
Oregon State University
Corvallis, Oregon 97331
and
Bruce A. McCarl
Department of Agricultural Economics
Texas A & M University
College Station, Texas 77843
With the assistance of
Henry A. Froehlich
Department of Forest Engineering
Oregon State University
Corvallis, Oregon 97331
and
Scott L. Johnson
Department of Agricultural and Resource Economics
Oregon State University
Corvallis, Oregon 97331
Cooperative Agreement No. CR811965-01
-------
CONTENTS
ACKNOWLEDGMENTS Mi
*
FINDINGS 4-1
CHAPTER 1: INTRODUCTION 4-3
OBJECTIVES 4-3
ORGANIZATION 4-4
CHAPTER 2: METHODOLOGY 4-5
PROCEDURE 4-6
SUPPORTING CROP YIELD DATA 4-8
SUPPORTING HYDROLOGIC DATA: ASSUMPTIONS AND DEFINITIONS 4-8
CHAPTER 3: RESULTS 4-12
EFFECTS OF CROP YIELD//WATER CHANGE ASSUMPTIONS 4-12
GISS RESULTS 4-14
SENSITIVITY ANALYSES 4-18
EFFECTS OF CHANGES IN TECHNOLOGY AND WORLD FOOD DEMAND 4-21
Technology Assumptions 4-24
Demand Assumptions 4-24
Economic Consequences 4-25
GFDL RESULTS 4-25
Technology and Demand Assumptions 4-27
Direct Effects of CO2 on Crop Yields 4-27
CHAPTER 4: IMPLICATIONS AND CONCLUSIONS 4-38
REFERENCES 4-41
APPENDIX A. 4-43
APPENDIX B 4-49
u
-------
ACKNOWLEDGMENTS
The research described in this document has been funded wholly by the UJS. Environmental
Protection Agency through a cooperative agreement (CR-811965-01) with Oregon State University. It has
been subjected to the agency's peer and administrative review and has been approved for publication as
an EPA document. The views expressed in the document do not necessarily reflect the views of the agency
and no official endorsement should be inferred. Many individuals contributed to the completion of this
research. The authors gratefully acknowledge the assistance of Jim Jones, Bruce Curry, Brian Baer and
Joe Ritchie hi providing critical plant science data. We are particularly grateful to Cynthia Rosenzweig for
her assistance in providing and interpreting plant science data as well as serving as the project officer in
this research. We appreciate the assistance of Bob House, Marcel Aillery, Glen Schaible and Terry
Hickenbothom of the USDA, ERS Policy and Soil and Water Groups for providing data on demand
elasticities, irrigation costs and use patterns, and the current FEDS budgets. Cynthia Rosenzweig, Joel
Smith, John Riley and three anonymous reviewers provided constructive comments on earlier versions of
this manuscript. Finally, we wish to thank Bette Bamford for her patience in typing the many drafts of this
manuscript.
111
-------
Adams
FINDINGS1
In this study a model of U.S. agriculture is used to measure the economic effects of changes in crop yields and
water availability arising from projected long-term changes in climate associated with a doubling of CO2- The
resulting economic effects are conditional on the validity of the data on crop yields and water availability. These
data are, in turn, a function of the validity of the general circulation models (GCMs) that forecast global climate
change and the crop yield and hydrologic models that translate climate change into its physical effects. Given the
uncertainties involved in each component of this analysis, and the general difficulties of forecasting the structure
of a dynamic industry such as agriculture over a 60 to 70-year time period, caution should be exercised in
interpreting and applying the economic estimates.
This analysis encompasses (1) a range of possible climate changes as predicted by the GISS and GFDL
climate models under a doubled CO2 environment; (2) a diverse set of yield and water availability assumptions
for each climate change; and (3) some fundamental assumptions concerning the structure of agriculture over the
next 70 years. To keep the analysis tractable, "common themes" are used to limit possible combinations of
assumptions and parameters. First, all crop yield changes are based on yield estimates from the CERES and
SOYGRO plant models. The models predict yield changes for corn, soybeans, and wheat in various regions of
the U.S. (for each GCM). An average of yield changes for these three crops is then used to develop surrogate
responses for other crops in each region for which no plant model estimates are available. Uncertainties in crop
yield data are captured in sensitivity analyses reflecting one standard deviation (plus and minus) around midpoint
yield responses. Changes in irrigation water availability and crop water demands (for irrigated crops) are based
on ratios of GCM forecasts of evaporation and of rainfall. Uncertainties are measured in sensitivity runs (higher
rain, lower evaporation and vice versa) for each GCM's grid point estimates. Finally, potential changes in
technology and in future UJS. and world food demand, as well as the direct effects of CO2 on crop yields, were
introduced into the climate change analyses.
As expected, these varying assumptions give rise to a range of probable economic effects. For the GISS
2xCO2 scenario, "midpoint" changes in crop yields and in water availability and demand result in an aggregate net
loss in economic welfare of approximately $6.0 billion per annum (in 1982 $). The distribution of these effects
is sharply skewed, as consumers lose over $7 billion while producers experience a net gain of $1.1 billion owing
to the generally inelastic demand for modeled crops. All crops except hay experience minor to moderate
reductions in production. Regionally, more southern production areas decrease land in crop production (e.g.,
Appalachia, Delta, and Southern Plains) while land use increases in northern and western regions (Lake States,
Corn Belt, Mountain, and Pacific). For most irrigated areas, slight increases in water availability offset increases
in evaporative water loss, except for the Pacific, where increases in availability are substantial. Nationally, irrigated
acreage increases by about 5 million acres (11 percent) owing to (1) an increase in the comparative advantage of
irrigated versus dryland yields under the climate change scenarios, and (2) the rise in commodity prices that makes
irrigated production economically feasible.
The GFDL midpoint case implies substantially greater potential losses from climate change. The net annual
loss in economic welfare is approximately $33 billion with consumers losing $37 billion and producers gaining $4
billion. Regionally, northern and western areas increase cropped acreage (Pacific, Mountain, Northern Plains)
while other areas experience reductions (Corn Belt, Appalachia, Southeast, Delta States, Southern Plains). As in
the GISS analyses, irrigated acreage increases in all regions (about 40 percent nationally), again due to rising
commodity prices and the increased comparative advantage of irrigated production.
Imposing long-term changes in technology and food demand on these analyses alters the above effects. For
GISS, potential technological change appears capable of offsetting yield reductions from climate yield change,
'Although the information in this report has been funded wholly or partly by the VS. Environmental
Protection Agency under Cooperative Agreement No. CR811965-01, it does not necessarily reflect the Agency's
views, and no official endorsement should be inferred from it.
4-1
-------
Adams
even with increased food demand. Under the larger yield reductions forecast by GFDL, 40 to SO years of
technology change is required to offset the yield reductions attributed to climate change. Finally, the potential
yield-enhancing effects of CO, may moderate the economic consequences of a doubling of CO2. When such
direct CO2 effects are combined with the generally adverse effects of climate change, the GISS analyses show slight
to moderate increases in economic welfare. For GFDL, however, CO, yield enhancement does not completely
mitigate climate change, resulting in a net loss of approximately $10 billion. As in the other analyses,
northwesterly shifts in crop production and increased irrigation use are observed across the combined CO2-climate
change evaluations.
The results of these varied analyses suggest that climate change is not a food security issue; the production
capacity of U.S. agriculture is adequate to meet domestic needs, even under the more extreme climate changes.
However, major resource and environmental quality adjustments are likely. Expansion .of irrigation and shifts in
regional production patterns imply more competition for water resources, greater potential for ground and surface
water pollution, loss of wildlife habitat, increased soil erosion, and major structural changes in local economies.
The costs of these changes on the well-being of future generations are not addressed here.
It should be noted that these results consider the effects of climate change on the U.S. in isolation. If other
areas of the world benefit from climate change/CO2 increases or are less severely affected than the VS., then ILS.
agricultural export trade could be changed from the patterns assumed in this analysis. Also, the analyses
performed here do not consider the effects of climatic variability. Greater variability in precipitation, for example,
would be expected to have greater adverse effects than measured here. Changes in crop pest and disease
infestations due to climate change are not considered in the plant yield forecasts used in these evaluations.
Finally, CO, increases beyond those forecast here could imply more severe implications for agriculture, as the
yield-enhancing effects of CO2 increases are likely to plateau, thus failing to mitigate increasingly adverse climate
effects.
4-2
-------
Adams
CHAPTER 1
INTRODUCTION
Global climate change arising from anthropogenic increases in atmospheric CO2 and trace gas
concentrations is an issue of international concern. The implications of climate change are complex, occurring
on a global scale with potential effects on virtually all ecosystems and ecosystem service flows (Wigley et al.,
1981). One ecosystem of particular importance to human welfare is agriculture. The consequences of crop
failure arising from unfavorable climate are apparent. Recent notable examples include the famine in Northern
Africa arising from prolonged drought and, within the VS., the 1983 and 1988 droughts. While the level of
human suffering manifested in these two examples are undeniably different, both provide dramatic evidence
of the susceptibility of agriculture and agriculture's constituents to changes in regional climates.
The direct and indirect effects of global climate change on agriculture are discussed in qualitative terms
in numerous studies (Decker et al., 1986; Rosenzweig, 1986; Callaway et al., 1982). For example, Rosenzweig
has identified the following consequences on agricultural productivity of climatic change: (1) changes in yield
due to increased atmospheric CO2 concentration, increased temperature, and the likelihood of increased pest
and pathogen populations arising from a warmer global climate; and (2) the indirect consequences on
agricultural productivity associated with potential reductions in irrigation water supplies. To this list, one could
add yield effects of increases in tropospheric ozone and UV-B radiation incidence at the Earth's surface arising
from the same trace gas emissions associated with climate change. In combination, some or all of these factors
may alter the yields of major food and fiber crops. Forecasting the exact magnitude and distribution of long-
term changes in yield and water availability from global climate change models is a challenging task for plant
scientists, hydrologists, and others. Once obtained, these physical and biophysical responses need to be
translated into economic or other measures of human welfare for use in policy formation. The latter task may
be even more challenging than forecasting the likely physical/biological manifestations of climate change, given
the adaptive behavior of producers, consumers, and other economic agents likely to occur over the long-time
horizons associated with climate change.
OBJECTIVES
The purpose of this research is to provide a preliminary measure of the economic consequences of long-
term climate change on VS. agriculture. The results provide a general impression of the importance of global
climate change with respect to the welfare of agricultural producers and consumers. Specific objectives include
the following:
(1) Estimate the regional and national economic implications of changes in yield, and water availability and
use, for a set of major ILS. commodities (e.g., corn, soybeans, and wheat) associated with alternative
global climate change scenarios arising from a doubling of CO2;
(2) Determine the sensitivity of these estimates to selected assumptions concerning critical biological, physical,
and economic dimensions of the analysis; and
(3) Based upon the outcome of objective 2, define a set of longer term research objectives aimed at improving
the ability of economists and others to provide policy analyses concerning global climate change.
4-3
-------
Adams
ORGANIZATION
The remainder of this report consists of three sections. Section two describes the methodology used to
measure the economic effects of climate change at different levels of aggregation. Critical data sources and
assumptions, along with limitations of the analysis, are discussed. The third section presents the results of
applying the methodology to crop yield and water availability changes associated with two global climate model
scenarios that reflect a doubling of CO, concentrations. The implications of these results are evaluated in
the fourth section. This last section draws some preliminary implications and conclusions from these climate
change analyses. Future research needs are also outlined.
4-4
-------
Adams
CHAPTER 2
METHODOLOGY
Performing the bioeconomic assessment defined in the above objectives requires critical input from several
disciplines. Such assessments typically involve a series of steps to link physical and biological phenomena to
an economic valuation model. The starting point in the current assessment is definition of likely changes in
global climate due to a doubling of CO, and, specifically, how they will be manifested in terms of changes in
temperature, precipitation, and other cumatic variables across agricultural production regions in the United
States. Two forecasts of climate change based on general circulation models (GCMs) are used here. One
comes from the Goddard Institute of Space Studies (GISS), the other from Princeton University Geophysical
Fluid Dynamics Laboratory (GFDL).
The GCMs forecast changes in regional temperature, precipitation, evaporation and other climate variables
due to the doubling of CO2- Such changes are likely to lead to changes in crop yields and water available for
irrigation. These measures of climate change become inputs into the second stage of this process, which
requires knowledge of how changes in regional climate affect items that people value, e.g., the quantity and
quality of food and fiber production. This information is provided by crop yield response models that
incorporate the likely mechanisms of yield change arising from climatic alterations. Specifically, crop yield
changes were predicted by plant scientists using the CERES family of plant physiology models and SOYGRO,
a soybean model of comparable design. In the assessment, information is also required on how long-term
climate changes will affect both crop water demand and water supply in irrigated areas of the US. Since water
is, perhaps, the most important input in the agricultural production process, forecasts of water demand and
availability are an important aspect of an economic assessment of climate change. These water forecasts were
derived from the GCM scenarios (see Appendix A). Once quantified, resulting yield and irrigation effects were
used to modify an economic model of the UJS. agricultural sector to translate the physical and biological effects
into economic consequences.
The primary focus of this research is on applying an "appropriate" economic model to the crop yield and
hydrologic assumptions arising from CO2 changes. Appropriate is defined in terms of the model's economic
credibility, as established by its theoretical and empirical content, coupled with how well the model captures
critical dimensions of global climate change. Specifically, since change in climatic variables is not homogeneous
across agricultural production areas in the U.S., the model must contain sufficient regional detail to account
for these variations in regional climate. Second, since global climate change is likely to affect both yields (as
measured by the CERES and SOYGRO crop response models) and the supply of irrigation water, the model
needs a detailed characterization of irrigation requirements, by crop and region, as well as regional water
availabilities. Third, since the model is to be used for policy analysis, it should measure economic consequences
at various levels of aggregation including producer welfare at the regional and national levels, as well as effects
on both domestic and foreign consumers.
The assessment model used here conforms to the above requirements and is based on an economic model
used in several recent analyses of the economic effects of tropospheric ozone on agriculture (e.g., Adams et al.,
1984; Adams et al., 1988). In general, the model and its application are conceptually similar to the numerous
induced change analyses found in the agricultural economics literature. Specifically, the economic model is a
spatial equilibrium model formulated as a mathematical programming problem (Takayama and Judge, 1971).
An alternative to this type of partial-equilibrium model formulation is a compatible general equilibrium (CGE)
model with an agricultural subsector. However, submodels are extremely consumptive of data and it is doubtful
that a tractable CGE curve could be constructed with the degree of resolution required to meet the needs of
this study. Further, Kokoski and Smith have shown that the differences in welfare measures between a partial
equilibrium analysis and a CGE solution are minor if the size and duration of indirect price effects are
consistent.
4-5
-------
Adams
The model represents production and consumption of 30 primary agricultural commodities, including
both crop and livestock products. Processing of agricultural products into 12 secondary commodities is also
included. The production and consumption sectors are assumed to be made up of a large number of
individuals, each of whom operates under competitive market conditions. This leads to a model which
maximizes the area under the demand curves less the area under the supply curves. Following Samuelson, this
area can be interpreted as a measure of economic welfare (ordinary consumers' plus producers' surplus)
equivalent to the annual net income lost or gained by agricultural producers and consumers as a consequence
of global climate change, expressed in 1982 dollars. Both domestic and foreign consumption (exports) are
included. The assumptions and methodology are discussed in Appendix B. Additional detail is provided in
Adams et al. (1984) and McCarl and Spreen (1980).
The model consists of two components, a set of micro or farm-level models integrated with a national
(sector) model. Producer-level behavior is captured in a series of technical coefficients that portray the physical
and economic environment of agricultural producers in each of the 63 homogeneous production regions in the
model, encompassing the 48 contiguous states. These regions are then aggregated to 10 macro regions, as
defined by the US. Department of Agriculture (USDA) (Figure 1). Of importance in this assessment is the
inclusion of both irrigated and nonirrigated crop production and of water supply relationships for each region.
Availability of land, labor, and irrigation water is determined by supply curves defined at the regional level
depicted in Figure 1. Farm-level supply responses generated from the 63 individual regions are linked to
national demand through the objective function of the sector model, which features demand relationships for
various market outlets for the included commodities. The model simulates a long-run, competitive equilibrium
as reflected in 1981-1983 economic and environmental parameters. This three-year base period was selected
because (1) a multi-year period was deemed more reasonable than a single year as a base against which to
assess long-term changes, and (2) while not a period of equilibrium in supply and demand, the period had more
stability than alternative recent multi-year periods. It should be noted that this period differs from the base
period used in the GCMs (1951-1980) to model climate change. Given the rate of change in the agricultural
sector (particularly since 1951), the use of a recent base period was viewed as more appropriate for economic
modeling.
PROCEDURE
To implement the assessment, specific scenarios of global climate change are required. Of these, we
focus on two scenarios of doubled CO2 equilibrium, where the base-level CO2 corresponds to the 1951-1980
period. Specifically, the climatic consequences of the 2xCO2 equilibrium scenario are estimated via the GISS
and the GFDL GCMs.
Estimated yield and water resource changes associated with regional climate changes forecast by each
GCM are introduced into the model through modifications in (1) regional crop yields; (2) crop water use
coefficients; and (3) regional water supply functions. The subsequent model simulations then generate a picture
of their economic effects, including shifts in regional market shares (i.e., comparative advantage), changes in
producers' returns, changes in consumers' well-being and other economic aspects. In addition to the model
solutions for the climate scenarios, sensitivity analyses are performed to establish "bounds" on these economic
estimates. The sensitivity analyses focus on uncertainties in the yield and water availability estimates, as well
as on the effects of uncertainty in economic assumptions, including changes in technology and demand. A final
set of analyses incorporates the direct (yield enhancing) effects of CO, in both the GISS and GFDL climate
change evaluations.
4-6
-------
Adams
DELTA
SOUTHERN v ESTATES
PLAINS
Figure 1. Farm production regions in the United States.
Source: USDA ERS.
4-7
-------
Adams
SUPPORTING CROP YIELD DATA
Crop yield implications of regional climate changes are provided by plant scientists at Goddard NASA's
Space Flight Center Institute for Space Studies, University of Florida, and Michigan State University.
Specifically, Cynthia Rosenzweig (at GISS), Jim Jones, Bruce Curry, and Bob Peart (at the University of
Florida) and Joe Ritchie (at Michigan State University) provided regional crop yield changes for wheat, corn,
and soybeans using the CERES and SOYGRO models. Coverage was limited to soybeans in the Corn Belt,
Lake States, and Southeast; corn in the Corn Belt, Lake States, and the Southeast; and wheat in the Great
Plains states. These regions represent major production areas for each crop. For cotton and other excluded
crops, responses obtained from the CERES and SOYGRO model predictions for corn, wheat, and soybeans
were averaged to develop surrogate responses. Yield changes were extrapolated to the Northeast and West
from the regions that are modeled. Because most of the production in the western regions is from irrigated
crops that were much less affected by the climatic shifts than their dryland counterparts, these extrapolations
may be less than they appear.
For most crops, the plant models indicate lower yields associated with the GCM forecast climate
conditions, with the GFDL climate projections resulting in substantively lower yields than for GISS. Potential
yield reductions are greater for dryland crops than for irrigated. In general, potential yield reductions are
greater in the South, the Southern Plains, and the Southwest. For soybeans, slight yield increases are predicted
in more northern latitudes (e.g., Minnesota). Crop yield projections and associated standard deviations for
various locations, along with details of the CERES and SOYGRO model estimation procedures, are reported
in studies by Rosensweig, Peart et al., and Ritchie et al. in this volume.
SUPPORTING HYDROLOGIC DATA: ASSUMPTIONS AND DEFINITIONS
Hydrologic information on potential changes in ground and surface water levels and crop water demands
associated with regional changes are another component of this assessment. Total precipitation and evaporation
estimates from the GISS and GFDL models are used to develop a first approximation to such potential changes
in water demand and availability. This section describes how these "first approximations" are developed and
applied in the subsequent analyses.
Any analysis of the hydrologic implications of climate change projections from the GCMs requires
numerous assumptions. Changes in temperature, evaporation, and precipitation specified for one or two grid
boxes extending over a large region may not adequately reflect changes within the region. The relative changes
in the climatic values estimated by GCM model grid points were assumed to adequately represent the changes
in the region in which they fall. For each grid point, both baseline and 2xCO2 values for each climate variable
(e.g., rainfall) were estimated. The ratio of the 2xCO2 estimate to baseline wen provides an indication of the
percent change in that particular variable. When more than one grid box falls within a region, a weighted
average of grid box values was assigned to the region, based on the proportion of that region's geographical
area contained in each grid box. Figure 2 shows the regional boundaries and the GISS and GFDL grid points
utilized for the hydrologic regions; note that they do not correspond exactly to the USDA regions in Figure 1.
Changes found in the hydrologic regions are adjusted to USDA definitions in the application of the economic
model. The hydrologic regions used here do not cover the northeastern quarter of the UJS. because the area
accounts for only 2% of irrigated acreage.
These hydrologic assumptions take a broad view of potential changes and, at best, are qualitative
generalizations. The economic model incorporates both water requirements, by crop and region, as well as the
supply function of water available for irrigation (ground and surface). To estimate changes in water demand
for each region in the model, the interaction of changes in evaporation and precipitation was considered. This
was done by calculating the ratio of the predicted values in the 2xCO2 scenarios to predicted current (IxCO,)
values for both evaporation and rainfall. The ratio of these two ratios was then calculated for each grid point
to arrive at a net change in crop water requirements. Thus, if evaporation is forecast to increase more (in
4-8
-------
Adams
relative terms) than local rainfall, water requirements (net evapotranspiration) are expected to increase. The
data on which these calculations are based are shown in Table 1 for both the GISS and GFDL GCMs.
The other component of interest is irrigation water. Like the estimates of water demand, potential
changes in irrigation water should reflect the interaction of forecasts of evaporation and rainfall. If changes
in long-term, mean rainfall are expected to be greater than long-term mean evaporation, some "surplus" should
result, leading to increased runoff or aquifer storage, shifting out the supply curve of irrigation water.
Conversely, if rainfall increases less than evaporation, then a decrease in the supply curve may be expected.
Using the estimated ratios for rainfall and evaporation, a ratio of net change in supply was estimated. This
ratio was then used to adjust the baseline water supply levels (specified for the period 1981-1983) for each
irrigated region in the economic model. For surface water, it is assumed here that irrigation use is a "senior"
water right within the applicable water doctrine for each state. Further, it is assumed that those irrigation rights
are currently oversubscribed (insufficient water to meet current irrigation rights). Therefore, any increases in
streamflow will be allocated to irrigation. Given the timeframe of this analysis, we assume that it is feasible
to build new dams to adjust to changes in the timing and quantity of runoff. Also, we assume that reservoir
management will adjust to new climatic and streamflow regimes, thus storing water earlier in the winter runoff
period. Specific hydrologic assumptions for the GISS and GFDL results are provided in Appendix A.
4-9
-------
n
""•• • • —• • • — • • w— • • J i
DAKOTA I •^-•,
MINNESOTA
sourHOAK ^"""
CLEARTYPE-
STATE OUTLINE
UNITED STATES
* GISS Grid Points
• GFDL Grid Points
Figure 2. Hydrologic Regions and GCM Grid Points.
-------
Adams
Table 1. Climatic Characteristics of Nine Agricultural Regions as *
Predicted by GISS and GFDL Models under the 2xC02 Scenario
Evaporation
Ratio
Northwest
California
Northern
Mountain
Southern
Mountain
Northern
Plains
Southern
Plains
Delta
Southeast
GISS
1.166
1.069
1.151
1.062
1.085
0.985
1.024
1.084
GFDL
1.099
0.970
1.097
1.031
0.989
1.018
1.016
0.927
Precipitation
Ratio"-7
GISS
1.230
1.062
1.180
1.050
1.070
0.922
1.024
1.105
GFDL
1.027
1.018
1.017
0.986
0.966
0.997
1.003
0.922
Temperature'-7
Increase , C*
GISS
4.4
4.9
4.8
4.9
4.7
4.4
5.3
3.5
GFDL
4.5
4.9
5.5
5.1
5.9
4.5
4.4
4.9
*J These are relative to values predicted by the models for lxC02.
b-/ Net relative changes in supplies of (demands on) water are obtained by taking
the ratio of rainfall to evaporation (evaporation to rainfall) for each
scenario.
c-/ GISS is about 1£ more in winter than in summer, GFDL is about 1£ less.
4-11
-------
Adams
CHAPTERS
RESULTS
The previous section presents some key features of the economic model and its application to this
assessment. As discussed, the economic model includes features that reflect the complex set of interactions that
underlie economic markets. It also integrates across diverse components of the UJS. agricultural sector to obtain
measures of aggregate and regional economic activity, including measures of producer and consumer welfare.
As with any model, however, complexity does not guarantee predictive ability. Thus, it is important to establish
that this model is a reasonable approximation to the agricultural sector over the period of interest. As described
earlier, the economic model is solved as a mathematical programming problem, where the optimal solution is
characterized by a set of prices and quantities that maximize the sum of producers' and consumers' surplus. To
validate the model, we test these endogenous prices and output for the base years (1981 through 1983) against
the actual price and output values for these years for the modeled commodities. Successful validation provides
one indication that the model is appropriate for evaluating the effects of climate change on agriculture.
Table 2 provides a comparison of actual average prices and quantities produced with those determined by
the model solution at 1981-1983 environmental and economic conditions. As is evident, the prices for aU
commodities match reasonably well, while the quantities generally understate actual levels by 5 to 10%. Overall
then, model prices and quantities for both crop and livestock commodities appear to capture the relative
magnitudes of equilibrium prices and quantities observed in the 1981 through 1983 period. It should be stressed,
however, that while the model is reasonably "tuned" to 1980s economic and technological conditions, there is no
assurance that it will be representative of the agricultural sector in 70 years. Indeed, a quick look at changes in
U.S. agriculture over the last 70 years (from 1918 to today) indicates that agriculture in 2060 is likely to be
dramatically different from today, even in the absence of global climate change.
EFFECTS OF CROP YIELD//WATER CHANGE ASSUMPTIONS
As is evident from Table 2, the economic model predicts current conditions fairly well. In addition to the
prices and quantities reported in Table 2, the model solution contains a number of measures of other economic
activity, including total social welfare (both consumers' and producers' surplus), regional crop acreage, regional
resource use (water, labor, land), exports, and other items. Together, this array of economic "indicators" for the
1981-1983 period make up the baseline solution of the economic model.
The interesting question in this assessment of global climate change is: How does the economic model
solution change as the model is altered to reflect differing climate assumptions? Thus, it is the change in various
economic measures reported by the model between the baseline case and the altered model that provides the
estimates of the economic effects of climate change. Each change in crop yields and/or water availability will
give rise to changes in economic measures in comparison with the baseline case. The direction and magnitude
of the changes in economic measures provides an indication of whether the agricultural effects of the underlying
climate change are trivial or substantial in terms of social welfare.
The model simulations (analyses) described in this report are derived from the GISS and GFDL 2xCO2
scenarios. We start first by examining the results of the GISS climate change projections. Results of the GFDL-
based climate changes are then presented. Finally, combined climate change and CO2 effects for both models
are discussed.
4-12
-------
Adams
Table 2. Model Prices and Quantities vs. Actual: 1981-1983
Commoditv
Cotton
Corn
Soybeans
Wheat
Sorghum
Rice
Barley
Oats
Silage
Hay
Milk
Pork
Fed beef
Nonfed beef
Prices (£ j
Model
284.09
2.68
5.78
3.58
2.53
8.25
2.24
1.54
21.95
65.64
13.39
169.24
232.27
131.99
>er unit)
Actual
281.90
2.68
5.65
3.50
2.50
8.01
2.20
1.67
n.a.
65.76
13.65
165.90
239.70
145 . 15
Quantities (
Model
10.25
6,703.26
1,974.86
2,260.78
555.68
118.96
431.14
509.56
48.02
77.58
1,288.61
139.38
135.81
87.60
millions)
Actual
11.79
6,839.00
1,915.00
2,419.00
730.00
145.00
498.00
526.00
n.a.
82.00
1,359.80'
151.70
156.25
78.37
Prices for all crops are dollars per bushel, except for cotton ($ per 480
pound bale), rice ($ per hundredweight), and silage and hay ($ per ton).
Meat prices are $ per cwt. and are average retail prices for finished meat
products.
Sources: USDA, ERS, Statistical Bulletin No. 715, Washington, D.C.
USDA, Agricultural Statistics. 1984. Washington, D.C.
4-13
-------
Adams
GISS RESULTS
The specific GISS climate change analyses are defined in Table 3. These consist of the baseline case (1981-
1983) and four analyses that reflect various combinations of crop yield changes, and water demand and supply
levels. As described in the table, the range of potential climate change effects increases across the various
analyses, from only three crops and no water changes in Analysis 1 to assumed changes in all crop yields as well
as water demand and supply in Analysis 4.
The first points of comparison among these climate analyses and the base solution are the measures of
direct economic welfare: producers' and consumers' surplus. Table 4 reports the changes in economic surplus
for consumers, producers, and the total for each analysis. The right-most column is the most important for this
discussion, as it provides evidence of the total per annum change in welfare from these various changes in yields
and water demand. The negative economic effects increase as the range of potential effects is expanded, from
approximately $3.5 billion for only three crops to $5.9 billion when all crops are assumed to experience climate
change effects. At the aggregate level, the economic effects of water change assumptions are small in comparison
to the yield effects. This relatively small effect of the water change is due to several factors, including (1)
irrigation expenses are only a modest portion of total cost of producing the modeled crops, and (2) there is a
balancing of increases in both supplies of and demands for water in most regions.
The four analyses reported in Table 4 reflect increasing magnitudes of yield and water changes. As noted
earlier, Analysis 1 uses actual crop yield estimates from the CERES and SOYGRO models for corn, wheat, and
soybeans. No changes in crop yields for other crops are included. While this analysis may be viewed as more
"defensible" because it reflects only crop yield changes predicted by plant models for each crop, it is not realistic
in terms of national level economic modeling. The reason is simple: by not changing other crop yields in the
model, one is implicitly assuming no yield effects from climate change on these crops. The economic implication
of this the base case. In Analysis 2, the production of corn, wheat, and soybeans is reduced, but the acreage and
production of all other crops increase. Analyses 4, while involving a surrogate response for all other crops, does
present a more realistic picture of the process likely to occur under climate change; that is, all crops are likely
to be affected, some more than others. As indicated in Table 5, production of most crops is now reduced.
Table 5 reflects another feature of the economic modeling process of potential importance. Specifically,
economic adjustments undertaken by producers e.g., crop substitution or substitution of summer fallow, can
often mitigate initial yield reductions due to environmental change. In Analysis 4, the net reductions in corn,
wheat, and soybean production are 12,10, and 12%, respectively. However, the average yield losses predicted
by the plant models and used in the analysis are approximately 20% for each crop. Thus, the production
consequences of these yield losses are partially mitigated by the economic adjustments included in the economic
model, including a relative expansion of irrigated production.
The distributional effects of these yield and water changes are quite distinct as almost all the aggregate loss
is borne by consumers, while producers (in the aggregate) experience relatively trivial losses or may gain in
welfare (net income). This is due to the generally inelastic nature of agricultural commodity demand, so that
when crop yields (and production) are reduced, there is a correspondingly greater increase in price. This does
not mean that all producers are unaffected, however, as producers of some commodities (e.g., livestock producers
who pay higher prices for feed grains) or producers in some regions are losers due to the yield changes. Within
the consumer groups, losses are also borne disproportionately because foreign consumers (Le., the export market)
typically absorb over half of the consumer losses, although exports make up less than 20% of the aggregate
consumers' surplus in the base solution. This loss in consumers' surplus generated in export markets is driven
by the crops that experience the greatest production declines, Le., soybeans, rice, and feed grains. Typically,
50% or more of the production of these commodities moves into export markets.
4-14
-------
Adams
Table 3. Description of Yields and Water Demand Availability Assumptions
Underlying Alternative Economic Analyses
Model Specification
Description
Baseline
Analysis 1:
Corn, Wheat
and Soybean
Analysis 2:
Corn, Wheat
and Soybeans
Plus Water
Adjustment
Analysis 3:
All Crops
Analysis 4:
All Crops
Plus Water
Adjustment
The base line case reflects economic, agronomic and environmental
conditions for 1981-1983. Yields and water availability specified
at actual 1981-1983 levels.
Analysis 1 involves yield changes for corn, wheat and soybeans in
each of the 63 regions of the economic model, based on
predicted crop yield changes from CERES and SOYGRO models.
Analysis 2 includes the corn, wheat and soybean changes
defined above plus changes in irrigation water demand (by
crop and region) as well as changes in regional ground
and surface water supplies for irrigation.
Analysis 3 includes the CERES and SOYGRO estimates for corn, wheat and
soybeans plus a yield adjustment for all other crops in the model (cotton,
barley, rice, sorghum, oats and hay) equal to the average change of corn,
wheat, and soybeans for each region.
Same as Analysis 3 but with the addition of water demand
and availability adjustments defined in Analysis 2.
4-15
-------
Adams
Table 4. Aggregate Economic Effects of GISS 2xC02 Global Climate Change
on U.S. Agriculture, in 1982 Dollars
Yield/Water
Economic Surplus Change in Surplus (from Base)
($ billion)
Base Model
Analysis 1
Analysis 2
Analysis 3
Analysis 4
77.318
73.990
73.688
70.627
70.009
17.259
17.191
17.417
18.370
18.715
94.577
91.181
91.106
88.997
88.724
($
—
-3.328
-3.630
-6.691
-7.309
billion)
—
-.068
-§-.158
+1.111
+1.456
—
-3.386
-3.471
-5.580
-5.853
4-16
-------
Adams
Table 5. Aggregate U.S. Crop Production: Comparison of Base and GISS
Analyses 2 and 4
Crop
Unit
Base Analysis 2 % Change Analysis 4 % Change
Cotton
Corn
Soybeans
Wheat
Sorghum
Rice
Barley
Oats
Hay
Bales
Bushels
Bushels
Bushels
Bushels
Cwt.
Bushels
Bushels
Tons
(million units)
10.25 10.20
6,703.26 5,724.47
1,974.86 1,736.99
2,257.48 2,066.68
555.68 657.15
118.96 112.88
431.14
509.56
77.58
458.81
525.94
86.45
0
-15
-12
- 8
+19
- 5
+ 6
-l- 3
+11
(million units)
9.57 - 7
5,883.90 -12
1,743.02 -12
2,036.42 -10
437.30 -21
104.01 -13
427.51 - 1
492.83 - 3
78.55 + 1
4-17
-------
Adams
It should be stressed here that while the economic model includes excess demand relationships for the major
export commodities, changes in consumers' surplus in this assessment are generated only by changes in VS.
production. Hence, any changes in production in the rest of the world due to climate change are not explicitly
accounted for in the analysis. If production areas outside the UJS. experience yield declines of the magnitudes
suggested by the GISS or GFDL scenarios, then the losses in foreign consumers' surplus reported in this analysis
will understate such impacts.
In addition to changes in aggregate economic welfare (as measured by producers' and consumers' surplus),
changes in climate may impose differential effects on a regional basis. To assess projected regional effects of
long-term climate change, Table 6 presents a comparison of two regional indices, total land use and gross
revenue, for the base case and Analysis 4.
Analysis 4 is selected for reasons described above - it includes the most complete combination of crop
changes and adjustment processes. As can be seen from Table 2, some regions increase total land use devoted
to model crops, such as the Lake States, Corn Belt, Mountain, and Pacific states, while others, such as
Appalachia, the Delta, and Southern Plains, experience reductions. This northward and westward shift in land
use is due to the relative advantage of some regions with respect to the predicted crop yield, water supply, and
demand changes, and the increasing relative advantage of irrigated production. Overall, there is a small net
reduction in total cropped acreage. Similarly, gross revenues for some regions increase, while others experience
slight decreases. Revenue changes do not always directly mirror changes in acreage (or economic welfare)
because the mix of crops varies by region and because inelastic demands can mitigate against acreage reductions,
as evidenced by the Southern Plains, which has reduced acreage but increased gross revenue.
A final feature of the GISS analyses concerns changes in irrigated acreage. Changes in precipitation and
temperatures under doubled CO2 tend to favor irrigated crop production relative to dryland activities. Also, the
rising commodity prices that result from general reductions in total output enhance the feasibility of irrigation
activities, particularly those associated with groundwater use. As a result of these factors, there is an increase
in irrigated crop acreage in most regions of the model, as shown in Table 7. The largest increases occur in the
Northern (+29%) and Southern Plains (+28%). Overall, irrigated acreage increases by about 5 million acres
or 11% from 1981-83 levels. Such an expansion could only be accommodated with increased overdraft of some
aquifers and a large investment in irrigation capital. The long-term feasibility of such overdrafts to accommodate
irrigation is open to question unless gains in water use efficiency are achieved.
SENSITIVITY ANALYSES
The preceding four analyses represent midpoint or "best guess" assumptions concerning crop yields and
water demand and availability. The economic consequences of the midpoint assumptions underlying Analysis
4 imply a possible loss in social welfare of approximately $6 billion (1982 dollars). However, the errors inherent
in these assumptions are considerable. It is reasonable, then, to inquire as to how errors in these "best guess"
assumptions may affect the economic estimates. The stability of such midpoint economic analyses is typically
tested by sensitivity analyses. In this case, the sensitivity of the midpoint economic estimates resulting from
Analysis 4 is tested against more extreme changes in yield/water assumptions.
For this sensitivity analysis, we focus on uncertainties in (1) crop yield assumptions, and (2) water change
assumptions. Within each general category of uncertainty, some possible upper and lower limits can be defined
by looking at standard deviations or other measures of variability. Table 8 provides a description of four
sensitivity analyses that represent alternative crop yield, water demand, and water availability characterizations.
These four are obviously a small subset of possible combinations of these uncertainty factors. They do, however,
imply a wide range of climatic conditions and, hence, may "bound" the economic effects of yield and water
manifestations of climate change.
4-18
-------
Adams
Table 6. Regional Land Use and Gross Revenue
Baseline and GISS Analysis 4
Values; Comparison of
Land Use
Region
Base
Analysis 4
Change
(million acres)
Northeast
Lake States
Corn Belt
3.956
33.786
95.457
North Plains 101.684
Appalachia
Southeast
Delta States
Southern Plains
Mountain
Pacific
Total
15.583
12.513
19.876
54.709
21.667
9.671
368.902
2.331
34.840
97.259
101.592
14.096
11.512
17.677
42.609
22.739
11.427
35.6.142
-1.625
+1.054
+1 . 802
-0.092
-1.487
-1.001
-2.199
-12.100
+1.072
+1.756
-12.760
Revenue
Base Analysis 4
Change
($ billion)
0.928
7.430
23.636
9.534
4.291
3.541
4.705
9.543
6.559
3.891
74.053
0.745
8.149
23.073
10.208
4.086
3.057
4.511
12.925
6.632
5.110
78.496
-0.183
+0.719
-0.563
+0.674
-0.205
-0.484
-0.194
+3.382
+0.073
+1 . 219
+4.438
4-19
-------
Adams
Table 7. Regional Irrigated Acreage: Comparison of Base and GISS
Analysis 4
Region
Base Case
Acreage
GISS Analysis 4
Acreage
Change Percent
from Base Change
Northern Plains 10.389
Southeast 1.715
Delta 3.087
Southern Plains 5.318
Mountains 16.139
Pacific 7.738
Total 44.387
(millions)
13.395
1.968
3.597
6.888
15.629
7.941
49.419
+3.006
+ .253
+ .510
+1.520
- .510
+ .203
+5.032
+29
+15
+17
+28
- 3
+ 3
+11
4-20
-------
Adams
The results of the four sensitivity analyses are presented in Table 9. For comparison purposes, changes as
measured against both the base case and Analysis 4 are provided. The latter information implies a set of bounds
around the assumed midpoint estimates for crop yield and water demand changes.
A sensitivity analysis of primary importance here is the effect of uncertainty in the yield estimates. As
noted in Table 8, standard deviations associated with the yield estimates are derived from the CERES and
SOYGRO models. The derivation of these standard deviations is described by Rosenzweig in this volume; these
deviations are representative of the year-to-year variations in weather conditions. However, they do not
necessarily capture the uncertainties inherent in extrapolations to crops of regions other than those analyzed by
plant scientists. Table 9 presents the aggregate economic effects of these sensitivity evaluations. Sensitivity
Analysis 1 (SA.1) represents the lower bound on yield effects (reduced from yield changes in Analysis 4 by one
standard deviation). The aggregate effect of this change is a slight increase in economic welfare from the base
case, where producers' gains are about equal to consumers' losses. Producer gains arise from the interaction of
the mix of affected crops and the associated inelastic demands for those crops. Consumer losses are due to
increases in most commodity prices. SA.2 addresses the case of more extreme yield losses, where midpoint yield
reductions used in Analysis 4 are further reduced (by one standard deviation). This leads to a net annual loss
in aggregate economic welfare of over $12 billion in 1982 dollars. While producers still experience a moderate
gain, consumers' losses are substantial (in excess of $13 billion). The magnitude of yield reductions in this
analysis (exceeding 60% for some crops in some regions) causes major adjustments in crop production, including
declines of 50% for rice, 25% for soybeans, 22% for corn, and 13% for wheat. This decline in feed grains
triggers 10 to 15% reductions in the production of many livestock commodities, including poultry and pork. Over
half of the welfare losses accrue to the foreign sector, which is a large consumer of US. feed grains.
Water adjustment uncertainties are portrayed in SA3 and SA.4. The changes in water supplies and demands
were adjusted in both an optimistic (SA3) and pessimistic (SA.4) scenario. These were obtained by increasing
(decreasing) rainfall and decreasing (increasing) evaporation by 50% of the change from the base scenario for
the optimistic (pessimistic) scenario of each GISS grid point. The optimistic assumption results in a slight
reduction in the economic losses suggested by Analysis 4. The pessimistic water scenario increases the losses
observed in Analysis 4 by approximately $.6 billion. These relatively modest responses in economic estimates
to rather large changes in water assumptions imply that the effects of water demand and availability are not as
serious as the direct yield assumption. Regionally, however, these water adjustments portend some major
adjustments in land acreage, particularly in the Southern Plains, which potentially faces sharp reductions in total
acreage, but an increase in irrigated acreage. Western regions, particularly in the Pacific region, may substantially
increase irrigated acreage and share of total US. production.
EFFECTS OF CHANGES IN TECHNOLOGY AND WORLD FOOD DEMAND
The midpoint and sensitivity analyses presented above provide an impression of the potential economic
effects of long-term climate changes when those changes are imposed on present day (1980's) agriculture. The
use of a model calibrated to 1980's conditions to measure such long-term effects is required to keep the
assessment problem tractable. However, if CO2 continues to increase over the next 70 years with its associated
climate effects, then the structure of agriculture on which those effects are ultimately imposed will differ from
that portrayed in the economic model
It is useful, then, to consider what the estimated economic effects of climate change might be if some
aspects of the agricultural structure were to change. In this section we investigate two fundamental adjustments
likely to occur in agriculture over the next 70 years, technological change and changes in US. and world demand
for agricultural commodities. Technological change, as embedded in such practices as genetic improvements,
chemicals, fertilizers, and mechanical power, has historically enabled agriculture to produce more output from
the same or less land, labor, and other resources. World food demand increases steadily with population growth.
Both forces are likely to continue over the next 70 years.
4-21
-------
Adams
Table 8. Description of Alternative GISS Sensitivity Analysis Reflecting Crop
Yield and Water Demand Uncertainties
Sensitivity
Analysis Description Source of Data
SA.l Midpoint crop yield changes decreased by one Derived from CERES
standard deviation. Applied to Analysis 4 and SOYGRO estimates.
yield adjustments. Results in lower poten- Procedure for estima-
tial yield losses (and absolute gains for tion described in
some crops) than used in Analysis 4. Rosenzweig.
SA.2 Same as SA.l but yields increased by Same.
one standard deviation. Results in
greater potential yield reductions
than used in Analysis 4.
SA.3 An optimistic water availability analysis; GISS grid point
the changes in water supply are increased ratios for evap-
by 50% and in water demand are decreased oration and pre-
by 50% from GISS grid point ratio estimates. cipitation.
SA.4 A pessimistic water availability analysis; Same
the changes in water supply are decreased
by 50% and in water demand are increased
by 50% from GISS grid point ratio estimates.
4-22
-------
Adams
Table 9. Sensitivity Analyses of GISS Aggregate Economic Effects
Sensitivity
Analysis
Analysis 4
SA.l
SA.2
SA.3
SA.4
Chanee in
Consumers
- 7.309
- 1.636
-13.048
- 6.154
- 7.765
JSconom^C. Surplus from RflcA
Producers
($ billion)
+1.456
+1.910
+ .883
+ .877
+1.274
Total
- 5.853
+ .276
-12.165
- 5.277
- 6.490
Change in Total Surplus
from Analysis 4
($ billion)
—
+6 . 129
-6.312
+ .576
- .637
4-23
-------
Adams
To examine how changes in technology and/or world food demand may alter the economic effects of climate
change, we construct four additional analyses. The first two focus only on technology. Of these, one analysis
represents potential agricultural yields in the year 2060 in the absence of climate change effects. The other
analysis combines the technologically induced increases in yields with the climate change yield reductions used
in Analysis 4. Comparing these two analyses provides a measure of the potential economic loss due to climate
change. These values are then compared with the 1981-1983 baseline case to estimate absolute changes in social
welfare.
The two food demand projection analyses include both changes in technology and food demand. The first
represents agriculture in 2060 under increased food demand in the absence of climate change. The second food
demand analysis includes changes in climate, as used in Analysis 4, imposed on agriculture under elevated levels
of technology and demand. The specific procedures used to develop the technological change and world demand
analyses are defined below.
Technology Assumptions
Projecting crop yield 70 years into the future is a speculative enterprise. For example, it is believed that
most yield gains observed during the 1955-1987 period occurred because of the rapid increase in pesticides and
fertilizer utilization in the 1950s and 1960s, coupled with an increase in irrigatioa Increases in fertilizer and
pesticide use and in irrigation acreage are unlikely to be major determinants of yield in the future changes. On
the other hand, the potential yield enhancements from biotechnology may replicate the average gains experienced
over the last three decades.
The projected yield changes were estimated with data on yields for the period 1955 through 1987 (USDA,
1987). Specifically, yield projections were obtained for each crop using the general Box-Cox functional form (Box
and Cox, 1964) to transform y'= f(y,A) = (y*-l)/A. This transformation contains the linear and log-linear
specifications as special cases (A = 1 and 0, respectively) without imposing it. Transformed yields were regressed
on time, solving for the optimal A's using maximum likelihood techniques. The resultant yield adjustments are
somewhat more regular across crops than using a standard log-linear regression model; they range from 41.2%
(0.72%/annum) for cotton to 128.8% (0.72%/annum) for corn. These estimates also were used to project yields
for each crop through 2060.
Demand Assumptions
Population is a major factor influencing the demand for agricultural commodities. Since the economic
model includes both domestic and foreign consumption, increases in U.S. and world population to 2060 were used
to shift domestic and export demands. A projected US. population increase of 42% was used to alter domestic
demands while a projected world population increase of 114% was used for exports (Merrick). Specifically, the
demand curves for all crops in the model were shifted equally to reflect changes in aggregate demand caused by
population increases. Elasticities were not changed (owing to a lack of information on which to base such
changes), but the price and quantity points that the demand curves pass through were increased. Ideally, crop
models in combination with projected climatic variables (as used in this chapter) could be used to model foreign
supply and demand, but such information is currently lacking.
Clearly, these adjustments are gross approximations. For example, foreign yields will be affected by changes
in technology, cultural practices, and economic organization. In addition, since CO2 increases are a worldwide
phenomenon, climate change will affect foreign as well as domestic producers. Changes in income and tastes
will also affect demands. Finally, one should note the changes which have occurred in the last 50-75 years to the
composition of demand and supply in the UJS. resulting in differential demand changes across crops. For
example, oat production decreased as horsepower was replaced with mechanical power. If similar changes (such
as fuel production from corn) occur in the next 70 years, then the uniform demand shifts in this analysis are
questionable.
4-24
-------
Adams
Economic Consequences
The aggregate economic effects of changing technology and world food demand in the context of this climate
change assessment are represented in Table 10. The base case and Analysis 4 are provided for purposes of
comparison. As the numbers indicate, changing technology (yields) has the potential to sharply change the
economic consequences predicted in the model. For example, if crop technology is assumed to continue to evolve
in the pattern of the past 30 years, the net social value of agricultural output per annum will rise over 34%, or
by over $32 billion. All of this gain accrues to consumers. When climate change is imposed on this "new"
agricultural setting, the effect is approximately a $2.1 billion annual loss in potential agricultural production.
However, the level of agricultural surplus win still be approximately $30 billion more than in 1983. Thus,
technology appears to have a much larger impact on agricultural production than the climate changes predicted
by the GISS model, though losses due to these changes still occur.
The last two analyses in Table 10 address the interactions of technological change and food demand
assumptions. That is, both technological change and increased demand are imposed on the economic model.
Such a combination gives rise to a new "base case" against which the effects of climate change (as imposed on
that same new "base case") can be measured. The total economic surplus values in these demand analyses should
not be compared with the total surplus values in the Base or Analysis 4 case, given the shift hi demand curves
(i.e., the interval of integration is not constant across analyses). The important point of comparison is the last
column, which records the difference between the new base and the climate change analysis. The effect of
climate change under the new demand level is approximately a $6.8 billion economic loss, which consists of $4.7
billion of gains to producers and $115 billion in consumers' losses, 63% in the foreign sector. The large foreign
component is a function of the large projected increase in world population. Hence, increases in U.S. and world
food demand increase the potential adverse economic effects of climate change.
GFDL RESULTS
The GFDL climate model predicts greater temperature and precipitation changes for a doubling of CO-
than does the GISS model. Thus, it is not surprising that the associated yield changes predicted by the CERES
and SOYGRO models are also greater (up to an 80% reduction - substantially larger yield reductions than
were obtained from the GISS climate forecasts). This section reports the economic consequences of those
GFDL-induced yield and water changes. The same set of analyses performed on the GISS data is repeated for
GFDL.
The first set of GFDL analyses follows the four analyses described in Table 3, i.e., various combinations
of crops and water assumptions: these are presented in Table 11. As expected, the aggregate economic
consequences associated with each analysis is greater than for GISS. For each case, there is about a fivefold
increase in economic losses compared with GISS. Analysis 4 results in annual economic losses of almost $34
billion in 1982 dollars. This is approximately 30% of the 1981-83 gross value of crop and livestock products in
the UJS. Even the three crop-no water adjustment analysis (No. 1) results in losses exceeding $24 billion or about
20% of the value of crops and livestock products. Thus, if a doubling of CO2 were to result in climate changes
as portrayed by the GFDL model, the adjustment costs to agriculture would be substantially greater than implied
by GISS.
These relatively large economic losses are the consequence of some rather severe changes in crop
production. These are depicted in Table 12 for Analysis 4, along with the base levels and those for the
comparable GISS case. As is evident, the reductions under GFDL for corn, wheat, and soybeans are from 80
to 300% greater than under the GISS Analysis 4 case. These large reductions in production are responsible for
the large consumer losses and producer gains noted in Table 11.
4-25
-------
Adams
Table 10. Economic Effects of GISS Climate Change with Increased Technology
and Food Demand, in 1982 Dollars
Analyses
Technological
Change to 2060
Technological
Change to 2060
with Climate
Change
Total Economic Change in Surplus
Surplus from 1981-83 Base
Change in Surplus
Due to Climate Change
Base
Analysis 4
94.577
88.724
($ billion)
—
-5.853
-5.853
126.987
124.854
+32.410
+30.277
-2.133
Demand and
Technological
Change to 2060
Demand and
Technological
Change to 2060
with Climate
Change
191.041
184.256
+96.464
+89.679
-6.785
4-26
-------
Adams
The effects of GFDL climate changes on crop acreage for Analysis 4, along with base and comparable GISS
results, are presented in Table 13. Figure 3 provides a graphical display of regional land uses changes relative
to the base. In total, both analyses show reductions in total cropped acreage (about 11 million acres or 3% for
GISS and 7 million acres or 2% for GFDL). However, the direction of acreage shifts is somewhat different.
The general shift for GFDL is slightly more northwesterly than for GISS. Specifically, under GFDL, the
Northern Plains experience a major increase in acreage (6%), along with the same general increases in the Pacific
(20%) and Mountain (10%) areas that were noted in the GISS case. Unlike the GISS case, however, Corn Belt
acreage declines (by 6%). Irrigated acreage expands even more in the GFDL case, by 18 million acres or 40%
from 1981-1983 levels, due to the even greater relative advantage of irrigated crop yields (vis-a-vis the base and
GISS) compared to dryland and the resulting rise in commodity prices that makes increased groundwater
pumping feasible. As in the GISS case, however, it is questionable whether the overdrafting involved hi this
adjustment could be sustained. Note, however, the impact of the hydrologic assumptions (Analysis 3 vs. 4) is
quite small.
Technology and Demand Assumptions
Inclusion of potential changes in technology and demand in the GISS analyses had a substantial effect
on possible economic consequences of climate change. These same technology and demand assumptions were
also imposed on the GFDL analyses. Specifically, Analysis 4 results derived from GFDL climate change
(reported in Table 11) were resolved, allowing for changes in either technology or food demand plus technology.
The results of these simulations are presented in Table 14.
The effects of technological change (i.e., a continued increase in crop yields over the next 70 years)
increase the base economic level in the model, as expected. When the adverse yield consequences of climate
change are combined with the positive yield adjustments associated with the technology assumption, the
resultant level of aggregate economic welfare is still greater than the 1981-1983 base value (by $12 billion).
However, the climate change effects reduce the potential increase by over $20 billion per annum. Put another
way, the adverse effects of climate change are approximately equal to almost 50 years of technological change.
Increases in US. and world population will increase the demand for food over the next 70 years. Within
the context of the economic model, such increases in demand may be viewed as increasing the value of food
to consumers. As a result, the level of aggregate economic activity in the model increases dramatically when
demand and technology are changed in the model. The effects of climate change hi this new characterization
of demand and yields are noted hi the last analysis reported in Table 14. In this setting, climate change imposes
an economic cost of over $44 billion in 1982 dollars. Thus, increasing food demand will aggravate the economic
losses of climate change as recorded under the no technology-no demand case (Analysis 4).
As discussed earlier, the technology-demand assumptions used here are at best gross approximations of
these factors. Their inclusion, however, does point to the potential importance of these factors on the
magnitude of the climate change effects derived from this (or any) economic model. The results also show that
technological change has the potential to mitigate against adverse consequences of environmental stress, like
long-term climate adjustments. Climate change does, however, reduce the overall potential welfare of society.
Finally, increases in demand will increase the potential costs of adverse climate change.
Direct Effects of CCv on Crop Yields
The GISS and GFDL climate changes evaluated in this report are driven by an assumed doubling of COg
According to the CERES and SOYGRO model simulations, these climate changes will result in a general
depression of crop yields of up to 80%. However, chamber studies with some field crops have documented that
increases in CO2 enhance plant growth and crop productivity. This suggests that the GISS and GFDL crop
yield adjustments, which do not include this direct yield-enhancing CO2 effect, may be overstating the adverse
yield consequences of the doubling of CO2>
4-27
-------
Adams
Table 11.
Aggregate Economic Effects of GFDL 2xC02 Global Climate
Change on U.S. Agriculture, in 1982 dollars
Yield/Water
Assumption
Analysis 1*
Analysis 2
Analysis 3
Analysis 4
Change
Consumers
-28.720
-29.234
-37.203
-37.461
ss in Economic Sun
Producers
($ billion)
+4.259
44.717
+3 . 845
+3.863
plus
Total
-24.462
-24.518
-33.358
-33.599
Analyses correspond to the definitions provided in Table 3.
4-28
-------
Adams
Table 12. Aggregate U.S. Crop Production: A comparison of Base, GISS and
GFDL Analyses 4
Crop
Unit
Base
GISS % Change
GFDL % Change
Cotton
Corn
Soybeans
Wheat
Sorghum
Rice
Barley
Oats
Hay
Bales
Bushels
Bushels
Bushels
Bushels
Cwt
Bushels
Bushels
Tons
10.25
6,703.26
1,974.86
2,257.48
555.68
118.96
431.14
509.96
77.58
9.57
5,883.90
1,743.02
2,036.42
437.30
103.011
427.51
492.83
78.57
- 7
-12
-12
-10
-21
- 7
- 1
- 3
+ 1
9.13
3,496.78
931.26
1,850.94
528.71
66.55
290.09
383.82
50.76
-11
-47
-53
-18
- 5
-44
-32
-25
-35
4-29
-------
Adams
Table 13. Regional Land Use: A Comparison of Base, GISS and
GFDL Analyses 4
Region
Northeast
Lake States
Corn Belt
Northern Plains
Appalachia
Southeast
Delta States
Southern Plains
Mountain
Pacific
Total
GISS
Total Change
Land from
Base Use Base
(mil.
3.956
33.786
95.457
101.684
15.583
12.513
19.876
54.709
21.667
9.671
368.901
acres)
2.331
34.840
97.259
101.592
14.096
11.512
17.677
42.609
22.739
11.427
357.083
(%)
-40
+ 3
+ 2
0
-10
- 8
-11
-22
+ 5
+19
- 4
GFDL
Total Change
Land from
Use Base
(mil. acres)
3.891
33.654
89.560
108.065
-12.882
-11.174
-16.178
51.392
23.834
11.568
358.657
(%)
- 2
0
- 6
+ 6
-18
-11
-19
- 6
+10
+20
- 2
4-30
-------
Adams
o
p
>—'
I
&
o
0)
en
o
O
CD
-40
NEst Appl Corn SPIn Mtns
SEst Lake Delta NPIn Pacf
''////////A
GISS
9999991
GFDL
Figure 3. Change in regional land use for Analysis 4, by climate forecast.
4-31
-------
Adams
Given the potential importance of a direct CO2 effect on this preliminary analysis of global climate change,
an additional set of CERES and SOYGRO analyses was performed that included both climate change and
COj-induced yield adjustments. The procedures by which these direct effects were included in these models
are described in Rosenzweig in this volume. The result of the CO2 addition was to dramatically alter the yield
consequences as originally predicted by the plant models and as used in the preceding analyses. Specifically,
for GISS, some crops now realized actual increases in crop yields under the combined climate change-CO2
effect. For GFDL, the large yield reductions reported earlier were moderated to levels more comparable to
the non-CO2-adjusted GISS yield changes, i.e., from 70 to 80% yield reductions to losses of 20 to 30%. Such
upward adjustments in yields would be expected to dampen or even offset the adverse economic consequences
reported in Tables 4 through 14.
The economic consequences of combined climate change and CO2 effects are evaluated for both the GISS
and GFDL GCM forecasts. These are reported in Table 15. In addition to the Analyses 4 "midpoint"
estimates, Table 15 also presents sensitivity analyses based on yield adjustments of plus or minus one standard
deviation around the midpoint case (SA.1 and SA.2, Table 8). These latter evaluations provide an indication
of the importance of uncertainties in the CO2 yield adjustments.
As expected, the inclusion of direct CO2 effects alters the magnitude (and direction for GISS) of the
earlier economic estimates. Specifically, the combined effects of climate change and CO2 now result in a net
increase in economic welfare under the GISS evaluations. For the Analysis 4 midpoint case, the previous loss
of approximately $6 billion per annum in 1982 dollars is now a $10 billion gain. Both sensitivity analyses around
the GISS midpoint also indicate net gains under the doubled CO2 environment. For GFDL, the direct effect
of CO2 is to reduce, but not totally offset, the economic losses measured in the climate change-only analyses.
In this case, the previous loss estimate is reduced from $33 billion to approximately $10 billion. Under both
sensitivity runs, the net economic consequences are still negative.
Although the inclusion of CO2 adjustments alter the nature and magnitude of aggregate economic
consequences, regional shifts in agricultural cropping activity still occur. Table 16 presents the regional acreage
adjustments for GISS and GFDL Analyses 4 under the CO-effect. Figure 4 displays regional land use changes
for GISS and GFDL with and without the direct effects of CO2. For the GISS analysis, slight increases are
observed in the Pacific and Lake States regions (8 and 1%, respectively). The Corn Belt and Northern Plains
are relatively unaffected. However, rather severe reductions in acreage are noted in Appalachia (-80%), the
Delta (-53%), and the Southeast (-30%). A somewhat similar pattern is observed under GFDL, where the
Pacific and Lake states regions each experience increases of 12%. The most severe reductions again occur in
the Appalachia (-51%), Southeast (-35%), and Delta (-13%) regions. In relative terms, the combined effects
of climate change and CO2 on regional crop acreage are consistent with the previous regional adjustments,
namely a northward and westward shift in crop production. A similar consistency occurs in irrigated acreage
response under the combined climate change-direct CO, effects. That is, an expansion of irrigated acreage over
1981-1983 levels is observed under both GISS (2 million acres or 5%) and GFDL (17 million acres or 38%).
Figure 5 presents changes in irrigated acreage under the GISS and GFDL cases, with and without direct CO2
effects. While somewhat less than for the non-CO, analyses, the GFDL adjustments still suggest a major
expansion of irrigated agricultural activity in the U.ST
These findings suggest the importance of the CO2 adjustments as well as the specific GCM used in the
assessment. If CO, has positive effects on yields of the magnitude modeled here, then the aggregate production
consequences of climate change may not be as severe as presented earlier in this report. However, if the
doubling of CO2 results in climate changes similar to those as suggested by GFDL, then even the mitigating
effects of CO2 are insufficient to prevent the occurrence of substantial losses in economic welfare.
4-32
-------
Adams
Table 14. Economic Effects of GFDL Climate Change with Increased
Technology and Food Demand, in 1982 Dollars
Change in Economic
Change in Economic Surplus Due to Climate
Analysis Surplus from Base Change
Analysis 4 -33.599 -33.599
Technological Change
to 2060 +32.822
Technological Change
to 2060 with Climate
Change +12.008 -20.814
Demand and Techno-
logical Change to
2060 +133.815
Demand and Techno-
logical Change to
2060 with Climate
Change +89.227 -44.588
4-33
-------
Adams
Table 15. Combined Effects of Direct C02 and Climate Change on Agriculture:
GISS and GFDL Analyses 4
Model/Assumption
GISS Analyses 4:
without C02
GISS Analyses 4:
with C02
a) Plus One Standard
Deviation in Yields
a) Minus One Standard
Deviation in Yields
GFDL Analyses 4:
without COj
GDDL Analyses 4:
with C02
a) Plus One Standard
Deviation in Yields
b) Minus One Standard
Deviation in Yields
Change
Consumer
- 7.309
9.354
14.199
3.338
-37.461
-10.291
- 2.649
-17.343
in Economic Surplus
Producer
1.456
1.291
.598
2.321
+3.863
.607
- .148
- .488
(from Base)
Total
- 5.853
10.646
14.797
5.659
-33.599
- 9.683
- 2.797
-17.832
4-34
-------
Adams
Table 16. Regional
Analyses
Land Use: A Comparison of Base, GISS and GFDL
4 with C02 Direct Effects
Region
Base
GISS
GFDL
Total Change Total Change
Land Use (%) Land Use (%)
(million acres)
Northeast
Lake States
Corn Belt
North Plains
Appalachia
Southeast
Delta States
Southern Plains
Mountain States
Pacific States
Total
3
33
95
101
, 15
12
19
54
21
9
368
.956
.786
.457
.684
.583
.513
.876
.709
.667
.671
.902
0.
33.
96.
100.
2.
8.
9.
43.
20.
10.
324.
038
923
985
560
818
665
275
813
439
491
505
(million
-49
+ 1
- 1
- 1
-80
-30
-53
-20
- 6
+ 8
-12
acres)
3.434
37.
89.
99.
7.
7.
16.
52.
21.
10.
346.
188
531
543
349
832
701
213
422
881
134
-13
+10
- 6
- 2
-53
-37
-16
- 5
- 1
+13
- 6
4-35
-------
Adams
120-j
110-
.
1 nn
I UU
-
m QO-
Qj CJVJ
in
r—\
m 80-
ill
*+— ~7n
0 70~
-
"£ 60-
CD
f \ r— *-*.
^ 50-
_
I
*
>
j;
-
„
-
^
-
/
/
\
\
V
H
«
-
«
"
«
U
"
H
5-
H
5
*•
j
^
:
;
•
H
^
_
«
0,
M
_
-
/
V
^
3c
M
M
M
"
^
;
~
;
-
-
M
5
»•
5
<
3
^
H
M
"
^
;
M
M
M
<
P
-
_
„
_
••
„
—
>.
••
V
^
3c
n
2
%
-
tol
H
M
M
M
M
-
•«
M
M
55
•-
*•
n
«
toa
5
tol
M
/
H* V« >5. V0x ^o O>
'* '%, ^ ^x \ ^ %
"d1
•*J»
- - Base
GISS
GISS+DE
GFDL
GFDL+DE
Figure 4. Changes in regional land use: Analysis 4 with and without direct CO2 effects.
4-36
-------
Adams
20 n
16-
OT* 12-1
b
8-
4-
0
X X
x
Sx
x
X
pqx
X
X X
x
x
X
%-
'o
Base
GISS
GISS+DE
GFDL
GFDL+DE
Figure 5. Regional irrigated acreage: Analysis 4 with and without direct CO2 effects.
4-37
-------
Adams
CHAPTER 4
IMPLICATIONS AND CONCLUSIONS
The results of the various analyses performed here suggest a range of possible economic effects associated
with climate change in agricultural production areas of the VS. The diverse set of assumptions explored in
these analyses (different GCMs, different crop and water assumptions) reflect uncertainties inherent in
predicting long-term changes in biological and economic phenomena. This, in turn, gives rise to a set of results
that can tell many stories. This section attempts to interpret those results with respect to common themes that
emerge. Implications of these changes are also drawn.
As a starting point, one should put the estimated aggregate economic effects into perspective. These
aggregate consequences of climate change vary substantially across analyses. Midpoint or Analyses 4 results
(in the absence of technology or CO, effects) range from approximately 6 to 30 billion 1982 dollars on an
annual basis. On a per capita basis, these net surplus losses range from $6 (GISS) to $65 (GFDL) per VS.
citizen. The larger relative losses per capita for GFDL are due to a higher proportion of consumer losses
falling on domestic consumers. As noted in the previous discussion, the aggregate loss estimates are from about
5 to 28% of the 1982 value of crop and livestock commodities produced in the VS.
To put these estimates into a somewhat different perspective, they are also compared to economic effects
of some other environmental stresses. For example, the economic consequences of tropospheric ozone on U.S.
agriculture are estimated to be about $2 to 3 billion per year in 1982 dollars (Adams et al., 1988). Similarly,
the effects of a 15% depletion in the stratospheric ozone column are estimated at $2.5 billion in 1982 dollars.
Thus, the effects of a doubling of CO2 imply economic costs 2 to 10 times greater than some other
environmental stresses. With the inclusion of technology assumptions or yield enhancements from COg, the
magnitude of economic effects is greatly reduced (e.g., to an economic loss about four times greater than the
other stresses for the GFDL analysis).
An important policy concern in addressing a major environmental adjustment, such as climate change,
is whether that adjustment threatens the food and fiber base of a society. The results of the analyses reported
here, even in the most extreme cases (e.g., GISS SA.2 or GFDL Analysis 4), indicate that the productive
capacity of agriculture may be reduced by climate change but not to a level that implies any major disruptions
to the supply of basic commodities. Domestically, consumers would face slightly to moderately higher prices
under some analyses but supplies would be adequate to meet current and projected domestic demand. Exports,
however, do experience a major reduction (up to 70% for most exported commodities). Changes in total world
food production due to climate change will influence the net effects of these VS. production changes on the
welfare of foreign consumers and producers. Allowing for technological change, or a yield enhancing effect of
COj, the productive capacity of agriculture will likely be greater in 70 years than it is today, even in the
presence of climate change, offsetting most or all of the adverse climate effects. For example, under GISS, both
technology and CO2 direct effects appear capable of offsetting climatic effects. For a GFDL-type of climate
change, the picture is not so comforting. Midpoint losses without technology approach 30% of current
agricultural value. While technology can offset these losses, continued and substantial improvements in yields
are required to realize such an outcome. However, even without major technological gains, it appears the U.S.
could still meet domestic needs but with little residual for exports. If the rest of the world experiences similar
yield reductions, then the welfare effects on major food importers could be severe.
One relatively unambiguous finding in this assessment is that shifts in U.S. agricultural production patterns
are highly likely. Specifically, all analyses show a north or northwest shift in production of major commodities
such as wheat, corn, and soybeans. This has implications for regional economies, with major changes in the
capital structure of agriculture and likely increases in input demands for areas of expanded crop acreage and
corresponding reductions in regions experiencing acreage declines. The changes in capital requirements will
be of particular importance if irrigated acreage expands as predicted in these analyses. For many rural
4-38
-------
Adams
communities, this may further weaken an economic base already under pressure from long-term structural
changes under way in US. agriculture.
Shifts in crop production also imply demands or pressure on environmental and natural resources,
including water quantity and quality, wetlands, soil, fish and wildlife, and other resources. For example, a
northward shift in corn and soybean production (through the Dakotas to southwestern Canada) may exacerbate
the loss of critical prairie wetlands by making drainage and conversion to crop production more profitable. A
westward shift may increase wind and water erosion of fragile soils. The substantial (2 to 18 million acres)
increase in irrigated acreage suggested in all analyses enhances the likelihood of ground and surface water
pollution. Recent evidence concerning selenium poisoning in California indicates that long-term irrigation poses
potential environmental problems. Obtaining water for increased irrigation also implies more and larger
reservoirs, which in turn implies greater pressure to develop remaining wild or scenic rivers. Groundwater
overdrafts would likely be required to accomplish this expansion. The current analyses do not address the issues
of whether the physical and institutional changes required to accommodate such an increase in irrigated acreage
are feasible.
Overall, the analyses reported here indicate possible directions of economic effects arising from climate
change. The directions of these economic changes are generally consistent with expectations. The relative
magnitudes of the economic estimates, which vary substantially between GISS and GFDL, may also suggest
whether the agricultural consequences of climate change are important from a policy perspective. Both GCMs
imply adjustments within agriculture, with GFDL implying some major adjustment problems, particularly for
consumers and specific regions and resources. Another purpose of this preliminary assessment, however, is to
indicate areas for future research; for example, a large potential effect may merit additional research to confirm
such a finding.
Given the crudeness of the data used to develop these analyses, and the wide range of results elicited
from alternative treatment of these data, a longer term research agenda seems warranted. An area in need of
improvement is the quality of the crop yield data used here. The extent of crop coverage, both genetically and
geographically, needs expansion. Only three crops are modeled in this current agricultural effects program and
then for only a limited number of sites or regions. Possible changes in genetic materials (i.e., cultivars) may
also be as important as climate changes. Such cultivar switching was not included in the plant science modeling,
suggesting that the yield effects may be overstated. Finally, the belated inclusion of CO2 effects in the CERES
and SOYGRO analyses, while a worthwhile effort in terms of providing a more complete treatment of the likely
consequences of climate change on agriculture, needs further refinement. The availability and adaptability of
the CERES and SOYGRO models could be further exploited within a relatively short period of time to improve
the quality of the crop yield estimates.
Measures of the hydrologic consequences of climate change require much more refinement. While the
analyses in this assessment attempted to incorporate some of these potential effects, it is not clear that the signs
of the adjustments are correct, let alone the magnitudes. Cooperative research involving hydrologists,
meteorologists, agriculturalists, economists, lawyers, and others is needed to define how climate change will
influence the long-term physical and institutional arrangements concerning irrigation.
Economic models in general do not forecast well when projecting over long time periods. The economic
model used here is a comparative static, spatial equilibrium model keyed to the 1980s. It is thus not intended
to address some of the dynamic or long-term effects inherent in climate change over 70 years. For example,
each region in the model uses 1980s cropping mixes and cultivars, which understates potential mitigation
adjustments in the long term. This implies that the economic effects may be biased upwards. Also, while we
attempted to include two important forces in shaping agriculture, changing technology and demand, the accuracy
of these adjustments is questionable. These specific adjustments can be refined but will not address the
problem of using models with fixed production (i.e., technological) relationships to measure long-term effects.
Improvements in model capabilities are possible, however, if one is willing to sacrifice the level of detail in the
assessment. The tradeoff between more accurate modeling of dynamic processes and loss of detail needs to
be weighed in terms of how the economic input may affect the regulatory process.
4-39
-------
Adams
Finally, the quality of the GCM forecasts that drive all of the above is critical. While the models are
"state of the art," the quality of the forecasts appears to deteriorate from the macro to the micro scale (grid box
level), where predictions seem highly variable within each model. Across the two GCMs, great differences in
climatic consequences are observed, with associated differences in economic effects. Therefore, improvements
in GCM performance appear to be a high priority, particularly if agencies are going to use forecasts of such
models to assess impacts on a relatively small geographical scale, such as at the state or regional level, as in
this economic analysis.
4-40
-------
Adams
REFERENCES
Adams, R.M. and BA. McCarl, 1985. Assessing the benefits of alternative oxidant standards on agriculture:
The role of response information. J. Environ. Econ. Management, 12, 264-276.
Adams, R.M., J.D. Glyer and BA. McCarl. 1988. The NCIAN economic assessment: Approach, findings and
implications. Chapter 20, in Assessment of Crop Loss from Air Pollutants. W.W. Heck, D.T. Tingey and O.C.
Taylor, eds. Elsevier Applied Science Publishers, (in press)
Adams, R.M., SA. Hamilton, and BA. McCarl, 1984. The economic effects of ozone on agriculture. EPA-
60013-84-090. US. EPA. October.
Box, G.E.P. and D.R. Cox, 1964. An analysis of transformations, J.Roy. Statist. Soc. Ser. B26, 211-243.
Callaway, J.M., FJ. Cronin, J.W. Currie, and J. Tawil, 1982. An analysis of methods and models for assessing
the direct and indirect economic impacts of CO2-induced environmental changes in the agricultural sector of the
VS. economy. PNL-4384, UC-11. Pacific Northwest Laboratory, Richland, Washington.
Chang, C.C. and BA. McCarl, 1988. The Agricultural Sector Model. Working Paper, Department of
Agricultural Economics, Texas A and M University, College Station.
Decker, W.L., V.K. Jones, and R. Achutuni, 1986. The impact of climate change from increased atmospheric
carbon dioxide on American agriculture. US. Department of Energy, DOE/NBB-0077, Washington, D.C.
Heady, E.O., and U.K. Srivistava. 1975. Spatial Sector Programming Models in Agriculture. Ames, Iowa: Iowa
State University Press.
Just, R.E., D.L. Hueth, and A. Schmitz, 1981 Applied Welfare Economics and Public Policy. New York:
Prentice-Hall.
Kokoski, M.F. and V.K. Smith, 1987. A general equilibrium analysis of partial equilibrium welfare measures:
The case of climate change. Am. Econ. Rev. 77:331-341.
Kopp, RJ., WJ. Vaughn, M. Hazilla, and R. Carson, 1985. Implications of environmental policy for US.
agriculture: The case of ambient ozone standards. J. of Environmental Management, 20, 321-331.
Manabe, S. and R.T. Wetherald, 1980. On the distribution of climate change resulting from an increase in CO2
content of the atmosphere. J. Atmos. Sci., 37, 99-118.
McCarl, B A. 1982. "Cropping activities in Agricultural Sector Models: A Methodological Proposal" American
Journal of Agricultural Economics. 62:87-102.
McCarl, BA. and T. Spreen, 1980. Price endogenous mathematical programming as a tool for sector analysis.
Am. J. Agric. Econ., 62, 87-95.
Merrick, T.W., 1986. World population in transition. Population Bulletin. Vol. 42.
Murtaugh, B., and M. Saunders. 1977. MINOS: Users Guide. Operations Research Technical Report, No. 77-
9. Systems Operations Laboratory, Stanford University.
Parry, M.L. and T.R. Carter, 1985. The effect of climatic variations on agricultural risk. Climatic Change, 7,95-
100.
4-41
-------
Adams
Rosenzweig, C., 1986. Effects on agriculture. Chapter III in Potential effects of future climate changes on forests
and vegetation, agriculture, water resources and human health. DA. Tirpak (ed.), USEPA Draft report.
October.
Rosenzweig, C., 1985. Potential CO2-induced climate effects on North American wheat-producing regions.
Climatic Change, 7, 367-389.
Sahi, R., and W.C. Craddock. 1974. "Estimation of Flexibility Coefficients for Recursive Programming Models:
Alternative Approaches." American Journal of Agricultural Economics 56:344-350.
Samuelson, PA., 1952. Spatial price equilibrium and linear programming. Am. Econ. Review, 42, 283-303.
Takayama, T. and G. Judge, 1971. Spatial and temporal price and allocation models. North Holland Publishing
Company, Amsterdam.
Thompson, L.M., 1975. Weather variability, climatic change, and grain production. Science, 188, 535-541.
USDA. U.S. Department of Agriculture. 1987. Agricultural Statistics, 1987. Washington, DC: VS. Government
Printing Office.
Waggoner, P.E., 1986. How changed weather might change American agriculture. Paper delivered at UNEP
and EPA International Conference on Health and Environmental Effects of Ozone Modification and Climate
Change, 16-20, June, 1986.
Wigley, T.M.L., M J. Ingram, and G. Farmer (eds.), 1981. Climate and history: Studies in past climates and their
impact on man. Cambridge University Press, Cambridge, Mass.
Willig, R.D. 1976. "Consumers' Surplus Without Apology." American Economic Review 66:589-597.
4-42
-------
Adams
APPENDIX A
HYDROLOGIC CHARACTERISTICS AND ASSUMPTIONS
In the six regions west of the Mississippi River, runoff water is typically routed over hundreds to
thousands of km. The mountainous regions of the western states supply most of the irrigation water used in
arable land. In particular, changes which take place in the vast chain of the Rocky Mountains influence water
supply more than do changes in local precipitation.
In the GISS scenario, regions east of the Mississippi are likely to experience a slight to significant
increase in water available for irrigation. The Southern Plains and California Region are both expected to feel
a significant decrease in water available from within the region, while the Northern Plains and Southern
Mountain Regions are likely to have no change to a slight decrease in available water. The Northern Mountain
and Northwest Regions are assumed to have slight to significant increases in available water.
The climate changes modeled by GFDL indicate a warmer climate with less rainfall east of the Rocky
Mountains, more evaporation and switch in precipitation from summer to winter. Because of the wanner
temperatures, less of the winter precipitation maybe available during the growing season. As a result, water
available for irrigation decreases in all areas except California.
Northwest Region
This region currently has the highest annual precipitation of any of the regions. A majority of the
precipitation falls as rain, but the snowpack accumulation above 900m elevation is a crucial source of irrigation
for the region. Large reservoirs on the Columbia and Snake Rivers receive most of their water from snowmelt
from the Rocky Mountains in Idaho, Montana and from as much as 350 km into Canada Irrigation is primarily
from surface water stored in reservoirs, but some groundwater is pumped from shallow aquifers.
GISS
The expected increase in summer temperature of 4.2°C is shown to be accompanied by an increase in
annual precipitation of 23 percent. The additional water is likely to make it possible to extend the irrigation
season to correspond with the longer growing season. As some of the increased precipitation will occur in the
growing season, drawdown of reservoirs will not be significantly affected.
There will be an increase in the proportion of precipitation as rain and the area covered by a snowpack
accumulation will be reduced. This means that the reservoirs will begin filling earlier in the winter and
streamflow should not be as drastically reduced in summer. Groundwater aquifers should be fully recharged as
a result of the increased summer and winter precipitation. Assuming reservoir capacity is sufficient, additional
water is likely to be made available for irrigation. Specifically, we assume that the new equili- brium level will
increase by approximately 7 percent. This is the largest increase in runoff projected for any area in the analysis.
Water demand will be reduced by approximately the same percent.
GFDL
The modest increase in rainfall of 1.7% will not be sufficient to offset the large (10%) increase in
evaporation. Some increase in runoff from the Rockies and an earlier start of the growing season may enable
better use of the remaining water, especially if the earlier runoff can be captured in the Columbia-Snake reservoir
system. Otherwise, the drawdown on the reservoirs may be rather severe, increasing pumping costs and
decreasing irrigated acreage.
4-43
-------
Adams
California Region
The major agricultural activity in California is located in the southern half of the state. These irrigated
lands typically receive little precipitation. Statewide, rainfall ranges from as little as 250mm to about 1200mm.
A 70- to 100-day dry season is common. Water for irrigation is supplied by reservoirs in the northern, wetter
part of the state, fed by snowmelt from large snow accumulations in the Sierra Nevada, Siskiyou and Klamath
Mountains. The Colorado River is also a major source of water for the southern desert region of Imperial
County which gains its flow from snowmelt in the Rocky Mountains. This region may be unique in that the
major sources of water for irrigation may be over a 1000 km away from the point of application. Some
groundwater is pumped from both shallow and deep aquifers probably recharged by water from the Sierra
Nevada Mountains.
GISS
Snowmelt will occur a few weeks earlier and a slightly higher proportion of precipitation as rainfall will
mean reservoirs will fill earlier, helping to accommodate a somewhat earlier start of the irrigation season. There
will be an irrigation water deficit within the Imperial Valley region of California which will be supplied by the
increase in precipitation in the Rocky Mountains.
The GISS grid point in southern California shows an expected increase in evapotranspiration of 6.9
percent. There is also an apparent decrease in precipitation by three percent. Thus irrigation demands may
increase by approximately ten percent. There is not likely to be any increase in length of the irrigation season.
However, most of the water for irrigation in California comes from mountainous regions far from the agricultural
lands. The expected increase in annual precipitation and somewhat higher expected increases in summer
precipitation in the northern mountainous areas are likely to provide for a slight increase in irrigation water of
2 percent. Similar increases in the Rocky Mountains will increase Colorado River flow, with slight increases in
California's appropriations.
GFDL
With evaporation decreasing by 3% and rainfall increasing by 1.8% California does better hydrologically
than any other region under the GFDL model prediction. In addition, increased rainfall in the Rockies and
extreme northern California should more than counter any decrease in the snow pack caused by warmer
temperatures. The already long growing season should be increased further, given the availability of additional
irrigation water.
Northern Mountain Region
This region experiences a cold, continental climate only slightly modified by marine air masses from the
Pacific Ocean. The climate is strongly influenced by air circulation over Canada. There is a relatively short frost-
free period Cold winters are followed by cool to warm summers. Extreme fluctuation in streamflows occur
between the peak snowmelt season
and the summer low flows. Most irrigation water is drawn from reservoirs but a significant amount is pumped
from shallow aquifers assumed to be recharged by subsurface flows from the Rockies.
GISS
The potential for evaporation is expected to increase by 15 percent under the GISS 2xCO2 scenario and
annual precipitation is expected to increase by 18 percent. A slightly higher proportion will come in summer and
a significant increase in winter precipitation will come as rainfall. This more uniform distribution should facilitate
4-44
-------
Adams
the management of water for irrigation. With a longer frost-free season, sites not now suitable for agriculture
will be within an acceptable climate for some crops.
The snowpack will accumulate at a higher elevation and will on average be deeper where it does
accumulate. The net effect of the climatic change will be an increase in water available for irrigation, both within
and outside of the region. The specific assumption here is for a 3 percent increase in available irrigation water.
Water demand by crops in this region will decrease by 2 percent.
GFDL
As in the Northwest, modest increases in precipitation (1.7%) will not be sufficient to overcome a large
(9.7%) increase in evaporation. In the western portion this will not be the case; but the eastern region with an
increase in whiter precipitation will not compensate for a large drop in the summer. However, both portions
will benefit by a substantial lengthening in the growing season occasioned by a temperature 5 to 7° wanner.
Effective capture of earlier runoff will be important as net water availability decreases by 7.9%.
Southern Mountain Region
While this region experiences a fairly cold winter, it differs from the Northern Mountain Region by
having a hot dry summer. The area of most intensive agricultural activity is located in the southern part of the
region and is characterized as typical of Sonoran climates. Precipitation ranges from under 300mm to 500mm
per year. Extremely low humidities prevail much of the year.
Large scale agriculture is sustained primarily by impoundment and diversion of water from the Colorado
River. Some water is provided by the Gila River and by pumping from deep aquifers. The southeastern portion
of the region is supplied by the Rio Grande River which is in turn primarily fed by snowmelt from the southern
Rockies. Even at the higher elevations in this region, the precipitation is typically under 750mm. Groundwater
is thought to be primarily from a slow process of recharge from water originating in the southern Rockies.
GISS
The GISS 2xCO, scenario for this region is a 6.2 percent increase in evaporation and a 5 percent
increase in precipitation, if the increase in precipitation is gained mostly in the summer it may be effective in
offsetting most of the increase in evaporation. However, the precipitation in this region is generally so low and
erratic that a 5 percent increase may not be of much use. A 6.2 percent increase in evaporation will need to be
made up by increased import of water from the Northern Mountain Region.
The runoff from the 18 percent increase in precipitation in the Northern Mountain Region, if not
claimed by other water rights, should be able to supply the need. Thus, in this study, it is assumed that there
is a modest one percent increase in available water in this region. Water demand, however, will be increased
by approximately 2 percent.
GFDL
As for the northern monitor area, under GFDL the eastern and western portions of the southern
mountains will see different precipitation patterns. The increased rainfall in the western half will compensate
for most of the increased evaporation, while the substantial rainfall decline in the east is not enough to offset
a modest decline in evaporation. Sustained increases in groundwater pumping are not feasible and the area will
be adversely affected by the net increase in water demand at almost 53.
4-45
-------
Adams
Northern Plains Region
This region experiences a cold, continental climate with a relatively short frost-free season. Winters are
dominated by a polar air mass circulating from the Canadian shield. Cold winters are followed by a relatively
short growing season, but May, June and July are the wettest months of the year. Much of the agricultural
production is dependent on the spring and summer rainfall. It also depends on winter snowfall and the practices
of water conservation (snow trap, summerfallow, etc.).
Where irrigated agriculture is practiced, the water is primarily supplied from reservoirs fed by rivers
originating in the Rocky Mountains. The headwaters of the Missouri, North Platte and Arkansas Rivers all
originate from high elevation snow fields. Groundwater is also used where aquifers are recharged by deep
percolation from the east slopes of the Rockies.
GISS
This region is expected to experience an 8.5 percent increase in evaporation. This is roughly equivalent
to 65-70mm annual increase in water consumption. This is partially offset by a 7 percent increase in
precipitation, the majority of which will come in the growing season. However, it is likely that a part of the
increased evaporation will have to be made up from water originating in the Rocky Mountains which presently
feeds the reservoirs of this region. Allowing for a moderate increase in run-off from the Rockies, we assume
a one percent increase in available water, but a 3 percent increase in water demand.
GFDL
Under the GFDL predictions, both precipitation (-3.5%) and evaporation (-1%) decline while the
temperature increases by about 6°C. This results in a net decrease in water availability of 2.4%, which will be
exacerbated by the shift from summer to winter precipitation. However, a longer growing season may make use
of winter precipitation more efficiently and plants may mature before available moisture disappears.
Southern Plains Region
This region is strongly influenced by weather patterns generated in the Gulf of Mexico, but is in a
transition from the continental steppe climate. Short, cool winters are followed by long hot summers.
Precipitation is dominated by rainfall which ranges from 250mm in the northwest corner to about 1500mm in
the southeast portion of the region. The precipitation is generally well distributed over all months of the year.
Irrigation water in the northern portion of this region is largely from impoundments on the Canadian
and Red Rivers which rise in the Southern Rockies. The southwest portion of the region is supplied by the Rio
Grande River which also drains from the southern Rockies. Extensive use of groundwater is made from deep
aquifers thought to be partially recharged by water from the southern Rockies.
GISS
The GISS forecast for this region is for it to be warmer by 3.9°C but that evaporation will decrease by
about 1.5 percent. Winter temperatures are expected to increase more than summer temperatures which may
help to moderate the effect of the overall temperature increase on crop water usage. The reduction in
precipitation, however, is estimated to be about 8 percent It appears that some water deficit will occur,
particularly in the later part of the growing season. This will have to be drawn from the increases in precipitation
in the mountainous areas to the northwest. Given that the southern Rockies are not likely to show any increase
in run-off, a 3 percent drop in available irrigation water, coupled with a 5 percent increase in water demand is
assumed.
4-46
-------
Adams
GFDL
As with the southern mountains, the wet, dry high plains region will see a decrease in rainfall, especially
during the growing season. The humid eastern portion is not similarly affected. Hence, the region's modest
decrease in available water of 2% will not be evenly distributed. Since the eastern Rockies will be drier,
reservoirs and aquifers which supply irrigation demands will not be refilled by winter precipitation.
Delta Region
Arkansas, Mississippi and Louisiana make up this region. The area has a warm, temperate rainy climate
which receives from 1000 to 2000mm of rainfall per year. The rainfall is quite consistent from year-to-year and
well distributed over all months of the year. Numerous major streams and rivers bisect or border this region.
Both surface water and shallow aquifers are used in irrigation. High summer humidities are common throughout
this region.
GISS
i increases
The GISS model shows that the Delta Region will be significantly warmer but that evaporation ir
only by 2,4 percent. Precipitation is predicted to increase by 2 percent. As precipitation is a larger component
in this region than is evaporation, the overall effect could be from none to a slight increase in water available
for irrigation. Specifically, we assume a one percent increase in run-off but a 2 percent decline in water demand.
GFDL
The GFDL projections are similar to the GISS model (rainfall + .2%, evaporation +1.6%, temperature
+4.5%) except a shift from summer to winter rainfall may increase irrigation demands and require more storage
capacity. Drought conditions may occur with greater frequency; increasing the efficiency of irrigation. The high
levels of moisture should support this transition.
South East Region
The warm, temperate rainy climate found in this region has precipitation amounts ranging from 1000
to 1500mm. Rainfall is nearly equally distributed throughout the year. Irrigation is largely from diversion from
the consistent streamflow, but shallow aquifers are also utilized. There is a remarkably small difference in winter
to summer temperatures, averaging around 21°C, plus or minus 5°C. High humidity prevails through several
months of the year.
GISS
This region is expected to have an 8.4 percent increase in evaporation and a 12 percent increase in
annual precipitation. It appears that this region will experience a warmer, wetter climate, with the change in
precipitation closely balancing the increased evaporation. We assume the interaction of evaporation and
precipitation changes translate into a 3 percent increase in water availability and a 4 percent decrease in water
demand.
4-47
-------
Adams
GFDL
Both evaporation and rainfall are assumed to decrease by 7 to 8% with net irrigation supply decreasing
marginally. As in the Delta, increased temperatures (+5°C) and change in the rainfall pattern from summer to
winter will increase the efficiency of irrigation and necessitate more seasonal storage of runoffs.
4-48
-------
Adams
APPENDIX B
AN OVERVIEW OF THE ECONOMIC MODEL
Economists devote considerable effort to assessing the consequences of changes in policy and technology
on the agricultural sector. Assessments of the benefits of such change are performed at both the farm and
sectoral levels. There are, potentially, major differences in the results of such evaluations depending upon the
level at which the evaluations are performed. The evaluation of changes at the sectoral (i.e., aggregate) level
often require one to sacrifice microeconomic detail in order to keep the problem tractable. This can have serious
consequences. For example, appraisals of induced changes with aggregate programming models often result in
extreme specialization in production (solutions where whole regions are devoted to a single crop). This situation
usually leads to the imposition of inflexible "flexibility" constraints.
McCarl (1982) recently argued that linking microeconomic considerations with the sector model through
a Dantzig-Wolfe decomposition scheme using heuristic procedures avoids this specialization in production within
a sectoral analysis. The analysis used in this assessment incorporates this approach. Implementing the
methodology requires both farm-level data and a macro (sector) model. Both the structure of the sector model
and the microeconomic detail embedded in the model are discussed in this appendix.
The Sector Model
In this study the agricultural sector model component is a price-endogenous mathematical programming
model of the agricultural sector; i.e., an activity analysis spatial equilibrium model (Takayama and Judge, 1971).
Such sector models are used extensively by agricultural economists to simulate the effects of alternative
agricultural policies or of technological change (Heady and Srivistava, 1975). Mathematical programming is a
particularly useful tool given its ability to simulate potential consequences of as-yet-unrealized policies. This
general methodology has been applied to air pollution effects by Adams et aL, 1986; Adams et al., 1988; and
Rowe et al., 1984.
The sector model features constant-elasticity demand relationships for the outputs (commodities) of the
micro models. The elasticities vary with end use and across domestic and export markets. Assuming supply and
demand functions which are integrable and independent of sector activity, first order conditions are then achieved
in the macro model specification. The objective function of this specification is:
maximize * - E g,(Z|) - £ ^(Xj) - E CmYm
where x is the sum of ordinary consumers' and producers' surplus and the integrals are evaluated from zero to
Z,*, the amount of i* commodity produced and sold to consumers; and from zero to X,*, the amount of the j^
factor used. The parameters are as follows:
4fi
g|(Z|) is the area under the demand function for the i_ product;
e,(X.) is the area under the supply function for the j* factor;
Cm is the miscellaneous cost of production;
subject to a set of technical and behavioral constraints. Given the micro and macro structure of the model, the
solution then simulates a long-run, perfectly competitive equilibrium.
Following Samuelson (1952), the objective function Or) may be interpreted as a measure of ordinary
consumers' and producers' surplus (quasi-rents) or net social benefit. Analytically, this is defined as the area
between the demand and supply curves to the left of their intersection. The demand functions are specified at
4-49
-------
Adams
the national level, as are aggregate production responses, providing national-level consumers' and producers'
surplus welfare measure. The use of economic surplus in policy analysis is well documented in the literature
(Willig, 1976; Just et al., 1982), and is particularly relevant to agricultural uses where aggregate distributional
consequences are of concern. The economic implications of alternative climatic scenarios are assessed by
measuring changes in consumers' and producers' surplus which result when crop yields and input supplies are
altered as predicted by the scenarios.
The sector model was solved under constant-elasticity demand curves using the MINOS software package
(Murtaugh and Saunders). The model works from a set of budgets for 30 primary crops and livestock activities.
For production purposes the VS. is disaggregated into 63 geographical sub-regions. Each region possesses
different endowments of land, labor and water as well as having different crop yields. This regionally specified
information is an important feature in this model. Details on the items mentioned above follow. The model
distinguishes between primary and secondary commodities with primary commodities being produced directly
by the farms while secondary commodities involve processing activities.
Primary Commodities
Thirty primary commodities are listed in Table B-l. The primary commodities are chosen so as to depict
the majority of aggregate agricultural production, land use and economic value. They can be grouped into field
crops and livestock.
Both supply and demand information (i.e., equilibrium prices, quan- titles, and elasticities) are required in
the model. The total supply consists of domestic production from all agricultural regions plus imports. Total
demand is made up of domestic and foreign (export) components. Domestic demand includes consumption,
stocks, government programs, livestock feeding and processing. Transportation costs to market are included in
the supply budgets. Livestock feed and processing are endogenously determined, derived demands. Price and
quantity data come from Agricultural Statistics. Agricultural Prices Annual Summary, and Livestock and Meat
Statistics Supplement. Elasticity, and other demand information were supplied by Bob House, Economic
Research Service, U.S.Department of Agriculture.
Secondary Commodities
The processing of secondary commodities is modelled at the sector level. Table B-2 lists the 18 secondary
commodities in the model. These are chosen based on their linkages to agriculture. Some primary commodities
are inputs to the processing activities and certain secondary products (feeds and by-products) are in turn inputs
to other agricultural activities. The main data sources are Agricultural Statistics. Agricultural Prices Annual
Summary. Livestock and Meat Situatioa and Livestock Slaughter Annual Summary.
National Inputs
The model contains 27 national inputs listed in Table B-3. For the most part these are specified in dollar
terms; for example, ten dollars worth of nitrogen, twenty dollars worth of repair costs. In doing so, the input
usage is converted into a homogeneous commodity. These inputs are assumed infinitely available at whatever
price was entered in the 1982 Farm Enterprise Data System (FEDS) budgets.
Regional Inputs
There are three inputs that are available in the regional level: land, farm labor and water. Production of
crops and livestock compete for these scarce resources in each state or region. Therefore, the price and
quantities of these inputs are endogenously determined on a regional basis.,
4-50
-------
Adams
Two types of land are specified. The first (type 1) is land suitable for crop production. The second (type
2) is suitable for pasture or grazing. The information on land utilization by states or regions was derived from
Agricultural Statistics; regional prices of land were derived from the information in Farm Real Estate Market
pevelQpmenl Cash rental prices of land were used to reflect annual opportunity costs to the owners.
The labor input also includes two components: family labor and hired labor. The model requires
specification of a maximal amount of family labor available and a reservation wage for family labor. Additional
hired labor is available, based on an upward-sloping supply schedule with a reservation wage higher than that
of family labor. The regional information on wages and employment was obtained from Farm Labor.
Water can be obtained from both fixed (surface) and variable (pumped groundwater) source. Surface water
is available at a constant marginal cost but groundwater has a rising supply schedule; increasing amounts of water
are available only at a higher price. The information on water is from USDA personnel, the Farm and Ranch
Irrigation Survey and other government sources.
Regional Disagereeation
The model operates with two levels of regional disaggregation. The fundamental unit of disaggregation is
63 state and/or substate areas. In addition, these 63 areas are grouped into the ten USDA production regions
for the purposes of land, labor and water supply. A list of these two levels of disaggregated regions and areas
are given in Table B-4.
Regional Production Activities
Currently a total of 1683 production possibilities (budgets) are specified to represent agricultural production.
These include major field crop production, livestock production and some miscellaneous transfer activities. Most
field crop activities are also divided into irrigated and nonirrigated according to the irrigation facilities available
in each state or area.
In some cases, the production activities produce more than one commodity. All commodities can be
produced by more than one set of input combinations. Most field crops (except rice) are produced by either
irrigated or nonirrigated production practices. Livestock production is somewhat more complicated. (See Chang
and McCari, 1988, for details.)
For each activity, information on yields and usages of national and regional inputs or other commodities
is required. The basic source of this information is the 1982 USDA FEDS budgets. The irrigated/nonirrigated
budget breakdown was developed by the USDA water group based on the FEDS surveys, the survey of irrigated
acreage, extension budgets and Soil Conservation Service budget sets. The Livestock budgets are from the FEDS
system for 1982.
Processing Activities
The secondary commodities are produced by three types of processing activities: soybean crushing;
combining feed ingredients into various livestock and poultry feeds; and conversion of livestock and milk into
consumable meat and dairy products. The processing cost of each commodity is calculated as the difference
between its price and the costs of the primary commodity inputs.
Soybean crushing involves conversion of soybeans into meal and oil. Two soybean crushing activities are
included, the model solution selects the more profitable one. Meat processing includes conversion from culled
animals to slaughter and from slaughter to meat. Dairy processing involves conversion of raw milk to five
4-51
-------
Adams
different dairy products. The conversion of feed and feed supplements involves more than one processing
activity, the model solution selects the least cost combination of feed ingredients.
Crop Mixes
The sector model is disaggregated into 63 internally "homogeneous" production areas. However, within each
region least-cost production is represented by few, often one, crop budget. This can lead to misleading results
since such representation cannot capture the full factor-product substitution possibilities in each of those areas.
This is avoided by requiring crop production in each region to fall within the mix of crops observed in cropping
records over the past 25 years.
4-52
-------
Adams
Table B-l. Primary Commodities
List of Commodities Units List of Commodities Units
1. Cotton Bales (480 Ibs.) 16. Hogs for slaughter Cwt. LW
2. Cora Bushel 17. Feeder pigs Cwt LW
3. Soybeans Bushel 18. Live (beef feeder) calves Cwt. Lw
4. Wheat Bushel 19. beef feeder yearlings Cwt. LW
5. Sorghum Bushel 20. Slaughtered calves Cwt LW
6. Rice Cwt 21. Slaughtered nonfed beef Cwt LW
7. Barley Bushel 22, Slaughtered fed beef Cwt LW
8. Oats Bushel 23. Culled sows Cwt LW
9. Other livestock (horses) GCAU 24. Poultry GCAU
10. Cull dairy cows Head 25. Slaughtered lambs Cwt LW
11. Cull beef cows Cwt. LW 26. Feeder lambs CwtLW
12. Cull dairy calves Head 27. Culled ewes Cwt. LW
13. Milk Cwt 28. Wool Cwt
14. Silage Ton 29. Wool incentive payments $
15. Hay Ton 30. Unshorn lamb payments $
Note: LW indicates live weight GCAU is in terms of grain consuming animal unit.
4-53
-------
Adams
Table B-2. Secondary Commodities
List of Commodities Units
1. Soybean meal 1000 Lbs.
2. Soybean oil 1000 Lbs.
3. Fluid milk Cwt
4. Feed grain 1000 Lbs.
5. Dairy protein feed 1000 Lbs.
6. High protein swine feed 1000 Lbs.
7. Low protein swine feed 1000 Lbs.
8. Low protein cattle feed 1000 Lbs.
9. Fed beef Cwt. CW
10. Veal Cwt. CW
11. Nonfed beef Cwt. CW
12. Pork Cwt. CW
13. High protein cattle feed 1000 Lbs.
14. Butter Lb.
15. American cheese Lb.
16. Other cheese Lb.
17. Ice cream Lb.
18. Nonfat dry milk Lb.
Note: CW means carcus weight.
4-54
-------
Table B-3. National Inputs
List of Inputs
1. Nitrogen
2. Potassium
3. Phosphorous
4. Lime
5. Other variable costs
7. Custom operation
8. Chemicals
9. Seed costs
10. Interest on operating capital
11. Repair costs
12. Vet and medical costs
13. Marketing/storage costs
14. Insurance (except crop)
15. Machinery
16. Management '
17. Land taxes
18. General overhead costs
19. Non-cash variable costs
21. Fuel and energy costs
22. Crop insurance
23. Land rent
24. Set-aside(conservation cost)
26. Processing Labor
27. Irrigation energy cost
4-55
-------
Adams
Table B-4. Assignment of States to Regions
NORTHEAST
Connecticut
Delaware
Maine
Maryland
Massachusetts
New Hampshire
New Jersey
New York
Pennsylvania
Rhode Island
Vermont
CORNBELT
North Illinois
South Illinois
North Indiana
South Indiana
North East Iowa
Central Iowa
South Iowa
West Iowa
Missouri
North East Ohio
North West Ohio
South Ohio
SOUTHERN PLAINS
Oklahoma
Texas Central Blacklands
Texas Coast Bend
Texas East
Texas Edwards Plateau
Texas High Plains
Texas Rolling Plains
Texas South
Texas Trans Pecos
LAKE STATES
Michigan
Minnesota
Wisconsin
SOUTHEAST
Arizona
Alabama
Florida
Georgia
South Carolina
MOUNTAIN
Colorado
Idaho
Montana
Nevada
New Mexico
Utah
Wyoming
NORTHERN PLAINS
Kansas
Nebraska
North Dakota
South Dakota
APPALACHIAN
Kentucky
North Carolina
Tennsessee
Virginia
West Virginia
DELTA STATES
Arkansas
Louisiana
Mississippi
Pacific
North California
South California
Oregon
Washington
4-56
-------
CLIMATE CHANGE IMPACTS UPON AGRICULTURE AND RESOURCES:
A CASE STUDY OF CALIFORNIA
by
Daniel J. Dudek
Environmental Defense Fund
257 Park Avenue South
New York, NY 10010
Grant No. NAGS-1025
-------
CONTENTS
FINDINGS 5-1
CHAPTER 1: INTRODUCTION 5-3
THE PROBLEM 5-3
CALIFORNIA AGRICULTURE 5-3
CALIFORNIA'S WATER SYSTEM 5-4
CHAPTER 2: METHODOLOGY 5-5
CROP PRODUCTIVITY ANALYSIS 5-5
CALIFORNIA AGRICULTURE AND RESOURCES MODEL 5-7
HYDROLOGIC MODELING 5-10
ASSUMPTIONS AND LIMITATIONS 5-12
Crop Productivity Model 5-12
Economic Model 5-13
Hydrologic Limitations 5-13
THE SCENARIOS 5-14
CHAPTER 3: RESULTS 5-16
CROP PRODUCTIVITY CHANGES 5-16
Climate Change Effects 5-16
Net Effects Including CO 5-16
WATER RESOURCE SUPPLY IMPACTS 5-19
ECONOMIC IMPACTS: AGRICULTURE AND RESOURCES 5-20
Aggregate Results 5-20
Regional Results 5-22
SOCIETAL RESPONSES TO CLIMATE CHANGE 5-25
INCLUDING CARBON DIOXIDE EFFECTS 5-28
CHAPTER 4: POLICY IMPLICATIONS 5-31
AGRICULTURE 5-31
WATER RESOURCES 5-31
Existing Legal and Institutional Setting 5-31
Status of Water Transfer Activities 5-32
Barriers to Water Transfers and Prescriptions for Change 5-33
THE ENVIRONMENT 5-34
CHAPTER 5: CONCLUSIONS 5-36
REFERENCES 5-37
-------
Dudek
FINDINGS1
This study assesses the impacts of climate change on California's agricultural and water resource systems.
The methodology employed explicitly links predictions of climate changes from general circulation models
(GCMs) of the atmosphere with an agricultural productivity model. These productivity impacts are introduced
into the California Agriculture and Resources Model (CARM), which determines the economic and market
implications of such changes. The climate changes assessed include temperature, evapotranspiration,
precipitation, and cloudiness. The implications of such changes for water resource supplies were separately
evaluated for the Sacramento and San Joaquin Valley Basins by a team of hydrologists and engineers
(Lettenmaier et al., Volume A; Sheer and Randall, Volume A).
To assess climate change driven productivity impacts, an existing agro-ecological zone model (De Wit)
was adapted. For several general crop groups, this method estimated that climate changes, in general, would
reduce yields. The greatest impacts are expected in interior southern regions on cool season crops such as
sugarbeets. Reductions in crop productivity from climate changes ranged from modest 3% declines to more
serious 40% impacts for the crop groups and regions studied. Including CO2 enrichment effects significantly
alters the overall result. Net productivity changes range from a 41% increase to a 27% decline. This radical
change reflects the differential potential ability of crops to utilize the increased CO2.
Our economic system of markets and private decisions exerts compensating influences which tend to
modulate the ultimate influence of direct physical productivity changes. Market incentives act to reallocate
production activities and resources to maximize value to society under the environmental conditions specified in
the scenarios. Scenarios for the economic model were constructed from the climate changes predicted by
alternative GCMs, by the specific set of productivity impacts analyzed, by the availability of water resources, and
assumed social response to such changes. Results from CARM indicate that even after market responses
function, statewide average yields would be significantly reduced for all crop groups as a result of climate
changes. Vegetables would be least severely impacted with average yields reduced from 6 to 15%. Fruit and
nut crops would be hardest hit with average declines from 23 to 33%. After CO, enrichment is factored in,
vegetable yields improved substantially to levels above the statewide average base, but most fruit, nut, and field
crop yields remained reduced.
Regional changes in the production of agricultural commodities and in the use of resources by agriculture
dominate the results. The hydrologic study team has estimated that surface water reductions under a changed
climate would be greatest in those regions served by California's state water project (SWP). For a 30-year
simulation of hydrologic flows under alternative future climates, SWP deliveries were reduced by 25 to 28% on
average. Overall, climate change effects reduced the net economic well-being produced from agricultural
operations between 14 and 17%. Statewide crop acreages were reduced between 4 and 6%, depending upon
scenario, from a base level of slightly more than 9 million acres. Regionally, acreage declines were greatest in
the Imperial Valley. Although no assessment was made of future groundwater stocks or pumping lifts, pumping
declined from roughly 20% statewide as a result of reduced crop profitability. Overall surface water use declined
roughly 16% as a result of both supply and demand changes. Regional water use changes were most dramatic
throughout the San Joaquin and Imperial Valleys.
Factoring in CO, effects produced significantly different results. Total crop acreage either slightly declined
or increased with net economic well-being, rising approximately 1.4% in each case. Surface water supplies
remained short in state water project service areas which required large increases in groundwater pumping to
'Although the information in this report has been funded wholly or partly by the US. Environmental
Protection Agency under Grant No. NAGS-1025, it does not necessarily reflect the Agency's views, and no official
endorsement should be inferred from it. The author gratefully acknowledges the contributions of Gerald L.
Homer, Cynthia Rosenzweig, John Ruston, and Zach Willey to this report.
5-1
-------
Dudek
compensate. Despite increasing marginal costs for groundwater withdrawals, increased pumping was
economically supported by the generally elevated yields.
The last phase of this study evaluated social responses to climate change and their potential to reduce the
impacts of such changes. This analysis focused on the introduction of irrigation water markets and improved
on-farm irrigation management. All publicly provided surface water supplies were offered for sale as long as
transfer charges were profitably covered. For all scenarios, including the base, water marketing produced net
economic benefits. In the scenarios emphasizing climate change effects, impacts were reduced. When CO2
enrichment effects were included as well, similar gains were recorded.
In summary, this case study has demonstrated the importance of including all related climate change
impacts within a single analytical framework. In particular, it is not possible to translate physical crop
productivity or water supply changes directly into impacts without accounting for the effect of market forces.
Commodity markets operate to induce shifts in crop locations producing average yields close to the minimum
biologic impact. Introducing markets for water resources similarly offsets supply reductions by improving the
efficiency of use of what is available.
Market forces play a crucial role in creating the flexibility to respond to climate changes and in mitigating
their ultimate impact. However, in the absence of radical changes in agricultural production technologies or
environmental management institutions, nonmarket effects such as nonpoint source pollution will be exacerbated
as agriculture relocates in response to climate changes. Some problems, such as drainage and salinity, may
improve marginally as acreage and average water use are reduced. However, groundwater overdraft problems
are likely to be exacerbated in the San Joaquin Valley although it is not known .fovi whether future energy prices
or groundwater levels would support these withdrawal rates.
Climate changes and increased competition for water likely will have negative impacts on existing aquatic
ecosystems. Increased temperatures and altered flow regimens in managed and free-flowing river systems may
change species composition from cold to warm water varieties. Altered precipitation patterns would reinvigorate
interest in large-scale public works including expansion of both the federal Central Valley Project and the State
Water Project. Increased reservoir capacity would also affect fishery resources and the mix of recreational
opportunities. Intensification of cropping and water use in the Sacramento Valley could negatively impact
migratory waterfowl in critical Pacific flyway habitats. In short, many of the most severe consequences of climate
change for California's environment remain to be studied.
5-2
-------
Dudek
CHAPTER 1
INTRODUCTION
THE PROBLEM
The increasing atmospheric concentration of radiatively active gases within the troposphere has been the
focus of substantial concern within the environmental and scientific community. These cumulating gases are
expected to produce a global climate change unprecedented in human history. Current atmospheric modeling
results indicate that a climate change equivalent to that from a doubling of CO2 concentration is expected in the
first half of the next century if current growth rates of atmospheric CO2 and trace gases continue. An actual
ambient doubling of CO^ concentrations is not expected until the later part of the 21s century, depending upon
the specific emissions trajectory produced by our use of fossil fuels. These changes are expected to produce up
to a 4-5° Celsius increase in the average global temperature. These global average increases underestimate the
extent of potential regional changes since the impacts will be differentially distributed from the poles to the
equator with the greatest effects predicted at the high latitudes.
Temperature changes are the most certain of climate impacts. For central and southern California, mean
temperatures are predicted to rise 3.8-4.4°C in an atmosphere with a doubled CO2 concentration. In the
northern part of the state, the increases range from 4.3 to 5.0°C. Other climate impacts will include changes in
the pattern and distribution of precipitation, changes in sea level and the total area of dryland, and changes in
the frequency of severe weather events. For example, increased temperatures in California would produce
more rain and less snow during the winter. In addition, the snowpack would melt earlier in the spring reducing
total effective reservoir storage and available supplies.
Of primary concern is the effect of climate change upon our food and fiber system and the resources used
to support it. Agriculture is one of the most weather-sensitive sectors of our economy. Previous attempts to
quantify this sensitivity to climate change have highlighted the particular sensitivity of irrigated agricultural
regions (Dudek, 1987c). This study is designed to explore that sensitivity in greater detail through a case study
of California's irrigated agriculture.
CALIFORNIA AGRICULTURE
California agriculture annually produces about 10% of total cash farm receipts in the United States. In
1986, California's farm income was first in the Nation at $14.5 billion, followed by Iowa and Texas with $9.1 and
$8.5 billion, respectively. Grapes, cotton, hay, lettuce, almonds, tomatoes, strawberries, oranges, broccoli,
walnuts, sugarbeets, peaches, and potatoes represent, in descending order, the major irrigated crops ranked in
terms of gross receipts. As a sector, California agriculture produces 3-4% of total state income. The major
sectors in the state's economy are petroleum and chemicals, banking and finance, real estate, electronics, and
services.
Acreage harvested in California has varied between 7.5 and 9.0 million acres during the 1980s (2.467 acres
are equivalent to 1.0 hectare). Agricultural irrigation accounts for approximately 80% of all consumptive water
uses annually in California. By hydrologic basin, the major agricultural water uses in 1980 were San Joaquin and
Tulare — 51%; Sacramento ~ 25%; and Colorado ~ 12%. Cotton, feed grains and hay, and pasture are the
largest water-using crops. Evapotranspirative water demands (ET) of crops vary by weather, soils, and locations.
They range from 7.9 acre-feet per acre of rice in the Sacramento Valley to 1.2 acre-feet for an acre of wine
grapes on the Central Coast.
5-3
-------
Dudek
CALIFORNIA'S WATER SYSTEM
California's water system is supplied by a statewide average annual precipitation of 23 inches, 60% of
which is evaporated and transpired by native trees, brush, and other vegetation. Approximately 71 million
acre-feet run off as stream flow and groundwater recharge in an "average hydrologjc year." An additional 1.4
million acre-feet is inflow from Oregon streams, and the Colorado River has contributed another 4.8 million
acre-feet in recent years.
The state's developed water supply system yields annual consumptive use of approximately 1/2 of annual
stream run-off. Current categories of supply sources by percentage of annual use are as follows: (1) local
surface water (27%); (2) groundwater safe yield (17%); (3) groundwater "overdraft" (6%); (4) Federal Central
Valley Project (20%); (5) State Water Project (7%); (6) Colorado River (15%); and (7) other (8%). The State
of California has jurisdiction over 1,188 dams and reservoirs with a gross storage capacity of 19.7 million
acre-feet, and the VS. has an additional 125 dams and reservoirs with 22.9 million acre-feet of capacity. Various
estimates can be made of the economic value of this installed capacity. A very rough indicator of the magnitude
of the investment in this system is that at an average cost of $500 per acre-foot of installed water supply delivery
capacity, a $15-$20 billion investment exists to deliver 30-40 million acre-feet of annual water applications.
The driest year in California's recorded history was 1977. That year was also the second successive dry
year of the worst drought California has experienced in over 100 years of record. The water year ending on
September 30, 1977, yielded precipitation 45% of average. The fourth driest year of record was 1976, with
precipitation at 65% of normal at the end of the water year. Run-off in streams and rivers was 47% and 22%
of average flows during 1976 and 1977, respectively. The state's storage in its surface reservoirs hit a record low
in 1977. Heavy precipitation and snow in early 1978 ended the drought. There have been no droughts since
1976-77, although 1987 was a critically dry year.
Groundwater was heavily pumped to offset reduced surface water supplies during the drought. In the San
Joaquin and Tulare basins alone, overdraft increased in 1977 by 3.6 million acre- feet over the 1975 overdraft
level of 13 million acre-feet. Saltwater intrusion in the Sacramento and San Joaquin River Delta was extensive.
Economic damages during the 2-year period were estimated to total $2.7 billion (1978 dollars), of which 55%
were agricultural. The second largest loss category was in forests, where fires and insects accounted for 25%
of total losses. The third significant loss was in energy production (oil, gas, and imports substituted for reduced
hydroelectric generation), which accounted for nearly 18% of total losses.
The remainder of this report will describe the methodology employed and its limitations, will present and
interpret the results, will address the environmental and socioeconomic implications of those results, and will
recommend policy changes.
5-4
-------
Dudek
CHAPTER 2
METHODOLOGY
The objective of this study is to characterize potential shifts in the demand for resources for agricultural
production that might occur in California under a changed climate. The general methodology is that employed
for the assessment of impacts of ambient pollutants upon agriculture (Wetzstein). Four types of models --
climate, yield, hydrologic, and economic -- are required for this assessment. The models used are the Goddard
Institute for Space Studies (GISS) and General Fluid Dynamics Laboratory (GFDL) GCMs described elsewhere,
crop productivity models, and the California Agriculture and Resources Model (CARM). A general description
of the crop productivity model and CARM follows.
CROP PRODUCTIVITY ANALYSIS
Physiologically, plant growth has been limited by the availability of atmospheric CO,-
:ased carbon dioxide concentrations in the atmosphere upon plants have been studied (
The effects of
increased carbon dioxide concentrations in the atmosphere upon plants have been studied* (Kimball, 1983).
These studies have emphasized productivity gains from enhanced CO2. However, crops are also influenced by
a host of other climatic variables. Phenological studies of crop response are necessary for an enhanced
understanding of the net physiologic and climatic effects upon yields. Temperature itself is a determinant of
yield. For example, the number of consecutive days above 95°F is a critical determinant of corn yield (Mearns
et al., 1984). Rosenzweig (1985) has demonstrated the importance of considering climatic change as well as
enhanced CO2 in studies of varietal wheat response.
In general, crop phenology models are the best choice for evaluating the impacts of general environmental
changes upon crop productivity (Ritchie and Otter, 1984). These semi-empirical yield models are designed for
large-area yield prediction using a whole-plant framework and including the effects of the major factors on crop
growth - climate, soil, and management. Requirements are few: daily solar radiation, minimum and maximum
temperatures, and precipitation. Crop phenology models are being employed in the national assessment of
climate change impacts upon agriculture, but limited crop coverage restricts the general application of this
approach to the California case study.
At the other extreme, general response assessments have been previously employed to assess climate
change effects upon crop productivity (Bolin et al., 1986). These general assessments have been culled from
existing literature and presented as potential response ranges for both C3 and G4 crops. C3 plants have
photosynthetic mechanisms which allow efficient exploitation of CO, concentration increases, while C4 types
benefit relatively less. The review by Bolin et al. indicated that yield reductions in the 3-17% range would be
associated with a 2°C temperature rise, while a doubling of CO2 would indicate a 10-50% yield increase for C3
crops, while C4s would only have increases from 0 to 10%. The productivity impact methodology applied in this
study attempts a balance between the extremes of the detail of crop phenology models and the generality of
literature reviews.
In order to estimate the effect of climate change upon crop production in California, the agro-ecological
zone method developed by Kassam (1977) was adapted. Given data on environmental conditions, this method
was originally developed to match crops to regions most likely to support high potential yields under the
assumption that "the maximum yield level of a crop is primarily determined by its genetic characteristics and how
well the crop is adapted to the prevailing environment" (Doorenboos and Kassam, 1979).
The three factors considered by Doorenbos and Kassam as generally contributing to maximum potential
yield are temperature, length of growing season, and incident solar radiation (which is in turn a function of cloud
cover, season, and latitude). By definition, this method does not account for the effect of soil conditions,
availability and quality of water, insect pests and plant pathogens, and farm management practices. Based on
empirical studies of plant yield response to environmental factors under good growing conditions (e.g., De Wit,
5-5
-------
Dudek
1965), Doorenbos and Kassam provide the following equations as means of relating ambient environmental
conditions to potential maximum yield (Y ). The following example uses coefficients for sugarbeets.
When ym > 20 kg/ha/hour
YmP - «L * CN * *H * G [F(0.8 + O.lym)y0 +
(1-FX0.5 + 0.025ym)yc]
When ym < 20 kg/ha/hour
Y = « ' ° * « * G
where:
CL = correction for crop development and leaf area.
CN = correction for dry matter production, 0.6 for cool and 0.5 for warm conditions.
CH = correction for harvest index.
G = total growing period.
F = fraction of daytime the sky is clouded.
ym = maximum leaf gross dry matter production rate of a crop for a given climate, kg/ha/hour.
y0 = gross dry matter production of a standard crop for a given location on a completely overcast
(clouded) day.
yc = gross dry matter production rate of a standard crop for a given location on a clear (cloudless) day.
In this study, data on average monthly temperatures and average monthly cloud cover for five California
weather stations and data on crop planting time and length of growing season for four indicator crops in various
California regions (see Table 4) were employed. Using these equations, estimates of maximum potential yield
for each crop (sugarbeets, tomatoes, cotton, and corn grown for grain and seed) by location were generated.
As expected, the maximum potential yield results obtained from this method are substantially higher than those
obtained in actual production. However, since only relative changes in yield are relevant for this study, the
equations are evaluated for both current and projected conditions. The projected condition results are then
ratioed to the baseline figures to yield an estimate of proportional productivity change. This procedure is
completely analogous to that employed for GCM predictions applied to observed data.
In order to estimate the effect of changes in climate, the baseline temperature data for each weather station
were converted into degrees Kelvin, multiplied by the average monthly temperature ratios obtained from the
GISS and GFDL gridboxes corresponding to northern and central/southern California, and converted back to
degrees Celsius. In the climate change scenarios, monthly average mean sky cover data in the baseline case are
replaced by cloud cover data from the GISS and GFDL 2xCO2 model outputs.
As expressed in the Doorenbos and Kassam equations, growing season is an important variable; if plant
growth is increased as a response to temperature, a reduced growing season could well eliminate any overall
increase in yield for a single crop season. For example, temperature triggering could be advanced as a result
5-6
-------
Dudek
of faster accumulation of growing degree day requirements with less time for production of harvestable yield.
As a consequence, yields may decline despite increased temperatures and seemingly more favorable growing
conditions.
Maximum leaf gross dry matter production rate of a crop for a given climate, y , is determined as a
function of the average daUy temperature over the growing season. For three crop types examined in this study,
the form of this function is displayed in Figure 1 from tabular data supplied by Doorenbos and Kassam (Table
5). These functions ignore differences between canopy and air temperatures.
While the primary focus of this study is the evaluation of climate change effects upon California agriculture,
increases in CO2 concentrations in the atmosphere will also attend the climate changes. As previously noted,
CO2 is a limiting factor in photosynthesis and its increased availability will tend to increase productivity
depending upon crop type. Consequently, yield response to both doubled COL concentrations and climate
changes is required to accurately assess impacts upon agriculture. Estimates ofproductivity increases under
doubled CO, conditions were taken from Cure (1985) and Kimball (1983). These response factors were used
to adjust Ym.
CALIFORNIA AGRICULTURE AND RESOURCES MODEL
Since the 1960s, researchers at the Department of Agricultural Economics at the University of California,
Davis, have been producing spatial optimization models of California agriculture. The earliest such model is a
linear program developed by Shumway (1970) and applied in federal planning studies of California's resource
base. Adams (1979) began with Shumwa/s general spatial production and resource structure converting the
model to a quadratic form which adjusts equilibrium market prices in response to underlying supply changes.
Howitt and Mean (1985) converted this model to one capable of calibrating model results to specific conditions.
The specification of CARM employed in this case study was derived from the latter research.
CARM is a quadratic programming model which maximizes the sum of producers' and consumers' surplus,
an approximate measure of social welfare, given available production opportunities and resources. More
formally, the model may be stated as:
Maximize Z = c'Xj + 1/2^*0^ - (k'x + l/2x'Sx)
subject to Ax <. b
where:
X| = vector of crop commodities produced and marketed.
x = vector of alternative regional crop production activities.
c = vector of intercept terms for linear demand functions.
D = diagonal matrix of demand function slope terms.
k = vector of variable production costs.
S = diagonal matrix of regional linear supply functions.
A = matrix of input-output coefficients for the regional production activities.
b = vector of resource availabilities.
The modeling of agricultural production activities begun with the crop productivity models previously
described must be translated into both economic and spatial dimensions in order to assess the implications of
crop productivity changes stemming from climate change. As currently configured, the spatial equilibrium
mathematical programming model used in this study has 7 production regions with 16 crop commodities (91 total
crop production activities). The crop commodities included and their base acreages are presented in Table 1.
5-7
-------
Dudek
Table 1. Crop Commodities in CARM
CROP
Fruits and Nuts
Almonds
Grapes
Oranges
Peaches
Walnuts
Field Crops
Cotton
Sugarbeets
Grains
Hay
Corn
Rice
Wheat
Alfalfa
1986 Acreage
(thousands of acres)
Vegetables
Broccoli
Cantaloupes
Lettuce
Potatoes
Tomatoes
106.4
79.1
145.5
49.4
239.0
412.7
670.8
174.7
54.0
179.3
1,037.5
188.0
250.0
363.0
675.0
1,680.0
Total
6,304.4
Source: California Department of Food and Agriculture (1987)
5-8
-------
Figure 1. Production Rates for Crop Groups by Temperature
I
ff
•o
o
a
8
o
Surface Air Temperature (degrees Celsius)
Source: Doorenbos and Kassam (1979), p. 12
D
I
-------
Dujek
These crops accounted for slightly more than 76% of California's total crop acreage. Each of these commodities
was used to represent a set of cropping activities not explicitly represented within the current model configuration
to ensure that resource use is accurately modeled. For example, wheat is used to represent barley and oat
production as well.
CARM has traditionally had between 14 and 17 production regions. The 14-region configuration is depicted
in Figure 2. For the purposes of this study, these 14 regions were collapsed into 7 locations. Table 2 presents
the regional correspondences between the 14-region specification and the more compact form employed in this
study. As indicated in Figure 2, CARM regions 1, 2, 5, and 6 have been aggregated into a single production
region, the Sacramento Valley. The other spatial aggregations are as defined by the shadings in Figure 2 and
the correspondences specified in Table 2. Given the spatial resolution of the GCMs, the increased level of
aggregation employed in CARM is not a serious compromise given the improved responsiveness and ability to
analyze alternative scenarios. Since all model results are expressed relative to a base model, little bias is
expected.
Costs of production for each of these commodities for each of the production regions were specified from
base budgets prepared by the California Cooperative Extension Service. Where necessary, these budgets were
indexed to a common 1985 basis. Other critical inputs into the programming model include the availability of
soil and water resources within each of the production regions. The rest of the agricultural economy of the
United States is presumed to be functioning normally - an analytic fiction given the climate changes under
analysis. However, since California agriculture is highly diversified and unique, this assumption is limiting only
for the major grain crops where national price changes would have the greatest effect. International trade in
agricultural commodities must also be exogenously assessed and specified as an input to the model. Since all
agricultural regions around the world will experience degrees of change, changes in agricultural trade demand
will be particularly difficult to project. Since trade is one of the elements of uncertainty expected to strongly
condition the demand for agricultural resource use in the United States (Homer et al., 1985), a range of
estimates covering a variety of demand scenarios should be evaluated. However, given the time and resource
limitations of this study, export market sensitivity was not included.
CARM has been modified to allow ease in aggregating production activities, regions, and resources. In
addition, in order to improve the accuracy of response under the water marketing scenarios, trade-offs between
irrigation efficiency and capital expense have been developed. Thus, growers may sell or lease "saved" water
generated from improved on-farm efficiencies with no potential diminution of yield, but with some capital and
management expense. In principle, deficit irrigation could be employed to trade-off between reduced crop
revenues and increased water marketing sales. Including a range of management choices avoids the all-or-none
responses typical of some previous studies.
HYDROLOGIC MODELING
The importance and complexity of water resource systems in California required the development of several
integrated case studies in order to assess climate change effects. Just as climate changes will affect the biologic
processes governing plant growth, changes in temperature, precipitation, and evaporation will impact hydrologic
processes. Given these impacts, it is not possible to evaluate the impact of climate change upon California
agriculture without estimates of impacts upon water supplies. Within California, the Sacramento Valley Basin
was selected as the focus for the hydrologic case study. This region is the site for major surface water supply
investments including central units of the federal Central Valley Project (CVP) at Shasta and the State Water
Project (SWP) at Oroville.
5-10
-------
Dudek
Figure 2. California Production Regions
1 North Coast
2 North Bay
3 Delta
4 South Bay
5 Sacramento Valley
6 Mountain Valleys
7 Central Coast
8 Northern San Joaquin
9 Interior Coast
10 Eastside San Joaquin
11 Westside San Joaquin
12 South Coast
13 High Desert
14 Low Desert
5-11
-------
Dudck
Table 2. Production Region Correspondences
14 Rerion Model 7 Region Model
1 North Coast Sacramento Valley
2 North Bay
5 Sacramento Valley
6 Mountain Valleys
3 Delta Delta
4 South Bay Central Coast
7 Central Coast
9 Interior Coast
8 Northern San Joaquin Northern San Joaquin
10 Eastside San Joaquin Southern San Joaquin
11 Westside San Joaquin
12 South Coast South Coast
13 High Desert Imperial Valley
14 Low Desert
The first component of the hydrology study is the simulation of basic rainfall and runoff processes
(Lettenmaier, Volume A). The results of this model are estimates of virgin or unimpaired flows under
alternative climate scenarios. Basically, precipitation, temperature, and wind changes predicted from individual
GCMs and scenarios are translated into proportional changes and then overlaid on a historic record. These
virgin flows were then adjusted for land use changes and analyzed in a simulation model of the joint operations
of the SWP and the CVP (Sheer and Randall, Volume A). On the basis of the existing reservoir operating rules
predicated on the existing pattern of water rights developed under California's appropriation doctrine, any water
shortages are allocated. The joint operating model then produces estimates of deliveries to users throughout the
state. It is these delivered quantities under alternative climate scenarios which are input into CARM.
ASSUMPTIONS AND LIMITATIONS
The outcomes produced from this set of linked models are strongly conditioned by the quality of the data
inputs flowing between the various analytical components. For each of the model subsystems employed, critical
assumptions or limitations are briefly noted below. In most respects, the limitations reflect either those imposed
by the time and resource restrictions imposed by the stringent study schedule or the limitations of current
science.
Crop Productivity Model
As indicated in the section describing the crop response methodology, the technique developed by Kassam
and adapted in this study is less than perfect. Detailed crop phenology models would have given better estimates
5-12
-------
Dudek
of crop response to climate changes, but even these detailed models have not yet incorporated CO, effects. The
agro-ecological zone method employed in this study has an empirical basis; however, it has not been explicitly
calibrated for California production regions. Further, a subset of crops were selected on the basis of data
availability to act as indicator crops in the response analysis. Percentage yield changes in these indicator crops
were proportionally applied to crops with similar characteristics.
While the differential productivity enhancement of CO2 increases were included, interaction between CO2
and climate factors in productivity response were not. This omission reflects the current status of knowledge
rather than any limitation inherent to the methodology. Changes in water use efficiency as a result of stomatal
response to elevated CO2 were not included. Water use efficiency responses by crop are neither completely
understood nor widely available (Acock and Allen, 1985).
Given the need to employ detailed weather data in the evaluation of yield impacts, first order weather
stations were employed These stations tended to be located near larger urban centers, not all of which were
well correlated with agricultural production regions (see Table 4). Lastly, the response methodology did not
include possible variations in productivity due to soils or management practices.
Economic Model
The following is not an exhaustive list of the assumptions underlying this research since many are common
to all forms of economic analysis. Rather, the intent is to highlight areas of particular limitation or import for
this issue. CARM is limited by the accuracy of both GCM predictions and the performance of the productivity
impact methodology employed in this study. CARM itself, however, has other limitations. Technical change has
not been included in any of the analysis or construction of the scenario experiments. Other than simple
exponential growth rates, technical change assessments are beyond the scope of this study. In the absence of
precise dating for the doubling of ambient CO2 concentrations, even the compound growth approach is
problematic.
As previously indicated, some critical demand and supply conditions for both crops and resources have
been omitted. The export demand situation under a changed climate is unknown. Changes in underlying
consumer tastes and preferences, even using extrapolations from current trends, have not been included. On the
supply side, no crop pest changes have been introduced and no assumptions concerning the status of ambient
pollutant levels such as tropospheric ozone have been made. No attempt has been made to predict the status
of groundwater resources at the doubling time, a potentially critical omission for the integrated water analysis.
In addition, only the relatively simple predictions of ET change from the GCMs were used to portray changes
in crop water demand. Lastly, the future status of current resource problems was not projected. For example,
no attempt was made to account for the extensive drainage and salinity problems of the San Joaquin Valley
Basin. Nor was any attempt made to project the conversion of agricultural land to urban uses or the availability
of land resources in general.
There are a few sources of explicit model bias within the current structure of CARM. First, to the extent
that there are differences in price responsiveness between indicator commodities and the underlying set of crops
represented, the results will be biased. Model results have a downward bias (greater costs of climate change)
because neither new crop varieties nor planting dates were included within the range of alternatives evaluated
by CARM. Nor were new crop/location combinations not currently in evidence in California considered. Each
of these production responses would act to further reduce the impact of climate changes. Lastly, with the
exception of the grain crops, there was no national context of climate changes conditioning price signals for
California producers. Since these are the crops for which prices are set in global or national markets, national
rather than regional elasticities were employed.
Hvdrologjc Limitations
From the vantage of assessing the implications of a changing climate for California agriculture, the most
serious limitation of the hydrologjc modeling is the absence of forecasts for groundwater resources which account
5-13
-------
Dudek
for roughly 25% of the state's developed supply sources. Changes in precipitation and runoff as well as project
deliveries will affect groundwater recharge and therefore its availability and cost. For much of California's
agriculture, groundwater is the marginal supply source since it is generally more expensive than surface sources.
Further, as surface supplies are altered by climate or weather, the intensity of groundwater extraction varies.
Thus, the availability and depth of groundwater resources is central to estimating the impact of surface water
supply changes on agriculture. In the absence of other information, groundwater availability was not altered in
the economic model. However, nonlinear cost functions were introduced to reflect the increasing cost of
pumping from greater depths as extraction increases.
Other concerns surrounding the hydrologic component include the representativeness of the historic period
1951-80 used in the simulation. In hydrologic terms, this is a relatively short span of time which may not capture
all of the variability in northern California's water resources. For example, the drought of record in California's
Central Valley occurred between 1927 and 1934. In addition, in the climate change scenarios, there was no
change in the underlying variability of precipitation as only the variability contained in the historic record was
used in the simulations. For a system designed to manage the extremes of weather, this assumption may be very
inadequate in describing the nature of a changed climate under doubled CO^ atmosphere. Further, the adequacy
of a highly engineered water system like that in California depends upon its performance under a sequence of
adverse years. The linked methodology used in this study is inherently static, but since the underlying variabilities
have not changed, this is not a serious limitation.
A final set of issues concern technical assumptions employed in the hydrologic study. For example, the joint
SWP/CVP operating model uses forecasts of the needs of end use sectors such as agriculture in determining the
quantity and pattern of releases. However, the agricultural demand for irrigation water is likely to differ in
magnitude and location under a changing climate (Dudek, 1987c). The lack of feedback between the economic
and hydrologic components of this analysis likely overstate the impact of surface water supply changes. Another
key ingredient in both the agricultural and hydrologic analyses is evapotranspiration (ET). In the present study,
ET changes are derived from GCM output. However, ET may depend critically upon local and regional
phenomena such as wind. Poor ET assumptions affect both the water supply results from the hydrology study
and the agricultural demands reported in this study.
THE SCENARIOS
This case study was designed to combine the results from several detailed analyses in order to produce a
more realistic view of the potential impacts of climate change upon an irrigated agricultural system. Unique to
this study is the use of detailed hydrologic assessments of surface water changes under specific climate scenarios.
To understand both the magnitude and source of changes in such a complex system, it is important to carefully
construct scenarios as sets of precise assumptions about uncertain variables. For this study, there are four main
sources of change, each of which is identified in Table 3.
Since there are a number of competing GCMs, there are a number of alternative projections of future
climate. As has been previously discussed in the crop productivity analysis, climate differences can be very
significant in crop productivity terms. There are at least four alternative climates. In general, crop productivity
changes have been driven by climate change impacts such as the effects of temperature and cloudiness changes.
These changes have been the easiest to assess with existing agronomic models. However, the changing
concentration of CO, will also affect yields. Assessing these latter effects is the subject of significant
contemporary research. Consequently, there will be at least two sets of productivity changes, one of which will
account for the effects of CO2 in addition to climate change impacts.
Assumptions concerning water and the institutions for its management complete the specification of
variables comprising a particular scenario experiment. The hydrology study which focused on the Sacramento
Valley produced 30-year simulations of surface runoff and flows under alternative climates. Since CARM is a
static single period model ill suited for time series simulations, the 30-year hydrologic simulations need to be
5-14
-------
Dudek
translated into meaningful estimates of surface water deliveries. The limitations of time and resources for this
study dictated a focus on the mean change within the simulated period. This narrow scope does not allow an
analysis of the implication of the extremes in altered water supplies, but it does emphasize the expected change.
While physical flows are important, the set of institutions and rules governing the allocation of water supplies
are a critical determinant of the economic outcome produced under a particular water supply assumption. One
important institutional innovation in California water management would be water marketing.
Overall, there are three alternative climates, two sets of productivity impacts, two water supply
characterizations, and two alternative institutional arrangements. The ten scenario experiments described in
Table 3 were analyzed.
Table 3. Scenario Experiments
VARIABLES ALTERNATIVES
Climate Base, GISS, GFDL
Productivity Impact Climate Change Only
Climate Change plus CO2
Water Supply Base, GISS, GFDL
Institutional Response None, Water Markets
5-15
-------
Cudek
CHAPTERS
RESULTS
The primary focus of this study is to evaluate the implications of a changing climate for the agricultural and
water resources of California To this end, a set of models were linked to conduct a set of scenario experiments
each describing different climatic conditions and responses. As previously indicated, climate change is a complex
phenomenon stemming from the acumulation of trace gases in the lower atmosphere which trap radiation.
The trapped radiation acts to alter temperature, weather, and climate with implications for crop productivity,
hydrology, and the agricultural economy. One of the gases causing this change, COy also has direct effects upon
plants in general and agriculture in particular. As a result, this case study separately analyzes the effects
stemming from climate changes and net effects which include CO2 impacts.
CROP PRODUCTIVITY CHANGES
Climate Change Effects
Table 4 presents the percentage yield reductions estimated for four indicator crops, five California locations,
and two GCMs. The productivity changes driven by climate change effects without accounting for CO2
concentration effects are reported under the column headings CC. The specific mapping of these productivity
changes to CARM production regions is presented in Figure 3. In general among the indicator crops, cool
season C3 crops represented by sugarbeets were estimated to suffer the most serious productivity declines.
Sugarbeet yield reductions from climate changes ranged from approximately 21 to 40%. Sugarbeets were
estimated to have the largest individual productivity decline (40.1%) when using Blythe weather station data
Elsewhere, corn yield declines ranged from 3 to 31 percent. Cotton, the warm season C3 crop, exhibited a very
similar pattern of yield impacts from climate change as corn. Vegetable crops, represented by tomatoes, were
least affected with reductions ranging from 5 to 16%.
As generally expected, the severity of these productivity impacts from climate change effects alone were
distributed along a south to north gradient (see Figure 3). In addition, with the exception of corn, yield impacts
were greatest in interior production regions where temperatures are less modulated by marine influences.
Overall, yield reductions under the GISS 2xCO2 scenario were greater than under projections from the GFDL
GCM owing to the greater temperature increases for California produced by the GISS model. Differences were
most severe for corn and least severe for vegetables. Other than the absolute magnitude of effect, the scenarios
from the two GCMs are in general agreement.
Net Effects Including CO,
When the effect of inadvertent fertilization due to the increased CO2 concentrations is introduced, however,
some of the productivity changes alter direction, i.e., yields increase. These results are displayed in the columns
labeled NET in Table 4. For example, the warm C3 crops such as cotton and the vegetable group all generally
exhibit increases over base period yields. Overall, among the crop groups evaluated, the warm C4 crops as
represented by corn exhibited the greatest yield declines relative to base conditions. As a C4 crop, com is less
able to take photosynthetic advantage of the increased CO2. Consequently, climate change effects continue to
dominate in productivity changes for that crop. Cotton, a C3 type, would benefit the most from CO2
concentration increases with yield changes ranging from a modest 1.5% decline to a 41% increase. Differences
between the climate change and net effects scenarios were also dramatic for sugarbeets.
5-16
-------
Dudek
Table 4. Predicted Yield Changes, Agro-ecological Zone Method
REGION
SCENARIO
CROP
Sugarbeets Corn Cotton Tomatoes
%%%%%%% mean percentage yield changes' %%%%%%%
CC NET CC NET CC NET CC NET
Coastal Regions
Los Angeles
(LAX) GISS
GFDL
Interior Regions
Red Bluff
•27.4 -3.2
(2.1) (3.0)
•21.4 4.8
(2.7) (3.8)
-22.1 -17.7
(3.7) (3.9)
-2.8 3.0
(5.1) (5.3)
-21.7 11.4
-3.5 40.9
-8.3 16.9
(3.9) (5.0)
-5.3 20.4
(3.5) (4.4)
Sacramento
Fresno
Blythe
GISS
GFDL
GISS
GFDL
GISS
GFDL
GISS
GFDL
-33
(4
-25
(3
-29
(3
-24
(3
-33
(3
-32
(3
-40
-39
.6 -11
.2) (5
.5 -0
.0) (4
.1 -3
.3) (4
.5 2
.8) (5
.7 -13
.3) (4
.1 -12
• 6) (4
.1 -1
.2 -0
.3
.9)
.2
.3)
.4
• 6)
.8
.4)
.5
.3)
.8
• 7)
.8
.1
-16.8
(2.8)
-14.4
(3.2)
-14.4
(3.4)
-8.4
(3.7)
-18.9
(4.1)
-12.9
(5.3)
-31.1
-13.5
-11.5
(3.0)
-8.8
(3.4)
-8.7
(3.6)
0.2
(3.9)
-13.5
(4.3)
-7.2
(5.6)
-26.9
-7.9
-30.3
(3.9)
-26.4
(3.6)
-34.1
(3.1)
-31.5
(3.8)
-29.2
(2.7)
-25.8
(2.8)
-27.8
-18.7
3.07
(5.7)
8.9
(5.4)
-1.5
(4.6)
2.1
(5.7)
6.4
(4.0)
11.1
(4.2)
5.5
20.6
-15
(1
-13
(0
-13
(1
-12
(1
-15
(1
-15
(2
-13
-11
.6
.1)
.5
.9)
.5
.4)
.1
• 3)
.4
.8)
.1
.1)
.4
.5
9.7
(1.4)
12.18
(1.2)
13.2
(1.9)
14.8
(1.8)
10.4
(2.4)
10.5
(2.6)
12.9
15.4
The mean values reported in this table are computed from 30 years of
model simulated yield effects for the historical period 1951-80. The
figures reported in parentheses below the mean values are the standard
deviations.
5-17
-------
Dudek
Figures. Regional Productivity Changes
LEGEND
GISS Climate Change
GFDL Climate Change
GISS Net Effect
GFDL Net Effect
cool C3 afl C4 warm C3 vegetables
5-18
-------
Dudek
Since the analysis of crop productivity impacts followed the general procedure established for all impact
studies in this series, i.e., superimposing new climatic values upon historic series, the results reported in this
section have focused upon the mean changes for the 30-year simulation. The standard deviations associated with
these mean values are reported in parentheses in Table 4. When comparing the climate change only versus
climate change plus CO2 scenarios, the standard deviations of yield changes increased when CO2 effects were
included. Yield variability has been used as one measure of risk for agricultural enterprises. Using this measure,
then, risk would incrase under climate change despite the positive effects of CO2 fertilization.
From the spatial view of the results presented in Figure 3, both the coastal-interior and south-to-north
change gradients previously observed persist. Coastal region productivity for sugarbeets and tomatoes in general,
and for cotton under GFDL, is less affected than that for interior regions. Corn, a warm season crop, fares more
poorly in the hotter southern production regions represented by Blythe and Fresno. As Figure 1 illustrated, corn
exhibits the greatest tolerance and increase in productivity with temperature, but also the sharpest declines
beyond 35°C. South-to-north changes were generally observed, although not as pronounced as when climate
changes only were assessed. Overall, the pattern of results under the two GCM scenarios were consistent in the
direction of change. Net productivity impacts varied most for cotton owing to the strong productivity increases
produced when CO2 effects were included. Corn and sugarbeet yields were also significantly affected. The corn
declines for all interior regions were due to the higher temperatures predicted by the GISS model and the
reduced ability of corn to benefit from increased CO2-
WATER RESOURCE SUPPLY IMPACTS
Table 5 presents estimates of surface water supply changes in percentage terms. These estimates are
summary statistics derived from the hydrologic simulations produced by Lettenmaier et al. and Sheer and
Randall. The estimates in Table 5 refer to changes in deliveries to various regions within the state. Refer to
Figure 2 for the geographic location of these CARM regions.
Table S. Surface Water Supply Changes
SCENARIO
Sacramento
Valley
Delta San Joaquin
Valley
(Federal)
San Joaquin
Valley
(State)
South
Coast
- percentage changes -
GISS 2xCO2
mean 0.000 -0.003 0.000 -24.677 -1.125
standard deviation 0.000 0.409 0.000 10.044 0.411
maximum 0.000 1.634 0.000 -2.696 -0227
minimum 0.000 -0.789 0.000 -34365 -1.583
GFDL 2xCO2
mean 0.000 -0.030 0.000 -28.109 -1250
standard deviation 0.000 0349 0.000 11322 0.467
maximum 0.000 -1535 0.000 -3.209 -0377
minimum 0.000 -0.233 0.000 -48537 -2.112
Source: Dan Sheer, Water Resources Management, Inc., personal communication, March 11,1988. Percentage
changes calculated from the 1951-80 simulations run for a changed climate according to the scenarios identified.
In each case, the reference scenario is the GISS lxCO2 run.
5-19
-------
Dudek
From Table 5, it is clear that the major supply impacts predicted from these hydrologic simulations would
occur in state water project delivery areas primarily in the San Joaquin Valley. Federal CVP service area
deliveries are not similarly affected because of differences in the seniority of water rights. The GISS scenario
produces a hydrology with increased surface water flows seasonally shifted. Flow is increased in winter months
and reduced in spring and summer. Effectively, the warmer temperatures result in the loss of storage provided
by snowpack.
The need to provide flood control storage in reservoirs particularly in the state water project facility at
Oroville does not permit retention of the increased winter flows with existing storage capacity. Consequently,
state water project deliveries would be reduced by approximately 384,000 acre-feet. Results under the GFDL
2xCO2 scenario were similar, but with slightly greater water supply impacts due to less severe temperature
changes and decreased precipitation. Total SWP deliveries are reduced by some 438,000 acre-feet under the
GFDL scenario.
ECONOMIC IMPACTS: AGRICULTURE AND RESOURCES
The preceding results have focused on climate change impacts upon biologic and hydrologic systems. This
section describes the implications of those productivity and water supply changes for California's agricultural
economy. This portion of the case study is a departure from previous sections since it emphasizes the collective
responses of both agricultural producers and consumers to the changes comprising each scenario. The
interaction of supply and demand responses in a market economy is critical to correctly estimating the nature
and magnitude of climate change impacts. Productivity declines first act to reduce farm revenues by decreasing
the quantity of produce available for sale. The precise impact upon farm revenues depends upon the
responsiveness of demand and supply, particularly from competing regions. In order to maximize farm
profitability, growers switch among crops matching their most profitable alternatives with available resources.
They also respond by altering management techniques and therefore production as well as costs. In this case
study, irrigation system options are assessed. Consumers respond to price changes both by altering consumption
and by substituting among food items. These price changes are also received by growers who adjust their
production plans accordingly.
Aggregate Results
The result of the operation of this market system is best illustrated by comparing the statewide average yield
changes from CARM (Figure 4) with the crop productivity results from the Doorenbos and Kassam model.
Figure 4 presents the statewide average yields produced by CARM under alternative scenario experiments - the
GISS and GFDL 2xCOy climates coupled first with productivity changes driven by those climate changes only
and then with CO, effects included (discussed in section E). In each case, evapotranspirational changes
predicted by those GCMs and the mean surface water deliveries under each climate were included.
Overall, the results for the two climate change scenario experiments are broadly similar with consistent
directional changes. In contrast to Figure 3 and Table 4, which displayed the direct estimated crop productivity
changes from the modified Doorenbos and Kassam model, Figure 4 presents the statewide average yields that
result after economic incentives have operated to reallocate production activities and resource use. As such it
is a good example of how the economic system operates to modulate direct productivity changes. Corn, for
example, was predicted to have productivity declines between 14 and 31% depending upon location under the
GISS 2xCO2 climate projections. CARM produced statewide average corn yields roughly 18% lower than the
base run, an overall result which is very close to the lower bound of direct productivity impacts under the GISS
scenario. Results were similar under the GFDL scenario. Corn productivity impacts ranged from reductions
of from 3 to 135%. The CARM analysis produced a statewide average decline of only 10%.
5-20
-------
Dudek
40
Figure 4. Statewide Average Yield Changes
£ 20
I
-20
-40
T
_L
_L
Broccoli Lettuce Tomatoes Grapes Peaches Cotton Com Wheat
Cantaloupes Potatoes Almonds Oranges Walnuts Sugarbeets Rice Alfalfa
Figure 5. California Production Changes
Broccoli Lettuce Tomatoes Grapes Peaches Cotton Com Wheat
Cantaloupes Potatoes Almonds Oranges Walnuts Sugarbeets Rice Alfalfa
LEGEND QIS8 Climate Change GFDL Climate Change GISS Net Effect
GFDL Net Effect
5-21
-------
Dudek
Figure 5 depicts the associated aggregate commodity production changes referenced to the 1985 base run.
Production changes largely follow the pattern of underlying yield changes as modified by alternative production
opportunities and price changes. Again, the trend in both climate change only scenario experiments was broadly
similar. Almonds, peaches, walnuts, sugarbeets, and alfalfa hay are particularly large losers (productivity changes
in these crops were represented by the sugarbeet effects listed in Table 4).
Figure 6 presents crop acreage changes in percentage terms. The pattern follows that discussed for average
yields and production with the exception of increases in selected crops. The increases in grain crops are
particularly noteworthy and stem from the use of national demand elasticities for these crops to reflect a national
context of simultaneous stress across the national agricultural economy. The result is significant price increases
for the grain crops as displayed in Figure 7. Figures 6 and 7 taken together are another measure of the
homeostatic powers of market systems to modulate direct physical impacts. Overall, the results presented in
Figures 5 and 7 are inversely related. Production declines are matched by price increases with the extent
depending upon the magnitude of decline and the price elasticity of demand.
CARM measures aggregate economic values in terms of changes in values received by both producers and
consumers, i.e., producers' and consumers' surplus (CPS). The changes described for the state as a whole under
the GISS 2xCO2 scenario with climate change effects resulted in a 17% decline in CPS. Losses under the
GFDL scenario were only slightly lower (see Figure 10). Total crop acreage was reduced by 6.7% under GISS
and by 3% under GFDL. Given these acreage reductions and the water supply impacts, both ground and surface
water use were reduced for the state as a whole under both scenarios. Groundwater use declined by roughly
20% in both GISS and GFDL scenario experiments. Surface water reductions, largely determined by hydrologic
changes, that reduced state water project deliveries, were closely matched at close to 17% for the two GCMs.
Regional Results
While the statewide average results are useful to appreciate the overall impacts of the individual scenarios,
one of the real strengths of CARM lies in its detail of regional production activities and resources. As the crop
productivity results indicated, there are significant differences in the strength of impacts on crop yields in
different locations. These impacts are further reinforced by water supply reductions which mainly affect the state
water project service area in the San Joaquin Valley. Figure 8 presents some of the regional detail underlying
the statewide changes. Crop acreage shifts for five major crop groups are displayed.
The greatest proportional changes under all scenarios occurred in the Delta and in the Imperial Valley.
In the Delta region, strong price increases for grains encourages switching away from other field crops and fruits
and nuts. Tree fruits and nuts move from the warmer interior valley regions to the cooler coastal production
regions such as the Central Coast where productivity impacts are smaller, but still substantial. For the field
crops, the large declines reflect the poor performance of cool season crops like sugarbeets, one of the principal
field crops modeled. Vegetable crop acreage is the least affected of all crop groups since productivity was least
affected under all scenarios. At the other extreme, hay is the most volatile crop, disappearing almost completely
from some production regions under some scenarios (the climate change scenarios for the Central Coast and
Imperial Valley regions). Hay is generally the marginal crop in irrigated regions and would be expected to
decline under productivity declines or water shortages. Certainly, this decline would depend upon demand from
the livestock sector and the availability of substitute feed from more distant regions.
In contrast, when viewed in absolute terms, the largest changes under the climate change scenarios occur
in the northernmost production region, the Sacramento Valley. Under the GISS climate change scenario, ma
acreage is reduced by 511,000 acres from base conditions. Increases in grain acreage make up for 337,000 acres
of this loss. Nonetheless, the region overall loses 325,000 acres of previously cropped land. The shifts between
hay and grain acreage in the southernmost Imperial Valley are second in size under the climate change scenarios
with 293,000 acres of hay lost and 168,000 new grain acres. After CO, effects are included, the largest change
still occurs as in hay acreage in the Sacramento Valley with a 235,000-acre loss under the GISS net scenario.
Hay acreage reductions in the Imperial Valley remain the second greatest impacts despite the potentially
beneficial direct effects of CO2. These results illustrate the fact that although productivity impacts fall along a
5-22
-------
Figure 6. California Acreage Changes Dudek
40
20
-20
•40
-60
1
I
j I I I I I
I i
Broccoli Lettuce Tomatoes Grapes Peaches Cotton Com Wheat
Cantaloupes Potatoes Almonds Oranges Walnuts Sugarfoeets Rice Alfalfa
Figure 7. California Commodity Price Changes
so
60
40
20
o
-20
r
I I
_ • •,_
r»
l
J I
Broccoli Lettuce Tomatoes Grapes Peaches Cotton Com Wheat
Cantaloupes Potatoes Almonds Oranges Walnuts Sugarbeets Rice Alfalfa
QlSS Climate Chanae GFDL Climate Change QISS Net Effect QFDL Net Effect
5-23
I
-------
Dudek
Figure 8. Regional Acreage Changes
Sacramento Valley
vegetables field crops hey
fruits &. nuts grains
vegetables field crops hay
fruits & nuts grains
Northern San Joaquin
Central Coast
vegetables field crops hay
fruits & nuts grains
vegetables < field crops hay
fruits & nuts grains
South Coast
Southern San Joaquin
vegetable* I eld crept hey
fruits & nuts groins
Imperial
Valley
LEGEND
GISS Climate Change
GFDL Climate Change
C3
GISS Net Effect
GFDL Net Effect
vngMbln lild crop.
5-24
-------
Dudek
south-to-north gradient, that fact is no cause to expect that northern regions will necessarily benefit in any way
from climate change.
Figure 9 portrays the spatial pattern of resource use changes that might result from climate changes
depicted by the GISS and GFDL GCMs. The values depicted are an index constructed from the ratio of
scenario results to the 1985 base period. Overall, the Imperial Valley is the region most severely affected,
experiencing nearly uniform reductions in crop acreage and ground and surface water use across all scenarios.
The Delta and South Coast regions exhibit the least amount of change, but they represent small proportions of
the state's agricultural economy.
The most dramatic changes occur in groundwater use in the Northern San Joaquin Valley between the
scenarios based upon climate change impacts only and those which include CO2 effects. These shifts are driven
both by crop acreage increases in general and by alfalfa hay acreage differences in particular (see Figure 8).
Alfalfa is only exceeded by rice in its water requirements. For those regions relatively less disadvantaged by
climate changes, in both productivity and water supply terms (northern and coastal regions), resource changes
may be beneficial. For example, in much of central California groundwater aquifers are routinely overdrafted,
i.e., pumped at a rate greater than recharge. In many of these regions, groundwater is more expensive than
surface water and so groundwater pumping is reduced. These reductions are accomplished through a
combination of cropping pattern and on-farm irrigation efficiency changes.
SOCIETAL RESPONSES TO CLIMATE CHANGE
Given the water supply stresses expected to be caused by climate change, this case study has focused
additionally on an evaluation of mitigation measures. Several alternatives ranging from the construction of new
reservoir capacity to improved on-farm irrigation efficiencies can be identified. This study examines the potential
of both water marketing and on-farm irrigation efficiency improvements to reduce the impact of climate changes.
Irrigation efficiency improvements are both technically and economically proven (Dudek and Homer, 1982). The
obstacles to more widespread adoption of such methods have included the lack of market opportunities for
conserved water (Willey, 1985). Specific barriers to water marketing and remedies are discussed in Chapter 4.
This linkage between on-farm investments in improving irrigation and the reform to allow water tranfers prompts
their joint consideration. Water market reform has already been identified as the single most cost-effective
means of meeting water demands (Howitt et al., 1980).
Previous studies of water marketing have used more aggregate methods with less detail in the agricultural
sector (Howitt et al., 1980). Since increases in urban demand are already factored into the WRMI model and
results from Sheer and Randall, this case study focuses on water transfers between agricultural regions. As
Howitt et al. note, there are substantial opportunities and gains possible from trading water between agricultural
and urban users (see Chapter 4 as well). To accomplish this analysis, CARM was modified to include a range
of alternative irrigation application systems as appropriate for each crop and location. The systems evaluated
included hand move sprinklers, 1/4 mile furrow with recycling, 1/2 mile furrow with recycling, each of the furrow
systems without recycling, drip irrigation, and a border system. Irrigation efficiencies were varied by application
system and location. They ranged from a high of 80% for drip irrigation in the South Coast and Southern San
Joaquin regions to 50% for some border systems and furrow irrigation with 1/2 mile runs (Sacramento Valley,
Delta, and Imperial Valley). Capital and operation costs, net of water costs, varied by system as well from a
low of $17 per acre for 1/2 mile furrow on some crops to a high of $80 for drip irrigation of grapes. For water
marketing, publicly developed surface water supplies were offered for sale F.O.B. Sacramento Delta to the
highest bidder.
Figure 10 portrays the statewide average changes for each scenario after water marketing is introduced.
Note that a base model is included for comparison as well. The combination of water marketing plus irrigation
efficiency improvements yields between 1.3 and 1.6% increases in consumers' and producers' surplus. Since the
water market is geared to sales to agricultural users only, the equilibrium prices prevailing under each scenario
reflect differing capacities to pay. Under base conditions, water marketing produced a price of $35.15 per acre-
5-25
-------
Dudek
Figure 9. Regional Resource Changes
GISS Climate Change
LJ
GFDL Climate Change
\
\ Southern San Joaquin
\
\
1
i
j
•*-
-
\
":
-T-,
\
\
|
1
1
1
-j
•
crop acreage groundwater surface water
5-26
-------
Figure 10. Overall Results
Dudek
ACREAGE
8
o
0)
14
o
*i
! 2
g
ro
§ 0
2 -2
•*- *•
O)
I -4
o
ACREAGE
GROUND
WATER
SURFACE
WATER
CPS
AVERAGE
WATER
USE
GROUND SURFACE
WATER WATER
CPS
AVERAGE
WATER
USE
LEGEND g
!•:
[•
S BASE
GISS-NET
B GISS-CC
HH GFDL-NET
111 GFDL-CC
5-27
-------
Dudek
foot. Under the two climate change scenarios, water market prices declined to $27.35 For the scenarios which
included CO2 effects, prices rose to $31.78. It is important to note that these prices do not presume recovery
of the substantial capital investments in the CVP or SWP. Further, these values are in sharp contrast to the
costs of alternative new surface water supplies, which are estimated to range between $220 per acre-foot for an
enlarged Shasta and $559 for Los Vaqueros in capital costs alone (Willey, 1985). These expansion projects
have been typical of the suggestions offered by the traditional water supply agencies such as the U.S. Bureau of
Reclamation and the California Department of Water Resources when confronted with the potential prospect
of dealing with climate change. Water marketing alone would allow California's agricultural economy to increase
the value it produces.
Overall, the results after water marketing are broadly similar: irrigated acreage decreases under climate
change impacts and increases slightly with CO2 effects, groundwater pumping is increased, and surface water use
declines. Average water use per acre changes very little when comparing the marketing scenarios with those
under which water is allocated by traditional means. These results are further amplified by the rough
displacement of surface water supplies with increased groundwater pumping as the relative cost of these
competing water supplies changes under water marketing.
Figure 11 displays the regional changes resulting from water marketing and irrigation efficiency investments.
Again, both the Sacramento and Imperial Valleys show the greatest responses. In each case, crop acreage is
increased, groundwater pumping is reduced, and surface water is increased. Trends in these two regions are
clearly in opposition to the overall result. The Sacramento Valley is the region with the greatest potential for
expansion of agricultural operations and it benefits from a market system of resource allocation which allows it
to bid and acquire resources and forces others to pay higher prices for previously subsidized water supplies. The
Imperial Valley benefits because it can substitute water purchased in the market, expensive though it is for more
expensive groundwater, and so expand crop acreage profitably.
The Northern San Joaquin region experiences the greatest proportional changes of any region in its
groundwater resources. Under the climate change scenarios with water marketing, groundwater pumping
virtually doubles. This region of California is one which has benefited from extremely low surface water prices.
When the region must suddenly face an economic competition for those supplies, it turns to groundwater and
nearly doubles extraction. Nonetheless, final pumping levels are still only roughly 60% of base period levels after
water marketing.
The Southern San Joaquin Valley has a pattern of changes which are the mirror image of those occurring
in the Sacramento Valley. Crop acreage declines while groundwater pumping increases and surface water use
decreases. This pattern reflects the region's diminished ability to compete for marketed public surface water
supplies which were previously provided through substantial subsidies. Further, it highlights the importance of
the future of groundwater resources in responding to the shifting resource potentials imposed by a changing
climate. In the Southern San Joaquin, average water use is reduced due to cropping pattern shifts to less water-
intensive crops and to more efficient irrigating methods. It is important to remember that the San Joaquin
Valley suffers current severe drainage, salinity, and toxic trace element problems which are irrigation related.
Improvements in irrigation efficiency would also have beneficial effects for these problems by reducing irrigation
return flows (Dudek and Homer, 1982).
INCLUDING CARBON DIOXIDE EFFECTS
Throughout the preceding discussion reference has been made to the differences between scenarios
evaluating climate change effects only and those which also included CO, (termed net). At the onset, it was
clear that with two different sets of crop productivity impacts under the two general scenario types that the
results would be different (see Figure 3). Most seriously divergent were the productivity results for vegetables.
However, vegetables represented one of the most durable cropping patterns owing to the fact that the impacts
were not as great in the first place. Overall, as illustrated in Figure 10, crop acreage was reduced for all
scenarios.
5-28
-------
Dudek
Figure 11. Resource Changes Under Water Marketing
Sacramento Valley
Northern
San Joaquln
South Coast
Southern San Joaquin
LEGEND
GISS Climate Change
GFDL Climate Change
GISS Net Effect
GFDL Net Effect
Imperial Valley
CrapAcrgi Qroundwatw Surtacs Water AvtrigtWanr
5-29
-------
Dudek
This case study highlights the importance of the accuracy of predicting crop responses to the set of
simultaneous changes that will accompany the greenhouse effect. The methodology used in this study is an
approximation developed from a model designed initially to assess irrigation needs. Its results seem to
demonstrate substantial downside risk from either climate change effects or those which also include CO2.
These results further emphasize the need to assess any physiologic interactions between the set of stresses
induced by climate change and the inadvertent fertilization provided by elevated CO2 concentrations.
5-30
-------
Dudek
CHAPTER 4
POLICY IMPLICATIONS
AGRICULTURE
While technologic change has not been directly analyzed in this case study, it is clear that improved
production techniques that reduce costs are critical. Clearly, improved irrigation application methods blunted
some of the shocks analyzed with the model. Productivity improvements at a pace sustained in recent decades
in agriculture would offset the impacts of even the more pessimistic climate change only scenarios. However,
while some believe that new developments in biotechnology will aid in sustaining this pace of research, these
gains have not been costless. Society has made huge investments in agricultural research and development and
its dissemination through an elaborate extension service. Public willingness to continue this level of support
has been questioned and is not assured, but climate change impacts require an invigorated and targeted research
program. Also uncertain is public acceptance of new agricultural technologies. Bovine somatotropin (bst) has
demonstrated the capability of boosting per animal milk output, yet it is not clear that consumers wUl accept this
production technique. Similarly, management practices which rely upon increasing chemical use are likely to
come under increased regulatory if not market pressure. Both the human health and environmental implications
of chemical use are likely to weigh heavily in determining the feasibility of alternative production responses.
Of particular importance is research on water use efficiency under climate change, an area of active inquiry
and controversy. Improvements in irrigation management are critical given the expected water supply reductions
and the uncertainties surrounding groundwater availability. In addition, irrigation runoff is a primary source of
environmental insults ranging from toxic trace elements to nutrients. Removing public subsidies to agriculture
whether provided directly in the form of price supports or indirectly in the form of subsidized water would
contribute to improved efficiency of water use. The improved responsiveness provided by market signals would
also facilitate the regional transitions that are likely to occur. The challenge of this magnitude of policy reform
looms large.
Further, agricultural operations do not have a benign impact on the environment (Dudek, 1987b). Policies
to manage nonpoint sources of pollution such as sediment, fertilzer, and pesticide residues, and salinity and toxic
trace elements have languished. Nonpoint source water quality problems have been identified as the single
greatest impediment to achieving the nation's water quality objectives. The regional shifts in agricultural
production intensity and cropping patterns described in this case study imply changes in the geographic
distribution of nonpoint source loadings and resulting damages as long as current policies are pursued.
Cross-compliance provisions requiring farmers to reduce off-farm loadings in exchange for the benefits provided
by other public programs should be strengthened.
WATER RESOURCES
The water resource component of this study dearly demonstrates the need to question the long-term climate
assumptions employed in water resource planning. Further, the results of this case study have illustrated the
importance of the flexibility provided by market incentives as a means of buffering the effects of climate changes.
They enable efficient reallocations of resources and provide growers with incentives for the adoption of more
efficient irrigation techniques. The following sections explore these issues in the context of evolving California
water conflicts.
Existing Legal and Institutional Setting
The California State Constitution states in Article X that the state's waters shall be put to "beneficial and
reasonable" uses. The system of state water rights together with entitlement water for the State Water Project
5-31
-------
Dudek
(SWP) and contract water for the Federal Central Valley Project (CVP) is loosely regulated according to this
criterion by the State Water Resources Control Board (WRCB) through its water rights permit system. Several
types of usufructuary water rights have been established for waterfront property owners (riparian), groundwater
pumpers (correlative), and diverters (appropriative). All state water rights, including those of the federal
government for CVP appropriative rights south of the Delta, are governed by the Board.
There is a considerable history of support for adjustments in this system of state and federal water supply
allocation to allow voluntary transfers and marketing. The Governor's Commission on Water Rights
recommended in 1978 that legislative changes be made to facilitate such transfers. During the 1979-86 period,
new legislation addressed, in a somewhat piecemeal fashion, various facets of state water law which had
supported the "use-it-or-lose-it" status quo and fear of transfers among water rights holders. Individual legislative
initiatives addressed such related topics as allowance of water rights maintenance with non-use resulting from
water conservation; defining voluntary water transfers as "beneficial and reasonable"; directing the State
Department of Water Resources (DWR) to facilitate water transfers through various supportive services for
water rights holders; and requiring public water supply conveyance facilities to wheel transferred water with fair
compensation where conveyance capacity exists.
At the same time, relevant policies supportive of water transfers have emerged elsewhere. The Western
Governors' Association recommended in 1986 that water transfers should be facilitated as a means of
encouraging efficiency of use. The U.S. Bureau of Reclamation recommended support for water marketing of
reclamation contracts in a new policy document released in late 1987. Other states, such as Nevada and
Colorado, have developed significant markets for water transfers in recent years.
Status of Water Transfer Activities
Despite recent legislative reforms, water transfer activities in California remain very limited. Inter-regional
and inter-basin transfers, other than those already established by the operations of the CVP, the SWP, the Los
Angeles Aqueduct, and the Colorado River Seven Party Agreement, remain in the planning/negotiation phase.
The history of conflict between rural water supply sources and California's growing urban demands is one factor
in this inactivity. The acquisition of "water ranches" in the Owens Valley by the Los Angeles Department of
Water and Power during the 1920's and 1930's and ensuing depletion of the common property groundwater
resources of the basin is perhaps the most representative episode in this conflict.
Despite this antagonism and other barriers to water trades noted below, there is an increasing amount of
discussion and negotiation regarding specific potential water trades in several regions of the state. The following
are some key examples:
o Imperial Irrigation District (TTD) to Metropolitan Water District (MWD) of Southern California or
others -- This is the first and potentially largest single transfer to be discussed during the legislative
reforms enacted during the 1980's. As much as 400,000 acre-feet have been estimated as potential water
supply availability in the Imperial Valley as a result of various irrigation system investments and
management programs. The State Board processed a petition charging violation of "beneficial and
reasonable use" in the IID in 1984, and required water conservation plans by the FID. Water marketing
to the Los Angeles metropolitan area, most likely to the MWD, would provide the financial ability to
invest in conservation facilities in the IID.
o Berrenda Mesa Water District (BMWD) to Marin Water District or others - BMWD announced in
1987 that it intends to sell 50,000 acre-feet of its SWP entitlements. Marin has agreed to discuss this
possibility, as have several other urban districts. Legal controversy regarding the respective rights to
SWP entitlements of BMWD, the Kern County Water Agency, and the SWP Contractors surrounds this
proposal.
o Kern County Water Bank ~ This proposal is being pursued by the SWP to deliver surplus water to
groundwater recharge areas in Kern County, thereby facilitating pumping of this stored water during dry
5-32
-------
Dudek
years. Negotiations for acquisition by the SWP of a large area of land presently held by the Tenneco
Corporation are underway. Issues of definition of surplus water and marketing rights to dry year
pumping remain.
o Los Angeles Department of Water and Power (LADWP) - In an attempt to resolve longstanding
disputes concerning its diversion of water from the Mono Lake Basin, a unique natural resource
providing critical habitat for waterfowl and shorebirds, the LADWP and the Mono Lake Group (MLG)
are seeking replacement water supplies of up to 100,000 acre-feet per year. A portfolio of water rights
acquisition options is presently being developed by the Environmental Defense Fund for LADWP and
the MLG, and negotiations with individual agricultural water rights holders began in 1988.
Barriers to Water Transfers and Prescriptions for Change
There are a number of reasons why a statewide market for water transfers is not developing rapidly in
California Some or all of the following sets of factors are often involved in protracted negotiation or failure of
seemingly beneficial water transfers. Proposals for removal of these barriers are noted as well.
o Groundwater rights - With several important exceptions, such as the groundwater basins of the Los
Angeles region, most groundwater in California has not been adjudicated. Since the resource remains
in a quasi-common property status in these areas, the interface between rights to sell private rights to
surface water on the one hand and to pump unadjudicated groundwater on the other is a problem which
can hinder approval of some water transfer proposals. Adjudication of groundwater rights in areas
selling water rights may be the ultimate answer to this problem, but this involves expensive and
time-consuming procedural and technical processes. In the interim, it may be necessary to include
stipulations about groundwater usage as part of an application for transfer of water rights before the
State Board. As indicated in this case study, this could be an issue of increasing contention under
climate change.
o State and federal water projects ~ Along with the complexities of the private water rights system in
California, the state's water system is also heavily influenced by the operations of two large public water
supply and distribution systems. The State Water Project (SWP) allocates its water supplies according
to a scheme of entitlements established nearly three decades ago. That scheme has remained fairly rigid,
and can be altered only under a consensus arrangement among the SWP contractors and the California
Department of Water Resources. Although entitlement holders have been repaying the costs of SWP
facilities, their legal right to sell their entitlements to the highest bidder has been questioned by state
water bureaucrats and influential contractors, most notably the MWD. The resistance to the Berrenda
Mesa Water District's proposal noted above is a case-in-point. The introduction of marketing flexibility
into the SWP entitlement scheme will probably require some combination of legislative reforms and
litigation.
The Federal Central Valley Project (CVP) is also a system based on Fixed contractual rights of the
reclamation districts which it was originally intended to serve. The right of these districts, and/or of the
individual farmers which they serve, to market federal contract water is an unsettled issue. While there
are a number of individual irrigators within reclamation districts who are interested in leasing or outright
sales of their contractual rights, there is often resistance from others within the district or from outside
political and bureaucratic forces. The U.S. Bureau of Reclamation would have to agree to such
transfers, and policies pronounced by the Department of Interior in 1987 are favorable to voluntary
transfers of reclamation water. Such transfers have occurred in other states, with approval of the
Secretary of the Interior. The barriers here appear to be political and not legal.
The state and federal systems reached an agreement in 1986 called the Coordinated Operating
Agreement (COA) which was ratified by Congress (H.R. 3113). The COA allows the SWP to purchase
up to 250,000 acre-feet per year from the CVP for about 20 years. The CVP is also allowed to sell
additional water, which subsequently has been determined to be up to 1 million acre-feet per year.
5-33
-------
Dudek
Since the water transfers and sales allowed by the COA will occur at rates which are consistent with the
financial requirements of the CVP, those rates will be cost-based and subsidized and, therefore, will be
considerably lower than water prices which would result from a value-driven market setting. The COA
water transfers will be based upon bureaucratic as opposed to market allocations. While this will have
beneficial effects for those fortunate enough to be able to acquire this cheap water, it will also have the
effect of stifling the development of market-based transfers for a considerable period of time. A
combination of allocation of some COA water for mitigation of the environmental costs of the CVP
together with a bidding process for COA water would alleviate this problem. This issue is now under
active discussion in the context of the U.S. Bureau of Reclamation's Environmental Impact Statement
process for its COA water marketing program.
o Third-party issues ~ Most proposed water trades involve impacts on other water users within the seller's
region. Surface return flows or groundwater recharge which in turn supplies other water uses are often
affected. Legal rights of these third parties are variable, but.their political power is sometimes
significant. Transfers of consumptive uses only may be the solution.
o Local economic impacts — Fear of adverse impacts on some elements of the local economies of potential
water-selling areas is frequently an additional political barrier. In addition, county property tax revenue
deterioration could result if agricultural lands are retired from production as part of a water transfer.
In-lieu taxation of retired lands ~ continued payment to the county of these taxes - and establishment
of rural economic trust funds are often proposed as compensation for these impacts. This amounts to
a severance tax placed on water transfers, and would result in reduced amounts of such transfers.
o Public trust issues ~ Court rulings during the 1980's have opened the possibility that existing water rights
may be usurped at least in priority of right by what are often undefined public rights to water for a
variety of uses. Important cases involving public trust rights for Mono Lake and the Delta have become
a consideration in defining and negotiating the transfer of private water rights. Public trust protection
based on acquisitions, as opposed to taking, may resolve this apparent inconsistency.
o Uncertainty of water rights and of transfer contracts -- Water rights do not have the status of private
property rights in California Appropriative rights, which are the prime candidates for transfer, are
evidenced by permits for time, place, and amount of diversion and manner and place of use which are
issued and enforced by the State Board. These permits are not equivalent to real estate deeds as they
are in other states, such as Nevada. Legislation or judicial findings defining water rights permits as real
property may be the only solution to these uncertainties.
Water policy reform is an active element of California's contemporary political scene. It offers substantial
benefits for many environmental and resource problems plaguing the state. The benefits that water marketing
offers in mitigating climate change impacts is only one more reason to expedite these reforms.
THE ENVIRONMENT
Changes in surface water supplies and the location of water use in California were highlighted in the results.
These changes are likely to have two main environmental impacts. One would be the demand for more dams
and reservoirs. The few remaining free-flowing rivers support natural ecosystems and recreational opportunities
increasingly prized for their scarcity. Thus, we would expect increased conflict over the allocation of water
resources to maintain ecosystem values and wildlife. Temperature increases will affect both managed and
unmanaged watercourses with possible impacts on cold water species distribution.
While changes in crop productivity, soil moisture availability, precipitation patterns, and water resource
availability are expected to have profound impacts upon the location of agricultural production, these impacts
may be even larger upon ecological systems. Areas currently devoted to wildlife habitat due to their marginal
5-34
-------
Dudek
potential for agricultural development may come under pressure as climate changes alter favorably their
suitability for crop production. The basic carrying capacity of existing habitats may be substantially altered.
We do not yet fully comprehend the consequences of all of the potential environmental changes taken
together. For example, natural systems will be stressed simultaneously by increased ultraviolet radiation which
is harmful to almost all life forms (Teramura), a changed climate with increased temperatures and altered
rainfall and weather patterns, as well as more traditional pollutants such as photochemical oxidants and acid
deposition. We have made huge investments in the conservation of natural environments and the species that
they are designed to protect in this country. We have committed vast sums for the maintenance of ecological
reserves, wildlife refuges, and national parks. Wild plant and animal species heavily dependent upon protected
areas may be severely disrupted as these biospheric changes cause ecological shifts. Land-use patterns outside
these protected areas may prevent species from "retreating" to other suitable habitats. Taken together, these
effects may mean that the considerable sums already "sunk" in conservation efforts will be lost. This is but a
single measure of the severity of the problem and the resources at risk. Shifts in the location of agricultural
production could play a substantial role in the future viability of these natural systems.
5-35
-------
Dudek
CHAPTERS
CONCLUSIONS
The conclusions from this case study are best presented against the backdrop of an overview of the results
of the scenario experiments. Figure 10 summarized these results with reference to the base period. From the
preceding sections, it is clear that the results differ in magnitude only depending on the basis of the treatment
of carbon dioxide effects. Each of the scenarios analyzing climate change productivity impacts alone (labeled
CC) show acreage, water use, and economic values reduced. Those which include CO2 effects (termed net)
have less severe economic impacts. Water use reductions persist since reduced SWP surface water deliveries are
unaffected by the treatment of COL. Consequently, in all scenarios water use is affected with resulting regional
impacts which may be amplified by California water politics. These results demonstrate the importance of
integrating agricultural impact assessments with studies of changes in the availability or suitability of critical
agricultural resources.
This study also investigated the potential of changes in both on-farm management and public policy to
mitigate the impacts of climate changes. As Figure 10 illustrates, scenarios which incorporate water marketing
produce greater economic surpluses and thus blunt the force of climate change impacts. As that analysis
indicates, the benefits of current reform would not be inconsequential. However, in each case groundwater
pumping also increases. For regions such as the San Joaquin Valley which already have overdraft problems, the
economic feasibility of continuing current trends is uncertain. Water marketing and irrigation efficiency
improvements are not a panacea for climate change impacts upon agriculture. Nonetheless, proposed
conservation investments such as those described for the Imperial Irrigation District could be right on the mark
given potential impacts on that region.
This study has demonstrated that economic feedback is a critical element of impact assessment. In addition,
the introduction of markets can facilitate adjustments and reduce costs. However, the scope of resources at risk
in California from climate change is narrowly construed in this study and there are a diversity of values not
captured in market processes. The potential for increased stress on natural systems and species is a likely
outcome and one for which market mechanisms are poorly developed.
5-36
-------
Dudek
REFERENCES
Acock, B. and L.H. Allen, Jr., "Crop Responses to Elevated Carbon Dioxide Concentrations," in Strain, B.R. and
J.D. Cure, editors, "Direct Effects of Increasing Carbon Dioxide on Vegetation," VS. Department of Energy,
DOE/ER-0238, pp. 99-116, December 1985.
Adams, R.M., A Quadratic Frogramniing Approach to the Production of C^ljfoniia Field and Vegetable Crops
Emphasising Land. Water, and Energy Use. Ph.D. thesis, Department of Agricultural Economics, University of
California, 1979.
Bolin, B., J. Jager, and B.R. Doos, "The Greenhouse Effect, Climatic Change, and Ecosystems: A Synthesis of
Present Knowledge," B. Bolin et al. (editors), The Greenhouse Effect, Climatic Change, and Ecosystems. SCOPE
29, John Wiley & Sons, New York, 99.1-32,1986.
California Department of Food and Agriculture, "California Agriculture Statistical Review 1986," Agricultural
Statistical Service, Sacramento, October 1987, 28 pp.
Cure, J.D., "Carbon Dioxide Doubling Responses: A Crop Survey." In Strain, B.R. and J.D. Cure, editors,
"Direct Effects of Increasing Carbon Dioxide on Vegetation," UJS. Department of Energy, DOE/ER-0238, pp.
99-116, December 1985.
De Wit, C.T., "Photosynthesis of Leaf Canopies," Agricultural Research Report 663, Pudoc, Wageningen, 57 pp.,
1965.
Doorenbos, J. and A.H. Kassam, "Yield Response to Water," FAd Irrigation and Drainage Paper No. 33, FAO,
Rome, 1979.
Dudek, D J., "A Preproposal to Research Climate Change Impacts Upon Agriculture and Resources: A Case
Study of California," Environmental Defense Fund, New York, 14 pp., May 1987a.
Dudek, D J., "The Ecology of Agriculture, Environment, and Economy," background paper submitted to the
Technical Workshop, "Developing Policies for Responding to Future Climatic Change," Villach, Austria, 25 pp.,
28 September - 2 October, 1987b.
Dudek, DX, "Assessing the Implications of Changes in Carbon Dioxide Concentrations and Climate for
Agriculture in the United States," paper presented at the First North American Conference on Preparing for
Climate Change: A Cooperative Approach, Washington, D.C., 28-29 October 1987c, 26 pp.
Dudek, D J. and G.L. Homer, "An Integrated Physical-Economic Systems Analysis of Irrigated Agriculture,"
K.H. Zwimmann (editor), Nonpoint Nitrate Pollution of Municipal Water Supply Sources: Issues of Analysis
and Control International Institute for Applied Systems Analysis, Laxenburg, Austria, Chapter 11, pp. 247-99,
September 1982.
Homer, G.L., D. Putler, and S.E. Garifo, "The Role of Irrigated Agriculture in a Changing Export Market," ERS
Staff Report AGES850328, Economic Research Service, USDA, 31 pp., June 1985.
Howitt, R.E. and P. Mean, "Positive Quadratic Programming Models," Working Paper No. 85-10, Department
of Agricultural Economics, University of California, Davis, 1985.
Howitt, R.E., D.E. Mann, and HJ. Vaux, Jr., "The Economics of Water Allocation," EA. Englebert (editor),
Competition for California Water. University of California Press, Berkeley, CA, 1980.
5-37
-------
Dudek
Kassam, A.H., "Net Biomass Production and Yield of Crops," Present and Potential Land Use by Agro-ecological
Zones Project, FAO, Rome, 1977.
Kimball, B., "Carbon Dioxide and Agricultural Yield: An Assemblage and Analysis of 770 Prior Observations,"
WCL Report 14, Water Conservation Laboratory, Phoenix, November 1983.
Mearns, L. et al., "Extreme High Temperature Events: Changes in their Probabilities with Changes in Mean
Temperature," Journal of Climate and Applied Meteorology, vol 23, pp 1601-13, 1984.
Manabe, S. and R.T. Wetherald, "Reduction in summer soil wetness induced by an increase in atmospheric
carbon dioxide," Science. 232:626-28, 1986.
Ritchie, J.T. and S. Otter, "CERES-Wheat -- A user-oriented wheat yield model," preliminary documentation,
AGRISTARS publication no. YM-U3-044420JSC-18892, 1984.
Rosenzweig, C., "Potential CO, induced Climate Effects on North American Wheat-Producing regions," Climate
Change. 7:367- 89, 1985.
Shumway, C.R. et al., "Regional Resource Use for Agricultural Production in California, 1961-65 and 1980,"
Giannini Foundation Monograph No. 25, Division of Agricultural Sciences, University of California, 1970.
Teramura, A.H., "The Potential Consequences of Ozone Depletion Upon Global Agriculture," J.G. Titus (editor),
Effects of Changes in Stratospheric Ozone and Global Climate, vol 2: Stratospheric Ozone, pp. 255-62, October
1986.
Wetzstein, M.E., "Methods for Measuring the Economic Impact of Ambient Pollutants on the Agricultural
Sector: Discussion," American Journal of Agricultural Economics, vol 67(2), pp. 419-20, May 1985.
Willey, Z., Economic Development and Environmental Quality in California's Water System. Institute of
Governmental Studies, University of California, Berkeley, 1985, 73 pp.
5-38
-------
EFFECTS OF PROJECTED CO--INDUCED CLIMATIC CHANGES ON IRRIGATION WATER
REQUIREMENTS IN THE GREAT PLAINS STATES
(TEXAS, OKLAHOMA, KANSAS, AND NEBRASKA)
by
Richard G. Allen
and
Francis N. Gichuki
Department of Agricultural and Irrigation Engineering
Utah State University
Logan, UT 84322
Contract No. CR-814887-01-0
-------
CONTENTS
Page
ACKNOWLEDGMENTS 6-1
FINDINGS 6-2
CHAPTER 1: INTRODUCTION 6-4
OBJECTIVES AND SCOPE OF THE STUDY 6-4
DESCRIPTION OF THE ECOLOGICAL SYSTEM 6-5
Geographic Area Covered 6-5
Irrigation and Water Issues 6-5
LITERATURE REVIEW 6-6
ORGANIZATION OF THE REPORT 6-7
CHAPTER 2: METHODOLOGY 6-8
THE ENVIRONMENT AND EVAPOTRANSPIRATION 6-8
Climatic Factors 6-8
Evaporative Energy 6-8
Air Temperature 6-8
Wind 6-9
Humidity 6-9
Effects of Changing Climates 6-9
Plant Factors 6-10
MODEL DEVELOPMENT 6-10
Governing Equations 6-10
Evapotranspiration 6-10
Aerodynamic Resistance 6-11
Canopy Resistance 6-12
Leaf Area Index 6-12
Water Balance-Irrigation Requirements Model 6-12
Predicting Evapotranspiration 6-12
Crop Evapotranspiration 6-13
Cropping Calendar 6-14
Soil Moisture Balance 6-14
Plant Surface Temperatures 6-14
MODEL DATA 6-15
Climatic Data 6-15
Adjustment of Air Temperature 6-15
GCM Climatic Scenarios 6-17
Control Variables 6-18
Cropping System 6-18
Bulk Stomatal Resistances 6-18
System Parameters 6-18
Soil Types 6-18
Irrigation System Types 6-20
CHAPTER 3: RESULTS AND DISCUSSION 6-21
CHANGES IN LENGTHS OF GROWING SEASONS 6-21
EVAPOTRANSPIRATION 6-23
IRRIGATION REQUIREMENTS 6-23
SURFACE TEMPERATURE 6-28
POTENTIAL AGRONOMIC ADJUSTMENTS 6-31
-------
CONTENTS (continued)
Page
CHAPTER 4: INTERPRETATION OF RESULTS 6-34
CLIMATIC AND STOMATAL RESISTANCE EFFECTS 6-34
Climate-induced Change 6-34
Stomatal Resistance-Induced Changes 6-34
CAVEATS AND LIMITATIONS OF THE STUDY 6-35
CHAPTER 5: IMPLICATIONS OF RESULTS 6-37
ENVIRONMENTAL IMPLICATIONS 6-37
SOCIOECONOMIC IMPLICATIONS 6-37
REFERENCES 6-39
111
-------
Allen
ACKNOWLEDGMENTS
This study was supported by the funding of the VS. EPA Contract CR-814887-01-0 and by funding from
the Utah Agricultural Experiment Station. The authors wish to acknowledge the time and interest invested in
this study by Dr. Cynthia Rosenweig, Goddard Institute for Space Studies, Columbia University, and by Dr.
Robert Hill, Utah State University, who assisted in writing the original research proposal. Dr. Roy Jenne and
Mr. Will Springer in NCAR provided the majority of weather data used in this study.
6-1
-------
Allen
FINDINGS1
In general, based on averages of the two general circulation model (GCM) 2xCO« climatic scenarios and
the likely occurrence of only moderate CO2-induced increases in bulk stomatal diffusion resistances of crop
canopies of about 20%, seasonal irrigation requirements for a mixture of alfalfa, corn, and winter wheat in the
Great Plains region will likely increase by about 15% under the 2xCO2 scenario. The following findings illustrate
the causes and effects of these changes.
1. Lengths of potential growing seasons were increased in all parts of the Great Plains region, with
magnitudes of increases ranging from 0 to 28%. Increases were less in the lower latitudes (less than 30 degrees
latitude), as much of this region currently has an almost all-year growing season. Although the two general
circulation models (GISS and GFDL) predicted different climatic scenarios, projected changes in potential
growing season lengths were not significantly different.
2. Changes in season lengths were heavily influenced by the timing of planting dates. Growing season
lengths for corn were projected to shorten in regions where current planting dates are in April and May
(northern latitudes). However, regions with com planting dates in February and March are expected to
experience increases in the lengths of growing seasons. These increases will likely occur when planting dates
under the GCM climatic scenarios shift to January and early February periods, during which air temperatures
and solar radiation levels are low. Crop development rates will be slower during these planting periods than
under current climatic conditions where crops are planted later when solar radiation levels are higher.
3. Growing seasons for winter wheat will likely be shortened under the GCM scenarios owing to later
planting dates in the fall when temperatures will be cool enough to eliminate moisture stress to seedlings, and
in most instances, owing to earlier harvest dates resulting from more rapid growth and earlier spring green-ups
resulting from higher air temperatures.
4. Major changes in irrigation water requirements were predicted at all 17 stations. The most significant
changes will be the persistent increases in seasonal net irrigation water requirements for crops that take
advantage of longer growing seasons such as alfalfa. These increases will be driven by projected increases in air
temperature, wind, and solar radiation, and the lengthening of growing seasons. Decreases in seasonal net
irrigation requirements were predicted for winter wheat and corn in most regions, especially if water vapor
diffusion resistances of crop canopies increase, as is postulated by some recent research. Decreases were also
caused by the shortening of growing seasons caused by projected increases in air temperature and solar radiation,
which accelerated crop phenologies.
5. Adaptation of a longer-season variety of winter wheat having a 20% greater growing season solar
radiation/degree day requirement would likely increase growing season lengths by 15 to 20 days in the Great
Plains and would increase seasonal irrigation requirements by about 10% in Nebraska, Kansas, and Texas and
by about 20% in Oklahoma as compared to current cultivars under the 2xCO2 settings. Thus, if fanners adapted
to crop varieties with higher seasonal environmental energy requirements to more fully utilize increased levels
of available solar radiation and temperature by changing to longer season crop varieties and/or increasing
cropping intensity, irrigation water requirements would increase.
6. Irrigation water requirements during peak periods increased in almost all areas. These increases may
require larger capacity irrigation systems and peak energy demands, but may not necessarily increase total
'Although the information contained in this report has been funded wholly or partially by the U.S.
Environmental Protection Agency under Contract No. CR-814887-01-0 under Dennis Tirpak, it does not
necessarily reflect the Agency's views, and no official endorsement should be inferred from it.
6-2
-------
Allen
seasonal water and energy requirements. Annual drafts of the Ogallala aquifer may also increase by 15% based
on projected increases in seasonal irrigation water requirements.
7. Predicted increases in peak and seasonal irrigation water requirements would be ameliorated somewhat
by postulated increases in values for bulk stomatal diffusion resistances resulting from elevated atmospheric CO2
concentrations.
8. Plant canopy (leaf) temperatures were predicted to increase above current baseline values for all crops
and sites studied. Projected increases in leaf temperatures may have detrimental effects on photosynthetic
activities and crop yields. They also make crops more sensitive to moisture stress.
6-3
-------
Allen
CHAPTER 1
INTRODUCTION
Carbon dioxide concentration in the atmosphere has been increasing and is expected to double by the year
2015 (Gribbin, 1981). Increase in carbon dioxide concentration will likely have a profound effect on the climatic
energy balance, as carbon dioxide (CO2) is fairly transparent to solar radiation but is largely opaque to thermal
radiation, causing an increase in net radiation (Kimball and Idso, 1983). The increase in atmospheric CO2 is
predicted to result in an increase in air temperature, net solar radiation, and humidity. Amounts and distribution
of wind and precipitation may also be affected. This modification of the plant environment is expected to
increase photosynthetic activities and influence stomatal regulation of water use. These changes are likely to have
a major effect on irrigation water requirements.
Irrigation is a major consumer of discretionary water in the United States. Any increase in irrigation
water consumption due to increased evaporative demands may accelerate depletion of limited ground and surface
water resources. In addition, in water-short areas or in areas under rainfed agriculture, increased evaporation
may increase differences between precipitation amounts and evaporative demands, thereby reducing available
water supplies in aquifers and streams and increasing crop water stress and decreasing yields.
OBJECTIVES AND SCOPE OF THE STUDY
The purpose of this study was to provide information for assessing environmental and socioeconomic
impacts of increasing CO2 concentration on the agricultural sector in general and water management in
particular. The objective of this study was to determine the effect of projected climatic and bulk stomatal
diffusion resistance changes on irrigation water requirements in Texas, Oklahoma, Kansas, and Nebraska.
Specific tasks undertaken included:
1. Computation of daily estimates of reference and crop evapotranspiration (Et) over a 30-year period
(1951-1980) using weather data collected from 17 locations in the region.
2. Computation of daily soil moisture-^ water balances for each weather station location for three
crops (alfalfa, corn, and winter wheat), three typical soil types representative of soils in the
surrounding areas, and for center pivot and surface irrigation methods.
3. Determination of baseline (current CO2 level) seasonal and peak monthly evapotranspiration and
net irrigation requirements for each of the site-soil-crop system combinations assuming irrigation
on demand, no leaching requirements, and 100% irrigation water application efficiency.
4. Repetition of tasks 1,2, and 3 for the same period of record with adjustments in climatic parameters
(air temperature, humidity, solar radiation, windspeed and precipitation) based on projected climatic
scenarios and conditions as provided by the National Center for Atmospheric Research (NCAR),
from General Circulation Models (GCMs) operated by the Goddard Institute for Space
Studies (GISS), and by the Geophysical Fluid Dynamics Laboratory (GFDL).
5. Repetition of task 4 using four levels of increased plant bulk stomatal vapor
diffusion resistance that may result from increased atmospheric carbon dioxide
content.
6-4
-------
Allen
DESCRIPTION OF THE ECOLOGICAL SYSTEM
Geographic Area Covered
This study covers four states of the Great Plains Region, namely Nebraska, Kansas, Oklahoma, and Texas.
This area is immense and varies in topography, soil, water resources, and climate. Although most of the region
is sparsely populated, its agricultural production is significant as it produces approximately 40% of the nation's
sorghum, 25% of cotton, and 17% of wheat (Schaffer and Schaffer, 1984).
Nebraska. The landscape of Nebraska changes from gently rolling prairie in the east to rounded sandhills
in the north-central part. The elevation rises from less than 300 meters in the south to 400 meters in the
northeast. The elevation also increases westwards to about 900 meters in the southwest and 1500 meters in
the northwest. The climate of the area is characterized by light rainfall, low humidity, hot summers, cold winters,
great variation in temperature and rainfall from year to year, and frequent changes in weather from day to day.
The precipitation of the state decreases from about 850 mm in the southeastern corner to about 350 mm near
the western border (Stevens, 1959). The economy of the state is dependent heavily on agriculture. The principal
crops are corn, soybeans, sorghum, hay, winter wheat, and oats. Approximately 8 million acres are now irrigated
in Nebraska
Kansas. The elevation across Kansas gradually rises from 240 to 300 meters above sea level in the eastern
counties to approximately 1100 meters at the Colorado line. Precipitation ranges from 1000 mm in southeastern
counties to 750-900 mm in the northeast, and decreases gradually westward to about 400 mm at the Colorado
line. The distribution of precipitation favors crop production with about 75% of the year's total occurring during
the crop growing season (Robb, 1959). The principal crops are wheat, corn, and sorghum.
Oklahoma The terrain of Oklahoma is mostly rolling plains, sloping downward from west to east. The
climate is characterized by long and occasionally hot summers and short and less rigorous winters. Precipitation
decreases sharply from east to west, about 1400 mm in the southeastern corner of the state, to 380 mm in the
extreme western areas (Curry, 1970). The principal crops are wheat, cotton, corn, and sorghum.
Texas. The changes in climate across Texas are considerable but gradual. The average annual rainfall
ranges from 1400 mm in the eastern border to about 200 mm in the western extremity of the state. Rainfall
occurs most frequently in late spring. East of 95 degrees longitude, the rainfall is fairly evenly distributed
throughout the year and exceeds average potential evapotranspiration, whereas west of this meridian, potential
evapotranspiration exceeds precipitation (Orton, 1969). The principal crops are cotton, sorghum, winter wheat,
and rice. Other important crops are corn, oats, peanuts, soybeans, potatoes, alfalfa, citrus, and vegetables.
Irrigation and Water Issues
Irrigated agriculture is the lifeblood of much of the Great Plains region. Researchers have shown that
whereas climatic, crop, and soil factors may limit crop productivity, availability of water is the most critical factor
in determining crop development, survival, and productivity (Dale and Shaw, 1965; Musick et al., 1976;
Rosenberg et al., 1983). In many locations, irrigated agriculture has replaced native vegetation, low-value crops
and dryland fanning, thereby improving the productivity of land and strengthening the regional and local
economies. Table 1 shows the extent of irrigated agriculture in the region.
The depletion of the water in the Ogallala aquifer threatens irrigated agriculture in this region. Schaffer
and Schaffer (1984) projected that Texas, New Mexico, and Oklahoma may experience annual water use declines
of 53% by the year 2020 due to aquifer depletion, and that annual water use may increase 33% in the more
northern states of Kansas, Colorado, and Nebraska.
6-5
-------
Allen
Table 1. Irrigated Land by State
State Irrigated crop land Irrigated land as a
as a percent of total percent of US total
cropland harvested irrigated land
Nebraska 34 11
Kansas less than 15 5
Oklahoma less than IS less than 1
Texas 31 14
Source: USDA, 1982.
Water issues in the region are complicated by socioeconomic and environmental factors. The development
and exploitation costs of surface and groundwater have soared over the last few decades making many technically
feasible projects unaffordable. In some areas deteriorating water quality brought about by increasing pollution
and diminishing flows is threatening recreation, wildlife, and fishery habitats.
LITERATURE REVIEW
Crop water use, irrigation water requirements, and precipitation adequacies are governed largely by the
evaporative process in which energy from radiative and convective heat sources is converted into the latent form
of energy through conversion of water from the liquid to vapor state. The modification of the atmospheric
environment by human activities can affect the evaporative process by varying amounts of energy available for
evaporation. This may occur through increased air temperatures resulting from increased levels of carbon
dioxide and, to a lesser extent, through changes in humidity levels and turbulent mixing (windiness) of the lower
atmosphere and through increases or decreases in solar radiation at the earth's surface as a consequence of CO2
increase.
Studies of the sensitivity of the evaporation or evapotranspiration (Ej) processes to changes in net radiation,
air temperature, dew point temperature, and windspeed have been conducted by Saxton (1975), Brockway et al.
(1985), and Rosenberg et al. (1988). Brockway et al. (1985) found that estimates of Ej increased by 4.5 to 5.0%
per degree Celsius increase in average air temperature, depending on which E^ estimating equation was used.
Estimates of E^ decreased by about 1.5% per degree Celsius increase in dew point temperature, and increased
by about 23% per kilowatt per square meter increase in solar radiation intensity. Estimated E. increased from
6 to 12% per m/s increase in windspeed. Results of the Brockway study are specific to climatic conditions in
southern Idaho, but indicate relative sensitivities of Ej to changes in climate or local environment. Rosenberg
et al. (1988) observed increases of about 4 to 8% in estimated Ej from a Kansas grassland per degree Celsius
for high and low values of Et demand and about 8% per degree for forests. They also used the Penman-
Monteith (Monteith, 1965) Et equation and predicted the following relationships:
1. One percent increase in net radiation increased E^ by 0.6%;
2. One percent increase in air vapor content increased Et by about 02 to 0.8%;
6-6
-------
Allen
3. Increasing temperature by 3°C and net radiation by 10% and decreasing vapor content by 10% increased
E( by 20 to 40%;
4. Increasing LAI by 15% increased Ej by 5%; and
S. Increasing stomatal resistance by 40% decreased E^ by 15%.
ORGANIZATION OF THE REPORT
The remainder of this report describes the evapotranspiration environment, the model used to estimate
the potential impacts of climate change and increased stomatal diffusion resistance on irrigation water
requirements in the southern and central Great Plains, the results and limitations of the study, and the
environmental and socioeconomic implications of the results.
6-7
-------
Allen
CHAPTER 2
METHODOLOGY
This section is divided into three parts. The first part is a review of material pertinent to the development
of the irrigation water requirement model. The second part describes the model development. The last part
describes data used.
THE ENVIRONMENT AND EVAPOTRANSPIRATION
Climatic Factors
Climatic variables which have a direct bearing on irrigation water requirements are evaporative energy,
temperature, windspeed, relative humidity, and precipitation. Identification of these variables, assessment of their
relative importance, and an appreciation of the interrelationship among the variables is essential in predicting
irrigation water requirements.
Evaporative Energy. The energy that is consumed by the evapotranspiration process is supplied mainly
from radiation energy and advective energy. The energy budget analysis method attempts to account for major
heat exchange processes and is represented by the equation:
Rn = H + LE + G + P [1]
where R_ is net radiation, H = vertical sensible heat flux to or away from the surface, LE = evapotranspiration
energy flux, G = heat flux into or out of the soil, and P is miscellaneous energy consumed by photosynthesis and
associated activities (usually negligible).
According to Eq. 1, energy reaching an evaporative surface by way of radiation is dispersed through
evapotranspiration, or is convected from the surface to the atmosphere or is conducted into the soil. Net
radiation is typically the major source of energy for evapotranspiration in humid regions. In more arid regions,
sensible heat flux to vegetation can provide as much energy for E^ as Rn. Advection is the process by which heat
is transported horizontally by the mass motion of the atmosphere. The origin of sensible heat in advective air
depends on surface conditions, especially the availability of water for evapotranspiration. The availability of
water determines the partitioning of the available energy among sensible, latent, and soil heat fluxes. For moist
surfaces, almost all net radiative energy is consumed as latent heat, whereas under dry surface conditions, latent
heat is reduced and sensible heat is generated. This sensible heat can then be advected to downwind areas,
thereby increasing evaporative demands there. Rosenberg (1969) noted that sensible heat advection was a major
component of energy balance in the Great Plains region. Rosenberg and Verma (1978) reported that during the
extended drought of 1976, Et from an irrigated alfalfa field at Mead. Nebraska, ranged from 5 to over 14 mm
day'1 even though Rn provided sufficient energy for only 7 mm day .
Air Temperature. Air temperature affects the evapotranspiration process as follows:
1. The amount of water vapor that the air can hold increases exponentially with temperature. As air
temperature increases, the vapor pressure deficit of the air increases, consequently increasing
evapotranspiration demand;
2. Higher leaf temperatures increase saturation vapor pressure inside leaf stomatal cavities, thereby forcing
more vapor out of the leaf and increasing evapotranspiration;
6-8
-------
Allen
3. Warm, dry air blowing toward the evaporative surface may supply adverted sensible heat energy to the
evapotranspiration process;
4. Slightly less energy is required to evaporate water at higher temperatures;
5. Increased temperature is accompanied by an increase in stomatal openings, indicating attempts by some
plants in some environments to increase evaporative cooling of plant tissues (Hofstra and Hesketh,
6. Increased air temperatures will generally increase the rate of crop growth and phenology, thereby
decreasing the length of time between plant emergence and maturity for annual crops.
All the above phenomena increase the evaporative demand The following phenomena would tend to
moderate evaporative demand:
1. Higher surface and air temperatures increase the net outgoing longwave radiation at the evaporating
surface, thereby decreasing net radiation; and
2. Decrease in season length resulting from increased rate of crop growth will tend to reduce the magnitude
of the increase in seasonal Ej requirement.
Wind. The presence of wind-induced atmospheric turbulence plays an important role in the
evapotranspiration process by facilitating the movement of moist air away from plants and the transport of
sensible heat from dry regions into the plant canopy. Air turbulence is by far the major transfer mechanism for
both sensible and latent energy exchange within and above the plant canopy.
Humidity. The humidity of the air near the plants is a rough indicator of the drying power of the
atmospheric air under a given condition. The inclusion of the air humidity effect in the Penman type of
evapotranspiration equation is typically as a saturation vapor pressure deficit (eg - ed) at a standard height above
the ground surface.
Effects of Chanyrjnfi Climates
Weather is an extremely stochastic process with many complexities and interactions among parameters.
This complicates farmers' decision-making and management strategies. Farmers of the Great Plains frequently
cope with variations in growing season precipitation of more than 60 mm and variation in length of season of
two weeks or more.
Climatic variability also plays an important role in selection of specific plant varieties and genotypes. In
1920, hard-red winter wheat production ranged from northern Texas to central Nebraska and from eastern
Colorado to central Illinois. Since then, production has been adapted to regions with lower mean annual
precipitation, lower average air temperature, and changes in growing season lengths as it has spread north-
westward to the Canadian border and southward to central Texas (Rosenberg, 1982). Rosenberg (1982) noted
that equivalent yields were obtained in western Nebraska and central Texas, the latter having 300 mm more
precipitation but an average annual temperature that is 8-5°C warmer. This example demonstrates agriculture's
capacity to cope with changes in climatic conditions by adapting genotype and crop husbandry.
Increased CO2 in the air is expected to increase photosynthetic activities and to decrease transpiration,
especially in C3 species (Allen et al., 1985). These changes will increase the water-use efficiencies2 (WUE) of
2Water-use efficiency is defined as the ratio of mass of usable biomas produced to mass of evapotranspirated
water (Allen, 1986a and Kimball and Idso, 1983).
6-9
-------
Allen
most agricultural crops. Increases in WUE will be beneficial from agronomic and economic points of view, as
less precipitation or irrigation water is required to produce normal yields. However, yield increases resulting
from increased pbotosynthetic activities may be partially offset by shortened growing seasons having less solar
radiation energy owing to accelerated crop phenologies.
Plant Factors
Leaf area index (LAI) is the ratio of leaf surface area (one side only) to the area of underlying ground,
and is an indicator of the amount of foliage in a stand of plants. Generally, as leaf area increases, the number
of stomata increase and evaporation increases because the bulk stomatal resistance of the canopy to vapor
diffusion decreases. Crop height has a direct influence on evaporation because tall plants are more
aerodynamically rough, thereby promoting more mixing of air in and above the plant canopy and causing
increased transfer of momentum, heat, and latent energy from or to the atmosphere. Broad-leafed crops may
transpire more than grasses of the same height and leaf area, as the size of the broad leaves tends to be less
efficient in dissipating heat thorough convective transfer and may therefore retain more energy for
evapotranspiration than does the size of leaves (Rosenberg et al., 1983).
Increases in atmospheric CO2 content are also likely to have pronounced effects on plant growth, yield,
stomatal regulation, and water-use efficiencies. The magnitudes of the effects may depend to some part on the
type of photosynthetic pathway regulating CO2 uptake. C, plants (including small grains, such as wheat and
leguminous species, such as alfalfa) have been found to be more responsive to elevated CO2 levels in the
atmosphere than C4 plants (corn, sorghum, millet, sugarcane) (Allen et al., 1985).
Stomatal resistance (r ) is a primary determinant of the transport of water vapor from plants to the
atmosphere. Any increase in r due to elevated COL levels will likely decrease evaporation from plant leaves.
However, several feedback mechanisms in the plant's microclimate partially offset the increase in rs (Allen et
al., 1985), including the following:
1. Elevation of internal leaf temperatures due to decreased conversion of net radiation into latent heat
causes increased vapor pressure inside leaves and increases the diffusion of water vapor through the
partially closed stomata;
2. Increases in air temperature due to transfer of sensible heat from the leaf surface to air result in an
increased vapor pressure deficit causing increased Et demand; and
3. Increased LAI from elevated levels of CO2 results in lower bulk stomatal resistance (r ) to water vapor
diffusion.
Kimball and Idso (1983) estimated that doubling CO, concentrations would increase stomatal resistances
and would reduce transpiration an average of 34%. They also noted that decreases in plant evapotranspiration
(if occurring) would enable wheat production to spread into more arid areas.
MODEL DEVELOPMENT
Governing Equations
Evapotranspiration. Estimates of evapotranspiration in this study were made using a Penman-Monteith
resistance model with variable bulk stomatal (canopy) and aerodynamic resistances. This model had been found,
during previous studies, to provide reliable and consistent daily estimates of alfalfa and grass reference E, when
proper heights and leaf areas of the evaporating surfaces are considered (Allen, 1986b; Allen et al., 1988).
The form of the Monteith version of the Penman equation used in this study for estimation of reference
Ej on a daily average basis is:
6-10
-------
Allen
_ A(Rn-G)+ 0.0864, cp(ea-ed)/ra
£u = •«*—»«—«-—«»»«««- L*J
A + 7 (1 + rc/ra)
where E^ is the evaporation flux (MJ m"2 d"1 ), RR is net radiation flux to the plant canopy (MJ m'2 d~1 X G is soil
heat flux (MJ m"2 d"1), A is the slope of the saturation vapor pressure curve (kPa °C ), p is the density of the
air (kg m ^X c is specific heat of the air (J kg"1 eC~1), ea is the mean saturation vapor pressure of the air at
the current air temperature (kPa), e^ is the saturation vapor pressure of the air at the dew point temperature
(actual vapor pressure of the air) (kFa), r is the aerodynamic resistance to vapor and heat diffusion (s m" X 7
is the pyschrometric constant (kPa "C"1), and rc is the bulk stomatal (canopy) resistance (s m*1).
Resistance. The aerodynamic resistance (rfl) is the aerial resistance to the flow of sensible
and latent heat and describes the external factors that affect evapotranspiration. Aerodynamic resistance depends
on windspeed and geometry and roughness of the plant community and can generally be derived from a
logarithmic wind profile when the change in temperature with height is dose to the dry adiabatic lapse rate
(the surface is only slightly hotter or cooler than the air at screen height). The aerodynamic resistance to heat
transfer from the surface to height z was approximated for neutral stability conditions as suggested by Garratt
and Hicks (1973) and Brutsaert and Strieker (1979) as:
" zm-di r *h-
-------
Allen
Canopy Resistance. Direct measurement of canopy resistance is difficult and therefore methods of
estimating canopy resistance have been developed (Szeicz and Long, 1969; Federer, 1979; Allen et al., 1988).
Canopy resistance for a well-watered, actively growing reference crop was approximated by dividing the minimum
stomatal resistance per single leaf area by the effective leaf area index of the canopy. Szeicz and Long (1969)
recommended considering only one-half of the leaf area index as being effective in evaporation from a fully
developed crop, since typically the upper half of the canopy of a dense crop absorbs the majority of net radiation
and is therefore more active in vapor and heat transport than is the lower half. In addition, Wright and Lemon
(1966), Tanner and Fuchs (1968), and Lemon and Wright (1969) observed that the majority of carbon dioxide
exchange occurred within the top half of a dense crop canopy. Vapor exchange through stomata within the
canopy is governed by processes similar to carbon dioxide, with similar flux gradient profiles (Wright and Brown,
1967; van Bavel and Ehrler, 1968). The one-half factor agrees with observations of effectiveness of leaf area in
transpiration as summarized by Allen et al. (1985). The assumption that only one half of the total leaf area is
effective hi evaporation of water also helps to correct for the use of 24-hr averages of weather and resistance
parameters (Allen, 1986b).
Average daily values for canopy resistance, r , of alfalfa under current levels of CO, were estimated as
(Allen et al., 1988):
rl
p
0.5 LAI
c [7]
where r, is an average minimum daytime value of stomatal resistance for a single leaf, approximated as 100 s
m" for alfalfa and grass canopies (Monteith, 1965; Monteith, 1981; Sharma, 1985). LAI is the leaf area index
and TC has units of s m'1. Equation 7 does not consider effects of temperature or net radiation on the value of
rc, as reported relationships have been contradictory. Equation 7 is best used for daily average values of rc
rather than for shorter time periods. Average daily values of rc for well-watered alfalfa or grass canopies
generally range between 40 and 70 s nf1
Leaf Area Index. Leaf area indices (LAI) vary with time, crop height, and cultural practices. In estimating
potential evapotranspiration from grass or alfalfa reference surfaces, the major variable affecting leaf area is
height, although many types of grasses can differ significantly in physiological composition and structure. For
a clipped grass, less than 15 centimeters in height, LAI was approximated (Allen et al., 1988) as
LAI = 0.24 h,. [8]
where hc is mean grass height (cm).
For alfalfa, LAI was approximated as
LAI = 1.5 InO) - 1.4 [9]
where hc is mean canopy height (cm), and hg was greater than 3 centimeters. The logarithmic relationship in
Eq. 9 results from stem extension with less leaf development with increasing height. Equation 9 predicted an
LAI of 4.5 for 50 centimeters tall alfalfa.
Water Balance-Irrigation Requirements Model
A computer model was developed, verified, and calibrated to compute daily evapotranspiration (EA soil
moisture balances and irrigation water requirements (IR).
Predicting Evapotranspiration. Evapotranspiration by an alfalfa reference crop (E^) was computed using
the Penman-Monteith and supporting equations (equations 2 through 9). Intermediate parameters required to
6-12
-------
Allen
complete the Penman-Monteith calculation (net radiation, air density, etc.) were computed as described by Allen
et al. (1988). Net radiation was estimated from vapor pressure, air temperature, and solar radiation as suggested
by Wright (1982).
Daily values of global, short wave solar radiation (Rs) were estimated by transforming the percent sky
cover values into percent sunshine (n/N) estimates according to procedures suggested by Doorenbos and Pruitt
(1977), and then estimating solar radiation (R9) as:
R8 = (0.25 + 0.5n/N) Rg [10]
where R is extraterrestrial radiation (MJ m^d"1), which was estimated according to procedures by Duffie and
Beckman (1980).
Soil heat flux was estimated using the temperature gradient and thermal conductivity method. Average
daily temperature gradient was estimated using the current day's average temperature and the mean temperature
of the previous three days and a constant soil thermal conductivity of 038MJ/m2/day/°C was used (Wright,
1982).
Crop Evapotranspiration. The alfalfa E. reference crop is commonly used in the agricultural community
to represent potential evapotranspiration3, and has been defined as a well-watered, actively growing alfalfa crop
with aerodynamically rough surface, and with at least 30 centimeters of top growth (Jensen et al., 1970). The
alfalfa reference differs from an alfalfa hay crop, in that the reference is always actively growing and is always
more than 30 cm in height because it is never harvested. Use of an E^ reference crop allows ELfrom other crops
OEfc) to be estimated as simple ratios of reference E^. These ratios, termed "crop coefficients" (k ) are
essentially lumped parameters that include differences between characteristics of specific crops and alfalfa (or
grass) reference parameters, including differences in height, roughness, stomatal resistance, leaf area, row
spacing, and albedo. Values of crop coefficients generally change with phenology and time through the growing
season. Crop coefficients allow the use of a standard reference that can be both verified and/or calibrated for
specific areas. The assumption of consistency and transferability in crop coefficients (ratios of E^ to E^) allows
estimation of E without the need to verify and/or calibrate equations for every crop of interest.
Crop coefficients used in the model were "basal" crop coefficients developed by Wright (1982) at Kimberly,
Idaho, for an alfalfa reference. The term "basal" indicates that the coefficients approximate the ratio of crop E^
to reference E* when direct evaporation from the soil surface is small, e.g., for periods that are more than 5 to
7 days after rainfall or irrigation. Values of the basal coefficients were increased to reflect evaporation from
soil surfaces for periods of 1 to 7 days following precipitation or irrigation using exponential decay functions
for specific soil types as suggested by Wright (1982). In addition, the cumulative sum of evaporation from the
soil surface was limited to specified values that were also a function of soil type.
The crop coefficient curves reported by Wright (1982) express basal kc values as functions of time through
a growing season. These curves were converted from the time base to an energy unit base by which to "dock"
crop development and change in value of k.. The energy unit (EU) selected for the k base was a solar
radiation-air temperature interaction term represented by a form of the Jensen-Haise evapotranspiration
equation (Jensen and Haise, 1963):
EU - (0.025Ta + 0.078)R, [11]
a 9
where units of EU and R are in mm d~1 of equivalent evaporation. Values for EU were limited to zero when
mean daily air temperature (Ta) was less than -3.1°C. Selection of this particular energy unit to "drive" crop
^Potential evapotranspiration is defined as the evapotranspiration from an extended surface of a short green
crop which fully shades the ground, exerts little or negligible resistance to the flow of water, and is always well
supplied with water (Rosenberg et aL, 1983).
6-13
-------
Allen
and kg development follows from observations of both solar radiation and air temperature as having direct and
interactive influences on crop growth. Use of both radiation and temperature were felt to improve the
transferability of crop coefficient curves to areas throughout the Great Plains region and to more readily simulate
the compression of growing season lengths brought on by higher air temperatures and perhaps greater levels of
solar radiation. The use of the Jensen-Haise EJ equation as a dock for kc development was first adapted by the
VS. Bureau of Reclamation (Buchheim and Brower, 1981) and has been used elsewhere. Threshold values of
EU were used to trigger the occurrence of effective full cover and maturity (harvest) for wheat and corn and
dates of anticipated cuttings for alfalfa hay.
Cropping Calendar. Average dates of planting for corn and spring wheat for specific years were assumed
to occur when the 10-day running average of mean daily air temperature increased to 13 and 5°C, respectively.
Fall planting of winter wheat was assumed to occur when the 10-day running average of mean daily air
temperature decreased to 17°C or December 1, whichever came first. A maximum threshold temperature for
winter wheat was used to represent conditions when air temperatures (and E^) would be low enough to maintain
sufficient soil moisture after planting to ensure germination and healthy crop development before winter. The
temperature thresholds used for corn, wheat, and alfalfa follow from observations by local county extension
agents and agree with values recommended by the USDA-SCS (1967).
Alfalfa hay was assumed to "greenup" in the spring after the last occurrence of a minimum temperature
of -5-5°C or lower (very hard frost). The alfalfa crop was also assumed to become dormant in the late fall after
the first occurrence of a minimum temperature below -5.5°C. These thresholds follow from those observed and
used in Idaho by Allen and Brockway (1982). If there were no occurrences of minimum daily air temperatures
below -55°C within a year, then the alfalfa crop was assumed to remain green throughout the winter months
for that year. Winter wheat was assumed to remain on a "standby" status during winter months, responding to
periods of warm temperatures with some growth advance.
Soil Moisture Balance. Beginning soil moisture at planting for all crops were assumed to be at points
halfway between field capacity (upper limit of soil retained moisture) and allowable depletion at irrigation.
The portion of precipitation infiltrating the soil during each rainfall event was estimated using the SCS
curve number method. The curve number method provides estimates of the portion of daily precipitation depths
that run off from the soil surface. The difference between precipitation and runoff includes interception and
infiltration. Since evaporation of interception reduces transpiration demands of crops, the total difference
between precipitation and runoff was assumed to infiltrate into the soil. The curve number method is a very
approximate method. However, it was selected for use since information on specific rainfall storm durations was
unavailable. Specific curve numbers were selected for each crop and soil type from tables presented by Hjelmfelt
and Cassidy (1975). In addition, the values of curve numbers were adjusted according to soil moisture levels to
reflect capacities of dry soils to infiltrate more precipitation than similar moist soils.
A portion of precipitation infiltrating the soil was assumed to drain below the root zone when the sum of
soil moisture before the precipitation event and infiltration exceeded the drainable upper limit of soil moisture
within the root zone. Drainage below the root zone - deep percolation -- was assumed to be lost to the system
and to be unavailable for fulfilling E^ requirements. The effective root zone for annual crops was assumed to
increase in depth proportional to the increase in the basal crop coefficient. This assumption follows from the
assumption that rooting depth and density are proportional to height and density of the crop canopy. Because
the value of the basal crop coefficient relates to the height and density of the crop canopy, it serves as a good
approximation of root growth from planting until a maximum rooting depth is reached at full canopy cover.
Plant Surface Temperatures. Fall-ofls in plant growth rates and yields when plant canopy (surface)
temperatures exceed certain threshold values have been noted (Hatfield, et al., 1987). Because of possible
physiological implications of increases in plant canopy temperatures, estimates of an "apparent" daily average
surface temperature, TS, were made from a combination of the energy balance and sensible heat equations using
equation 12:
6-14
-------
Allen
where E^. is the estimated value of daily evapotranspiration from crop "c."
MODEL DATA
The model inputs are climatic data, controllable variables, and system parameters. Seventeen locations in
the Great Plains, five in Nebraska, three in Kansas, two in Oklahoma and seven in Texas were selected based
on the availability of long periods of climatic data
Climatic Data
The climatic parameters used in this study included maximum and minimum air temperature, windspeed,
mixing ratio, cloud cover, and precipitation. The 1951-1980 weather data set for the 17 first-order NOAA stations
was assembled for this study by R. Jenne and W. Springer at NCAR. Data set parameters included percent sky
cover for both opaque and total clouds for the period from 9 am. to 3 p.m., 24-hr windspeed, and 24-hr average
mixing ratios. Daily maximum and minimum air temperatures and precipitation were obtained from separate
temperature-precipitation data sets also furnished by NCAR.
Adjustment of Air Temperature. Irrigation of a local area or region affects the characteristics of the lower
boundary layer of the atmosphere. These effects typically include cooling of the air (reduced air temperature),
increased moisture content, and reduced wind (less macro-scale buoyancy mixing due to reduced sensible heat
transfer). The elevation of air temperatures at dryland settings relative to air temperatures at adjacent irrigated
setting commonly exceeds 5°C during summer months in areas of southern Idaho (Allen et al., 1983; Allen and
Pruitt, 1986). Changes in the lower boundary layer include changes in absolute values and shape of the vertical
profiles of air temperature, humidity, and windspeed. Values of air temperature and vapor pressure deficits at
measurement heights imply information concerning the value of surface temperature and the relative partitioning
of radiation energy into sensible and latent heat at the evaporating surface. Therefore, inclusion of air properties
measured over non-irrigated surfaces having less than full crop canopies or sufficient soil moisture to fulfill E^
demands into an E^ estimating equation will cause overestimation of E^. The majority of modem reference E^
equations, including equation 2, assume that weather measurements are made over a well-watered crop with
characteristics similar to the reference crop for which EJ is being estimated. These assumptions are usually
embedded in empirical constants in the equations and in the underlying theory.
Because weather data used in this study were collected from station located in airport settings, and because
these settings were typically surrounded by nonirrigated, rainfed agriculture, air temperatures and windspeeds
recorded at these locations lie above levels and air humidities below levels that would have occurred over an
irrigated reference crop. Because it was desirable to obtain reasonably accurate estimates of current, baseline
evapotranspiration and irrigation water requirements, a method was developed to make adjustments to maximum
and minimum recorded air temperatures at all weather sites. No adjustment was made to humidity and wind
data, as insufficient information and adjustment techniques were available and differences in these parameters
between irrigated and dryland settings are likely to be less pronounced.
The background, theory, and validation of the air temperature adjustment procedure is described in more
detail by Allen (1988). As a starting point in the adjustment procedure, an apparent, average daily temperature
of the evaporating surface of the rainfed region around the weather site (T ) was estimated as:
6-15
-------
Allen
((Rn - G) -
"
3d
where E^ is the estimated, actual daily evapotranspiration from the rainfed region. Equation 13 is a solution
of the sensible heat transport equation, where H, the sensible heat flux, has been replaced with the other three
major components of the energy balance equation, namely, Rn, G, and E^. It should be noted that the value
of Tgg has no direct physical equivalent, e.g., it cannot be measured in the field, as it is the product of using
daily averages of Ep windspeed and air temperature. However, it does function as an index of average surface
temperature properties that would be required to "drive" the amount of sensible heat transfer from the crop
surface to the atmosphere to close the energy balance.
Actual daily evapotranspiration, E^, was computed with a daily soil water balance with precipitation as
the only moisture input. The value of E was calculated as:
(VfU
Leu-eLj
for 0, - 0L < R ' (0y - 9L) and
^ = 0.8 Efc [15]
for 0, - 0L > R (Qy - 0L)
where 6, is the root zone soil moisture on day i, 0L is the lower limit of plant extractable water in the root zone,
0y is the drainable upper limit of plant extractable water in the root zone, and R is a stress threshold indicator,
set equal to 0.5 in this study. Equation 13 has been used successfully in relatively simple crop yield simulation
models to predict the effect of soil moisture on EL (Hill et al., 1982a). E^ represents evapotranspiration from
a grass reference crop. This parameter was calculated using the Penman-Monteith equation with a roughness
height of 12 cm and an LAI value of 2.9. The product of 0.8 E^ in equations 14 and 15 was used to represent
the average potential E. of the region (integrated over all crops and surface types) surrounding the airport
weather sites under conditions of adequate soil moisture. The daily soil moisture balance for the rainfed area
was of the form:
0, = 0 M - E^ + P,- SR, DP, [16]
where P, is precipitation on day i, SR, is estimated surface runoff on day i, computed using the curve number
method, and DP, is deep percolation, which was assumed to occur when soil moisture exceeded the drainable
upper limit. Initial values of 0 , at the beginning of the season were assumed to be at levels that lay 75% of
the way between the lower limit of plant extractable moisture and the drainable upper limit.
The second step in the adjustment procedure was to calculate an additional apparent, average surface
temperature, T ,, for an adequately watered, irrigated reference crop, which in this case was alfalfa. This crop
is the type of evaporating surface over which the E. reference equation assumes that weather data were
measured. The value of Tg was computed by numerically searching for the value that solved the following
energy balance equation:
6-16
-------
Allen
1\_ ™ \J ^ "«"«•••••«•••»••«•»»« ^ » « «• • • •« • BB« • L ^ J
T ra
where e°[T] is the saturation vapor pressure at temperature T and Td is average daily dew point temperature.
The values of ra and r in equation 17 represent resistances for an alfalfa reference.
If Egg and E^ were equivalent (the region surrounding the airport was an irrigated alfalfa reference), then
the values of T and Tj, would be equivalent. Deviations between TW and T^ then indicate differences between
Ej_ and Ej. resulting from lack of precipitation and/or to differences in characteristics of the evaporating surface.
These differences are largely due to differences in sensible heat fluxes from the rainfed and irrigated surfaces.
Because the implied values of sensible heat are numerically "anchored" to the same value of air temperature, Ta,
in equations 16 and 17, differences in implied levels of sensible heat are manifested as differences in apparent
surface temperatures.
The actual adjustments to maximum and minimum values of air temperature measurements were made
following a concept similar to the theory of "complimentary" relationships between actual and potential estimates
of evapotranspiration as formulated by Bouchet (1963), where any deficit between potential evaporative demand
and actual evapotranspiration in an area would manifest itself in nearly equivalent magnitude as an increase in
predicted potential evaporation by a Penman type of equation. In other words, the increased sensible heat flux
generated by a moisture-stressed surface would provide positive feedback to the atmosphere, thereby increasing
the evaporative demand by a like amount. One of the manifestations of the increased sensible heat flux would
be elevated air temperature, as previously discussed. Because the sensible heat fluxes and corresponding apparent
surface temperatures computed for the rainfed and irrigated conditions "bracketed" the actual sensible heat flux
and surface temperatures that would have occurred over an irrigated surface with a boundary layer at equilibrium
(similar to the bracketing of the complimentary Ej theory), adjusted temperatures were computed as follows:
TXa = Tx - 030^ - T,,) [18]
Tna ' Tn - Waa ' Tsl>
Values of T^ were almost always less than values for T^ owing to predicted downward transfer of sensible
heat to the irrigated surfaces (Tg - Ta was negative). The coefficients 03 and 0.2 (in contrast to a coefficient
of 0.5 indicated by the complementary theory) were found to best predict the necessary adjustment to air
temperatures to represent conditions over an alfalfa reference surface. This procedure was evaluated and
calibrated using irrigated and airport or rangeland weather data from southern Idaho and Scotts Bluff, Nebraska,
as described by Allen (1988). Results of the adjustment were found to be quite good, with adjusted temperatures
closely approximating air temperatures above an irrigated alfalfa surface on a daily basis. The deviation from
the theoretical 0.5 value are due to other effects, assumptions, and numerical idiosyncrasies lumped into
equations 13 through 17. Minimum daily air temperatures were adjusted less than maximum temperatures owing
to lower values of sensible and evaporative heat fluxes at night and the proximity of minimum daily air
temperatures to dew points in many areas even when regional Ej was less than potential values.
Gf.M Climatic Scenarios. Daily values of adjusted air temperature, estimated solar radiation, mixing ratio,
and precipitation from the baseline 1951-1980 data sets were multiplied by monthly ratios generated by the
GISS and GFDL models to reflect long-term, steady-state relative increases or decreases in the 30-year period
for a 2xCO2 atmosphere.
Absolute changes were necessary for modifying windspeeds, because applying ratios of monthly windspeeds
from the GCMs on windy days predicted winds that were unrealistically high. Daily values of windspeed from
the baseline data sets were modified by adding or subtracting absolute changes in windspeeds as projected by
the GCM models.
6-17
-------
Allen
Trends in changes in the climatic parameters air temperature, humidity (mixing ratio), solar radiation, wind,
and precipitation were similar between the GISS and GFDL scenarios, with the GFDL scenario having more
extreme changes than the GISS scenario. Averages of GCMs projected percent change in parameters for
stations within each of the four states are listed in Table 2. In general, air temperatures were projected to
increase during all months, with the increases averaging 1 to 2% of a Kelvin scale (approximately 3 to 6*0,
with the GFDL model projecting temperatures during summer months which were about L5°C higher than the
GISS (53°C versus 3.8°C above baseline (1951-1980) temperatures). Changes in projected precipitation were
widely scattered under both scenarios, with an average 2.9% increase in annual precipitation projected by the
GISS model and a 5% increase by the GFDL model. Changes commonly alternated in sign from month to
month under both scenarios and the magnitudes of monthly fluctuations were quite high. Projected changes in
air movement (wind) fluctuated widely from month to month in both models, with some months projected to
have increased windspeeds and some projected to have lower. Vapor content of the air (humidity), expressed
in the GCMs as mixing ratios, was projected to increase. One effect of increased humidity would be to reduce
the evaporative demand of the air and consequently evapotranspiration, all other factors being constant. Solar
radiation was predicted to increase. Higher changes in solar radiation were projected in the winter months,
especially in the northern states. This increase was most likely due to projected decreases in cloudiness.
Control Variables
Cropping System. Corn, winter wheat, and alfalfa were selected to represent typical crops grown in the
Great Plains region. Corn represents row crop production, winter wheat represents a drilled, overwinter crop,
and alfalfa represents a perennial crop that is able to take full advantage of extended growing seasons.
Bulk Stomatal Resistances. Leaf area indices and stomatal resistances of single leaves are the major
determinants in values of bulk stomatal resistance. These two parameters are also both felt by many researchers
to be affected by elevated CO2 levels. Rosenberg et al. (1988) hypothesized that values of LAI may increase by
about 15% and single leaf stomatal resistances by 40% for doubled CO2 levels. Allen et al. (1985) and Allen
(1986a) summarized evidence that LAI may increase by about 20 to 75% and stomatal resistances may increase
by about 75% for doubled CO2 levels. According to the relationship presented in equation 7, equivalent
increases in LAI and r( would effectively cancel each other out. Because of uncertainties in the literature
concerning the nature and magnitudes of increases in both LAI and r(, the water balance model was rerun
several times for four levels of increased bulk stomatal resistance, rQ. These levels represent increases of 20,40,
60, and 80% in the values of rc.
System Parameters
Soil Types. Three soil types were considered in the water balance model for each weather location to
evaluate the effect of soil type on evapotranspiration (primarily wet soil evaporation), rooting depth, effectiveness
of precipitation, and irrigation water requirements. Particular soil types at each site were selected according to
information furnished by local county extension agents and other agricultural personal. This information was
collected by C. Rosenzweig (1988, personal communication). Soil type information in the model was for "generic"
soil types ranging from shallow, medium, and deep silty clays to shallow, medium, and deep sands. Parameters
describing attributes of these soils were obtained from the IBSNAT data base. The three parameters in the
IBSNAT data base that were used directly in the water balance model were the lower limit of plant-extractable
water (wilting point), the drainable upper limit of soil moisture (field capacity), and upper limit of stage 1
evaporation. The lower and upper limits of plant-extractable water were significantly reduced for the IBSNAT
silty clay soil before use in the water balance model to better reflect observed field data.
6-18
-------
Allen
Table 2. Percentage Changes from Baseline Weather Parameters during the growing
season
Temperature Precipitation Windspeed Mixing Ratio Solar Radiation
GISS GFDL GISS GFDL GISS GFDL GISS GFDL GISS GFDL
Nebraska
Oct 1.5
Feb 1.6
Mar 2.0
Apr 1.7
May 1.1
Jun 1.2
Jul 1.4
Aug 1.6
Sep 1.4
Oct 2.0
Nov 2.0
Feb 1.4
Mar 1.9
Apr 1.7
May 1.0
Jun 1.1
Jul 1.5
Aug 1.6
Sep 1.2
Oct 2.2
Nov 1.9
Jan 1.8
Feb 1.3
Mar 1.8
Apr 1.7
May 1.0
Jun 1.3
Jul 1.4
Aug 1.6
Sep 1.3
Oct 2.0
Nov 1.9
Dec 2.1
1.9
1.6
1.4
2.8
2.7
2.3
2.1
1.8
32.7
10.1
15.0
-2.6
-11.0
21.3
-4.8
-30.4
5.0
37.7
1.3
-51.6
-30.6
-39.1
4.9
-15.6
-11.0
12.3
-76.1
3.4
10.5
-1.9
-2.0
15.9
-9.0
-15.5
-8.8
-13.8
22.2
20.0
-8.1
-15.8
53.1
28.1
32.1
33.8
15.4
29.6
34.0
32.5
45.3
38.4
27.7
26.1
6.4
15.8
10.1
43.0
7.0
4.4
1.0
3.3
-0.4
8.7
11.8
36.0
23.9
-9.4
7.1
21.0
9.6
7.1
5.3
23.7
Kansas
1.9
1.6
1.7
2.0
2.1
2.1
1.5
1.6
1.7
1.7
11.4
-9.3
-7.9
-2.8
-13.7
-16.7
33.4
85.0
-6.2
-15.6
38.7
17.8
48.8
-4.5
-50.7
-35.8
-31.8
7.1
-5.1
48.3
-5.0
3.1
-3.0
-9.3
10.3
3.7
-1.0
-5.0
4.2
8.9
-27.6
-9.6
1.5
-10.4
15.9
25.3
7.2
0.9
-4.1
-10.7
43.9
48.7
34.4
23.8
20.0
22.0
32.8
31.5
38.4
40.8
49.2
34.3
34.5
29.0
8.2
4.6
19.9
11.6
42.4
46.7
2.1
6.1
6.8
3.2
4.3
4.9
7.8
2.8
10.9
6.9
6.4
7.1
-1.0
12.2
9.5
-0.5
8.1
2.2
8.0
10.4
Oklahoma
1.7
1.6
1.6
2.0
1.5
1.6
1.2
1.6
1.7
1.6
7.3
-28.2
-15.1
-9.2
-18.1
-20.3
35.5
130.4
4.1
-37.4
24.0
39.9
16.0
-10.0
-33.2
7.9
8.0
-16.6
-12.5
35.0
-8.5
10.7
-11.2
23.0
16.4
1.5
-0.8
-8.1
-1.0
• 9.0
-27.8
-10.4
8.7
-5.0
18.2
35.9
15.6
-7.0
2.3
-13.6
35.6
46.5
37.7
19.7
13.1
25.3
34.4
30.3
41.4
35.1
42.3
33.4
30.3
23.6
0.8
4.8
22.3
13.3
39.7
47.2
3.1
12.2
10.8
4.7
4.3
7.7
9.3
2.9
11.2
10.5
-4.3
0.7
2.3
11.0
2.1
0.1
5.4
4.6
10.2
7.1
Texas
1.4
1.9
1.6
1.5
2.1
1.4
1.6
1.3
1.5
1.7
1.5
1.8
-19.7
8.9
-21.9
-3.5
10.5
-15.8
-4.6
21.2
67.0
-8.4
-29.5
-2.7
-5.9
13.2
17.6
-20.7
-31.2
70.1
-1.0
26.4
-34.2
-3.8
8.4
-14.6
-1.0
-1.3
10.9
-24.3
12.9
8.2
-4.4
-3.0
-6.3
3.6
3.5
5.8
-0.9
-15.8
-3.2
0.4
-2.8
19.1
20.6
12.7
-0.6
2.6
2.1
2.4
44.2
33.4
44.4
38.6
24.0
20.8
26.2
33.0
29.3
38.0
35.7
51.4
31.6
46.3
30.3
25.4
21.0
6.8
12.8
24.5
19.7
30.1
36.7
41.1
4.1
5.2
11.1
8.8
3.6
4.7
5.9
8.3
2.9
9.2
10.6
13.0
8.4
2.0
3.3
7.4
16.4
-1.7
3.1
4.8
9.8
10.9
4.7
15.3
6-19
-------
Allen
Irrigation System Types. The frequency of irrigation affects the portion of time during which the soil
surface is wet and is contributing moisture to fulfill the evapotranspiration demand Center pivot systems, which
are quite common in the Great Plains region, typically irrigate every 2 to 3 days during the peak of the growing
season, whereas periodically moved sprinkler systems such as wheel lines and hand-moved systems and surface
irrigation systems such as furrow, basin, and border are generally managed to maximize the time periods
between irrigations in order to minimize irrigation labor. Therefore, crops grown under center pivot systems
generally have increased evapotranspiration rates due to increased wet soil evaporation. However, center pivots
may provide for higher values of effective precipitation (that precipitation entering and remaining in the root
zone), since the soil moisture is generally maintained at levels below the drainable upper limit so that root
zones under center pivots have more capacity, more of the time, to retain infiltrated precipitation and thereby
limit deep percolation losses. Periodic systems (other types of sprinklers and surface systems) generally have
lower evapotranspiration demands than do center pivots because the soil surface is likely to be dry over a larger
portion of time.
Two system types were simulated in the water balance model. The first type represented center pivot
systems where net irrigation depths of 15 mm were applied whenever soil moisture in the root zone was
depleted by EL to a point halfway between the upper limit of plant-extractable moisture and the allowable
depletion level. This level was generally when 25% of the soil moisture between the drainable upper limit
and lower limit of plant-extractable moisture had been depleted. The second system type represented the
periodic systems, where irrigation water was assumed to be added when soil moisture was depleted to the
allowable depletion level. The allowable depletion level was defined as the level to which soil moisture can be
depleted by Ej before the lower availability of remaining soil moisture would cause E^ to decrease from its
potential amount. This level ranged from 50 to 55%, depending on the soil type.
6-20
-------
Allen
CHAPTERS
RESULTS AND DISCUSSION
The Water Balance-Irrigation Requirements model was run for a 30-year period (1951-1980) of baseline
climatic data and under GISS and GFDL scenarios for the same period with 2xCO, concentrations. Additional
simulation runs were made to evaluate the effects of increased growing seasonTor annual crops under the
GISS and GFDL scenarios. These additional analyses were made to evaluate potential changes in crop cultivars
that would take advantage of longer potential growing seasons. For each of the GISS and GFDL scenarios,
effects of five hypothetical levels of bulk stomatal resistance values on irrigation water requirements were
evaluated.
CHANGES IN LENGTHS OF GROWING SEASONS
The growing seasons for alfalfa were used as general indicators of maximum lengths of potential growing
seasons for most crops. Significant shifts in green-up and frost-induced dormancies of alfalfa were observed
at higher latitudes owing to earlier occurrences of last killing frosts in the spring and later occurrence in the
fall. The lower latitudes (less than 30°) experienced more modest changes in season lengths, as many of the
regions currently have an almost year-round growing season. Although the two general circulation models
(GISS and GFDL) predicted different climatic scenarios, the changes in lengths of growing seasons for alfalfa
obtained for the two scenarios were not significantly different.
The following regression equations were found to predict growing season lengths:
Baseline scenario (R2 = 0.98)
S, - 377 - 1.74d for 16 < d < 106 [20]
GISS Scenario (R2 - 0.88)
S, = 400 - 1.43d for 24 < d < 106 [21]
GFDL Scenario (R2 « 0.86)
S, = 402 - 1.47d for 25 < d < 106 [22]
where S( = season length and d is the day of year when alfalfa regrowth begins under current climatic
conditions Oast occurrence of -5-5°C). For regions where alfalfa regrowth began before January 25th, year-
round growing seasons were predicted under both GCM scenarios.
Increased air temperatures predicted by the GCM model scenarios increased rates of crop development
according to phenology, solar radiation, and air temperature relationships used in the water balance-E^ model
These increases resulted in increased numbers of bay cuttings per growing season for alfalfa and reductions in
the lengths of growing seasons for other annual crops. Figure 1 shows the changes in lengths of growing
seasons from the baseline values.
The changes in season lengths for corn were heavily influenced by the occurrence of planting dates.
Decreases in the season lengths were observed in regions where planting dates were in April and May (northern
latitudes). However, regions with corn planting dates in February and March experienced modest increases
in the lengths of the growing seasons. The longer growing seasons for February and March planting dates
occurred because planting dates under the GCM climatic scenarios shifted to January and early February
6-21
-------
Allen
&
ou
Af\
*HJ
0^*1
—2O
c^\
— DU
or\
OU
-lOO
1
\
1
|
1
1
pn
v::^_ ^t
11 1 i i
I l l
ii i
Alf-OSS
/LF-GFDL
V/////////A
COR-GISS
COR-GFDL
WHE-QSS
WHE-GFDL
Nebraska
Kansas
OMahoma
Texas
Figure 1. Changes in season lengths (days) from baseline values.
6-72
-------
Allen
periods, during which air temperatures and solar radiation levels were low. Thus, crop development rates were
slower during these periods than under current climatic conditions where crops are planted later when solar
radiation levels were higher. The large decreases in season lengths for com in Nebraska were due to more
rapid maturing of the crop, thereby eliminating extension of the season into the months of October and
November when levels of temperature and solar radiation are low.
The growing seasons for winter wheat were always shortened under the GCM scenarios owing to the later
planting dates in the fall when temperatures were cool enough to eliminate moisture stress to seedlings, and
in most instances, to earlier harvest dates due to more rapid growth and earlier green-up resulting from higher
air temperatures and higher solar radiation levels.
EVAPOTRANSPIRATION
The combined effects of the changes in climatic conditions and in lengths of growing seasons influenced
peak and seasonal evapotranspiration demands as shown in Figure 2. Conditions under the GFDL scenario
predicted larger changes in both seasonal and peak month evapotranspiration than did the GISS scenario,
mainly owing to the higher air temperature changes predicted by the GFDL scenario. The increases in seasonal
evapotranspiration of alfalfa were due to the increases in the lengths of the growing seasons coupled with the
effects of climatic changes. Seasonal evapotranspiration requirements of com and winter wheat crops were
reduced primarily because of decreases in lengths of growing seasons and because of changes in cropping
calendars. Winter wheat was predicted to have been planted later in fall and harvested earlier in the summer
under the GCM scenarios (late September to early June), the period in which climatic changes were generally
modest. Increases in seasonal evapotranspiration from corn were predicted in these regions where GCM
scenarios predicted increases or only slight decreases in lengths of growing seasons (around 30 to 40° latitude).
Estimates of peak monthly evapotranspiration increased under the GCM scenarios for all crops and all
sites evaluated, except in Brownsville, Texas (see Table 3). The increases in peak monthly E^ are attributed
mainly to the increase in evapotranspirative demand of the atmosphere under the GCM scenarios (higher air
temperatures, wind, and solar radiation in most months). In Brownsville, Texas, the peak evapotranspiration
of corn and winter wheat decreased by 11 and 2%, respectively, under the GISS scenario with no increase in
bulk stomatal diffusion resistance. The decrease was attributed to the fact that under the modified climate,
crops were predicted to be grown in relatively cooler months with lower solar radiation (January to June),
thereby avoiding months with higher evaporative demands.
The upper ranges in Table 3 represent increases in peak monthly Ej predicted if there were no changes
in bulk stomatal diffusion resistance of plant leaves (no change in LAI or stomatal resistance). The lower ranges
represent changes predicted if bulk stomatal diffusion resistances of plant canopies were increased by 80%.
Sensitivities of both peak and seasonal E^ to projected changes in bulk stomatal diffusion resistance are
very pronounced Results indicate that increases in evapotranspiration due to changes in climate may be
moderated or even negated by increases in bulk stomatal diffusion resistances under the high CO, scenarios.
The increases in bulk stomatal resistances required to nullify any projected increases in seasonal
evapotranspiration due to climatic change vary with GCM scenario, geographic location, and the type of crop
as evidenced in Figure 2. Trends in increases in peak E^ requirements were very similar to those for peak
irrigation requirements shown in Figure 5, with the largest changes occurring for the GFDL scenario and in
the northern latitudes.
IRRIGATION REQUIREMENTS
Irrigation requirements were computed by incorporating the effects of climate, cropping patterns and
planting schedules, soils, and irrigation methods with estimated evapotranspiration and precipitation. A summary
of the ranges of change in net irrigation requirements resulting under the GISS and GFDL scenarios are shown
6-23
-------
Allen
-3O
GISS GFDL GISS GFDL GISS GFDL GISS GFDL
NEBRASKA KANSAS OKLAHOMA TEXAS
Figure 2. Percent change in seasonal evapotranspiration from baseline values for alfalfa, corn and winter
wheat vs. postulated increases in bulk stomatal diffusion resistance.
6-24
-------
Allen
Table 3. Ranges of percent change in peak evapotanspiration from baseline values over the 17 stations.
Scenario
Crop GISS GFDL
Alfalfa Corn Wheat Alfalfa Corn Wheat
Nebraska -10 - 17
Kansas -10 - 17
Oklahoma -1 19
Texas -10 - 20
-5 -29
-11 19
-11 14
-28-17
-20-10
-21- 8
-11-15
-23-16
38-72
21-58
11-66
3-46
29-80
12-55
0-41
4-47
1-31
-15-18
1-35
-7-37
in Figure 3. Predicted changes in seasonal irrigation water requirements were consistently higher under the
GFDL than GISS scenarios, with estimated increases in seasonal irrigation requirements for alfalfa under the
GFDL scenario averaging about 90% greater than under the GISS scenario.
•
Net seasonal irrigation water requirements increased with the increased temperature and decreased
precipitation during growing seasons predicted under the GCM model scenarios. The predicted increases in
net irrigation requirement were generally higher than predicted increases in evapotranspiration. This is
attributed to the effect of lower rainfall during growing seasons predicted by the GCMs. Computed changes
in irrigation water requirements were greatest in central Nebraska, Kansas, Oklahoma, and northern Texas.
Percent changes in irrigation requirements were greatest in eastern Nebraska since baseline values were lower
owing to higher precipitation.
Shortened growing season lengths reduced predicted irrigation water requirements for corn under the
GISS scenario, even with small changes in bulk stomatal diffusion resistances. Under the GFDL scenario,
however, the effects of reduction in growing season lengths of corn were negated by larger increases in
evaporation demands and decreases in growing season precipitation. Seasonal requirements of winter wheat
were decreased under both GCMs scenarios owing to shortening and advancement of growing seasons, although
changes would be very slight if bulk stomatal diffusion resistances were to remain constant.
The effect of bulk stomatal diffusion resistance on net seasonal irrigation water requirements is dramatic.
Projected net seasonal irrigation requirements decreased as projected bulk stomatal diffusion resistances were
increased as shown in Figure 3. Under the GISS scenario, the effect of expanded growing seasons for alfalfa
would be nearly balanced by increases in bulk stomatal diffusion resistances of 80%. Under GFDL scenario,
seasonal irrigation requirements of alfalfa would increase by an average of 20% even with an increase in bulk
stomatal diffusion resistance of 80%. The increase would be about 85%, otherwise.
The predicted variations between stations were dramatic, especially for alfalfa and corn, primarily owing
to the larger variations in season lengths. The decreases were almost linear for the incremental values of bulk
stomatal diffusion resistance evaluated in this study. These decreases varied with latitude, longitude, altitude,
local conditions, soil, and crops. The decreases were, however, persistent in all cases. Isograms of net seasonal
irrigation water requirements of alfalfa are presented in Figure 4. Only the results for baseline, zero, and 40%
increases in bulk stomatal resistance for scenarios for GISS and GFDL are presented. In almost all cases, the
water requirements increased from east to west, which is consistent with recognized precipitation patterns for
the region.
6-25
-------
Allen
-30
—4O
GISS GFDL GISS GFDL GISS GFDL GISS GFDL
NEBRASKA KANSAS OKLAHOMA TEXAS
Figure 3. Percent change in net seasonal irrigation requirement from baseline values for alfalfa, corn and
winter wheat vs. postulated increases in bulk stomatal diffusion resistance.
6-26
-------
(a)
Isogram of Seasonal Irrigation Requirements
(Alfalfa).
a) baseline
b) GISS with no increase in bulk stomatal
resistance
c) GFDC with no increase in bulk stomatal
resistance
d) GISS with 40% increase in bulk stomatal
resistance
e) GFDL with 40% increase in bulk
stomatal resistance
(d)
Figure 4. Isograms of seasonal net irrigation water requirements for Alfalfa.
n
a
-------
Allen
Figure 5 show changes in peak monthly irrigation water requirements. These trends are similar to trends
in change in peak monthly E^ requirements, with projected increases under the GFDL scenario being more than
double those projected under the GISS scenario due primarily to higher projected increases in air temperatures
and in general, smaller projected increases in humidities under the GFDL scenario. Trends in peak irrigation
requirements for corn and wheat under the GISS scenario would be dose to zero if bulk stomatal diffusion
resistances were increased to about 40%. Otherwise, peak requirements for these two crops would be expected
to increase by 5 to 20% under GISS scenario if bulk stomatal diffusion resistance values did not change.
SURFACE TEMPERATURE
Plant temperature is a major component of the plant energy balance and provides an indication of the
energy exchanges between the plant and the atmosphere. Temperature exerts a profound influence on plant
metabolic activities (Hatfield et al., 1987). Predicted plant surface temperature (T^) represents only an index
of the actual average plant canopy temperature, owing to the use of averages of daily parameters R^ G, E^.,
ra and T., and incompleteness of plant canopies during parts of the growing season. However, differences
between TS and Ta and relative changes in the value of TS for different climatic scenarios and values of canopy
resistance may reveal information that will be helpful in assessing crop productivity in the changing climatic
environments. Values of TS and Tg - T were computed for the month of each year having the highest value
of Ete, as this would likely be a period when the plant canopy is nearest full development (T more closely
represents canopy temperature than an integration of canopy and soil surface temperature). Also, the peak Et
month is likely to coincide with a period when the effects of temperature stress on crop growth and yield are
most pronounced.
Surface temperatures were predicted to increase above baseline (current climatic conditions) values for
all crops and sites studied. It is reasonable to speculate that surface temperatures would increase for all
irrigated crops as the atmosphere warms up. Higher surface temperatures and higher variations between study
sites were predicted for the GFDL scenario. Linear increases in surface temperatures over baseline values
were predicted to occur with increases in bulk stomatal diffusion resistances under both scenarios (GISS and
GFDL) as shown in Figure 6.
The increases in surface temperatures as bulk stomatal diffusion resistances increase were due to
reductions in Eg and corresponding increases in sensible heat fluxes where sensible heat fluxes were positive
(away from the crop) and to decreases in sensible heat where sensible heat fluxes were negative (advection of
heat into the crop). Increases in surface temperature were greatest for alfalfa. The variability among locations
in the changes in surface temperatures over the baseline values were greatest for wheat, followed by corn, and
were least for alfalfa These variations may be explained by the spatial (geographic) variations in the months
in which peak evapotranspiration occurred and in climatic differences and projected changes in air temperatures
for each GCM cell for peak months. The increases in surface temperatures agree with effects projected by
Slatyer and Bierchuizen (1964) and Polyakoff-Mayber and Gale (1972).
Projected increases in plant canopy temperatures were fairly constant between the two GCMs scenarios
for alfalfa and corn crops (see Figure 6). However, projected increases in canopy temperature for winter wheat
were 2 to 6°C greater under GFDL scenario as compared to the GISS scenario. These increases in canopy
tempreature were likely due to later projected fall plantings of winter wheat under the GFDL scenario due to
higher fall temperatures and extension of the growing seasons into hotter summer months as compared to the
GISS scenario.
The air-surface temperature differences serve as sensible heat transfer indices. The air-surface
temperature differences predicted for the scenarios decreased with increases in bulk stomatal resistance. The
6-28
-------
100
80
6O
4O
2O
01
0)
0
fl)
0)
w
(0
8
-------
Allen
o
o
(0
a)
3
.H
10
>
0)
c
0)
(0
(0
ja
e
o
to
\
' \
' \
' \
'\
/ j
^
1
7
/
/
/
/
/
/
/
/
/
/
/
/
/
y
(c) Winter Wheat
Figure 6.
GISS GFDL GISS GFDL GISS GFDL GISS GFDL
NEBRASKA KANSAS OKLAHOMA TEXAS
Change in surface temperature from baseline values for alfalfa, corn, and winter wheat vs. postulated
increases in bulk stomatal diffusion resistance.
6-30
-------
Allen
decreases indifferences were due to increases in surface temperatures, which resulted from increased bulk
stomatal diffusion resistances. These decreases reduced the transfer of sensible heat from the air to plant
surfaces. Plant canopies were estimated to be cooler than air temperatures because of evaporative cooling and
the extraction of sensible heat from the air.
POTENTIAL AGRONOMIC ADJUSTMENTS
The postulated climatic change scenarios were predicted to increase potential growing season lengths by
approximately 40 days and to compress actual growing seasons for current cultivars of corn and wheat by 0 to
80 days and 30 to 40 days, respectively (see Figure 1). Irrigators are likely to make changes in their cropping
systems to take advantage of the new climate regimes. This may take the form of either growing longer season
cultivars or by increasing cropping intensities.
Winter wheat was selected to study the effects of substituting longer season varieties in place of current
cultivars. This simulation was accomplished in the model by increasing the phenology energy units of winter
wheat by 10 and 30% for the crop development and crop maturation stages, respectively. The resulting mean
differences in season lengths between the current cultivars and longer season cultivar were 19 and 17 days for
the GISS and GFDL models, respectively (see Figure 7). Results indicate that growing season lengths for
winter wheat would still be shortened, as compared to current baseline conditions, even when phenology-energy
(growing degree radiation) requirements were increased by about 20%. It should be noted that some of this
reduction for winter wheat was caused by the reduction in lengths of dormant periods during winter months,
rather than by compression of growing periods.
Seasonal irrigation water requirements for winter wheat were predicted to increase by about 8 to 33%
across the region under the GISS scenario and by about 15 to 35% under the GFDL scenario as compared
to baseline values when longer season varieties were used (assuming a 20% increase in value of bulk stomatal
diffusion resistance). Increases in seasonal irrigation water requirement in Figure 8 for extended wheat cultivars
contrast with predicted reductions in seasonal irrigation water requirements for current cultivars as shown in
Figure 3.
Increases in potential growing seasons and compressed season lengths for annual crops may encourage
farmers to grow a second crop in regions with sufficient water supplies. This would result in increased irrigation
water requirements as evidenced in the alfalfa case study.
6-31
-------
Allen
1O
-30
-4O
-50
I
r
i
f \ \ \—
Nebraska Kansas Oklahoma Texas
QSS-C
QSS-L
V///////////A
GFDL-C
GFDL-L
Figure 7. Changes in season length (days) from the baseline values for current (C) and postulated longer-
season (L) varieties of winter wheat.
6-32
-------
Allen
W/////////A
40%
60%
V///////////A
BOX
-1O
GISS GFDL GISS GFDL GISS GFDL GISS GFDL
NEBRASKA KANSAS OKLAHOMA TEXAS
Figure 8. Percent change in net seasonal irrigation requirements from baseline values for longer-season
varieties of winter wheat vs. postulated increases in bulk stomatal diffusion resistance.
6-33
-------
Allen
CHAPTER IV
INTERPRETATION OF RESULTS
CLIMATIC AND STOMATAL RESISTANCE EFFECTS
The water balance-^ model was applied to baseline and GCM data to evaluate effects of a series of
climate change scenarios and variations in bulk stomatal diffusion resistances on irrigation water requirements.
Major changes in irrigation water requirements were observed in all 17 stations with requirements for alfalfa
significantly increasing and requirements for corn and wheat decreasing, depending on the climatic scenario and
projected increases in bulk stomatal diffusion resistances used. It should be noted that exact magnitudes of
changes cannot be predicted with complete accuracy owing to uncertainties in model input (GCM predictions
and projected cropping systems), the simplifying assumptions inherent in the model development, and the
complexities in the interactions of crops, farmers, and the environment. Results do, however, show definite
trends and relative changes in water requirements and local plant environment.
Climate-Induced Change
The results of the model indicate that the postulated climatic changes would have a significant effect on
seasonal net irrigation water requirements. The increases in irrigation requirements were mainly due to
increased evaporative demands driven by increased magnitudes of climatic parameters and changes in lengths
of growing seasons and in precipitation patterns.
Increases in temperatures, solar radiation, and windspeeds under the GCM scenarios provided the major
impetus for increased evaporative demands. Predicted increases in humidity and shifts in growing seasons to
months with lower levels of solar radiation ameliorated the majority of increase for annual crops. The lengths
of growing seasons have a major impact on irrigation water requirements. The largest increases in net seasonal
irrigation water requirements occurred in alfalfa owing to increased lengths of growing seasons and higher
evapoiative demands. Decreases or modest increases in seasonal irrigation requirements for corn and winter
wheat are attributed to the shortening of growing seasons and lower evaporative demands during growing
seasons due to earlier planting dates.
Stomatal Resistance-Induced Changes
Projected increases in bulk stomatal diffusion resistances brought about by increased levels of atmospheric
CO, had tremendous effect in ameliorating the impacts of climatic change on irrigation water requirements.
In all cases, irrigation water requirements were predicted to decrease with increasing levels of bulk stomatal
resistance. The levels of increase in bulk stomatal resistances required to negate the effect of climatic change
varied with location and crop. Surface temperatures were predicted to increase by one to two degrees as bulk
stomatal resistances increased by 80%. The increases in surface temperatures resulting from the climatic and
bulk stomatal resistance changes could have significant effects on crop metabolism. They may also make crops
more sensitive to moisture stress.
Because of uncertainties in the literature concerning the nature and magnitudes of increases in both LAI
and r., no specific conclusions are drawn concerning the probable magnitude of changes in bulk stomatal
diffusion resistances or amelioration of climatic impacts on irrigation water requirements. Any increases in
stomatal resistance are likely to increase water-use efficiencies. If the values of bulk stomatal diffusion
resistance do not change, then values of water-use efficiency for most crops will likely decrease, because of
increased Et demand resulting from advection of sensible heat.
6-34
-------
Allen
CAVEATS AND LIMITATIONS OF THE STUDY
1. Air temperature data measured at airport locations were adjusted to reflect temperature profiles expected
to have occurred over irrigated crops. However, wind and humidity data still contain effects of airport
settings (nonirrigated). Therefore, humidity levels are probably lower than those over irrigated crops and
wind levels are probably higher. These two effects have likely caused an overestimation of true baseline
crop EL by about 3 to 10 %. However, relative changes due to GCM scenarios are probably only slightly
affected by this bias.
2. Air temperature and humidity profiles in the boundary layer above a cropped surface are a reflection of
the energy balance at the surface, with higher positive fluxes of sensible heat generating steeper
temperature profiles for fixed levels of wind. Therefore, weather measurements are indicative of specific
evaporative and energy balance conditions at the ground or canopy surface. Any change in the condition
of the surface will cause a change in the temperature and vapor profiles and will change corresponding
weather measurements. The Penman and energy balance equations utilize logarithmic characteristics of
temperature, vapor, and wind profiles in estimating evaporative fluxes from crops. As the hypothesized
values of canopy resistances are increased, less energy at crop surfaces is converted into latent heat, with
more energy converted into sensible heat. Consequently, when sensible heat transfer is from crops to the
air, both the leaf temperatures and steepness of temperature profiles above crops will increase with
increasing canopy resistance, and air temperatures and vapor pressures at weather measurement heights
will be affected. This modification will provide a type of feedback control affecting energy balances at
leaf surfaces, much the same as the feedback of increased leaf temperature due to increased stomatal
resistance, which increases the vapor pressure inside the leaf, thereby increasing the evaporative flux out
of the leaf and reducing the effect of the increased resistance. The magnitudes and signs of the effects of
increased canopy resistance on air temperatures and vapor contents at weather measurement heights are
unknown, as the changes in surface temperatures and profile slopes are generally of opposite signs.
Therefore, there is some uncertainty embedded in the Et and irrigation water requirement estimates owing
to "mismatching" of weather profiles and crop characteristics.
3. The evapotranspiration analyses in this study assumed that alfalfa (used as a reference E| crop) and com
and wheat would all respond in similar manners to changes in carbon dioxide contents of the atmosphere,
primarily in changes in canopy resistance and leaf areas. This assumption permitted the use of "basal" crop
coefficients to estimate corn and wheat E^ from reference estimates for alfalfa.
4. Soil evaporation constitutes about 10 to 30% of the total evapotranspiration requirement of crops,
depending on the frequency of precipitation and irrigation and crop canopy development. The estimates
in this study are for a mixture of 50% center pivots with 2- to 3-day irrigation frequencies and 50 % other
system types with generally 7- to 28-day irrigation frequencies. Therefore, estimates in this study are
average values for a mixture of various system types.
5. The portion of daily rainfall infiltrating the soil and becoming available to reduce irrigation requirements
was estimated using values of surface runoff predicted by the SCS Curve number method. This method
is approximate. Therefore, estimations of effective precipitation, especially for large rainfall events, may
have significant error.
6. This study assumed that crop development was a function of the product of solar radiation and air
temperature (R8 T). Therefore, season lengths were shortened either by increased radiation (decreased
cloudiness) or by increased mean daily air temperature.
7. Bulk stomatal diffusion resistance (rc) includes both r and LAI components. Therefore, results of the
sensitivity analyses reflect integrated changes in these two parameters.
8. The effect of elevated CO, concentrations on phenological development (times of flowering, maturity, root
development, etc) or yield were not evaluated.
6-35
-------
Allen
9. Planting dates were significantly earlier and growing seasons for com and winter wheat were shorter under
the 2xCO2 global climatic scenarios. However, phenological or morphological changes or yield increases
or decreases due to changes in photoperiod lengths and/or increased vulnerability to spring frosts were
not evaluated.
10. Farmers may shift to crop varieties with higher environmental energy requirements to more fully utilize
increased amounts of solar radiation and temperature available. The majority of results presented in this
study assumed that varieties would not change, resulting in "compressed" growing seasons due to more
rapid crop growth and phenology. Two reasons why fanners may be reluctant to shift to longer season
varieties are the common lack of precipitation during later summer months, which would increase irrigation
water requirements or moisture stress for dryland crops, and elevated leaf temperatures during late
summer months, which may exceed optimum temperatures required for high productivity.
6-36
-------
Allen
CHAPTER V
IMPLICATIONS OF RESULTS
The postulated climatic changes imply future increased peak demands and potentially increased seasonal
demands on water resources in the Great Plains region. Because irrigated agriculture is the largest water user
in this region, policies and strategies for stretching water supplies will be required.
ENVIRONMENTAL IMPLICATIONS
Increases in irrigation water requirements for full-season crops such as alfalfa or for improved annual
cultivars that will be developed to take advantage of longer growing seasons will result in increased extraction
of water for irrigation purposes (assuming that the irrigated acreages remain the same or increase as dryland
farming gives way to irrigated agriculture). Increased extraction of groundwater may pose serious environmental
and economic problems, especially in areas where groundwater "mining" is currently being practiced. Reduced
streamflows resulting from increased extraction of surface and groundwater may aggravate water quality
problems which may in turn affect fish, wildlife, and recreational activities.
Water shortages brought on by increased water requirements by some crops may lead to increased salinity
problems if leaching requirements are not met. Recharge from irrigation seepage may also be diminished. Some
positive impacts may result from improved farming practices allowing for better control of water, soil, fertilizer,
and pesticides, thereby reducing agricultural pollution.
If irrigation requirements are reduced for com and winter wheat, as indicated for some locations, especially
if bulk stomatal diffusion resistances increase, then the changes will tend to benefit water resources and
economics as less water and energy will be required. Groundwater drafts from the Ogallala aquifer will likely
decrease, barring expansion of irrigated acreage, shifting to cultivars having longer growing seasons, or
implementation of double cropping under irrigation.
Water-use efficiencies may decrease if bulk stomatal diffusion resistances remain constant and advection
of sensible heat within the region increases. Water-use efficiencies may increase if bulk stomatal diffusion
resistances increase, thereby reducing the ratio of Ej to photosynthetic activities.
SOCIOECONOMIC IMPLICATIONS
Sodoeconomic impacts are difficult to assess because of the numerous external factors that influence the
supply and demand for agricultural produce. It is, however, obvious that increases in evaporative demands and
greater variability in rainfall will result in the following:
1. Potential reduction in crop yields due to reduced lengths of growing seasons for some annual crops.
However, some or all of the reduced yield from less radiation may be offset by increased photosynthesis
due to higher CO2 levels.
2. Increase in the need for irrigation in present dryland farming regions owing to increases in peak monthly
EJ and irrigation water requirements. Thus, the amount of capital invested in the irrigated sector may
increase as irrigated areas are increased, even though seasonal irrigation water requirements for some
crops may be less than at present
6-37
-------
Allen
3. Sizes of farm irrigation systems may need to be increased to meet increased peak demands. These
increases will also require larger peak drafts of groundwater and electric supplies.
4. Increased competition for water and energy among users and uses during peak irrigation months. This may
spur the development of technologies that promote water and energy conservation practices (technical and
not technical) and breeding of crop varieties better adapted to the climatic conditions.
5. Increased yields for full-season crops such as alfalfa owing to longer growing seasons.
6. Decreased lengths of growing seasons for corn may reduce energy costs for grain drying, since the corn
crop will have more opportunity to dry in the fields after reaching an earlier maturity.
7. Farmers may shift to crop varieties with higher seasonal environmental energy requirements to more fully
utilize increased levels of available solar radiation and temperature. Two reasons why farmers may be
reluctant to shift to longer season varieties are the common lack of precipitation during later summer
months, which would increase irrigation water requirements or moisture stress for dryland crops, and
elevated leaf temperatures during late summer months, which may exceed optimum temperatures required
for high productivity. However, other economic considerations are the possibility of lower yields due to
reduced lengths of growing seasons for some annual crops. Some, or all, of the yield reductions due to
lower amounts of solar radiation may be offset by increased photosynthesis due to increased CO2 levels.
Based on the assumption that the effects of climatic change will be spread over space and will occur
gradually through time, we can expect modern agriculture to learn how to cope with the changes from the
experiences gained in regions earlier impacted. During the transition stage, irrigators, planners, policy-makers,
and research communities could develop strategies to cope with the transient and final steady-state conditions.
6-38
-------
Allen
REFERENCES
Allen, L.H., 1986a. Plant responses to rising CO2. Paper presented at the 79th Annual Meeting of the Air
Pollution Control Associatioa 33 pages.
Allen L.H., P. Jones, and J.W. Jones 1985. Rising atmospheric CO2 and evapotranspiration, Proceedings of
the National Conference on Advances in Evapotranspiration, ASAE, held at Hyatt Regency, Chicago, Illinois,
Dec 1985 pp 13-27.
Allen, R.G. 1986b. A Penman for all seasons. Journal of Irrigation Vti PfiUllW Engineering. ASCE,
112(4):348-368.
Allen, R.G. 1988. Adjustment of historical weather data to reflect an irrigated environment. Paper in
preparation for submittal to Water Resources Research.
Allen, R.G. and C. E. Brockway. 1982. Estimating consumptive irrigation requirements for crops in Idaho.
Idaho Water and Energy Resources Research Institute Research Completion Report. 183 pages.
Allen, R.G., C.E. Brockway and Ji. Wright 1983. Weather station siting and consumptive use estimates.
Journal of Water Resources Planning and Management ASCE, 109(2):134-146.
Allen, R.G. and W.O. Pruitt. 1986. Rational use of the FAO Blaney-Criddle formula Journal of Irrigation
and Drainage Engineering. ASCE, 112(2): 139- 155.
Allen, R.G., M.E. Jensen, J.L. Wright and R.D. Burman. 1988. Operational estimates of daily
evapotranspiration. Paper accepted for publication to the Agronomy Journal.
Beven K. 1979. A sensitivity analysis of the Penman-Monteith actual evapotranspiration estimates, Journal of
Hydrology, 44, 169-190.
Bouchet, R J. 1963. Evapotranspiration re'elle et potentielle, signification climatique. Int. Assoc. Sci. Hydrol.,
Publ, No. 62, pp 134-142.
Brockway, C.E., GJS. Johnson, J.L. Wright and AX. Coiner. 1985. Remote sensing for irrigated crop water
use. Phase 1. Water Resources Research Institute, University of Idaho, Moscow, Idaho 128 pages.
Brutsaert, W.H. 1975. Comments on surface roughness parameters and the height of dense vegetation. L
Meteorol. Soc. Japan, (53):96-97.
Brutsaert, WJH. and H. Strieker. 1979. An advection-aridity approach to estimate actual regional
evapotranspiration, Water Resources Research. Vol. 15, No. 2, pp. 443-450.
Brutsaert, W.H. 1982. Evaporation into the atmosphere. Theory, history and applications. D. Reidel
Publishing Company, Boston, UJSA.
Buchheim, J. and A. Brower. 1981. Crop stage versus summation of ETP. Mimeographed departmental
report, United States Bureau of Reclamation, Denver, Colorado. 20 pages.
Curry B.R. 1970. Climates of the states: Oklahoma, Climatographv of the United States No. 60-34, VS.
Department of Commerce US. Government Printing Office, Washington, D.C.
6-39
-------
Allen
Dale, R.P. and R.H. Shaw. 1965. Effect on corn yields of moisture stress and stand at two fertility levels,
Agronomy J^TP^Ii 57:475-479.
Doorenbos, J. and W.O. Pruitt 1977. Crop Water Requirements. United Nations Food and Agriculture
Organization (FAO) Irrigation and Drainage Paper Number 24, 144 pages.
Duffie, J A. and WA. Beckman. 1980. Solar Engineering nf Thermal Processes. John Wiley and Sons. New
York, pp 1-109.
Federer, C. A. 1979. A soil-plant-atmosphere model for transpiration and availability of soil water, Water
Resources Research, (15)555-562.
Garratt, J.R. and B.B. Hicks. 1973. Momentum, heat and water vapor transfer to and from natural and
artificial surfaces. Quart.!. Roval Met. Soc.. 99:680-687.
Glantz, M. H. and J. H. Ausubel. The Ogallala aquifer and carbon dioxide: comparison and convergence,:
Environmental Conservatioa 11(2):123-131, 1984.
Gribbin J. 1981. The politics of carbon dioxide, New Science, 90:82-84.
Hatfield J.L., D.F. Wanjura and G.L. Barker. 1985. Canopy temperature response to water stress under partial
canopy, Trans. ASAE. 28:1607-1611.
Hatfield, J.L., JJ. Burke, J.R. Mahan, and D.F. Wanjura. 1987. Foliage temperature measurements: A link
between the biological and physical environment, Proceedings of the International Conference on Measurement
of Soil and Plant Water Status, Vol. 2, Utah State University, Logan, UT, pp. 99-112.
Hill, R.W., R J. Hanks, and J.L. Wright. 1982a. Crop yield models adapted to irrigation scheduling programs.
Research Report 99, Utah Agricultural Experiment Station, Utah state University, Logan, Utah, 180 pages.
Hill, R.W., AA. Keller, and B. Boman. 1982b. Crop yield models adapted to irrigation scheduling programs,
Appendix F: CRPSM user manual and sample input and output. Research Report 100, Utah Agricultural
Experiment Station, Utah state University, Logan, Utah, 94 pages.
Hjelmfelt, A.T. and J J. Cassidy. 1975. Hydrology for engineers and planners. Iowa State University Press,
Ames, la, 210 p.
Hofstra, G. and J.D. Hesketh. 1969. The effect of temperature on stomatal aperture in different species, Can.
J. Bot.. 47:1307-1310.
Jensen, M.E., and H.R. Haise. 1963. Estimating evapotranspiration from solar radiation, J. Irrig. and Drainage
Div.. ASCE. 89(1):15-41.
Jensen, M.E., D.C.N. Robb and C.E. Franzoy. 1970. Scheduling irrigations using climate-crop-soil data, J_.
Irrig. and Drainage Qjv.. ASCE. 96(l):25-28.
Kimball B. A. and S.B. Idso. 1983. Increasing atmospheric CO2: Effects on crop yield, water use and climate,
Agricultural Water Management, 7(1983) 55-72.
Lemon, E.R. and J.L. Wright. 1969. Photosynthesis under field conditions. XA. Assessing sources and sinks
of carbon dioxide in a corn (Zea mays L.) crop using a momentum balance approach.
Agronomy Journal 61:405-410.
Leopold, A.C., 1964, Plant growth and development. McGraw-Hill, New York, 465pp.
6-40
-------
Allen
Monteith J.L. 1965. Radiation and crops.:Experimental Agriculture Review. 1(4):241-251.
Monteith, J.L. 1981. Evaporation and surface temperature. Quart. J. Rov. Meteorol. Soc. 107:1-27.
Musick, J.T., L.L. New, and DA. Dusek, 1976. Soil water depletion-yield relationship of irrigated sorghum,
wheat, and soybeans, Trans. ASAE. 19:489-493.
Orton R.B. 1%9. Climates of the states: Texas, Climatographvof the United States No. 60*34, UJS. Department
of Commerce UJS. Government Printing Office, Washington, D.C.
Plate, E J. 1971. Aerodynamic characteristics of atmospheric boundary layers. AEC Critical Review Series,
UJS. Atomic Energy Commission, Div. Tech. Info. 190 pp.
Poljakoff-Mayber, A. and J. Gale. 1972. Physiological Basis and practical problems of reducing transpiration.
Water deficit and plant growth (T. T. Kozlowski ed.), Academic Press, New York, pp. 277-306.
Robb A.D. 1959. Climates of the states: Kansas, nimatographv of the United States No. 60-34, UJS.
Department of Commerce UJS. Government Printing Office, Washington, D.C.
Rosenberg, NJ. 1969. Advective contribution of energy utilized in evapotranspiration by alfalfa in the east
central Great Plains, Aerie. Meteorology. 6:179-184.
Rosenberg N. J. 1981. The increasing CO2 concentration in the atmosphere and its implication on agricultural
productivity -1. Effects on photosynthesis, transpiration and water use efficiency, Climatic Change 3(1981)265-
279.
Rosenberg N. J. 1982. The increase in carbon dioxide concentration in the atmosphere and its implication on
agricultural productivity U - Effects through CO2-induced climatic change, Climatic Change 4(1982)239-254.
Rosenberg NJ., B. L. Blad, and S. B. Verma. 1983. Microclimate: The Biological Environment Wiley and Sons,
New York.
Rosenberg N. J., B. Kimball, P. Martin and C. Cooper. 1988. Greenhouse warming, CO2 enrichment of the
atmosphere and evapotranspiration, Draft report submitted to the American Academy for the Advancement
of Science.
Rosenberg NJ. and S.B. Verma 1978 Extreme evapotranspiration by irrigated alfalfa: A consequence of the
1976 midwest drought, J. Appl. Meteorology. 17:934-941.
Saxton, K.E. 1975. Sensitivity analyses of the combination evapotranspiration equation. Agricultural
Meteorology. 15:343-353.
Schaffer A. and R. C. Schaffer. 1984. Social impacts on rural communities, Water Scarcity: Impacts on western
agriculture. (Engelbert and Scheming eds.), University of California Press, pp. 484.
Sharma, MJL 1985. Estimating evapotranspiration. Advances in Irrigation. D. Hillel, Ed, Academic Press,
Inc., New York, N.Y., p 213-281.
Slatyer, R.O., and J J7. Biercuizen, 1964. The influence of several transpiration suppressions on transpiration,
photosynthesis, and water use efficiency of cotton leaves, Austr. J. Biol. So. 17:131-146.
Stevens WJA. 1959. Climates of the states: Nebraska, Climatographv of the United States No. 60-34, UJS.
Department of Commerce UJS. Government Printing Office, Washington, D.C.
6-41
-------
Allen
Szeicz, G. and I.P. Long. 1969. Surface resistance of canopies. Water Resources Res. 5:622-633.
Tanner, C.B. and M. Fuchs. 1968. Evaporation from unsaturated surfaces: A generalized combination method
J. Geophvs. Res.. 73:1299-1304.
Temple P J. and L. F. Benoit. 1988. Effects of ozone and water stress on canopy temperature, water use, and
water use efficiency of alfalfa, Agronomy Journal 80:439-447(1988)
U.S. Department of Agriculture. 1982. Basic statistics, 1977 National Resources Inventory. SB-686.
USDA-Soil Conservation Service. 1967. Irrigation water requirements, Technical release No. 21, Engineering
Division, Revised 1970. 88 pages.
van Bavel, C.H.M. and WX. Ehrler. 1968. Water loss from a sorghum field and stomatal control. Agronomy
Journal.. 60:84-86.
Wright, J.L. 1982. New evapotranspiration crop coefficients. Journal of the Irrigation and Drainage Division.
ASCE, 108(2):S7-74.
Wright, J.L. and E.R. Lemon. 1966. Photosynthesis under fielotonditions. DC. Vertical distribution of
photosynthesis within a corn crop. Agon. J.. 58:265-268.
Wright, J.L. and K.W. Brown. 1967. Comparison of momentum and energy balance methods of computing
vertical transfer within a crop. Apron. J.. 59:427-432.
6-42
------- |