United States
Environmental Protection
Agency
Policy, Planning,
And Evaluation
(2122)
EPA 230-B-94-003
June 1994
Implications Of Climate Change
For International Agriculture:
Crop Modeling Study
-------
-------
Implications Of Climate Change For
International Agriculture:
Crop Modeling Study
Cynthia Rosenzweig and Ana Iglesias, Editors
United States Environmental Protection Agency
Office of Policy, Planning and Evaluation
Climate Change Division
Adaptation Branch
Ronald Benioff and Joel Smith, Project Officers
June 1994
-------
-------
IMPLICATIONS OF CLIMATE CHANGE FOR INTERNATIONAL AGRICULTURE:
CROP MODELING STUDY
Editors
Cynthia Rosenzweig and Ana Iglesias
-------
-------
TABLE OF CONTENTS
PREFACE
EXECUTIVE SUMMARY
SECTION 1: INTRODUCTION
THE USE OF CROP MODELS FOR INTERNATIONAL CLIMATE CHANGE IMPACT
ASSESSMENT: STUDY DESIGN, METHODOLOGY, AND CONCLUSIONS
Cynthia Rosenzweig and Ana Iglesias
SECTION 2: NORTH AMERICA
EFFECTS OF GLOBAL CLIMATE CHANGE ON WHEAT YIELDS IN THE CANADIAN
PRAIRIE
M. Brklacich, R. Stewart, V. Kirkwood, and R. Muma
THE EFFECTS OF POTENTIAL CLIMATE CHANGE ON SIMULATED GRAIN CROPS IN
THE UNITED STATES
C. Rosenzweig, B. Curry, J.T. Ritchie, J.W. Jones, T.-Y. Chou, R. Goldberg, and A. Iglesias
POSSIBLE IMPACTS OF CLIMATE CHANGE ON MAIZE YIELDS IN MEXICO
Diana Liverman, Max Dilley, Karen O'Brien, and Leticia Menchaca
SECTION 3: SOUTH AMERICA
POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE FOR BRAZILIAN AGRICULTURE:
APPLIED SIMULATION STUDIES FOR WHEAT, MAIZE, AND SOYBEANS
Otavio Joao Fernandes de Siqueira, Jose Renato Boucas Farias, and Luis Marcelo Aguiar Sans
IMPACTS OF GLOBAL CLIMATE CHANGE ON MAIZE PRODUCTION IN ARGENTINA
O. E. Sala and J.M. Paruelo
IMPACT OF CLIMATE CHANGE ON BARLEY IN URUGUAY: YIELD CHANGES AND
ANALYSIS OF NITROGEN MANAGEMENT SYSTEMS
Walter E. Baethgen
SECTION 4: EUROPE
POSSIBLE EFFECTS OF INCREASING C02 CONCENTRATION ON WHEAT AND MAIZE
CROPS IN NORTH AND SOUTHEAST FRANCE
R. Del6colle, D. Ripoche, and F. Ruget, G. Gosse
POTENTIAL EFFECTS OF GLOBAL WARMING AND CARBON DIOXIDE ON WHEAT
PRODUCTION IN THE FORMER SOVIET UNION
Gennadiy V. Menzhulin, Larisa A. Koval, Alexander L. Badenko
-------
SECTION 5: AFRICA
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT AND MAIZE YIELDS IN EGYPT
H.M. Bid
IMPLICATIONS OF CLIMATE CHANGE FOR MAIZE YIELDS IN ZIMBABWE
Paul Muchena
SECTION 6: ASIA
IMPLICATIONS OF GLOBAL CLIMATE CHANGE FOR AGRICULTURE IN PAKISTAN:
IMPACTS ON SIMULATED WHEAT PRODUCTION
Ata Qureshi, and Ana Iglesias
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT PRODUCTION IN INDIA
D.Gangadhar Rao and S.K.Sinha
IMPACT OF CLIMATE CHANGE ON THE PRODUCTION OF MODERN RICE IN
BANGLADESH
Z. Karim, M. Ahmed, S.G. Hussain, and Kh.B. Rashid
IMPACT OF CLIMATE CHANGE ON SIMULATED RICE PRODUCTION IN THAILAND
C. Tongyai
CLIMATE IMPACT ASSESSMENTFOR AGRICULTURE IN THE PHILIPPINES: SIMULATION
OF RICE YIELD UNDER CLIMATE CHANGE SCENARIOS
Crisanto R. Escano and Leandro V. Buendia
EFFECTS OF CLIMATE CHANGE ON RICE PRODUCTION AND STRATEGIES FOR
ADAPTATION IN SOUTHERN CHINA
Zhiqing Jin, Daokou Ge, Hua Chen, and Juan Fang
IMPLICATIONS OF CLIMATE CHANGE FOR JAPANESE AGRICULTURE: EVALUATION
BY SIMULATION OF RICE, WHEAT, AND MAIZE GROWTH
Hiroshi Seino
SECTION 7: AUSTRALIA
POSSIBLE EFFECTS OF GLOBAL CLIMATE CHANGE ON WHEAT AND RICE
PRODUCTION IN AUSTRALIA
Brian D. Baer, Wayne S. Meyer, and David Erskine
-------
PREFACE
This publication presents the crop modeling research conducted over the period 1989-1992 for the
U.S. Environmental Protection Agency Climate Change Division project "Implications of Climate Change for
International Agriculture: Global Food Production, Trade and ^Vulnerable Regions." Additional support was
provided by the US Agency for International Development. Principal Investigators of the project were Cynthia
Rosenzweig, of Columbia University and Goddard Institute for Space Studies, and Martin Parry, of the
Environmental Change Unit at Oxford University.
At U.S. EPA Climate Change Division, we thank Dennis Tirpak, Joel Smith, Ron Benioff, and Joel
Scheraga; and at U.S. AID, Tej Gill. Dr. Roy Jenne provided the GCM climate change scenarios. At Columbia
University, we thank Ruth Levenson, Chris Shashkin, and Rich Goldberg. We especially acknowledge the
IBSNAT crop modeling team, Bruce Curry, Gerrit Hoogenboom, Jim Jones, Joe Ritchie, Upendra Singh, and
Gordon Tsuji, for worldwide scientific and technical support.
Finally, the editors and authors are grateful to the reviewers of the reports: E. Alocilja, D. Bachelet,
T. Carter, S. Cohen, Y. Congyi, E. Cooler, G. Edmeades, J. Gerber, J. Goudriaan, A Hall, H. Hamdy, T.
Hodges, S. Hollinger, E. Kanemasu, J. Kiniry, G. Knapp, R. Loomis, R. Luxmore, H. Nix, S. Obrien, D.
Pottker, R. Puluektov, G. Rocca da Cunha, J. Russell, B. Smit, Z. Uchijima, J. Wescoat, E. Wheaton, and D.
Wilkes.
Special thanks to Shelly Preston for her editing prowess and dedication to the project.
-------
-------
EXECUTIVE SUMMARY
The central aim of the study was to provide an assessment of potential climate change impacts on
world crop production, including quantitative estimates of yield and water-use changes of major crops.
Agricultural scientists in 18 countries estimated potential changes in water use and crop growth, using
compatible crop models and consistent climate change scenarios. The crops modeled were wheat, rice, maize
and soybean. Wheat, rice and maize account for approximately 85% of world cereal exports; soybean accounts
for about 67% of trade in protein cake equivalent. Site-specific estimates of yield changes were aggregated to
national levels for the modeled major crops for use in a world food trade model, the Basic Linked System.
The study assessed the implications of climate change for world crop yields taking into account:
(a) Uncertainty in the level of climate change expected. Sensitivity tests were conducted with arbitrary increases
in temperature and changes in precipitation. The effects of three climate change scenarios were also tested
using doubled CO2 equilibrium climates from three general circulation models: the Goddard Institute for Space
Studies (GISS), the Geophysical Fluid Dynamics Laboratory (GFDL) and the United Kingdom Meteorological
Office (UKMO) models. These climates are assumed to occur in 2060. Finally, a transient projection of climate
change was tested, based on the GISS transient run A, for the 2010s, 2030s, and 2050s.
(b) Physiological effects of CO2. The climate change scenarios were tested with and without the direct effects
of CO2 on crop growth and water use, as reported in experimental literature.
(c) Different adaptive responses. Climate change impacts on crop yields incorporating farm-level adaptation
were simulated, based on different assumptions about shifts in crop planting dates, changes in crop variety,
level of irrigation, etc.
The principal results that emerged from the study were:
Sensitivity tests. A 2 ° C temperature rise increased aggregated crop yields on a global basis, while a 4 ° C
rise led to decreases in globally aggregated crop yields. Nevertheless, in semi-arid and subtropical regions, a
2°C temperature increase caused yield declines. The greatest yield decreases are caused by a 4°C temperature
increase and 20% precipitation decrease.
GCM climate change scenarios without adaptation. Without physiological CO2 effects, production of
all three crops decreased compared to baseline climate conditions on a global basis. With CO2 effects, yields
were positive at middle and high latitudes, and negative at low latitudes for the scenarios with lower
temperature increases (~4°C global surface air temperature increase). For the warmest scenario (~5°C
temperature increase), crop yields declined almost everywhere. Thus, increases in potential yield depend
strongly on full realization of the direct effects of CO2 on crop growth.
GCM climate change scenarios with adaptation. Farm-level adaptation (shifts in planting dates, changes
in crop variety, application of irrigation) reduces the negative effects of climate change. However, even when
farmer adaptation is taken into account, climate change may decrease yields in semi-arid, subtropical regions.
Successful adaptation often implies significant changes to current agricultural systems.
In order to minimize possible adverse consequences to climate change worldwide, the agricultural
sector should continue to develop crop breeding and management programs for heat and drought conditions
(these will be immediately useful in improving productivity in marginal environments today). Another
important activity is to enlarge, maintain, and screen crop genetic resources at established seedbanks.
Resilience of the agricultural production sector also depends on improved use of systems for monitoring
weather, soil moisture, nutrient requirements, and pest infestations. Finally, strong communication links among
the agricultural research, production, and policy sectors are essential.
-------
-------
SECTION 1: INTRODUCTION
-------
-------
THE USE OF CROP MODELS FOR
INTERNATIONAL CLIMATE CHANGE IMPACT ASSESSMENT:
STUDY DESIGN, METHODOLOGY, AND CONCLUSIONS
Cynthia Rosenzweig
Columbia University and Goddard Institute for Space Studies
New York, USA
and
Ana Iglesias
Inst. Nacional de Investigaciones Agrarias (INIA)
Madrid, Spain
INTRO-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
BACKGROUND AND PREVIOUS STUDIES
CLIMATE CHANGE SCENARIOS
Sensitivity Tests
GCM Equilibrium and Transient Scenarios
Limitations
CROP MODELS
Description
Physiological CO2 Effects
Limitations
CALIBRATION AND VALIDATION
CROP MODELING PROCEDURES
ESTIMATION OF NATIONAL YIELD CHANGES
DESIGN OF THE ADAPTATION STUDY
Planting Date
Irrigation
Fertilizer
Crop Variety
Cropping Area
SOURCES OF UNCERTAINTY
RESULTS
Sensitivity Tests
GCM Scenarios
Transient Scenarios
Adaptation Studies
CONCLUSIONS AND FUTURE RESEARCH NEEDS
INTRO-2
-------
SUMMARY
The methodology for an assessment of potential climate change impacts on world crop production,
including quantitative estimates of yield and water use changes for major crops, is described. Agricultural
scientists in 18 countries estimated potential changes in crop growth and water use using compatible crop
models and consistent climate change scenarios. The crops modeled were wheat, rice, maize and soybean. Site-
specific estimates of yield changes for the major crops modeled were aggregated to national levels for use in
a world food trade model, the Basic Linked System. The study assessed the implications of climate change for
world crop yields for arbitrary and GCM equilibrium and transient climate change scenarios. The climate
change scenarios were tested with and without the direct physiological effects of CO2 on crop growth and water
use, as reported in experimental literature. Climate change impacts on crop yields incorporating farm-level
adaptation were simulated, based on different assumptions about shifts in crop planting dates, changes in crop
variety, and level of irrigation.
INTRODUCTION
Scientists predict significant global warming in the coming decades due to increasing atmospheric carbon
dioxide and other trace gases (IPCC 1990a; 1992). Substantial changes in hydrological regimes are also forecast
to occur. Understanding the potential effects of these changes on agriculture is an important task, because
agriculture provides food for the world's population, now estimated at 5 billion and projected to rise to 10
billion in the coming century. Despite technological advances such as improved crop varieties and irrigation
systems, weather and climate are still key factors in agricultural productivity. For example, weak monsoon rains
in 1987 caused large shortfalls in crop production in India, Bangladesh, and Pakistan, contributing to reversion
to wheat importation by these countries (World Food Institute 1988). Despite adequate supplies elsewhere,
the 1980s also saw the continuing deterioration of food production in Africa, caused in part by persistent
drought and low production potential. This resulted in international relief efforts to prevent widespread
famine. These examples emphasize the close links between agriculture and climate, the international nature
of food trade and food security, and the need to consider the impacts of climate change in a global context.
Recent research has been focused on regional and national assessment of the potential effects of climate
change on agriculture (IPCC 1990b). The methodology for regional and national climate impact studies has
thus been developed and tested. However, the studies have, for the most part, treated each region or nation
in isolation, without relation to changes in production in other places. The purpose of this study was to
increase understanding of potential simultaneous changes in production in all major food-producing regions,
because such changes may lead to altered world supply and demand (and prices), and hence competitiveness
in any given region. Such understanding should aid in the meaningful interpretation of regional climate change
impact studies.
The study1 was an international collaborative effort of agricultural scientists in 18 countries. A suite
of dynamic process models, climate change scenarios, and simulation experiments was assembled, comprised
of climate sensitivity tests and climate change scenarios devised from global climate models (GCMs), dynamic
crop growth models, and a world food trade model (Figure 1). Common methodology was developed for the
'The study was commissioned by the U.S. Environmental Protection Agency Climate Change Division.
The U.S. Agency for International Development provided support and additional funding for the crop
modeling simulations.
INTRO-3
-------
simulation of climate change impacts on major agricultural crops and for the analysis of the results. The
International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT1989) compatible crop growth
models were utilized at over 100 sites (Figure 2). The crop models can simulate the direct physiological effects
of increased atmospheric CO2 on crop growth and water use. They allow for the simulation of both rainfed
and irrigated agricultural systems and other potential farmer adaptations to climate change. Results of the
dynamic crop growth simulations were then used to estimate global changes in crop yields for use in a world
food trade model, the Basic Linked System (Fischer et al. 1988). Results of the entire study are described in
Rosenzweig and Parry (1994).
BACKGROUND AND PREVIOUS STUDIES
Research on the effects of climate on agriculture has been extensive for many years. Much of the
previous research on climate impact assessment sought to isolate the effects of climate on agricultural activity,
whereas lately there has been a growing emphasis on understanding the interactions of climatic, environmental,'
and social factors in a wider context (Parry et al. 1988). An example of the earlier approach is the U.S.
Department of Transportation's study on the possible effects of atmospheric ozone depletion which (in part)
used regression models to determine statistical relationships between climate and agriculture, a method which
does not explain the processes underlying the relationships. The National Defense University also studied long-
term effects of climate change on crop yields and agricultural production with a relatively simple cause-and-
effect approach (NDU 1980).
More integrated approaches to climate change impact assessments have been described by Callaway
et al. (1982), the Carbon Dioxide Assessment Committee, Nix (1985), and Warrick et al. (1986). These studies
and reviews advocate comprehensive research methods that integrate economic and political factors as well
as biophysical ones. Some integration has been achieved in national impact studies completed in the United
States (Adams et al. 1990; Smith and Tirpak 1989), Canada (Smit 1989), Australia (Pearman 1988), the UK
(UK Department of the Environment 1991), and New Zealand (Martin et al. 1990), and in regional studies
including high-latitude and semi-arid agricultural areas (Parry et al. 1988) and the U.S. Midwest (Rosenberg
and Crosson 1991). 5
These regional and national studies have been summarized in the Intergovernmental Panel on Climate
Change (IPCC) Working Group II Report (IPCC 1990b), but integrated global assessments of climate change
impacts have been few to date. Kane et al. (1991) have analyzed the sensitivity of world agriculture to potential
climate changes and found that the overall effect of moderate climate change on world and domestic
economies may be small as reduced production in some areas is balanced by gains in others. These estimates
did not consider the agricultural impacts of climate change associated with the higher end of the predicted
IPCC-predicted range of warming (1.5 to 4.5 °C mean global surface air temperature rise) (IPCC 1990a).
Leemans and Solomon (1993) found that climate change would affect the yield and distribution of world crops
for one GCM doubled CO2 climate change scenario, leading to production increases at high latitudes and
production decreases in low latitudes.
CLIMATE CHANGE SCENARIOS
Because of the uncertainties surrounding prediction of climate change, it is common to employ climate
scenarios (Wigley 1987; Lamb 1987), in order to estimate the impacts of climate change on a system (in this
case agricultural production, both potential and actual, and world food trade). Climate scenarios are sets of
climatic perturbations which are used with impact models to test the sensitivity of the system to the projected
changes. The design of an impact study should include more than one scenario, so that a range of possible
effects may be defined. Realism is augmented when the climate change scenarios are internally consistent, i.e.,
INTRO-4
-------
the climate variables within the scenario should vary in a physically realistic way (Wigley 1987). Common
doubled CO2 GCM climate change scenarios and sensitivity tests were devised for the global crop modeling
analysis.
Sensitivity Tests
An approach to analyze the possible impacts of different climate on crop yield is to specify
incremental changes to temperature and precipitation and to apply these changes uniformly to the baseline
climate. Arbitrary climate sensitivity tests were conducted to test crop model responses to a range of
temperature (+2° and +4°C) and precipitation changes (+/- 20%). The scenario, +2°C with a precipitation
decrease of -20%, is interesting because it assumes a possible temperature increase combined with a large
decrease in the amount of rain, which is not simulated under the GCM climate scenarios. Sensitivity studies
allow the consideration of the question: "What type, magnitude, and rate of climate change would seriously
perturb the agricultural system in question?"
GCM Equilibrium and Transient Scenarios
Scenarios are often devised by changing an original set of climatological data by prescribed anomalies.
These anomalies may be derived from historical climate or from global climate models. GCMs provide the
most advanced tool for predicting the potential future climatic consequences of increasing radiatively active
trace gases in a consistent manner. However, climate models have not yet been validated to project changes
in climate variability, such as changes in the frequencies of drought and storms, even though these could affect
crop yields significantly.
Mean annual changes in climate variables from doubled CO2 simulations of three GCMs—Goddard
Institute for Space Studies (GISS, Hansen et al 1983), Geophysical Fluid Dynamics Laboratory (GFDL,
Manabe and Wetherald 1987), and United Kingdom Meteorological Office (UKMO, Wilson and Mitchell
1987)—were applied to observed (baseline) daily climate to create climate change scenarios for each site (Table
1). GCMs were used to create climate change scenarios because they produce climate variables that are
internally consistent; thus they allow for comparisons between or among regions.
The method used the differences between lxCO2 and 2xCO2 monthly GCM temperatures, and the ratio
between 2xCO2 and lxCO2 monthly precipitation and solar radiation amounts (lxCO2 refers to modeled
current climate conditions and 2xCO2 refers to the modeled climate that would occur with an equivalent
radiative forcing of doubled CO2 in the atmosphere). Temperature ranges of the three GCMs are near the high
end of the IPCC range of predicted warming for doubled CO2 (IPCC 1990a). In general, the GCMs predict
increases in global precipitation associated with warming because warmer air can hold more water vapor.
Most of our knowledge concerning the climate response to greenhouse-gas forcing has been obtained
from equilibrium response GCM experiments. These are experiments which consider the steady-state response
of the model's climate to step-function changes in atmospheric CO2. Recent evidence from a few GCM
experiments incorporating time-dependent greenhouse-gas forcing suggest that there may be important
differences between the equilibrium and transient responses (Hansen et al. 1988, Bryan et al. 1988; IPCC
1992).
This study also considered a set of transient climate scenarios (as opposed to the atmospheric
equilibrium scenarios), derived from the GISS transient climatic simulations (Hansen et al. 1988) for the
2010s, 2030s, and 2050s, and assuming CO2 concentrations of 405, 460, and 530 ppm, respectively. Transient
scenarios for each site were developed by using the same procedure as that used for the equilibrium scenarios.
Limitations
INTRO-5
-------
Current global climate models which have been used for CO2 studies employ grids on the order of 4°
latitude by 5° longitude, or greater. At this resolution, many smaller scale elements of climate are not properly
represented, such as warm and cold fronts and hurricanes, as well as the diversities of ecosystems and land-use.
Accurate modeling of hydrological processes is particularly crucial for determining climate change impacts on
agriculture, but GCM simulation of infiltration, runoff, and evaporation, and other hydrological processes is
highly simplified. Precipitation, in particular, is sometimes poorly represented in GCMs results.
There is also uncertainty in the prediction of rate and magnitude of climate change. Ocean heat
transport is a key, but not well understood, process that affects how fast the climate may warm. The doubled
CO2 climate change scenarios used in this study have assumed an abrupt doubling of the CO2 concentration
in the atmosphere and then have allowed the simulated climate to come to a new equilibrium. This step
change in the atmosphere is unrealistic, since trace gases are increasing gradually. The transient scenario does
simulate the response of gradually increasing radiatively active gases and is more realistic in this regard.
CROP MODELS
Description
The crop modeling study estimated how climate change and increasing levels of carbon dioxide may
alter yields and water use of world crops in both major production areas and vulnerable regions. The crops
modeled were wheat, rice, maize, and soybeans. Table 2 shows the percentages of world production modeled
in this study for wheat, rice, maize, and soybean. Even though only two countries (Brazil and USA) simulated
soybean production, their combined output accounts for 76% of world total. Less of the total world rice
production was simulated than total production of the other crops. This is because India, Indonesia, and
Vietnam have significant rice production not included in the study. Together, these crops account for more
than 85% of the world traded grains and legumes, although only approximately 4% of rice produced is traded
compared to about 20% of wheat. Rice is included in the study because of its importance to the food security
of Asia.
The IBSNAT models employ simplified functions to predict the growth of crops as influenced by the
major factors that affect yields, i.e., genetics, climate (daily solar radiation, maximum and minimum
temperatures, and precipitation), soils, and management (IBSNAT 1989). The models used were CERES-
Wheat (Ritchie and Otter 1985; Godwin et al. 1989), CERES-Maize (Jones and Kiniry 1986; Ritchie et al.
1989), CERES-Rice (both paddy and upland) (Godwin et al. 1993), and SOYGRO (soybean) (Jones et al.
1989).
The IBSNAT models were selected for use in this study because they have been validated over a wide
range of environments (e.g., Otter-Nacke et al. 1986) and are not specific to any particular location or soil
type. Thus they are suitable for use in international studies in which crop growing conditions differ greatly.
The validation of the CERES and SOYGRO models over different environments also serves to enhance
predictive capability concerning the climate change scenarios, in cases when predicted climates are similar to
existing climates in other regions. Furthermore, because management practices, such as cultivar, planting date,
plant population, row spacing, and sowing depth, may be varied in the models, they permit experiments that
simulate management adjustments by farmers to climate change.
Modeled processes include phonological 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 and SOYGRO
models also have the capability to simulate the effects of nitrogen deficiency and soil-water deficit on
photosynthesis and pathways of carbohydrate movement in the plant.
INTRO-6
-------
Physiological CO2 Effects
Most plants growing in atmospheric CO2 that is higher than ambient levels exhibit increased rates of
net photosynthesis (i.e., total photosynthesis minus respiration). High CO2 also reduces the stomatal openings
of some crop plants. By so doing, CO2 reduces transpiration per unit leaf area while enhancing photosynthesis.
Thus it often improves water-use efficiency (the ratio of crop biomass accumulation or yield and the amount
of water used in evapotranspiration). Experimental effects of CO2 on crops have been reviewed by Acock and
Allen (1985) and Cure (1985). In a compilation of greenhouse and other experimental studies, Kimball (1983)
estimated a mean crop yield increase of 33 +/- 6% for a doubling of CO2 concentration from 300 to 600 ppm
for a range of important agricultural crops.
In order to project the impact of increasing CO2 on agricultural production, these beneficial direct
effects should be considered along with the climatic effects of the radiatively active trace gases. The assessment
of the relative contributions of the direct effects of CO2 and the predicted climate changes to agricultural crop
responses remains a crucial research question. GCM predictions show that climate warms in virtually all
regions, while hydrologic regimes may become either wetter or drier. Thus, the climatic effects on crop yields
may thus be either negative or positive depending on location; this study tested whether the beneficial direct
effects (as estimated from experimental studies) may compensate for negative climate change impacts in a
variety of crop-growing environments around the world.
The IBSNAT models have been modified to simulate the changes in photosynthesis and
evapotranspiration caused by higher levels of CO2. These modifications (based on methods derived from Peart
et al. (1989)) were used in the crop yield/climate change scenario modeling to study the relative magnitudes
of the direct physiological and the climatic effects of increased CO2. Ratios were calculated between measured
daily photosynthesis and evapotranspiration rates for a canopy exposed to a range of high CO2 values, based
on published experimental results (Allen et al. 1987; Cure and Acock 1986, and Kimball 1983). Instantaneous
midday values were then modified to give daily integrated increases, allowing for lower light intensities in
morning and evening. In the crop models, the photosynthesis ratios (Table 3) were applied to the maximum
amount of daily carbohydrate production which is based on incoming solar radiation.
To account for the effect of elevated carbon dioxide on stomatal closure and increased leaf area index,
and hence on potential transpiration, the evapotranspiration formulation of the IBSNAT models was changed
to include a ratio of transpiration under elevated CO2 conditions to that under ambient conditions. To derive
the ratio, Peart et al. (1989) applied the Penman-Monteith equation (as written in France and Thornley 1984)
to the same canopy and environment, except for differing CO2 concentrations. The leaf resistances were
calculated as a function of the differing CO2 concentrations using equations developed by Rogers et al. (1983)
based on experimental data for maize and soybean (used for all C3 crops) (Table 3). The ratio procedure
results in a lower transpiration rate for higher CO2 levels on a daily per unit leaf area basis. Seasonal
evapotranspiration, however, may not change proportionately, and may even increase, because of the greater
leaf area grown under elevated CO2 conditions.
The simulation of direct CO2 effects for soybeans, wheat, and maize under current climate conditions
have been compared to experimental results (Peart et al. 1989; Jones and Allen, pers. comm.; Rosenzweig
1990). The wheat and soybean results compare well with experimental results, but maize simulations tended
to overestimate yield increases due to high CO2 at sites with low annual precipitation.
Rates of future emissions of trace gases, as well as when the full magnitude of their effects will be
realized, are unknown. For this study, CO2 concentrations are estimated to be 555 ppm in 2060 (based on
Hansen et al. 1988). Because other greenhouse gases besides CO2 (e.g., methane (CH4), nitrous oxide (N2O),
and the chlorofluorocarbons (CFCs)) are also increasing, an "effective CO2 doubling" has been defined as the
combined radiative forcing of all greenhouse gases having the same forcing as doubled CO2 (usually defined
as 600 ppm). The effective CO2 doubling will occur around the year 2030, if current emission trends continue.
INTRO-7
-------
The climate change caused by an effective doubling of CO2 may be delayed by 30 to 40 years or longer, hence
the projections for 2060 in this study.
Limitations
The IBSNAT models contain many simple, empirically-derived relationships that do not completely
mimic actual plant processes. These relationships may or may not hold under differing climatic conditions,
particularly the higher temperatures predicted for global warming. For example, 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 35 or even 40 °C during the growing period. Other
simplifications of the crop models are that weeds, diseases, and insect pests are controlled; there are no
problem soil conditions such as high salinity or acidity; and there are no catastrophic weather events such as
heavy storms. The crop models simulate the current range of agricultural technologies available around the
world; they do not include potential improvements in such technology, but may be used to test the effects of
some potential improvements, such as improved varieties and irrigation schedules.
CALIBRATION AND VALIDATION
Individual investigators calibrated and validated the IBSNAT crop models using local experimental
data, where possible. Where such procedures were not possible, previous calibrations of cultivars based on the
IBSNAT minimum dataset methods were relied on, as well as previous validations. The IBSNAT crop models
have been created with the express purpose of broad applicability across a wide range of environments.
The validation procedures and experimental data used for the 18 country studies are described in the
following chapters. In general, genetic coefficients for different crop genotypes were estimated from data
gathered in local agricultural experimental stations. Each parameter was calibrated directly from observed
results (phase duration, biometric ratio, and growth rates) in order to obtain rough parameter values. These
values were then used in model runs and adjusted in order to attain a pseudo-best fit of observed data. Overall,
validation showed acceptable results in the experiments conducted.
CROP MODELING PROCEDURES
The participating agricultural scientists carried out a set of crop modeling simulation experiments for
baseline climate, GCM doubled CO2 and transient climate change scenarios with and without the physiological
effects of CO2 and sensitivity tests. This involved the following tasks:
1. Define the geographical boundaries of the major production regions of the country, and estimate the
current production of major crops in those regions.
2. Provide observed climate data for representative sites within these regions, for the baseline period
(1951-1980), or for as many years of daily data as are available, and specify the soil, crop, and
. management inputs necessary to run the crop models at the selected sites.
3. Validate the crop models with experimental data from field trials.
4. Run the crop models with baseline data and climate change scenarios, with and without the direct
effects of CO2 on crop growth, with irrigated production, sensitivity tests, and adaptation responses
— for example, shifts in planting date and crop varieties. Report modeled yield changes and other
INTRO-8
-------
results, e.g., changes in crop growing season arising from climate change.
5. Identity and evaluate alterations in agricultural practices that would lessen any adverse consequences
of climate change.
The chapters that comprise the following sections of this volume describe the agricultural system that
was modeled, the methods and results of the crop modeling work, including adaptation responses, and the
implications of the projected climate changes, yield and water use changes, and adaptation strategies for
agriculture in the 18 individual countries.
ESTIMATION OF NATIONAL YIELD CHANGES
In order for the crop modeling results at individual sites to be used in the world food trade study, the
initial task was to scale up from site results to changes in national crop yields. This was done first for the crops
and countries in the crop modeling study. Then these results were extended to other crops based on agronomic
characteristics, and to other countries and regions based on similarities in agro-ecological zones, previous
climate change impact studies, and on comparison of GCM climate change scenarios for a full complement
of global estimates of potential impacts of climate change on crop yields.
Crop model results from over 100 sites in the 18 countries were aggregated by weighting current
regional production to produce national yield change estimates. This is an essential intermediate step that
permits the extrapolation of results from crop modeling experiments at individual sites to national yield
changes for the food trade model. The agricultural scientists in each country selected sites representative of
major agricultural regions, described the agricultural practices of the regions, and provided regional and
national production data for estimation of regional contributions to the national yield changes. All the crop
modeling aggregation results used to develop the BLS estimates were either calculated by the agricultural
scientists themselves or developed jointly with them.
A region is defined as an "area within a country where there are homogeneous agricultural practices,
soils, and climate." In the most complete national studies, enough sites were modeled to represent all the
major agroecological regions. In other cases, modelers were asked to analyze the sites modeled and the regions
in their country in order to extend the results as appropriately as possible. The regional yield estimates
represent the current mix of rainfed and irrigated acreage, the current crop varieties, nitrogen management
and soils, as provided by the country participants. In most cases only one crop variety and one soil were
modeled at each site.
A database was created for the current regional and national production for the 18 countries with crop
model results, primarily provided by the participants. Another source of production data was the FAO (1988).
For the USA, the data source was the USDA (Crop Production Statistical Division); for the former Soviet
Union the data source was the USDA International Service.
Results were aggregated similarly for the 11 countries where wheat was modeled (Table 4). There are
large differences among national results; for example, Brazil, Egypt, Pakistan, and Uruguay show significant
decreases in wheat yields even with the direct effects of CO2, while simulated wheat yields in Canada, and the
USSR primarily increase under the climate change scenarios with the direct effects of CO2. The other crops
were aggregated using the same methodology.
The crop yield estimates incorporate some major improvements: 1) consistent crop simulation
methodology and climate change scenarios; 2) weighting of model site results by contribution to regional and
national, and rainfed and irrigated production; and 3) quantitative foundation for estimation of physiological
CO2. effects on crop yields. Another set of estimates incorporating the effects of farmer adaptation to climate
change was also produced. All results forwarded to the world food trade model were in terms of percent
INTRO-9
-------
change from current yields, rather than absolute values. Analysis of relative changes in yields are more
appropriate given the many uncertainties involved in analysis of climate change impacts. It is important to
note, however, that percent change in yield depends on absolute value of base yields which are different for
the three crops, e.g., soybean base yields are low, so that a high percentage change does not represent a large
absolute decrease in yield.
DESIGN OF THE ADAPTATION STUDY
Farmers will react dynamically to changing environmental conditions. Country participants tested the
efficacy of several types of adaptations in crop simulation experiments. The adaptation strategies tested
involved changes in current management practices, e.g., planting date, fertilizer, and irrigation; and changes
in crop variety either to existing varieties or hypothetical new varieties. Several participants considered the
expansion of crops (e.g., winter wheat in the former Soviet Union and Canada and rice in China) to areas that
are temperature limited under the current conditions. The primary adaptation strategies tested are described
below.
Planting Date
The most likely response of farmers to warmer temperature would be to plant earlier to utilize the
cooler early season and to avoid high temperatures during the grain-filling period. All the crop model study
participants suggested earlier planting as a strategy for adapting to global warming. A relatively small change
in planting date, perhaps up to four weeks, should be easily supportable, but longer shifts in sowing date may
alter the soil moisture (it may be either too wet or too dry) and change the solar radiation (it may be lower)
at planting. In some country studies (Argentina, Philippines, and China) the sensitivity tests on planting date
suggested large changes in seasonal agricultural production, implying major changes in the agricultural systems.
In Argentina, planting date shifts of up to 4 months earlier or one month later were suggested by adaptation
simulations.
Irrigation
The crop model simulations demonstrated that climate change may bring significant increases in the
need for irrigation. In Petrolina, PE, Brazil, where soybeans are currently produced under rainfed conditions,
soybean response to the UKMO climate change was negative even with direct CO2 effects, but positive with
full irrigation. Climate change may increase demand for irrigation water for crops already under irrigation and
may encourage installation of new irrigation systems, if economic resources are available. Currently only about
15% of the world's agriculture is under irrigation. The potential problems associated with increased irrigation
as an adaptive strategy are the questionable availability of water resources, the associated costs, and the
environmental problems of soil salinization and water pollution.
Fertilizer
An increase in the amount of fertilizer applied can compensate in some cases for yield losses caused
by climate change. Studies in Uruguay and Mexico used the nitrogen module in the IBSNAT crop models to
test adaptation to climate change via fertilizer applications. In Uruguay, the barley response to nitrogen
fertilizer under baseline and UKMO conditions was compared. For the baseline runs the currently available
cultivar was used at the normal planting date. The same cultivar was used for the UKMO runs, but sown at
an earlier date with the physiological effects of increased CO2. The planting date was changed for the UKMO
INTRO-10
-------
runs, because the warmer temperatures of the scenario resulted in a shorter growing season. Four nitrogen
fertilizer rates were used, and the response curve was adjusted with linear regression analysis. Consequently,
the baseline maximum yield was more than 1 t ha"1 higher than the corresponding yield under UKMO
conditions. Also, the amount of N fertilizer needed to attain the maximum grain yield under UKMO was 2.6
times larger than the amount required to attain the same yield under current climatic conditions. These results
may have significant implications for future fertilizer use under climate change conditions at the high end of
the range of predicted warming and indicate an important area for future research.
Crop Variety
Since most regions are predicted to experience substantial increases in growing season temperature,
country participants tested substitution of existing varieties with higher thermal requirements for currently
grown varieties. In Mexico, use of existing cultivars produces slightly higher yields with the GISS climate
change scenario (with the direct effects of CO2), but does not overcome the negative climate change effects.
Some researchers also tested hypothetical new varieties in the crop model simulations, a technique useful for
establishing new breeding objectives. The crop models also allow testing of differing crop types, such as winter
and spring wheats. In the former USSR, a comparison of winter and spring wheat simulations indicates that
winter wheat will respond more favorably under the GCM climate change scenarios tested.
Cropping Area
With further analysis of crop-climate classification, estimation of changes in cropping systems may be
made. In China, climate change as projected by the three GCM scenarios would bring significant shifts in the
rice cropping pattern, based on the extension of the growing period and increased thermal regime during the
rice growing season (Gao et al. 1987). The regions where triple, double, and single-rice crops per year could
be grown would move northward and there would be increased sowing of "indica" rice now grown in southern
China, replacing the current "japonica" types.
SOURCES OF UNCERTAINTY
The primary uncertainties in the crop yield modeling depend on the assumptions embedded in the
IBSNAT crop models, as discussed earlier; the methods by which the IBSNAT models were used in the climate
change impact study; and the difficulty of estimating future technological improvements in agriculture.
Regarding the use of the IBSNAT models for the climate change study, there are several key points.
Climate data were taken from differing numbers of years and of differing quality in the various country studies,
and artificially generated daily solar radiation were used in the absence of observed data. Changes in climate
variability were not simulated.
Furthermore, the generalized soil characteristics do not encompass all the wide variety of global
agricultural soils. The resetting of the initial profile of soil water content, in most cases to full, at the
beginning of each cropping season leads to underestimation of the impacts of changes in the hydrological cycle
on crop production. Varying levels of nitrogen fertilization were not considered in most of the country studies,
exceptions being Argentina, Uruguay and Mexico. Consistent high levels of fertilization are especially
unrealistic in developing countries. Limited (often only one) cultivars were simulated at each site, although
in common practice several to many cultivars are planted in most cropping regions, which might respond
differently to climate change.
Finally, the crop models simulate the current range of available agricultural technologies. They do not
include potential improvements in such technology, although they may be used (as shown in this study) to test
INTRO-11
-------
the effects of some potential improvements, such as improved varieties and irrigation schedules.
RESULTS
Sensitivity Tests
While the arbitrary sensitivity tests are dissociated from the processes that influence climate, they
simulate a controlled experiment and provide better understanding of the factors affecting crop model
responses. They can also help to identify climatic thresholds of critical impacts. Climate sensitivity tests were
carried out in 13 countries for combinations of 0,2, and 4°C temperature increases coupled with precipitation
changes of 0%, +20%, and -20%. Changes were considered relative to the baseline yield at 330 ppm CO2.
Without the direct effects of CO2, crops averaged over all sites showed an increasingly negative
response to increased temperatures, with percent decreases in yields approximately doubling from the +2 to
+4°C cases. When direct CO2 effects are included, wheat, soybean, and rice yields increase about 15% with
a 2°C temperature rise, but turn negative at +4°C, indicating a possible threshold of compensation of direct
CO2 effects for temperature increases between 2 and 4°C as simulated in the IBSNAT crop models (Figure
3).
GCM Scenarios
Climate change scenarios without the direct physiological effects of CO2 caused decreases in simulated
crop yields in many cases, while the direct effects of CO2 mitigated the negative effects primarily in mid and
high latitudes (Table 4). Potential changes in national yields (averaged over all commodities in the BLS) varied
for the GISS, GFDL, and UKMO climate change scenarios with the physiological effects of CO2 (Figure 4).
However, latitudinal differences were apparent in all the scenarios; high latitude changes were less negative
or even positive in some cases, while lower latitude suffered more detrimental effects of climate change on
agricultural production.
The GISS and GFDL climate change scenarios produced a range of yield changes from +30 to -30%,
although there were regional differences. The GISS scenario is, in general, more detrimental to crop yields
in Asia and S. America, and GFDL is more harmful in North America (USA and Canada) and the former
USSR. The UKMO climate change scenario which has the greatest warming (5.2 °C global surface air
temperature increase) generally causes the largest yield declines (up to -50%).
The magnitudes of the estimated yield changes vary by crop. Maize production is most negatively
affected, probably due to its lower response to the physiological effects of CO2 on crop growth, while soybean
is least affected because it responds significantly to increased CO2, at least in the climate change scenarios with
lower estimated mean global surface air temperature warming. Simulated yield losses are caused
by a combination of factors, depending on location and nature of the climate change scenario. Primary causes
of detrimental impacts on yield are:
1. Shortening of the growing period (especially grain filling stage) of the crop. This occurred at some
sites in all countries.
2. Decrease of water availability caused by increased evapotranspiration and loss of soil moisture and
in some cases a decrease in precipitation in the climate change scenarios. This occurred in Argentina,
Brazil, Canada, France, Japan, Mexico, and USA.
3. Poor vernalization. Many temperate crops require a period of low temperature in winter to initiate
INTRO-12
-------
or accelerate the flowering process. Low vernalization results in low flower bud initiation and
ultimately reduced yields. This caused decreases in winter wheat yields in Canada and the former
USSR.
Simulated yield increases in the mid- and high-latitudes were caused primarily by:
1. The positive physiological effects of CO2. At sites with cooler initial temperature regimes, increased
photosynthesis more than compensated for the shortening of the growing period caused by warming.
2. The lengthened growing season and the amelioration of cold temperature effects on growth. At some
sites near the high latitude boundaries of current agricultural production, increased temperatures
extended the frost-free growing season and provided regimes more conductive to greater crop
productivity.
Transient Scenarios
When the crop models were run with transient climate changes projected for the 2010s, 2030s, and
2050s from the GISS transient run A, yield responses were non-linear over time (Figure 5). Aggregated
national wheat yields exhibited the widest range of effects and were the most non-linear of the three crops,
displayed differing trajectories of change in different regions of the world. Soybean yield changes in the U.S.
and Brazil were the most positive of the three crops, while maize yield changes tended to be the most negative.
Adaptation Studies
The adaptation studies conducted by the project participants suggest that ease of adaptation to climate
change is likely to vary with latitude (Figure 6). With the existing pool of cultivars and current resources of
water and fertilizer, agricultural adaptation to climate change seems likely in high and mid-latitude countries,
but seems out of reach for nations in the low latitudes. In tropical and semi-tropical regions, especially semi-
arid zones, the temperature changes suggested by the GCM scenarios tested are problematic. While soil
moisture deficits can easily be made up with simulated automatic irrigation, the economic and environmental
costs of establishing irrigation systems can be high.
CONCLUSIONS AND FUTURE RESEARCH NEEDS
Climate change induced by increasing greenhouse gases is likely to affect crop yields differently from
region to region across the globe. Under the climate change scenarios adopted in this study, the effects on crop
yields in mid- and high-latitude regions appeared to be less adverse than those in low-latitude regions.
However, the more favorable effects on yield in temperate regions depended to a large extent on full
realization of the potentially beneficial direct effects of CO2 on crop growth. Decreases in potential crop yields
are likely to be caused by shortening of the crop growing period, decrease in water availability due to higher
rates of evapotranspiration, and poor vernalization of temperate cereal crops. When adaptations at the farm
level were tested (e.g., change in planting date, switch of crop variety, changes in fertilizer application and
irrigation), compensation for the detrimental effects of climate change was found to be more successful in
developed countries.
Future research needs include determining how countries, particularly developing countries, can and
will respond to reduced yields. More detailed adaptation studies in many different locations will help to
address this need. In order to minimize possible adverse consequences to climate change worldwide, the
INTRO-13
-------
agricultural sector should be encouraged to continue to develop crop breeding and management programs for
heat and drought conditions (these will be immediately useful in improving productivity in marginal
environments today). Another important activity is to enlarge, maintain, and screen crop genetic resources at
established seedbanks. Resilience of the agricultural production sector also depends on improved use of
systems for monitoring weather, soil moisture, nutrient requirements, and pest infestations. Finally, strong
communication links among the agricultural research, production, and policy sectors are essential.
INTRO-14
-------
REFERENCES
Acock, B. and L.H. Allen Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds). Direct Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of
Energy. DOE/ER-0238. Washington, D.C. pp. 33-97.
Adams, R.M., C. Rosenzweig, R.M. Peart, J.T. Ritchie, B.A McCarl, J.D. Glyer, R.B. Curry, J.W. Jones, K.J.
Boote, and L.H. Allen, Jr. 1990. Global climate change and US agriculture. Nature 345(6272) :219-22.
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.
Callaway, J.N., FJ. Cronin, J.W. Currie, and J. Tawil. 1982. An Analysis of Methods and Models for Assessing
the Direct and Indirect Impacts ofCO2-induced Environmental Changes in the Agricultural Sector of U.S.
Economy. Pacific Northwest Laboratory, Battelle Memorial Institute. PNL-4384. Richland, WA.
Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure (eds). Direct
Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of Energy. DOE/ER-0238.
Washington, D.C. pp. 33-97.
Cure, J.D. and B. Acock. 1986. Crop responses to carbon dioxide doubling: A literature survey. Ag. and For.
Meteor. 38:127-145.
Fischer, G., K. Frohberg, M.A Keyzer, and K.S. Parikh. 1988. Linked National Models: A Tool for International
Food Policy Analysis. Kluwer. Dordrecht. 227 pp.
FAO. 1988.1987 Production Yearbook. Food and Agriculture Organization. United Nations. Statistics Series
No. 82. Rome. 351 pp.
France, J. and J.H.M. Thornley. 1984. Mathematical Models in Agriculture. Butterworths. Boston. 335 pp.
Gao, L., L. Lin, and Z. Jin. 1987. A classification for rice production in China. Ag. and For. Meteor. 39:55-65.
Godwin, D., J.T. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES-Wheat - v2.10. International
Fertilizer Development Center. Muscle Shoals.
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1993. A User's Guide to CERES-Rice. International
Fertilizer Development Center. Muscle Shoals.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient three-
dimensional global models for climate studies: Models I and II. Monthly Weather Review 111(4):609-
662.
Hansen, J., I. Fung, A. Lacis, 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 93(D8):9341-9364.
INTRO-15
-------
IPCC, 1990a. Climate Change: The IPCC Scientific Assessment. Houghton, J.T., G.J. Jenkins, and J.J. Ephraums
(eds). Intergovernmental Panel on Climate Change. Cambridge University Press. Cambridge. 365 pp.
IPCC, 19905. Climate Change: The IPCC Impacts Assessment. Tegart, W.J. McG., G.W. Sheldon, and D.C.
Griffiths (eds). Intergovernmental Panel on Climate Change. Australian Government Publishing
Service. Canberra.
IPCC, 1992. Climate Change 1992. The Supplementary Report to the IPCC Scientific Assessment. J.T.
Houghton, B.A. Callander, and S.K. Varney (eds). Intergovernmental Panel on Climate Change.
Cambridge University Press. Cambridge. 200 pp.
International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT). 1989. Decision Support System
for Agrotechnology Transfer Version 2.1 (DSSATvZl). Dept. of Agronomy and Soil Science. College
of Tropical Agriculture and Human Resources. University of Hawaii. Honolulu.
Jones, C.A. and J.R. Kiniry. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development. Texas
A&M Press. College Station. 194 pp.
Jones, J.W., K.J. Boote, G. Hoogenboom, S.S. Jagtap, and G.G. Wilkerson. 1989. SOYGRO v5.42: Soybean
Crop Growth Simulation Model. Users' Guide. Department of Agricultural Engineering and
Department of Agronomy. University of Florida. Gainesville.
Kane, S., J. Reilly, and J. Tobey. 1991. Climate Change: Economic Implications for World Agriculture. U.S.
Department of Agriculture. Economic Research Service. AER-No. 647. 21 pp.
Kimball, B.A. 1983. Carbon dioxide and agricultural yield. An assemblage and analysis of 430 prior
observations. Agronomy Journal 75:779-788.
Lamb, P.J. 1987. On the development of regional climatic scenarios for policy-oriented climatic impact
assessment. Bull Amer. Met. Soc. 68:1116-1123.
Leemans, R. and AM. Solomon. 1993. Modeling the potential change in yield and distribution of the earth's
crops under a warmed climate. Climate Research 3:79-96.
Manabe, S. and Wetherald, R.T. 1987. Large-scale changes in soil wetness induced by an increase in CO2.
Journal of Atmospheric Science, 44, 1211-1235.
Martin, R.J., M.J. Salinger, and W.M. Williams. 1990. Agricultural Industries in Climate Change: Impacts on
New Zealand. Ministry for the Environment. Wellington, pp. 165-173.
NDU. 1980. Crop Yields and Climatic Change for the Year 2000. Vol. 1. Fort Lesley and J. McNair (eds).
National Defense University. Washington, DC.
Nix, H.A. 1985. Agriculture. In R.W. Kates (ed). Climate Impact Assessment. SCOPE 27. John Wiley & Sons.
pp. 105-130.
Otter-Nacke, S., D.C. Godwin, and J.T. Ritchie. 1986. Testing and Validating the CERES-Wheat Model in
INTRO-16
-------
Diverse Environments. AgGRISTARS YM-15-00407. Johnson Space Center No. 20244. Houston.
Parry, M.L., T.R. Carter and N.T. Konijn (eds). 1988. The Impact of Climatic Variations on Agriculture. Vol 1.
Assessments in Cool Temperate and Cold Regions. Vol 2. Assessments in Semi-arid Regions. Kluwer.
Dordrecht. Netherlands. 876 pp. and 764 pp.
Pearman, G. 1988. Greenhouse: Planning for Climate Change. CSIRO. Canberra. 752 pp.
Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, and L.H. Allen, Jr. 1989. Impact of climate change on crop
yield in the southeastern U.S. A. In J.B. Smith and D.A. Tirpak (eds). The Potential Effects of Global
Climate Change on the United States. Report to Congress. U.S. Environmental Protection Agency.
EPA-230-05-89-050. Appendix C. Washington, D.C.
Ritchie, J.T. and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
model. In W.O. Willis (ed). ARS Wheat Yield Project. Department of Agriculture, Agricultural
Research Service. ARS-38. Washington, D.C. pp. 159-175.
Ritchie, J.T., U. Singh, D. Godwin, and L. Hunt. 1989. A User's Guide to CERES-Maize - v2.10. International
Fertilizer Development Center. Muscle Shoals.
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. Journal of Environmental Quality 12:569-574.
Rosenberg, N.J. and P.R. Crosson. 1991. Processes for Identifying Regional Influences of and Responses to
Increasing Atmospheric CO2 and Climate Change: the MINK Project. An Overview. Resources for the
Future. Dept. of Energy. DOE/RL/01830T-H5. Washington, D.C. 35 pp.
Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study. Prof.
Geog. 42:20-39.
Rosenzweig, C. and M.L. Parry. 1994. Potential impact of climate change on world food supply. Nature
367:133-138.
Smit, B. 1989. Climatic warming and Canada's comparative position in agricultural production and trade. In
Climate Change Digest. CCD 89-01. Environment Canada, pp. 1-9.
Smith, J.B. and D.A. Tirpak (eds). 1989. The Potential Effects of Global Climate Change on the United States.
Report to Congress. U.S. Environmental Protection Agency. EPA-230-05-89-050. Washington, D.C.
423 pp.
UK Dept of the Environment. 1991. The Potential Effects of Climate Change in the United Kingdom. United
Kingdom Climate Change Impacts Review Group. HMSO. London. 124 pp.
Warrick, R.A., R.M. Gifford, and M.L. Parry. 1986. CO2, climatic change and agriculture. Assessing the
response of food crops to the direct effects of increased CO2 and climatic change. In B. Bolin, B.R.
Dos, J. Jager and R.A. Warrick (eds). The Greenhouse Effect, Climatic Change, and Ecosystems.
SCOPE 29. John Wiley & Sons. New York. pp. 393-473.
INTRO-17
-------
Wigley, T.M.L. 1987. Climate Scenarios. Prepared for the European Workshop in Interrelated Bioclimate and
Land Use Changes. National Center for Atmospheric Research. NCAR 3142-86-3.
Wilson, C.A. and Mitchell, J.F.B. 1987. A doubled CO2 climate sensitivity experiment with a global climate
model including a simple ocean. Journal of Geophysical Research 92(13) :315-343.
World Food Institute. 1988. World Food Trade and U.S. Agriculture, 1960-1987. Iowa State University. Ames.
90pp.
INTRO-18
-------
Table 1.
GCM climate change scenarios.
GCM
Year*
Resolution**
Change in Average Global
Temperature Precipitation
GISS1
GFDL2
UKMO3
1982
1988
1986
7.83" x 10°
4.4° x 7.5°
5.0° x7.5°
4.2°C
4.0° C
5.2°C
11%
8%
15%
*When calculated.
**Latitude x longitude.
'Hansen, J. et al. 1983.
2Manabe, S. and R.T. Wetherald. 1987.
3Wilson, C.A. and J.F.B. Mitchell. 1987.
INTRO-19
-------
Table 2. Current world crop yield, area, production, and percent world production aggregated for
countries participating in study.
Crop
Yield
Area
Production
Study Countries
Wheat
Rice
Maize
Soybeans
tha'1
2.1
3.0
3.5
1.8
ha x 1000
230,839
143,603
127,393
51,357
txlOOO
481,811
431,585
449,364
91,887
%
73
48
71
76
Source: FAO, 1988.
INTRO-20
-------
Table 3. Photosynthetic ratios and stomatal resistances used to simulate direct physiological CO2
effects in the IBSNAT models (555 ppm CO2/330 ppm COJ.
Soybean
Wheat
Rice
Maize
Photosynthesis*
Ratio
1.21
1.17
1.17
1.06
Stomatal Res.**
s m"1
49.7/34.4
49.7/34.4
49.7/34.4
87.4/55.8
*Based on experimental work reviewed by Cure (1985).
**Based on experimental work by Rogers et al. (1983).
INTRO-21
-------
Table 4. Current production and changes in simulated wheat yields under GCM 2 x CO2 climate
change scenarios, with and without the direct effects of CO2l.
CURRENT
PRODUCTION
CHANGE IN SIMULATED YIELDS
Country
Australia
Brazil
Canada
China
Egypt
France
India
Japan
Pakistan
Uruguay
Former
USSR
winter
spring
USA
WORLD4
Yield
tha'1
1.38
1.31
1.88
2.53
3.79
5.93
1.74
3.25
1.73
2.15
2.46
1.14
2.72
2.09
Area
haxlOOO
11,546
2,788
11,365
29,092
572
4,636
22,876
237
7,478
91
18,988
36,647
26,595
231
Prod.
txlOOO
15,574
3,625
21,412
73,527
2,166
27,485
39,703
772
12,918
195
46,959
41,959
64,390
482
% GISS2 GFDL2 UKMO2 GISS3 GFDL3 UKMO3
Total % % % % % %
3.2
0.8
4.4
15.3
0.4
5.7
8.2
0.2
2.7
0.0
9.7
8.7
13.4
72.7
-18
-51
-12
-5
-36
-12
-32
-18
-57
-41
-3
-12
-21
-16
-16
-38
-10
-12
-28
-28
-38
-21
-29
-48
-17
-25
-23
-22
-14
-53
-38
-17
-54
-23
-56
-40
-73
-50
-22
-48
-33
-33
8
-33
27
16
-31
4
3
-1
-19
-23
29
21
-2
11
11
-17
27
8
-26
-15
-9
-5
31
-31
9
3
-2
4
9
-34
-7
0
-51
-9
-33
-27
-55
-35
0
-25
-14
-13
'Results for each country represent the site results weighted according to regional production. The world estimates
represent the country results weighted by national production.
*GCM 2xCO2 climate change scenario alone.
3GCM 2xCO2 climate change scenario with direct CO2 effects.
4World area and production x 1,000,000.
INTRO-22
-------
Crap Mod**
V«iMl.Rle».Miiz«, Soy
CropYMd
by Slta and Scenario
ET. Irrigation
Seaton Length
c
Aggregation of SRaRacuni
Agroecotogical Zone Analyila
Pravtou* Impact Studiee, QCM Output
J
YWdCtwng«EstbnatM
Commodfty Group and
Cou*y/R«glan
WortdFood
Trad* Moda)
ShfftsinTnria
Kldanc* of Food Pwarty ,
Figure 1. Key elements of crop yield and world food trade study.
-------
Figure 2. Crop model sites.
-------
% YIELD CHANGE
WITH DIRECT CO2 EFFECTS
T +2
T +4
Wheat
Rice
Soybean
Maize
Figure 3. Aggregated ffiSNAT crop model yield changes for +2°C and +4°C temperature increase.
Country results are weighted by contribution of national production to world production.
Direct effects of CO2 on crop growth and water use are taken into account.
-------
WITH DIRECT CO2 EFFECTS
GFDL 2XC02
WITH DIRECT COj EFFECTS
•<»*>
UKM02XC02
WITH DIRECT COj EFFECTS
-4510-30
-2910-15
-14 to 0
1to15
cn
18 to 30
Figure 4. Estimated change in average grain yield (wheat, rice, coarse grains, and protein feed) for the
GISS, GFDL, and UKMO climate change scenarios with direct CO2 effects.
-------
POTENTIAL CHANGE IN WHEAT YIELD
G1SS TRANSIENT A and 2XCO2
PERCENT
WITH DIRECT CO2 EFFECTS
-30
CANADA
=SU
CHINA
USTRALIA
FRANCE
NDIA
APAN
USA
'AKISTAN
JRUGUAY
EGYPT
RAZIL
BASE
2010s
2030s
2050s
2XCO2
Figure Sa Estimated change in average national yields of wheat for the GISS transient
run A and doubled CO2 climate change scenarios. Direct effects of CO2 on
crop growth and water use are taken into account
-------
POTENTIAL CHANGE IN MAIZE YIELD
GISS TRANSIENT A and 2XCO2
PERCENT
WITH DIRECT CO2 EFFECTS
-30
i JAPAN
RANGE
RGENTINA
RAZIL
HINA
MBABWE
SA
GYPT
EXICO
BASE
2010s
2030s
2050s
2XCO2
Figure 5b. Estimated change in average national yields of maize for the GISS transient
run A and doubled CO2 climate change scenarios. Direct effects of CO2 on
crop growth and water use are taken into account
-------
POTENTIAL CHANGE IN SOYBEAN YIELD
G1SS TRANSIENT A and 2XCO2
PERCENT
WITH DIRECT CO2 EFFECTS
JSA
RAZIL
2050s
2XCO2
Estimated change in average national yields of soybean for the GISS transient
run A and doubled CO2 climate change scenarios. Direct effects of CO2 of
crop growth and water use are taken into account
-------
UKUO 2XCO2
WITH DIRECT COj EFFECTS
UKMO 2xC02
UKMO 2XC02
WITH ADAPTATION 1
WITH ADAPTATION 2
-4510-30
-29 to -15
-14 to 0
1 to 15
I I %
18 to 30
Figure 6. Estimated change in average grain yield (wheat, rice, coarse grains, and protein feed) under
two levels of adaptation for the UKMO climate change with direct CO2 effects.
-------
SECTION 2: NORTH AMERICA
-------
-------
EFFECTS OF GLOBAL CLIMATE CHANGE ON WHEAT YIELDS
IN THE CANADIAN PRAIRIE
M. Brklacich
Department of Geography, Carleton University
Ontario, Canada
R. Stewart
Bureau for Environmental Sustainability
Agriculture Canada, Ontario, Canada
V. Kirkwood
Centre for Land & Biological Resources Research (CLBRR)
Agriculture Canada, Ontario, Canada
R. Muma
Horticulture & Special Crops Division
Agriculture Canada, Ontario, Canada
CANADA-1
-------
ACKNOWLEDGEMENTS
The authors wish to express their collective thanks to Cynthia Rosenzweig, the IBSNAT
team (especially Jim Jones, Upendra Singh, Bruce Curry, and Gerrit Hoogenboom), Anne Mane
Label, and Awegechew Teshome. Our appreciation is also extended to Julian Dumanski, Jim Dyer,
and three anonymous reviewers for their comments on an earlier version of this report.
CANADA-2
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Objectives
DATA, PROCEDURES, AND ASSUMPTIONS
Baseline Climate Data
Climatic Change Scenarios Derived from GCMs
Transient Scenarios
Sensitivity Analysis
Crop Model, Cultivars, Management Variables, and Soils
Simulation of the Physiological Effects of CO2
Performance of the CERES-Wheat Model in the Canadian Prairie
IMPLICATIONS OF GLOBAL WARMING AND CO2 INCREASES FOR SPRING WHEAT
Global Warming in Isolation
CO2 Increases and Global Warming
Sensitivity Analysis
The GISS Transient Scenarios
ADAPTIVE STRATEGIES: RESPONDING TO GLOBAL CLIMATE CHANGE
Irrigation
Earlier Seeding
Winter Wheat Conversion
CONCLUSIONS
REFERENCES
CANADA-3
-------
SUMMARY
Climate change, as projected by three General Circulation Models (GCMs), caused
simulated spring wheat yields to decrease at all sites selected for the study in the Canadian prairie.
Yield decreases were caused primarily by significant increases in temperature, which shortened the
growing season and resulted in less time for biomass accumulation. When the direct effects of CO2
were considered, the results varied with the scenario. Under the GISS scenario, wheat yields
increased at most sites, while under the more severe UKMO scenario (hotter and drier), wheat
yields declined at most sites.
Adaptation strategies designed to offset the negative effects of climate change were
successful in some areas. Higher temperatures allowed regions in southern Canada to plant winter
wheat instead of spring wheat, taking advantage of more favorable climatic conditions to produce
higher yields, but northern regions were still too cold to grow winter wheat. Irrigation was found
to be an effective method of adaptation at the driest sites.
INTRODUCTION
Background
Agriculture in Canada is highly diversified, an important influence on the national economy,
and a major contributor to international export markets. In the Atlantic provinces, livestock, dairy,
poultry, and potatoes are the major commodities. Central Canada benefits from many advantages,
including superior soils and climatic conditions, proximity to urban markets, and highly developed
processing and transportation sectors. The agriculture in this region is diversified and is a major
contributor to Canadian livestock, field crop, and vegetable production. In British Columbia,
agriculture is a relatively small industry, focusing on dairy, poultry, vegetable, and fruit production.
In the Canadian prairie provinces (Manitoba, Saskatchewan, and Alberta), the agri-food industry is
a major component of the economy (51% of the total Canadian cash farm receipts). In
Saskatchewan, agriculture accounts for about 16% of the provincial gross domestic product (the
highest percentage of any Canadian province). In the prairie provinces, spring wheat is the main
commodity (96% of the national production), and most of the crop is destined for export markets.
Canada is the third largest wheat exporter in the world.
Extensive dryland agriculture is the dominant characteristic of farming systems in the prairie
region. Annual production for the 1981-85 period averaged more than 21 million t, and more than
11 million ha are used for spring wheat production each year. Spring wheat is important in all three
prairie provinces, but approximately 55% of total production is from Saskatchewan.
The potential impacts of a greenhouse-induced global climate warming on agroclimatic
resources, crop yield, and regional production potential have been previously reported in regional
studies (Arthur 1988; Bootsma et al. 1984; Brklacich and Smit 1991; Singh and Stewart 1991; Smit
et al. 1989; Stewart 1990; Williams et al. 1988; Bootsma and de Jong 1988b). Current agroclimatic
conditions for many regions of Canada are characterized by relatively short periods without frost,
and therefore a majority of the Canadian research on the agricultural impacts of global warming has
focused on possible alterations in the growing season properties for annual crops with concomitant
adjustments to productivity levels. As knowledge of climatic processes, climatic change, and climate-
agriculture relationships improve, these assessments need to be refined. In addition, a study that
goes beyond regional impact studies, considers the agri-food sector in the broader context of
CANADA-4
-------
national and international economies, and analyzes the extent to which climatic change might alter
comparative advantages among various regions in Canada and its agricultural trading partners is
essential.
Objectives
This study is the Canadian contribution to a project designed to assess the potential
implications of global warming on major world agricultural commodities and their international
trade. It was coordinated by the U.S. Environmental Protection Agency and the U.S. Agency for
International Development. Contributing countries were requested to appraise the impacts of a
range of scenarios for global warming on production opportunities for a major export or import
crop. This study focussed on spring wheat production in the Canadian prairie provinces. The primary
analytical tool used in this study was the CERES-Wheat model (Godwin et al. 1989). Spring wheat
was selected because it has regional and national significance to the Canadian economy and because
Canada is one of the top three wheat exporters in the world. Consequently, changes in climatic (or
other) conditions which would alter Canada's potential for wheat production or the country's
comparative advantage, could have repercussions on national and international levels.
Wheat production occurs over a vast area in the prairie region which transverses several soil
and climatic zones. The projected global climatic change scenarios suggest that long-term climatic
alterations may not be distributed uniformly across the prairie region. In order to capture some of
the spatial variability in the region's biophysical properties, seven indicator sites, which include all
major biophysical zones in the prairie provinces, are considered in this study (Figure 1 and Table
1). The extent to which these indicator sites represent the region as a whole is not known, and
therefore this report does not attempt to generalize the findings across the region.
DATA, PROCEDURES, AND ASSUMPTIONS
Baseline Climate Data
Daily maximum and minimum temperatures and daily total precipitation data (from 1951-
80) for the seven sites were obtained from the Canadian national weather archive maintained by the
Atmospheric Environment Service (AES), Environment Canada. Daily solar radiation data were not
available for the full baseline period at all sites and therefore were estimated using the procedure
described in Doorenbos and Pruitt (1977). The method uses daily observations of bright sunshine
hours (obtained from the national archive), estimates of solar radiation at the top of the
atmosphere, and daylength (according to Russelo et al 1974). The observed and estimated solar
radiation values obtained by this method are comparable.
Climate Scenarios Derived from GCMs
Climate change scenarios for each site were generated from three equilibrium General
Circulation Models: the Goddard Institute for Space Studies Model (GISS), (Hansen et al 1983),
the Geophysical Fluid Dynamics Laboratory Model (GFDL), (Manabe and Wetherald 1987), and
the United Kingdom Meteorological Office Model (UKMO), (Wilson and Mitchell 1987). The
method for creating scenarios for each site involved the following procedures: (a) the GCM grid
point located closest to each indicator site was identified; (b) for each of the identified grid points,
the changes in mean monthly temperature were calculated as the difference between the 2XCO2 and
CANADA-5
-------
the control (IXCOj) GCM runs, and for mean monthly precipitation and solar radiation, the change
was calculated as the ratio of the 2XCO2 run to the control run; (c) changes in the monthly mean
values derived under step (b) were then applied to the observed daily baseline record.
Table 2 shows the temperature and precipitation changes used in this study. Temperature
increases considerably under the three GCMs at all sites, with the largest increases occurring in the
winter. The GISS and GFDL scenarios present similar temperature changes. In the southern
Canadian prairies, temperature increases are 3°C-6.5°C, and the largest increases correspond to the
northern site in Alberta (Fort Vermillion). Summer temperature increases are 2°C-3°C lower than
winter increases. The UKMO scenario is characterized by the largest temperature increases (5°C-
10°C). Winter increases are 3°C-5°C higher than summer increases and are largest in the Alberta
and Manitoba sites.
The three GCMs estimate precipitation increases for most seasons, but the magnitude of
the increases varies considerably among GCMs and sites. For the critical growing period of spring
wheat (May to August), the GISS model predicts precipitation increases up to 25% above the
current average; winter precipitation also increases. The direction of the precipitation changes
predicted by the UKMO model are similar to those projected by the GISS model, but the increases
in the winter are larger than those in the summer. The GFDL model predicts smaller precipitation
increases than the other two GCMs, and no trend was discernable for the Canadian prairie sites.
Late winter, spring, and summer precipitation changes for the sites in Manitoba tracked 50%-70%
above current averages. For the sites in Saskatchewan, winter and spring precipitation increases up
to 50%, but there are no significant changes in summer precipitation.
Solar radiation changes under the 2XCO2 scenarios (not shown) are not very large in
comparison to the current levels.
Transient Scenarios
This study also considered a set of transient climate scenarios (as opposed to the
atmospheric equilibrium scenarios), derived from the GISS transient climatic simulations (Hansen
et al. 1988) for the 2010s, 2030s, and 2050s, and assuming CO2 concentrations of 405, 460, and 530
ppm, respectively. Transient scenarios for each site were developed by using the same procedure as
that used for the equilibrium scenarios.
In general, each step in the transient scenario implies a further temperature increase, with
winter increases higher than the summer increases. Except for Fort Vermillion, the largest estimated
increase in temperature occurs between the 2030s and the 2050s. For the sites in Saskatchewan and
Alberta, precipitation changes mostly during the 2010s. Solar radiation changes little under the
transient scenarios, with only slight decreases during the winter.
Sensitivity Analysis
An alternative approach for incorporating changes to the current climate is to specify
incremental adjustments to selected climatic variables and to apply these changes uniformly to the
daily observed weather record. While this approach is removed from the processes that influence
climate, it has the advantages of simulating a controlled experiment and thereby providing a better
understanding of the factors affecting responses. In addition, the approach can also identify climatic
thresholds that could ultimately imply substantial impacts for agriculture. Combinations of
temperature increases of 0°C, +2°C, and +4°C and precipitation changes of 0%, -20%, and +20%
were considered in this study.
CANADA-6
-------
Crop Model, Cultivars, Management Variables, and Soils
The crop model used in this simulation study is the CERES-Wheat model (Godwin et al.
1989). The model simulates crop responses to the major factors driving plant growth and
development. Simulated processes include soil moisture balance, phonological development, yield
and biomass production.
Cultivars. The area sown with different spring wheat varieties has changed over the past fifty
years. In 1988, the most important varieties were Katepwa (49% of the total area of spring wheat
in the prairie), Neepawa (21%), and Columbus (18%) (Prairie Pools, Inc. 1988). In the first half of
the 1980s, Neepawa and Benito were the most popular varieties. The differences among varieties are
usually related to resistance to pests and diseases, resistance to shattering, and susceptibility to root
rot. For example, the major difference between Katepwa and Neepawa is that the Katepwa "has
better stem and leaf rust resistance and is easier to thresh" (Saskatchewan Agriculture and Food
1990).
These important factors, which ultimately affect crop yield, are not considered by the
CERES-Wheat model. Therefore, given that a few varieties account for the majority of the area
sown with spring wheat and that many of the differences among varieties are beyond the scope of
the CERES-Wheat model, a single variety can adequately represent a substantial proportion of the
varieties currently used in the Canadian prairie. We selected the Manitou variety for the simulation
study, with the genetic coefficients derived by Godwin et al. (1989). Manitou was an important
variety used in the 1960s and 1970s, and many of the varieties used in the 1980s are derived from
it. The major differences between Manitou and other varieties used in the 1980s are confined to
characteristics not considered in the CERES-Wheat model, and therefore, the genetic coefficients
associated with the Manitou are still applicable (Morrison, personal communication).
The genetic coefficients for winter wheat used in the adaptation section of this study were
the representative winter wheat genetic coefficients for varieties grown in the northern plains of the
United States (Godwin et al. 1989).
Management variables. The seeding dates for spring wheat were derived from Bootsma and
de Jong (1988a) (Table 3). These seeding dates were estimated using the observed weather record
(1951-80) and represent the average dates for this period. Yearly variability in the weather
conditions result in a considerable range of seeding dates. For example, the earliest and the latest
seeding dates estimated for Lethbridge between 1951 and 1980 were April 24 and May 23,
respectively. At Winnipeg the estimated seeding dates ranged from April 24 to June 9. The
importance of an early or late spring on seeding date, crop growth, and grain yield is beyond the
scope of this study.
For winter wheat (considered in the Adaptation section of this study), seeding dates were
derived from provincial field crop production guides (Manitoba Agriculture 1988; McLelland 1985a;
Saskatchewan 1981).
Field crop production guides and agronomists determined the planting densities for the
spring and winter wheats (Manitoba Agriculture 1988; McLelland 1985a, 1985b; Saskatchewan
Agricultural Services Coordinating Committee 1981). The midpoint in the seeding rate was selected,
and seeding rates in kg ha"1 were converted to plants m"2 using an average seed weight of 30 mg (Fei
and Ripley 1985).
The adaptation study considers the effects of irrigation on wheat yields. In this study,
irrigation was triggered when the soil moisture estimated for the 1.2-meter rooting zone dropped
below 50% of the moisture-holding capacity. It was assumed that the amount of water required to
CANADA-7
-------
return the rooting zone to field capacity was applied at that time, with an irrigation efficiency of
100%.
Soils. A representative soil for each site was identified in consultation with a prairie-soil
expert (Shields, personal communication). For all sites except for Fort Vermillion, data on horizon
depths, texture, bulk density, organic carbon, coarse fractions, pH, and soil classification were
extracted from the Canada Soil Information System (CanSIS) (Canada-Alberta 1989; Canada-
Manitoba 1989; Canada-Saskatchewan 1989). Permeability, drainage, slope, quantity of roots, and
soil color were estimated by Shields (personal communication). Soil surveys for the Fort Vermillion
area are incomplete, and data on soils for this area are not available from CanSIS. The Donnelly
series has been characterized for the neighboring Grinshaw and Notikewin areas (Scheelar and
Odynsky 1968), and these data were used in this study.
The CERES-Wheat model estimates the soil moisture storage capacity from the input soil
profile. This estimate represents the maximum amount of soil moisture that could be available to
the crop at seeding. The version of the CERES-Wheat model used in this study does not estimate
soil moisture recharge between harvest and seeding, but it does allow users to specify soil moisture
conditions at seeding as a percentage of the estimated storage capacity. Since soil moisture reserves
of prairie soils at seeding are typically well below the moisture-holding capacity, the initial soil water
conditions were adjusted. The estimates of soil-moisture conditions at seeding presented in Table
3 were derived from de Jong and Bootsma (1987). These estimates are long-term, average values for
the entire soil profile and were calculated using the observed weather record for 1951-80, assuming
a continuous wheat-farming system.
Simulation of the Physiological Effects of CO2
Higher levels of atmospheric CO2 have been shown to increase photosynthesis and water-use
efficiency, resulting in yield increases in experimental settings (Acock and Allen, 1985). Because the
climate change scenarios are associated with levels of CO2 that are higher than the current climate
GCM simulations (330 ppm), the physiological effects of alternative CO2 levels on crops were
included in the crop model simulations.
Other gases such as CH4, N2O, and CFCs are expected to contribute to the greenhouse
effect, and therefore, the equivalent of a 2xCO2 atmosphere would be reached before the actual
doubling of CO2. To account for this effect, an atmospheric CO2 concentration of 555 ppm was used
to represent the equivalent of a 2xCO2 atmosphere.
Performance of the CERES-Wheat Model in the Canadian Prairie
Previous studies. The CERES-Wheat model has been used previously to estimate wheat
yields in the Canadian prairies. Fei and Ripley (1985) used the model to estimate wheat yields from
1964-84 for the Saskatoon crop-reporting district (more than 2 million ha of cropland). After
accounting for technological advances, their main conclusions were: (a) for the entire period, the
CERES model overestimated observed yields by 24%; and (b) the model tended to overestimate
yields in the good years and underestimate yields in the poor years.
Fei and Ripley suggested that the following factors could have contributed to the observed
and simulated yield discrepancies: (a) the weather and soil data used as input for the CERES-Wheat
model did not accurately represent the range of weather and soil conditions occurring in the area;
(b) the model does not account for yield losses due to pest and disease damage; and (c) the model
underestimates root growth, and therefore, moisture-deficit stress was exaggerated in dry years, as
CANADA-8
-------
there was insufficient root growth to exploit water reserves from the deeper layers. In wet years, the
model would produce excessive above-ground growth, greater leaf area development, and higher
yields. In addition, there are the usual concerns regarding the reliability of the reported yields. The
estimates are based on samples of farmers' estimates of yields, and the extent to which these
estimates are influenced by unreported crop failures and/or portions of fields not used for crop
production is not known.
Current study. For this study, observed and simulated yields and season lengths were
compared (Table 4). These parameters were closely related at all sites, although the modeled season
lengths tended to be longer, with the greatest discrepancies occurring at Prince Albert and
Letherbridge. The simulated grain yields at Winnipeg, Dauphin, Swift Current, and Prince Albert
were comparable to previous estimates (Fei and Ripley 1985). On average, the model tended to
overestimate yield by about 25%. The model was considerably less reliable for the three Alberta
sites, especially for the central and northern Alberta sites (Ellerslie and Fort Vermillion), suggesting
that the version of the model used in this study has not adequately captured the influence of longer
days and lower light intensities associated with these northern latitudes.
For all the sites, the CERES-Wheat model simulated season length reasonably well. For
Winnipeg, Dauphin, Swift Current, Prince Albert, and Letherbridge, the model provides reasonable
estimates of long-term grain yields. However, for Ellerslie and Fort Vermillion, the yield-predicting
capability of the CERES-Wheat model needs to be refined.
These performance characteristics suggest that results from the Canadian analysis should
be applied in similar fashion to the results obtained from GCMs, i.e., as percent differences between
baseline and scenario runs. Relative changes from baseline simulations rather than absolute values
from crop model runs should be used. Impacts on yield in a particular region thus can be estimated
and compared to results from other scenarios and regions. The value of the CERES-Wheat model
is that it provides estimates of the direction and magnitude of yield change among different sets of
conditions. It is not intended to provide an exact measure of yield nor absolute predictions of
climate change impacts.
IMPLICATIONS OF GLOBAL WARMING AND CO2 INCREASES FOR SPRING
WHEAT
Global Warming in Isolation
The 2xCO2 scenarios for climate change provide a basis for comparative static assessments
of the implications of an altered climate on crop yields. This section evaluates the effect of three
GCM global climate change scenarios on agro-climatic conditions and wheat yields at seven sites
throughout the prairies.
Maturation Time. The large temperature increases associated with the GISS, GFDL, and
UKMO scenarios imply a decrease in the time required for spring wheat to mature (Figure 2). For
the indicator sites in the southern portion of the Canadian prairies, the GISS and GFDL scenarios
simulated a growing period that is 11 to 14 days shorter than the current growing period. The
impacts were more pronounced for the northern sites, with the GISS and GFDL scenarios
shortening the growing period by approximately 3 and 4 weeks, respectively. The UKMO scenario
showed the largest decreases in the growing period due to its higher temperature predictions. A
three-week reduction in the amount of time required for spring wheat to mature was typical in the
southern prairies under this scenario, with a 4-5-week reduction estimated for the northern parts
of the Canadian prairie.
CANADA-9
-------
Crop Moisture Stress. Each of the climate change scenarios had a different impact on crop
moisture stress (Table 5). (Crop moisture stress is the difference between precipitation and
simulated evapotranspiration during the crop growing period.) Under the current conditions, deficits
in crop moisture characterize wheat production throughout the Canadian prairies. The GISS
scenario had little impact on the magnitude of the moisture deficit accumulated over the maturation
period of the crop. However, these accumulated deficits, coupled with considerable declines in the
time required for wheat to mature, resulted in an increase in average deficits per day. Under the
GFDL and UKMO scenarios, precipitation increased in the eastern sites and decreased in the
western sites. These altered precipitation patterns eliminated moisture deficits for the sites in the
eastern prairie, while the deficits became somewhat more severe in the western sites. Under the
UKMO scenario, precipitation increases provided some relief to moisture stress, but the shortened
maturation period of the crop offset this potential benefit, and an increase in the average daily crop
moisture deficit was estimated.
Yields. The overwhelming trend of the impacts of all three climate change scenarios on
wheat yields was negative, but the magnitude of the impacts varied with scenario and site (Figure
3). The simulated shortening of the season length caused a decrease in the time available for the
grain filling, and thereby contributed to a decline in crop yields. The GISS and UKMO scenarios
had considerably smaller impacts on the driest part of the prairie (Swift Current and Lethbridge),
with larger declines in crop yields elsewhere. The yield impacts under the GISS scenario in the dry
areas were negligible, but in the rest of the regions yield decreased about 20%-30%. Under the
UKMO scenario yield decreases showed similar trends, but the substantially shorter simulated season
length had a larger effect on yields. Yields decreased about 20% in the dry areas and 45%-60%
elsewhere.
In the eastern sites of Winnipeg and Swift Current, there were modest yield increases under
the GFDL scenario, probably due to the scenario precipitation increases. In the western sites,
however, precipitation decreases, coupled with higher temperatures, created a less favorable regime
for wheat and led to substantial yield declines.
CO2 Increases and Global Warming
This section evaluates the combined effects of global climate change with the beneficial
effects of increased CO2 on crop yield-improved water-use efficiency and increased net
photosynthesis (Figure 3).
Under the GISS scenario, the direct effects of CO2 compensated for the yield decreases
under the scenario of climate change alone at all sites. The most noticeable changes were at the
driest sites (Swift Current and Lethbridge), where the benefits of increased water-use efficiency were
larger, and the direct effects of CO2 caused simulated yields to increase by about 40%-50% above
the current level. Yield increases were about 15% at the other sites.
The GFDL scenario projected precipitation increases for the eastern sites. The combination
of a more favorable moisture regime and additional CO2 caused simulated wheat yields to increase
above the current level. The increase was more pronounced in areas that are currently very dry.
However, at the sites where the GFDL scenario projects precipitation decreases (the western sites),
simulated wheat yields decreased, even with the direct effects of CO2.
Under the UKMO scenario, yields increased at the driest sites and decreased elsewhere.
Once again, the benefits of increased CO2 on enhanced water-use efficiency were more noticeable
in the driest areas. At the rest of the sites, the benefits of elevated CO2 levels only partially offset
CANADA-10
-------
the negative impacts of the UKMO scenario conditions on simulated yields, and yields decreased
25%-40%.
Sensitivity Analysis
Maturation Time. A 2°C temperature increase caused the simulated maturation period of
the crop to decrease 7-12 days in the southern sites, and 2-3 weeks in the northern sites (Figure 4).
A 4°C increase reduced season length by 2-3 weeks in the southern sites and by more than 30 days
in the northern sites.
Crop Moisture Stress. Increased evapotranspiration rates caused by the 2°C and 4°C
temperature increases added to crop moisture stress at all sites (Table 5). These higher
temperatures, coupled with a 20% reduction in precipitation, increased the severity of moisture
deficits accumulated over the crop-growing period by up to 41%. On the other hand, a 20%
precipitation increase more than offset the negative impacts of a 2°C temperature increase, but
seasonal moisture deficits persisted. Increases in evapotranspiration due to a 4°C temperature
increase tended to offset a 20% increase in precipitation.
Yields. A temperature increase alone caused decreases in simulated yields at all sites (Table
6). At Swift Current and Lethbridge, relatively severe moisture deficits currently cause wheat yields
to be low. As a result, the impacts of a shorter time for crop growth and drier conditions were not
as pronounced at these sites. Of course, the opposite trend is anticipated if the moisture regime
becomes more favorable for wheat production. The yield benefits from a 20% precipitation increase
were greater at the driest sites.
For all sites except those in the driest areas, a 2°C temperature increase offset the benefits
of a 20% precipitation increase, and therefore, simulated yields decreased. A 2°C temperature
increase tempered the yield benefits of additional moisture in the driest areas, and simulated yields
increased about 20% under these conditions.
A 20% precipitation decrease caused substantial declines in wheat yields at all sites. Prairie
agriculture currently suffers from a deficit of moisture and the simulated yield losses due to the
additional stress associated with a 20% reduction in precipitation ranged from 22%-39%. Higher
temperatures exacerbated the consequences of reductions in precipitation (-40% and -60% if
precipitation declines are coupled with a +2°C and a +4°C, respectively).
The direct beneficial CO2 effects offset the yield decreases in some of the scenarios
considered in this sensitivity analysis (Table 6). In the least favorable of the scenarios considered (a
+4°C increase combined with a 20% precipitation decrease), yield decreases were estimated at all
sites, even with the direct effects of CO2. In contrast, under the +2°C and +20% precipitation
scenario, wheat yields increased substantially at all sites.
The GISS Transient Scenarios
Simulated wheat yields under the transient scenarios (Figure 5) were nonlinear. For all sites
except Fort Vermillion, there was an initial increase in yields through the 2010s, followed by a
decrease from the 2010s to the 2030s, and then a recovery of the yields from the 2030s to the 2050s.
This suggests that under the GISS transient scenarios, the temperature increase until the 2010s
would have a relatively small impact on yield, and the beneficial effects of CO2 (405 ppm) would
stimulate grain productivity compared to base levels. However, the influence of temperature
increases with concomitant reductions in the maturation period of the crop dominated the impact
by the 2030s, and in some cases outweighed the benefits of the CO2 increase to 460 ppm.
CANADA-11
-------
Temperature increases between the 2030s and the 2050s were small, and therefore, the yield
increases associated with this period were a consequence of the increased beneficial effects of rising
CO2 levels (530 ppm in the 2050s).
ADAPTIVE STRATEGIES: RESPONDING TO GLOBAL CLIMATE CHANGE
The analysis in the previous section has isolated the impacts of climate change on wheat
yields, but it is reasonable to assume that many other biophysical and socioeconomic conditions will
also change during this time period, and that these adjustments will act in concert with the climate
change. In this section three possible responses to climate change are considered: irrigation, changes
in the planting date, -and a shift from spring to winter wheat production. These adjustments
represent different levels of economic adaptation, with irrigation being the most expensive. The
results presented in this section include the beneficial CO2 effects on simulated wheat yields. An
appraisal of economic and technical feasibility of each option is beyond the scope of this study, as
are the changes in other conditions, including the technological advances or adaptive measures taken
by farmers or public institutions.
Irrigation
Under all three climate change scenarios and at all sites, irrigation was an effective adaptive
strategy to improve wheat yields under the climate change scenarios. The sites that benefited the
most from irrigation are the driest sites: Swift Current and Lethbridge.
Earlier Seeding
Sowing spring wheat earlier in the season would take advantage of cooler temperatures
during the earlier part of the frost-free period and therefore lessen the negative impacts of climate
change on yield. This option was most effective at the driest sites. Under the UKMO scenario, which
predicts the largest yield decreases, earlier seeding compensated for the yield decreases only in the
driest areas. Under the GISS and GFDL scenarios, earlier seeding compensated for yield losses at
all sites except Fort Vermillion and Ellerslie. Overall, the earlier seeding option is the easiest to
implement, but it was the least effective of the three options considered, and in some cases it did
not fully offset the negative impacts of climate change on wheat yields.
Winter Wheat Conversion
Winter survival is the main, factor that currently deters winter wheat production in the
Canadian prairies. Global warming (especially if there is a substantial increase in winter temperature
and sufficient snow cover to ensure winter survival) would allow cereal grains to take better
advantage of spring moisture reserves and thereby contribute to a more favorable set of conditions
for winter cereals. A shift from spring to winter wheat increased yields under GCM climate change
scenarios at the southern sites. Winter wheat would allow a better use of spring moisture supplies
and would diminish the impacts of relatively high summer temperatures. For the northern locations
(Fort Vermillion, Ellerslie, and Prince Albert), the temperature increases under the GCM scenarios
during the winter were not enough to eliminate cold damage due to winter kill and therefore limited
the effectiveness of a conversion to winter wheat at these sites.
CANADA-12
-------
CONCLUSIONS
This study considered the extent to which global climate change, increases in atmospheric
CO2 concentrations to 555 ppm, and selected adaptive strategies would alter simulated wheat yields
at seven sites in the Canadian prairie. The climate scenarios were derived from three General
Circulation Models (GCMs); additional climate scenarios were created by altering the observed
temperature and precipitation records by fixed amounts. An assessment of the crop model used
(CERES-Wheat) under the Canadian prairie conditions indicated that the model can provide an
estimate of the direction and magnitude of yield shifts stemming from a potential climatic change.
However, the model is not intended to predict reliable yield estimates for a particular year.
The conclusions of the simulation study were:
1. Each of the three GCM scenarios used (GISS, GFDL, and UKMO) imply a different set
of agro-climatic conditions for the region.
2. There is no uniform response to climate change throughout the Canadian prairie. Each
climate change scenario results in different wheat yield response at each site.
3. Without the beneficial effects of increased CO2, wheat yields decreased under the climate
change scenarios in comparison with current yields. The shortening of the time required for
wheat to mature under the climate change scenarios was the main factor responsible for the
yield decreases.
4. When the direct effects of CO2 are included in the simulation, wheat yields increased
compared to current yields at all sites under the GISS scenario; yields increased only in the
northern sites under the GFDL scenario; and yields decreased at all sites except two under
the UKMO scenario.
5. The beneficial effects of increased CO2 have the greatest effect on the driest sites in the
prairie under the climate change scenarios.
6. Although the overall impact of global climate change on Canadian wheat yields may not be
negative, it appears that a considerable redistribution in yield patterns is possible.
7. The effectiveness of possible adaptation strategies varies from region to region, and
therefore it is reasonable to expect that no single response strategy will adequately mitigate
the possible negative impacts of climate change on all of the Canadian prairie provinces.
CANADA-13
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In
B.R. Strain and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation.
DOE/ER-0238. U.S. Department of Energy. Washington, D.C. pp. 53-97.
Arthur, L. 1988. The implication of climatic change for agriculture in the prairie provinces. CCD 88-01,
Atmospheric Environment Service, Environment Canada, Downsview.
Bootsma, A., and R. de Jong. 1988a. Estimates of seeding Dates of Spring Wheat on the Canadian
Prairies From Climatic Data. Canadian Journal of Soil Science, 68:513-517.
Bootsma, A., and R. de Jong. 1988b. Climate Risk Analyses of the Prairie Region, in J. Dumanski
and V. Kirwood (eds). Crop Production Risks in the Canadian Prairie Region in Relation to
Climate and Land Resources. Technical Bulletin 1988-5E, Research Branch, Agriculture
Canada, Ottawa.
Bootsma, A., W. Blackburn, R. Stewart, R. Muma, and J. Dumanski. 1984. Possible Effects of
Climatic Change on Estimated Crop Yields in Canada. LRRI Contribution No. 83-64,
Research Branch, Agriculture Canada, Ottawa.
Brklacich, M., and B. Smit. 1991. Implications of Changes in Climatic Averages and Variability on
Food Production Opportunities in Ontario, Canada, Climatic Change.
Canada-Alberta 1989. Soil inventory Map Attribute File - Alberta: Soil Layer Digital Data. CLBRR,
Research Branch, Agriculture Canada, Ottawa.
Canada-Manitoba 1989. Soil inventory Map Attribute File - Manitoba: Soil Layer Digital Data.
CLBRR, Research Branch, Agriculture Canada, Ottawa.
Canada-Saskatchewan 1989. Soil inventory Map Attribute File - Saskatchewan: Soil Layer Digital Data.
CLBRR, Research Branch, Agriculture Canada, Ottawa.
de Jong, R., and A. Bootsma. 1987. Estimated Long-term Soil Moisture Variability on the Canadian
Prairies. Canadian Journal of Soil Science, 68:307-321.
Doorenbos, J., and W. Pr uitt. 1977. Guidelines for Predicting Crop Water Requirements. FAO Irrigation
and Drainage Paper 24, FAO, Rome.
Fei, Q., and E. Ripley. 1985. Simulation of Spring Wheat Yields in the Saskatoon Crop District 1960
to 1984 Using the CERES-Wheat Growth Model. SRC Publication No. E-906-46-B-85,
Saskatchewan Research Council, Saskatoon.
Godwin, D., Ritchie, J., Singh, U., and Hunt, L. 1989. A User's Guide to CERES Wheat-V2.W.
International Fertilizer Development Center. Muscle Shoals. AL
CANADA-14
-------
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
Efficient Three-Dimensional Global Models for Climate Studies: Models I and II. April
Monthly Weather Review, Vol HI, No. 4:609-662.
Hansen, J., I. Fung, A. Lacis, 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
93(D8):9341-9364.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Manitoba Agriculture. 1988. Field Crop Production Guide for Manitoba 1988-1990. Manitoba
Agriculture, Winnipeg.
McLelland, M. 1985a. Winter Wheat Production. AGDEX 112/20-3, Field Crops Branch, Alberta
Agriculture, Lacombe.
McLelland, M. 1985b. Soft White Wheat Production. AGDEX 112/20-2, Field Crops Branch, Alberta
Agriculture, Lacombe.
Prairie Pools Inc. 1988.1988 Prairie Grain Variety Survey. Prairie Pools Inc., Regina.
Russelo, D., S. Eddy, and J. Godfrey. 1974. Selected Tables & Conversions Used in Agrometeorology
& Related Fields. Publication 1922, Agriculture Canada, Ottawa.
Saskatchewan Agricultural Services Co-ordinating Committee. 1981.1981 Guide to Farm Practices
in Saskatchewan. The University of Saskatchewan, Saskatoon.
Saskatchewan Agriculture and Food. 1990. Varieties of Grain Crops for Saskatchewan 1990. Soils and
Crops branch, Saskatoon.
Scheelar, M., and W. Odynsky. 1968. Reconnaissance Soil Survey of the Grimshaw and Notikewin
Area. Report No. 88, Research Council of Alberta, Edmonton.
Singh, B., and R. Stewart. 1991. Potential Impacts of a CO2-Induced Climate Change Using the
GISS Scenario on Agriculture in Quebec, Canada. Ecosystems and Agriculture, 35:327-347.
Smit, B., M. Brklacich, R.B. Stewart, R. McBridge, M. Brown, and D. Bond. 1989. Sensitivity of
Crop Yields and Land Resource Potential to Climatic Change in Ontario, Canada, Climatic
Change 14 (2):153-174.
Stewart, R. 1990. Possible Effects of Climatic Change on Estimated Crop Yields in Canada: A
Review. In G. Wall and M. Sanderson (Eds.) Climate Change: Implications for Water and
Ecological Resources. Department of Geography, Occasional Paper No. 11, University of
Waterloo, Waterloo (pp 275-284).
CANADA-15
-------
Williams, G., H. Jones, E. Wheaton, R. Stewart, and R. Fautley. 1988. Estimating the Impacts of
Climatic Change on Agriculture in the Canadian Prairies, the Saskatchewan Case Study, in
M. Parry, T. Carter, and N. Konjin (eds). The Impact of Climatic Variations on Agriculture.
Volume 1. Assessment in Cold Temperate and Cold Regions, Kluwer, Dordrecht. (PP221-
379).
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a
Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
CANADA-16
-------
Table 1. Recent production and land use data (1981-85) for Canadian spring wheat in the
areas represented by the indicator sites in the Canadian prairie provinces.
Province/Site
Estimated
Wheat contribution
Lat. Production to National
Long. T x 1000 Production (%)
Average
Yield
(T/Ha) Soil
MANITOBA
Winnipeg
Dauphin
SASKATCHEWAN
Swift Current
Prince Albert
ALBERTA
Lethbridge
Ellerslie
Fort Vermillion
3,882
+49.5 N 15
-97.1 W
+51.1 N 4
-100.3 W
11,869
+50.2 N 32
-107.4 W
+53.1 N 23
-105.4 W
5,661
+49.4 N 16
-112.5 W
+53.3 N 7
-113.3 W
+58.2 N 3
-116.0 W
2.24 Black Soils
2.1 Other Soils
1.65 Brown Soils
1.95 Black Soils
1.93 Brown Soils
2.06 Black Soils
1.9 Northern
Soils
CANADA-17
-------
Table 2. Temperature differences (°C) and precipitation changes (%) between GCM lxCO2
and 2xCO2 climate change scenarios, at selected sites in Canada.
Site/GCM
Winnipeg
GISS
GFDL
UKMO
Dauphin
GISS
GFDL
UKMO
Swift Current
GISS
GFDL
UKMO
Prince Albert
GISS
GFDL
UKMO
Lethbridge
GISS
GFDL
UKMO
Ellerslie
GISS
GFDL
UKMO
Spr.
4.6
5.0
9.3
4.6
5.0
9.5
3.9
5.5
7.1
3.9
5.5
7.1
3.9
5.2
5.5
3.9
5.2
9.0
Temp.
Sum.
2.8
2.9
7.4
2.8
2.9
6.5
3.2
3.6
5.9
3.2
3.6
5.9
3.2
4.2
5.7
3.2
4.2
5.4
Diff. (
Fall
4.8
5.4
8.5
4.8
5.4
8.0
5.9
5.3
7.6
5.9
5.3
7.6
5.9
5.0
7.3
5.9
5.0
7.2
:°c)
Win.
5.5
5.0
10.2
5.5
5.0
9.6
6.2
4.9
7.5
6.2
4.9
7.5
6.2
4.9
6.7
6.2
4.9
5.9
Ann.
4.4
4.6
8.8
4.4
4.6
8.4
4.8
4.8
7.0
4.8
4.8
7.0
4.8
4.8
6.3
4.8
4.8
6.9
Spr.
7
70
33
7
70
55
19
59
30
19
59
30
19
38
27
19
38
58
Precip.
Sum.
17
46
-3
17
46
15
9
13
9
9
13
9
9
1
18
9
1
20
Changes
Fall
5
27
2
5
27
24
5
15
17
5
15
17
5
0
19
5
0
34
(%)
Win.
12
-3
22
12
-3
42
30
18
25
30
18
25
30
6
24
30
6
28
Ann.
11
36
9
11
36
27
15
24
17
15
24
17
15
9
21
15
9
31
Fort Vermillion
GISS
GFDL
UKMO
2.9
5.2
8.6
2.7
3.6
5.3
5.1
5.5
7.2
6.9
6.3
7.3
4.4
5.1
7.1
28
19
53
13
-9
23
26
13
41
56
22
47
26
12
35
CANADA-18
-------
Table 3. Management variables for spring and winter wheat used in the simulation at
different sites.
% Soil moisture
Site
Spring Wheat
Winnipeg
Dauphin
Swft Current
Prince Albert
Lethbridge
Ellerslie
Ft. Vermillion
Winter Wheat
Winnipeg
Dauphin
Swft Current
Prince Albert
Lethbridge
Ellerslie
Ft. Vermillion
Seeding
May 13
May 11
May8
May 6
May 7
May8
May 13
Sept 15
Sept 6
Sept 21
Sept 8
Sept 29
Sept 8
Sept 8
Plants/m2
333
333
190
111
172
390
390
283
283
224
277
224
283
283
at seeding1
63
71
49
57
42
61
63
% of water holding capacity
CANADA-19
-------
Table 4.
Simulated and observed yield (kg ha"1) and season length (days) at selected sites.
Site
Season Length
Days % Diff.
Yield
kg ha'1 % Diff.
Winnipeg
Dauphin
Swift Current
Prince Albert
Lethbridge
Ellerslie
Ft. Vermillion
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
89
93
90
98
98
100
96
106
90
102
106
111
119
119
+6
+9
+2
+10
+13
+4
0
2,275
2,682
2,531
3,210
1,385
1,612
2,223
3,020
1,930
1,550
2,060
5,450
1,900
3,850
+18
+21
+16
+36
-20
+164
+102
CANADA-20
-------
Table 5.
Impacts of changes in climate on crop moisture stress* (mm).
Scenario
WINN
DAUP
Crop Moisture Stress (mm)
P.A. S.C. LETH ELLE
F.V.
BASE
GISS
GFDL
UKMO
T+2, PO
T+4, PO
T 0, P+20
T+2, P+20
T+4, P+20
T 0, P-20
T+2, P-20
T+4, P-20
-49
-48
15
-44
-50
-59
-13
-26
-40
-61
-64
-70
-67
-60
2
-55
-66
-69
-38
-48
-54
-77
-78
-79
-53
-55
-45
-56
-57
-61
-27
-41
-50
-60
-64
-65
-48
-47
-36
-40
-50
-51
-35
-43
-44
-53
-54
-55
-35
-37
-30
-32
-40
-43
-14
-27
-32
-43
-47
-48
*Precipitation during the crop season length less evapotranspiration simulated
model during the crop season length).
-69
-66
-77
-39
-64
-71
-19
-34
-46
-83
-84
-86
by the CERES- Wheat
-55
-54
-68
-46
-57
-61
-30
-42
-49
-62
-66
-69
CANADA-21
-------
Table 6.
Sensitivity analysis of CERES-Wheat to climate and CO2 changes at selected sites.
VARIABLE
CHANGE
Temp. Precip.
change change
330
+2
+4
0
+2
+4
0
+2
+4
555
+2
+4
0
+2
+4
0
+2
+4
ppm CO2
0
0
+20
+20
+20
-20
-20
-20
ppm CO2
0
0
+20
+20
+20
-20
-20
-20
SIMULATED YIELD CHANGES FROM CURRENT (%)
WINN DAUP P.A. S.C. LETH ELLE F.V.
-22
-44
24
-4
-30
-31
-45
-59
11
-16
71
33
0
3
-19
-39
-29
-30
24
-10
-36
-27
-48
-63
4
-24
69
26
-7
9
-23
-45
-21
-43
27
-2
-28
-32
-45
-58
13
-17
74
33
2
5
-18
-40
-14
-27
40
19
-4
-39
-43
-48
35
10
112
79
42
-1
-12
-24
-12
-26
31
14
-6
-39
-43
-48
31
7
95
62
33
-2
-13
-25
-25
-43
14
-13
-34
-23
-43
-57
2
-21
46
13
-13
11
-16
-37
-25
-47
23
-9
-37
-26
-43
-59
5
-26
66
25
-13
7
-16
-42
CANADA-22
-------
Table 7.
Impact on simulated wheat yield of adaptation strategies to GCM climate change.
Strategy
Changes in simulated yield (% from current)
WINN DAUP P.A. S.C. LETH ELLE
IRRIG = changes from dryland to irrigation
-2WK, -4WK and -6WK = changes in the planting date two, four and six weeks earlier
WIN WH = shift to winter wheat.
F.V.
GISS
no change
IRRIG
-2WK
-4WK
-6WK
WIN WH
GFDL
no change
IRRIG
-2WK
-4WK
-6WK
WINWH
UKMO
no change
IRRIG
-2WK
-4WK
-6WK
WINWH
10
68
14
22
24
39
40
60
51
56
58
62
-40
11
-28
-15
-2
13
2
56
3
5
9
21
31
48
39
41
45
49
-32
9
-31
-28
-25
2
14
82
11
12
16
23
13
63
20
27
31
35
-26
47
-25
-25
-24
-25
50
219
66
75
84
132
66
183
103
117
131
167
19
155
25
24
25
48
40
234
72
92
110
200
-7
218
46
90
116
145
17
181
55
68
84
139
-2
18
-6
-10
-12
-1
-26
11
-23
-20
-21
-23
-25
-7
-36
-41
-43
-39
6
76
3
6
13
22
-35
57
-33
-31
-25
-25
-31
39
-37
-42
-46
-42
CANADA-23
-------
CANADA
FT. YERMILLION
ELLERSLIE
PRINCE ALBERT
LETHBRIDGE •
DAUPHIN
SWIFT CURRENT __ WINNIPEG
Figure 1. Location of the indicator sites in the Canadian prairie provinces: Manitoba
(Winnipeg, Dauphin), Saskatchewan (Swift Current, Prince Albert) and Alberta
(Lethbridge, Ellerslie, Fort Vermillion).
-------
IMPACT OF CLIMATE CHANGE ON
SEASON LENGTH
CHANGE FROM CURRENT (DAYS)
•I WINN
OOH LETH
Figure 2. Impacts of GCM climate change on simulated wheat season length at selected sites
in the Canadian prairie provinces.
-------
CLIMATE CHANGE ALONE
CHANGE FROM CURRENT YIELD {%)
•i WlNN
OH LETH
DAUP
CD ELLE
P.A.
CD F.V.
CLIMATE CHANGE WITH
PHYSIOLOGICAL CO2 EFFECTS
CHANGE FROM CURRENT YIELD (%)
Figure 3. Impacts of GCM climate change on simulated spring wheat yields (% change from
current) at selected sites in the Canadian prairie provinces.
-------
SENSITIVITY OF DAYS TO MATURITY TO
INCREMENTAL TEMPERATURE CHANGE
CHANGE FROM CURRENT (DAYS)
TEMP +2
TEMP +4
WINN
LETH
DAUP
CD ELLE
P.A.
F.V.
S.C.
Figure 4. Sensitivity of simulated wheat season length (days to maturity) to incremental
temperature changes at selected sites in the Canadian prairie provinces.
-------
QISS TRANSIENT SCENARIOS: IMPACTS ON
WHEAT YIELDS
CHANGE FROM CURRENT (%)
WINN DAUP
P.A.
2010s
!2030s
LETH ELLE F.V.
I 2050s
Figure 5. Impact of GISS transient scenarios on simulated wheat yield changes at selected
sites in the Candian prairie provinces.
-------
THE EFFECTS OF POTENTIAL CLIMATE CHANGE ON SIMULATED
GRAIN CROPS IN THE UNITED STATES
C. Rosenzweig
Columbia University and Goddard Institute for Space Studies
New York, USA
B. Curry
University of Florida, Gainesville, USA
J.T. Ritchie
Michigan State University, USA
J.W. Jones
University of Florida, Gainesville, USA
T.-Y. Chou
Taiwan National University, Taiwan
R. Goldberg
Columbia University and Goddard Institute for Space Studies
New York, USA
and
A Iglesias
Int. Nacional de Investigaciones Agrarias (INIA), Spain
USA-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Aims and Scope of Study
Agricultural Regions and Crops
STUDY DESIGN
Climate
Baseline Climate
Climate Scenarios
Crop Models
Input Data for Crop Models
Physiological CO2 Effects
Crop Model Validation
Simulations
Limitations of the Study
EFFECTS ON CROP YIELD AND SEASON LENGTH
Sensitivity Analysis
GCM Climate Change Scenarios
Transient Scenarios
EFFECTS ON EVAPOTRANSPIRATION AND IRRIGATION WATER DEMAND
EFFECTS UNDER DIFFERENT MANAGEMENT STRATEGIES- ADAPTATION
CONCLUSIONS ^riAiiuiN
REFERENCES
USA-2
-------
SUMMARY
The study considered the potential effects of global climate change on wheat, maize, and soybean
production in the United States. Climate scenarios derived from three General Circulation Models (GCMs)
were used in combination with crop growth models to characterize yield and irrigation water demand changes
of the three main crops in major agricultural regions. Under the present management system, projected climate
change caused simulated wheat, maize, and soybean yields to decrease at most sites even when the direct
effects of CO2 were included. These decreases were caused primarily by temperature increases which shortened
the duration of the crop life cycles, particularly the grain-filling periods. At some northern sites, yields
increased, probably because crop growth is temperature-limited at these high latitudes. Yield decreases varied
among GCM scenarios. Adaptation strategies were identified that compensated for the negative effects of
climate change at some but not all sites. These strategies included changing planting date and shifting to
cultivars more adapted to the projected future climate. Patterns of agriculture in the U.S. are likely to shift
as a result of changes in regional crop yields and in crop irrigation requirements.
INTRODUCTION
Background
The enhanced greenhouse effect of increased atmospheric concentration of CO2 and other trace gases
could lead to higher global surface temperatures and changed hydrological cycles (IPCC 1990a). Most previous
climate impact assessments suggest significant consequences for agriculture, including shifts in agricultural
zones, changes in irrigation demand, and loss of fertile lands in deltas due to sea-level rise (Parry et al. 1988;
IPCC 1990b). The heavy dependence on North America for world grain reserves (almost 80% of the 1975-77
global marketable surplus) has increased the sensitivity of the world food supply to the climate of that region.
Crop impact studies in the U.S. suggest a range of outcomes depending on scenario and region (Smith and
Tirpakl989.
Aims and Scope of the Study
The purpose of this study was to investigate the potential effects of climate change on crop yield and
irrigation demand of the three major U.S. crops, based on 19 crop modeling sites that characterize the main
agricultural regions. The results generated from this study have been used in an economic world trade model
to determine the possible impact of climate change on world cereal production and prices (Rosenzweig and
Parry, 1994). The approach taken here is to compare output of crop simulations under three equilibrium
GCM, transient, and sensitivity climate scenarios to simulations of present climate and growing conditions.
The study incorporated simulation results of the three major crops using consistent methodology so
that comparisons can be made between crops and regions. Since agricultural production and systems are not
regionally isolated processes, projecting differential regional effects is important. While some crops are not
significant in a particular region at present, this study tests whether climate change may alter their zonation
in the future. The simulations also test the differential effects of climate change on winter (wheat) and spring
(maize and soybean) crop production.
Because the driving force for a greenhouse-gas-induced climate change is the observed increase in CO2
and other gases, the crop models have been modified to simulate the physiological effects of higher
atmospheric CO2 on crop growth and water use, based on experimental literature. Thus the simulated changes
in crop parameters under the climate change scenarios are driven by two interacting effects, changes in climate
USA-3
-------
and CO2 enrichment. It is interesting to compare the responsiveness of each crop to higher levels of
atmospheric CO2. Finally, this study evaluates changes in crop management that may represent ways in which
farmers adapt to changed climate.
Agricultural Regions and Crops
The three most important U.S. crops both in domestic and export markets are wheat, maize, and
soybean. The U.S. ranks first in world maize and soybean production and third in wheat production. The U.S.
raises about half of the world total of both maize and soybean. We selected 19 sites in major agricultural
regions to represent different agroclimatic and production conditions (Figure 1 and Table 1).
Although wheat is produced in most states, the most productive regions is the Great Plains (Figure
2). About 80% of U.S. wheat production comes from winter wheat; therefore we only simulated winter wheat
in the study. Two different practices were simulated: rainfed and irrigated. The only major irrigated zone for
wheat is the southern area of the Pacific Northwest; irrigated wheat areas in California and Arizona are not
significant contributors to the national production (Table 2).
U.S. maize production is centered in the Midwest (Figure 2 and Table 2). Soybean production is
mainly located in the eastern two-thirds of the country, with the heaviest concentration in the Corn Belt
(Figure 2). Most soybean production occurs in a rotation plan with maize on the best soils in a given area;
the highest yields are in Iowa. We estimate that in the soybean and maize growing areas only about 10% of
the land area is irrigated (Table 2). Although only a small area of grain production in the U.S. is irrigated,
some regions (especially the southeastern part of the Pacific Northwest and California) rely heavily on
irrigation (Table 2).
STUDY DESIGN
The approach taken was to compare output from the crop models under climate change scenarios
(either arbitrary changes in climate variables or derived from global climate models) to model output from
simulations of current climate. The scenarios derived from GCM results included both equilibrium doubled
CO2 and transient projections of climate change. Results for both yield and irrigation water use are analyzed
as percent change from baseline simulations in order to provide indications of possible direction and
magnitude of responses to changed climate conditions. Additionally the study evaluated possible adaptation
strategies to climate change.
Climate
Baseline climate. Observed daily climate data (1951-80) were obtained for the 19 sites, consisting of
daily maximum and minimum temperatures and precipitation. Daily solar radiation was simulated using a
weather-generating program, WGEN (Richardson and Wright, 1984). Figure 3 shows the baseline (observed)
monthly precipitation and temperature regimes at six of the 19 sites.
Sensitivity scenarios. In order to test the crop models' sensitivity to climate change, fixed combinations
of temperature (0°C, +2°C, +4°C) and precipitation changes (-20%, 0%, +20%) were tested.
GCM equilibrium scenarios. Climate change scenarios were created from GCMs because they produce
climate variables which are internally consistent and because they allow for comparisons among regions.
Output from the GCMs is not accurate enough for direct use in crop models on a daily basis. Therefore, mean
monthly changes in climate variables from doubled CO2 simulations of three GCMs: Goddard Institute for
Space Studies (GISS); Geophysical Fluid Dynamics Laboratory (GFDL); and United Kingdom Meteorological
Office (UKMO), were applied to observed daily climate records to create climate change scenarios for each
USA-4
-------
site. The GCMs compute mean climatic variables for large (—100,000 km2) gridboxes and they do not explicitly
account for variation of these quantities within the gridbox. Therefore no interpolation was made of the GCM
output.
The projected monthly GCM precipitation and temperatures changes for the U.S. are shown in Figure
3. All three models predict significant warming: the GISS and GFDL scenarios produce comparable
temperature increases (3°C to 5°C), while the UKMO scenario produces more drastic rises (6°c to 9°C).
Annual precipitation generally increases at most sites. (GCMs predict increases in global precipitation
associated with warming since warmer air can hold more water vapor.) However, the projected precipitation
shows considerable seasonal and geographical variation and is very uncertain.
Transient scenarios. Most of our knowledge concerning the climate response to greenhouse-gas forcing
has been obtained from equilibrium response GCM experiments. These are experiments which consider the
steady-state response of the model's climate to step-function changes in atmospheric CO2. Recent evidence
from a few GCM experiments incorporating time-dependent greenhouse-gas forcing suggest that there may
be important differences between the equilibrium and transient responses (Hansen et al. 1988, Bryan et al.
1988; IPCC 1992).
Crop Models and Inputs
Potential changes in crop yields at the selected sites in the U.S. were estimated with the CERES-
Wheat, CERES-Maize, and SOYGRO crop models (Ritchie and Otter 1985; Jones and Kiniry 1986; Jones
et al. 1988). The models simulate physiological crop responses (water balance, phenology, and growth
throughout the season) on a daily basis to the major factors of climate (daily solar radiation, maximum and
minimum temperature, and precipitation), soils, (albedo and a variety of measures relating to water in the
profile) and management (cultivar, planting date, plant population, row spacing and sowing depth).
The CERES and SOYGRO models have been validated with experimental data from many locations
encompassing a wide range of environments (Otter-Nacke et al. 1986; Jones and Kiniry 1986; Egli 1992; Jones
and Ritchie 1991).
All simulations were made using DSSAT, v2.5 (Jones et al. 1990). Changes were computed in the
mean and standard deviation of yield, evapotranspiration (ET), water applied for irrigation, and crop maturity
date.
Soils. Representative agricultural soils for each site were chosen with reference to the Major Land
Resource Area descriptions; State Soil Conservation Stations, and information provided by County Agricultural
Extension Agents. Soil characteristics for the representative soils were specified by 12 generic soil types (Table
1). The generic soil most representative of each site was used. Soil characteristics included in the model are:
albedo, water drainage, soil evaporation, runoff, and characteristics describing each layer such as depth, lower
and upper limit of plant extractable water, saturated water content, organic carbon, ammonium, nitrate, and
pH values. Since the present model did not simulate a soil water balance during the entire year, we assumed
that the soil moisture level was full at the beginning of each growing season. While this assumption represents
crop planting conditions at most sites in most years in the U.S., it overestimates soil moisture at planting for
dry years.
Crop varieties and management. In the CERES and SOYGRO models, crop varieties are defined by
a set of coefficients that represent characteristics such as photoperiodism, vernalization, and crop maturity
type. The varieties used in the simulations are representative of common varieties grown in each region (Table
3). The management variables for the CERES and SOYGRO models were determined for each location
according to information on current practices provided by the County Agricultural Extension Agents as well
as state publications (Table 3).
For the irrigation simulation, the water demand was calculated assuming 100% efficiency of the
USA-5
-------
automatic irrigation system; a 1-meter irrigation management depth; and automatic irrigation when the
available soil water is 50% or less of capacity.
Physiological Effects of CO2
Higher levels of atmospheric CO2 have been found to increase photosynthesis and stomatal resistance,
resulting in yield increases in experimental settings (Acock and Allen 1985). Because the climate change
scenarios are associated with concomitant higher levels of CO2 and other trace gases, we have included the
physiological effects of 555 ppm CO2 in the crop model simulations (Peart et al. 1989).
Limitations of the Study
Climate change scenarios. While GCMs are useful for climate change studies, current climate models
oversimplify many aspects of the climate system, especially ocean dynamics, cloud physics, and land-surface
hydrology. GCMs do not simulate current climate well at regional scales and they were not designed for
ptedictive regional studies. Therefore, the climate change scenarios created from GCM output must not be
considered as predictions, but only as examples of possible future climates for the regions under study.
As configured, the climate change scenarios do not alter the patterns of events in the base climate.
Therefore they dp not simulate changes in the underlying variability (e.g., extended periods of high
temperature, droughts, etc.) that can be vital for crops. Dryland yields may be considerably different depending
on whether a change in precipitation results from a change in mean, frequency or intensity. However, the
scenarios created for this study do result in an increased frequency of temperatures above certain thresholds.
Crop models. In the crop model simulations, technology is held constant; nutrients are not limiting;
weeds, diseases and insect pests are controlled; and there are no problem soil conditions or catastrophic
weather events. All these assumptions tend to overestimate simulated yields. In this study an important
assumption is that climatic tolerances of cultivars do not change. It is also important to note that the
physiological effects of CO2 in the crop model may be overestimated because experimental results from
controlled environments, used to calibrate the model, may not be accurate under windy, and pest-infected field
conditions.
EFFECTS ON CROP YIELD AND SEASON LENGTH
Sensitivity Analysis
The results of the sensitivity analysis for wheat, maize and soybean for three sites without the direct
effects of CO2 are presented in Figure 4. The main conclusions of the sensitivity study are:
Increased temperatures cause simulated yield decreases. The greatest percentage yield decreases caused
by high temperature are found in soybean.
The three crops are slightly more sensitive to a decrease in precipitation than to an increase.
In general, a 20% increase in precipitation mitigates yield decreases caused by a 2°C temperature
increase, resulting in yield changes that are close to zero. With the same 20% precipitation increase
and a 4°C temperature rise, yield decreases are still significant.
The greatest percent yield decreases are caused by a 4°C temperature increase and a 20% precipitation
USA-6
-------
decrease.
GCM Climate Change Scenarios
Tables 4,5, and 6 show the simulated changes in crop yields under the GCM climate change scenarios,
both with and without the direct effects of CO2. Since the driving force for a greenhouse temperature change
is the increase in CO2 and other gases, the most important comparison from an agricultural system response
point of view is between the base (current climate under 330 ppm CO2) and the GCM scenarios with increased
CO2 (555 ppm CC-2).
Soybeans. Yield changes under the climate change scenario vary with location and scenario considered
(Table 4), but in general soybean yields decrease under the three climate change scenarios. The largest
decreases occurred under the UKMO scenario and the smallest decreases under the GISS scenario. In some
cases the effect of increased CO2 on photosynthesis was enough to compensate for the yield decreases under
climate change alone. When the crop was simulated under irrigated conditions, many locations showed an
increase in yield compared to the base (see section on Adaptation). The main cause of the yield decrease is
the shortening of the growing season (by about two weeks) (Table 7).
Maize. Maize yield decreased significantly at almost every site with the climate change scenarios alone
(Table 5). When the direct effects of CO2 were included, significant yield decreases were still projected for the
GFDL and UKMO scenarios, but negative yield effects of the GISS scenario were completely mitigated. The
yield decreases were caused by the combined effects of high temperature shortening the grain-filling period
and by increased moisture stress. Maturity dates advanced by an average of over two weeks under the three
scenarios (Table 7). In the highest latitude site (Spokane) there was a large increase in maize yield as
simulated with all three GCM scenarios; the maize crop appears to be temperature-limited in the current
climate at this relatively high latitude.
In the case of climate change with the physiological effects of CO2, simulated maize yields increased
only slightly compared to climate change alone; this is consistent with the lower photosynthetic response of
C4 crops to increased CO2 levels (Acock and Allen 1985). At all sites except Spokane, irrigated maize yields
decreased significantly even when the direct effects of CO2 were considered.
Wheat. Wheat was simulated under dryland and irrigated conditions at all locations except Fresno
(California), where dryland wheat produces negligible yields under current climate. Overall, there were large
changes in simulated wheat yields under the climate change scenarios compared to the baseline yield (Table
6). Under the GISS scenario, modeled wheat yields decreased in every location except in the northernmost
sites, with the largest decreases at the lower latitude sites. Some of the major areas of wheat production, (e.g.,
Dodge City, Abilene) showed very negative yield changes under each of the climate change scenarios.
The yield decreases were driven primarily by increased temperatures, which caused the duration of
crop growth stages (particularly the grain-filling period) to be shortened (Table 7). Shortening of the grain-
filling period reduces the amount of carbohydrates available for grain formation and harvestable yield. Maturity
dates of wheat occurred, on average, about two weeks earlier under the scenarios. In the sites of the Pacific
Northwest region (Spokane, Yakima and Boise), the unusually large simulated dryland wheat yield increases
were probably due to the large increase in winter precipitation projected by the GISS climate change scenario.
Under the GFDL and UKMO scenarios, yields decreased at all but a few sites. The yield increases in the
northern sites may be due to the positive effects of higher temperatures on growth in colder, high-latitude
sites. In contrast, in the Southern Great Plains, the temperatures are already high to begin with, and the
additional stress of higher temperatures can lead to drastic yield decreases.
The addition of the physiological CO2 effects to the simulations caused a significant increase in wheat
yield at all sites. This was due to the beneficial increases in simulated dry matter conversion and stomatal
resistance. Nevertheless, in the lower latitude sites substantial yield decreases were simulated even when the
USA-7
-------
direct effects of CO2 were taken into account.
Aggregated Results
The simulated changes in crop yield at different sites suggest that regional and national production
may be significantly affected by climate change. In order to estimate these possible changes, we developed an
aggregation scheme based on actual regional production from 1979-1989 as reported by the USDA Crop
Reporting Service. A relative "production weighting factor", based on the amount of production of a particular
crop, was assigned to each site. We also produced a "dryland/irrigation weighting factor" for each site, which
is an estimation of the percent of harvested area associated with each site which is dryland production (Table
2).
Aggregated U.S. crop yield results are shown in Figure 5. The magnitudes of the estimated yield
changes varied by crop and scenario. When the direct CO2 effects were included, maize (a C4 crop) production
was most negatively affected, probably due to its lower growth response to higher levels of atmospheric CO2;
soybean and wheat (C3 crops) production responded significantly to increased CO2. The aggregated production
results also imply:
Without physiological CO2 effects, production of all three crops decreased. The warmest scenario
(UKMO) produced the largest decreases, with soybean the most severely effected.
Under the GISS and GFDL scenarios with direct CO2 effects included, soybean production increased,
wheat production remained unchanged, while maize production was reduced. Under the UKMO
scenario all three crops showed decreases in yields.
Simulated U.S. maize production was negatively affected in all scenarios, both with and without the
physiological CO2 effects.
Transient Scenarios
When the crop models were run with transient climate changes projected for the 2010s, 2030s, and
2050s from the GISS transient run A, yield responses were non-linear over time (Figure 6). Aggregated wheat
yield was the most non-linear of the three crops, showing a positive response to climate change in the 2010s
and 2030s, and a negative response in the 2050s. Soybean yield changes were positive throughout the time
trajectory, and maize yield changes were slightly negative.
EFFECTS ON EVAPOTRANSPIRATION AND IRRIGATION WATER DEMAND
The temperature increases under the GCM climate change scenarios are likely to lead to increased
daily potential evapotranspiration and therefore changes in crop water use and irrigation demand. The
following simulation results considered both the effect of climate change and the effect of increases in CO2
level on crop growth, water use, and irrigation demand. Because the calculation of the irrigation water in the
CERES and SOYGRO models assumes nonlimiting situations, only relative changes in crop water use
requirements are analyzed (Tables 8 and 9).
Soybeans. Water us.e efficiency (WUE), defined as yield/total crop evapotranspiration (kg/ha mm'1),
declined at most locations under the GFDL and UKMO scenarios, because yield reductions are relatively
greater than ET reductions (Table 8). Table 9 shows that irrigation demand increased substantially under the
climate change scenarios, with the warmer scenarios producing the largest increases.
USA-8
-------
Maize. The maize WUE increased almost everywhere under the GISS scenario, in about half of the
sites under the GFDL scenario, but decreased under the UKMO scenario except in the northernmost sites
(Table 8). Under the GISS scenario, irrigation demand decreased at most of the study sites, while under the
more severe GFDL and UKMO scenarios, irrigation demand increased at some sites, while decreasing at other
sites. The decreases in irrigation water requirements (Table 9) have to be carefully interpreted, since they
primarily reflect the decreases in total crop evapotranspiration due to the significant shortening of the crop
growing season and are often accompanied by significant yield reductions.
Wheat.The wheat WUE values decreased at most sites under the GISS scenario, but increased under
the GFDL and UKMO scenarios. Irrigation water demand for wheat shows very mixed results (Table 9).
EFFECTS UNDER DIFFERENT MANAGEMENT STRATEGIES: ADAPTATION
Farmers and agricultural systems will, in reality, try to adjust to changing environmental conditions.
This study evaluated changes in planting dates, cultivars, and irrigation as possible changes in crop
management under climate change conditions. Estimates of yield changes were based on crop model
simulations. Changes in economics or domestic agricultural policies were beyond the scope of the adaptation
estimates. Neither the costs of adaptation nor changes in water availability under the climate change scenarios
were considered.
Soybeans. Rainfed soybean yields decreased by more than 50% under climate change scenarios (Table
4), but adding irrigation tripled yields. Even though irrigation was found to increase yield, it may not be
possible to use it as a widespread amelioration tool, since the demand would probably exceed the water
resources available. Only about 10% of U.S. cropping regions is currently irrigated. Establishing irrigation as
an adaptation to climate change could be costly.
The soybean adaptation results using different cultivars and planting dates showed that adaptation to
temperature changes can be met with existing cultivars in the U.S. Midwest. In the southern part of the U.S.
where temperatures are higher, potential temperature changes suggested by some GCMs may exceed crop
physiological tolerance and increase water stress.
Maize. Changes in planting date (10, 20, and 30 days earlier), changes in variety, or combined
strategies of both were examined for Des Moines. With an earlier planting date, the yield, with the direct CO2
effect included, could not be restored to current levels. The selection of alternative varieties helped ameliorate
some of the yield losses, but not completely (Figure 7). The combination of earlier planting dates and a
different variety resulted in yield increases for the GISS scenario, and for the irrigated GFDL scenario.
However, under the UKMO scenarios, yields still decreased.
Wheat. When the automatic irrigation is applied, wheat yields improve over the baseline in all
locations, except in the southernmost latitudes, with combined climatic and direct effects of CO2 in both GISS
and GFDL scenarios.
DISCUSSION
Simulated yield decreases occurred in the Southern Great Plains and the Southeast in response to the
climate change scenarios tested in this study, even when the direct effects of CO2 were included. In those
regions, high temperatures shortened the duration of crop stages, especially the grain-filling period, and
increased water stress, resulting in moderate to severe yield decreases. The hotter UKMO model produced the
most severe yield reductions, while the GISS model was the most benign of the three GCM scenarios.
Increased temperatures had a beneficial effect (due to a longer frost-free period) on simulated yields at the
northern sites.
Farmer and agricultural system adaptations (change in planting date, crop variety, and irrigation)
USA-9
-------
helped mitigate yield losses. However, successful adaptation to climate change often implied significant changes
to current agricultural systems, and so some adaptations to climate change may be costly. A potential need
for increased irrigation is suggested, particularly at lower latitudes. Currently only a small percentage of the
area of wheat, maize and soybean is irrigated. Whether this would actually change depends on many factors
such as the availability and cost of water, which are beyond the scope of this study.
Adjusting to a need for increased irrigation water may be difficult. The country's hydrological system
has a complicated legal framework that limits flexibility; in addition the existing irrigation systems may be
subjected to climatic conditions for which they were not designed. Even without climate change, U.S. water-
allocation institutions need to evolve toward programs that encourage a more efficient use of water. Currently
there is little incentive to irrigate without wasting water. The environmental problems (salinity, erosion and
water pollution) derived from increased irrigation also need to be further addressed.
This study has not addressed the issue of changes in climate variability. Possible changes in climatic
variability under GCM climate change scenarios (such as the magnitude and frequency of droughts, storms,
and heat waves) are important factors in determining production amounts. Since crop yields exhibit nonlinear
responses to heat and cold stress, changes in the probability of extreme temperature events beyond critical
thresholds can be significant. Differential effects on minimum and maximum temperatures also need to be
studied. A farmer can adapt to gradual changes in mean weather variables, but he or she is highly vulnerable
economically to increased variability, e.g., crop failure in one or several years in a row with only adequate yield
in others. Government policies regarding crop insurance also need to be examined.
Overall, while national production does not appear to be at risk, shifts in regional patterns of
agricultural are implied by this study. The northern states could become much more productive for annual
crops such as corn and soybeans because of the lengthening of the frost-free period, while the southern states
could become less productive, due to heat and moisture stress.
USA-10
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R.
Strain and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238.
U.S. Department of Energy. Washington, D.C. pp. 53-97.
Egli, D.B. and W. Bruening. 1992. Planting date and soybean yield: evaluation of environmental effects with
a crop simulation model: SOYGRO. Agric. Forest. MeteoroL 62, 19-29.
Godwin, D., J. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES-Wheat - v2.10. International
Fertilizer Development Center. Muscle Shoals, USA
Hansen, J., I. Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1988. Global Climate Changes
as Forecasted by Goddard Institute for Space Studies Three-Dimensional Model. Journal of
Geophysical Research, 93:9341-9364.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
Three-Dimensional Global Models for Climate Studies: Models I and II. Monthly Weather Review, Vol.
Ill, No.4.
IPCC, 1992. Climate Change 1992. The Supplementary Report to the IPCC Scientific Assessment. J.T.
Houghton, B.A Callander, and S.K. Varney (eds). Intergovernmental Panel on Climate Change.
Cambridge University Press. Cambridge. 200 pp.
IPCC. 1990a. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums
(eds). Intergovernmental Panel on Climate Change. Cambridge University Press. Cambridge.
IPCC, 1990b. Climate Change: The IPCC Impacts Assessment. Tegart, W.J. McG., G.W. Sheldon, and D.C.
Griffiths (eds). Intergovernmental Panel on Climate Change. Australian Government Publishing
Service. Canberra.
Jones, C.A, and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development
College Station. Texas A&M University Press.
Jones, J.W. and J.T. Ritchie. 1991. Crop growth models, p.69-98. Management of Farm Irrigation Systems.
American Society of Agricultural Engineers, St. Joseph, Michigan.
Jones, J.W., K.J. Boote, S.S. Jagtap, G. Hoogenboom, and G.G. Wilkerson. 1988. SOYGRO v5.41: Soybean
Crop Growth Simulation Model User's Guide. Florida Agr. Exp. Sta. Journal No. 8304, IFAS. Univ. of
Florida.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. In
International Benchmark Sites Network for Agrotechnology Transfer. Proceedings of the IBSNAT
Symposium: Decision Support System for Agrotechnology Transfer. University of Hawaii.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
USA-11
-------
Otter-Nacke, S., D.C. Godwin, and J.T. Ritchie. 1986. Testing and Validating the CERES-Wheat Model in
Diverse Environments. AgGRISTARS YM-15-00407. Johnson Space Center No. 20244. Houston.
Parry, M.L., T.R. Carter and N.T. Konijn. 1988. The Impact of Climatic Variations on Agriculture. Vol 1
Assessments in Cool Temperate and Cold Regions. Vol 2 Assessments in Semi-arid Regions. Kluwer,
Dordrecht, Netherlands. 876 pp. and 764 pp.
Peart, R.M., J.W. Jones, R.B. Curry, K.T. Boote, and L.H. Allen. 1989. Impact of climate change on crop yield
in the southeastern United States: A simulation study. In: J.B. Smith and D.A. Tirpak (eds). The
Potential Effects of Global Climate Change on the United States, Appendix C, Agriculture, Vol. 1, EPA-
230-05-89-053. U.S. Environmental Protection Agency, Washington, DC.
Richardson, C.W., and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Ritchie J., U. Singh, D. Godwin, and L. Hunt. 1989. A User's Guide to CERES-Maize v2.10. Michigan State
University-IFDC-IBSNAT, Muscle Shoals, AL.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-wheat: A user-oriented wheat yield
model. In Willis, W.O., ed. ARS Wheat Yield Project. Washington, DC: USDA Agricultural Research
Service. ARS-38.
Rosenzweig, C. and M.L. Parry. 1994. Potential impact of climate change on world food supply. Nature, 367,
133-138.
Smith, J.B., and D. Tirpak (eds.): 1989. The Potential Effects of Global Climate Change on the United States.
Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency. Washington, DC.
423pp.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
USA-12
-------
Table 1.
Sites selected for the study and representative soils for each site.
Site
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
Long.
+47.63
+46.57
+43.57
+36.77
+33.43
+46.90
+41.08
+39.00
+37.46
+32.43
+41.53
+39.73
+35.05
+30.53
+33.95
+32.42
+30.38
+37.33
+41.25
Lat.
-117.53
-120.53
-116.21
-119.72
-112.02
-96.97
-100.41
-96.00
-99.58
-99.68
-93.65
-86.28
-89.98
-91.15
-81.12
-83.39
-84.37
-79.20
-79.92
Soil
DSL
DSL
DSL
DSL
DSnL
DSC
DSn
DSL
DSL
DSnL
DSL
MSL
MSL
MSC
MSnL
MSnL
OrSnL
MSL
MSC
DSL: deep silty loam; DSnL: deep sandy loam; DSC: deep silty clay; DSn: deep sand; MSL: medium silty loam;
MSC: medium silty clay; OrSnL: Orangesburgh sandy loam.
USA-13
-------
Table 2. Regional wheat, maize and soybean production (% of total national production) and area
irrigated (% of total grain area irrigated at site or in the region).
WHEAT
Region
Pacific NW
South West
N Mnt Plains
N Great Plains
S Great Plains
Corn Belt
Delta
Southeast
Northeast
Site % p
Spokane 12.0
Yakima
Boise
Fresno 3.0
Phoenix
Fargo 27.0
RPlatte 20.0
Topeka
Dod. C.
Abilene 13.0
Des Mo. 11.0
Indiana.
Memph. 3.0
B.R.
Colum. 2.0
Macon
Tallahas.
Lynch. 2.0
William.
%l
0
100
100
100
100
0
0
0
0
0
0
0
0
0
0
0
0
MAIZE
%P
0.4
0.6
0.4
1.5
2.0
71.0
16.0
0.5
1.5
4.1
2.5
%I
100
100
100
100
100
20
50
50
50
10
10
10
0
0
0
0
0
0
0
SOYBEANS
%P
7.0
7.0
1.0
37.0
28.0
8.0
5.0
2.0
2.0
0.5
2.0
2.0
%I
20
25
25
25
5
15
15
10
15
10
5
0
USA-14
-------
W
fcjj
O
CO
S
s
•*-£
q
1
03
Q
AH
el
3
o
EH
variables.
WHE
€
1
w
s ^~
<§ -^
J i
M ^-
|
•*— »
Q
PM
1
•4-*
"3
0
s
CO
« 0
s
t
a
o o
AH AH
« §
1 1
s-S
^^
i
T-H
ll
•1
s
o
T-H
0 3
cs
VN
C3
CM
vq
t~-
T-H
s
O
' VO
l.a
« "2
on S
0 §
11
>,
T-H
«
1^
^5
T— i
3
CM
CN
C3
CM
t>
§
T-H
o
VO
T-H
.s
00
Phoenix
03 Cd CO
»0 O <0
CM C^ rH
en
6
8 o i
66^
oJ"
^H *3
T-H ^,
T-H 10
u 0
f~ oo
C5 C5
rJ tD
C*^ "4"
C3 f"^
S S
8- 8-
C/3 00
T-H T-H
03 CA
.s .s
C3 CO
AH £C
^ 5S
1 JS
1 ^ 1
CH Z EH
C3 P3 CC
>o >o >o
T-< T-H CM
6 t2 6
S, -8 &
H^ ^T ( ^3
T— ) *O f^j
T-I CM >O
CM CO CM
0 0 0
oo oo oo
c5 c5 d
•* vq o
CM oo oo
oo vo o
CM T-H CM
111
co co "Z
o 0|
^^ O O
« c 5
OJJ ^ l*-4
0? §^
>0 >0
T-H T-4
CA
William
Forrest
^ .
u ^
^1 */^
rH 0
d d
oo co
CO ^H
S R
CO CM
VO VO
T-H T-I
S 1
U p
Indianapolis
Memphis
C5 CO
o >o
CM T-H
5^ 18
cQ cd
E^ «
"8 iS
[T j ^q
>O >O
CM T-H
3 0
>0 <0
d d
O\ 10
00 CM
>o o
CM 00
CM T-H
11
vo vo
T-I T-H
1 1
w w
B. Rouge
Columbia
03 W
>o o
T-H CM
II
OQ O
•8
PH
OO
^ **^
T-H f^4
CM T-H
8 Q
10 10
f^i f^l
>o oq
CM CO
O O
oo vo
T-H CO
1 §
Z CO
VO \D
T-I T-H
t2 w
4_*
Lynchburg
Williamspor
.
°3
s
OH
OH
£1
1
2
T3
Ji
s §
— ^ *r~*
.2 «
5 •£?
'S 'fe
51
gO **•*
O w
3S
CO
-o t--
Cj ri
•3 §
S *o if*
3 <3 T-H
OH 03 L
It |
H1^
i«r
3 a
•O _03
O OH
^ of
a **
o, §
2
-------
Table 4.
Effects of GCM climate change on dryland and irrigated soybean yields.
Site
DRYLAND
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)
0.79
0.34
0.38
0.45
0.30
1.35
0.74
2.49
1.42
0.92
2.72
2.43
2.06
2.49
2.41
1.75
3.44
2.42
2.11
GCM
GISS
-11
-26
-16
-53
-57
-10
-12
-31
-49
-62
-7
-12
-40
-43
-19
-24
-23
2
-5
Scenario
GFDL
-32
-35
-47
-49
-93
71
-18
-36
-57
-63
-26
-37
-63
-43
-35
-61
-21
-65
-55
alone
UKMO
-51
-65
-29
-98
-77
-46
-31
-40
-58
-84
-76
-43
-80
-82
-22
-86
-69
-71
-37
GCM Scenario +
GISS
52
12
26
-33
-40
58
50
-6
-26
-32
26
16
-21
-17
4
1
-3
34
34
GFDL
13
-9
-32
-24
-93
150
42
-11
-40
-41
8
-14
-46
-17
-14
-45
0
-48
-29
D.E. CO,
UKMO
-20
-47
-5
-96
-57
-4
16
-18
-42
-73
-64
-23
-72
-72
7
-74
-50
-58
-9
D.E, = Direct physiological effects of 555 ppm CO2
USA-16
-------
Table 4.
(Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)
2.66
3.24
2.99
3.32
3.93
2.48
2.55
2.90
3.03
4.15
3.42
3.47
4.03
3.60
3.90
3.90
3.69
3.61
2.86
GCM
GISS
16
-7
-12
-37
-38
15
-4
-14
-65
-36
-1
-2
-27
-15
-12
-14
-13
0
8
Scenario alone
GFDL
13
-3
-28
-36
-37
22
-10
-13
-25
-40
-5
-1
-18
-15
-6
-9
-10
-2
7
UKMO
-2
-40
-31
-68
-37
2
22
-20
-36
-63
-46
-15
-48
-41
-12
-25
-30
-16
-5
GCM Scenario +
GISS
58
20
28
-18
-20
65
47
8
-26
-18
25
21
-10
5
6
4
5
24
38
GFDL
58
28
14
-14
-3
66
44
10
-5
-19
21
22
1
5
14
10
9
22
39
D.E. CO2
UKMO
39
-19
-13
-56
-7
50
30
0
-19
-46
-30
5
-36
-26
7
-6
-2
6
23
D.E. = Direct physiological effects of 555 ppm CO2.
USA-17
-------
Table 5.
Effects of GCM climate change on dryland and irrigated maize yields.
Site
DRYLAND
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)
10.25
6.25
8.36
5.89
11.58
9.74
7.82
6.58
7.00
9.31
8.30
8.55
GCM Scenario alone
GISS
-10
-22
-8
15
-21
-7
-6
10
-28
-5
-58
-21
GFDL
-6
-17
-14
-9
-27
-59
-27
-45
-90
-41
-61
-51
UKMO
-47
-57
-18
-17
-42
-20
-44
-14
-28
-34
-21
-23
GCM Scenario +
GISS
2
11
7
22
-18
1
5
26
-3
0
-20
7
GFDL
3
16
1
8
-21
-32
-13
-23
-75
-22
-35
-29
D.E. CO,
UKMO
-28
-33
-10
-6
-34
-12
-34
-4
-14
-21
-3
-2
D.E. = Direct physiological effects of 555 ppm CO2.
USA-18
-------
Table 5. (Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha1)
9.88
12.92
14.58
12.19
7.78
13.37
12.94
12.66
10.49
13.68
13.04
11.53
11.47
10.75
11.65
12.67
12.86
GCM Scenario alone
GISS
40
-5
-22
-4
-35
-8
-22
-9
-2
-25
-20
-19
-9
-16
-10
-24
-21
GFDL
59
7
-20
1
-28
-12
-22
-15
-1
-28
-6
-14
4
9
-13
-3
-8
UKMO
32
-28
-24
-3
-57
-24
-38
-21
-36
-38
-25
-42
-39
-23
-9
-11
-14
GCM Scenario +
GISS
48
0
-17
-3
-31
-3
-17
-5
1
-21
-15
-14
-4
-11
-5
-19
-17
GFDL
62
13
-16
1
-24
-7
-17
-10
0
-24
-3
-10
5
12
-9
0
-14
D.E. C02
UKMO
39
-24
-20
-2
-55
-21
-34
-21
-32
-34
-21
-39
-36
-19
-6
-7
-10
D.E. = Direct physiological effects of 555 ppm CO2.
USA-19
-------
Table 6.
Effects of GCM climate change on dryland and irrigated wheat yields.
Site
DRYLAND
Spokane
Yakima
Boise
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)
2.62
1.48
1.38
0.89
4.13
2.79
3.56
2.60
4.76
4.95
4.41
3.86
4.38
3.58
4.72
4.88
GCM Scenario alone
GISS
76
103
97
-99
-18
-18
-48
-58
-4
-3
-25
-57
-22
-56
-6
-2
GFD
L
-6
-37
-20
-93
-14
-36
-44
-33
-12
-6
-10
-58
-19
-80
-2
0
UKM
O
29
-7
43
-99
-38
-33
-32
-67
-15
-16
-37
-98
-35
-100
-25
-6
GCM Scenario +
GISS
126
180
172
-99
8
11
-29
-43
16
14
-12
-48
-8
-46
13
17
GFDL
33
-84
14
-91
10
-15
-26
-10
7
10
4
-50
-5
-76
17
19
D.E. CO,
UKMO
71
31
102
-99
-13
-10
-10
-57
1
-1
-26
-97
-22
-100
-10
12
D.E. = Direct physiological effects of 555 ppm CO2.
USA-20
-------
Table 6.
(Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)
5.41
5.34
5.67
7.29
4.85
5.12
4.91
5.59
5.24
4.90
5.26
4.42
4.53
3.93
5.15
5.35
GCM Scenario alone
GISS
6
-3
2
-37
-92
-8
8
-7
-30
0
-8
-24
-22
-45
-12
-8
GFD
L
-6
-4
8
-39
-67
-15
12
-8
-24
4
-12
-12
-20
-80
-9
-6
UKM
O
10
-3
13
-63
-96
-8
1
-12
-48
-11
-20
-36
-34
-100
-27
-9
GCM Scenario 4
GISS
25
14
24
-16
-91
8
28
9
-18
17
7
-11
-9
-35
4
8
GFDL
10
12
27
-15
-62
0
31
8
-11
13
3
4
-6
-76
7
11
D.E. C02
UKMO
29
14
32
-45
-96
9
19
3
-40
4
-6
-25
-22
-100
-15
7
D.E. = Direct physiological effects of 555 ppm CO2.
USA-21
-------
<
CJ
S
6
1
•I w
.1 M
g S
1
i
i
•&
•o
H. 5
1£ w
>§ P"!
^ S
•S O
00 CO
a
_«
e
o
CO
_«>
Q
rS
j
0
VI
E
O
W
^ji
03
0
1
CO
S2
0
ttl
0
i
£J
O
8
23
«
CO
O OO CM TH c^ l> •*
IO CO *H rH ^- VO U")
CM CM CM TH CM CM CM
V> OO CO CM t^ IO TH
CM CM CM TH csi CM CM
CO TH 00 00 CO >n VO
vo •* TH CM co oo w->
CM CM CM TH CM CM CM
^ vg vo tj ^ TH 00
CS C^l C^ TH rH CO 0s!
TH
OO 00 C"** O\ 00 Ox O\
00 00 O\ IO Tj* rH t^
TH oo oo o\ CT\ ON o\
TrH rH . ( . i
O\ O ON O O 00 OO O
gNCQc^cNi^cot^oo
ONTHOCOOOONOOO
^_f ^r( i, ^ _^
CSrH^cOOOTHOOOO
OTHOCO^-ONOOO
CO CD TH £"*• O\ TH CP CO
CO *^* C*^ CO *O TH C3 t~H
| 1 0 S | 0, I |
o-i-^So£?^a-
CU 45 5 r"H J3 r«I ^ O
COr^OQCi^OnbZE-1
CM TH CM
•* 00 VO
CM TH CM
Tj* -H^ VO
co ON in
CM TH CM
CO CM "*
•not*-
CM CM CM
"si* co in
ON co 00
TH
TH
S C§ ON
oo in r-
O TH O
•* ON o
CM CM O
•* VO CM
CM CM O
in in co
CM CM O
00 CM ON
CM CM O
^
• CH
o a 1
So S 2>
in CM TH ON
VO •* O O
in vo oo i>
CM TH TH TH
•* O O ON
>n c~- oo vo
CM TH TH TH
f^ f^^ C1^ C1^
CM TH TH TH
TH
s * * s
VO U") CO VO
ON O\ O ON
rH
B 8 TH cB
ON O CO OO O
O CO CO Tf in
ON in r~ i> vp
O CM CM •* -^
§ 8 8 £ $
vo >n >n TH c~-
TH CM CM in •<*•
.53
1 a & 1 a
1 S 1 1 8
s i S g s
rH VO
r> co
rH CS
i
i
i
£ in TH
TH TH CM
I> 00 ON
VO O in
TH TH CM
•g g CM
TH TH CM
O ON ON
CO vo C"^
o oo oo
TH
t-» CO rf
tO ON C3
TH TH
TH 2 TH
in CM o
S S 2
5 3 S
^H T^ ^H
O TH in
in co TH
C
» £P EL
8 5 g
2 £> a
•s -s a
151
<
s
-------
.
1
o
u
1
•o
^3
'£
s — s
g
e
o
i
cl
en
C
fi
0
OH
I"
b
T3
a
,2
•o*
3
2
^^
'c3
si
00
^
•o
"w
•5.
a
CO
^
s>.
e
.22
'o
1
3
2
1
oo
n c^
CM rH rH rH
0 ca
W '^ ' CQ £
0
o
CM
0
0
c5
^
CM
ON
0
CM
CM
O
r-5
vq
VO
O
rH
O
t~
0
ON
C3
Phoenix
oo
vd
oo
VO
oo
oo
t^
t>
i— i
CM
in
vd
CM
H
CM
•n
CO
VO
00
oo
in
m
CO
$
PH
o\ in
in vd
ON in
oo 1>
t> 10
•* oq
vd vd
rH CO
CO •*
1— 1 1— 1
in t>
CJ ON
CM rH
•5t VO
ON ON
i— ( . i— i
10 ON
vd vd
rH rH
0 ON Tt-
CO CO rH
ON Tf VO
CO ** r-5
rH OO ON
^ Tf rH
l> 10 t-
CM 10 CM
&
2 « °
* • -S a
— « nj &0
o o
a
VO
IO
VO
CO
ON
•n
vq
i—5
i— i
vq
i— i
ON
i— i
CO
rH
VO
O
CM
r-5
CO
rH
0
CM
Abilene
CM
oo
•<*
00
^H
o
l-H
i— 1
00
CM
f^
1— 1
ON
CM
IO
CO
CM
3
C*^
oo
rH
£-^
10'
00
VO
VO
10'
1
1
CO
oo
ON
CM
c5
rH
-*
OO
CM
ON
i— i
oo
•n
i— i
vq
CM
vq
CM
l>
CO
in
ON
•n
i— i
iri
Indianapolis
00
in
p
r-5
rH
CO
00
00
00
oo
ON
CM
in
rH
l>
CM
vq
rH
0
l-H
CM
CM
oo
CM
0
.a
1
1
CM
0
•rj-
CO
in
CO
VO
t>
oo
CO
l-H
00
o
1— 1
CM
CM
m
>*
rH
CM
rH
^f
CO
VO
co
vq
u
I
CH
CQ
in
vd
oo
00
CM
00
00
00
ON
rH
rH
f;
CM
rt
VO
in
rH
0
•*
^f
CO
CD
"*
CM
•*
Columbia
ON
0
f — ,
r-5
ON
CM
CM
CO
i
r-5
^J.
CO
f-
vd
^.
^
rH
CO
rH
CM
CM
CM
ON
l-H
^
CM
l-H
10
CM
•n
oo
•n
Tallahassee
od
oq
r-5
rH
^J.
r-5
rH
CM
ON
O
ON
i— 1
CO
CO
1— 1
CM
ON
rH
in
oo
rH
C3
CM
f^
CM
CO
vd
o
in
Lynchburg
in
ON
ON
Tj.
C3
rH
Tt-
00
p
£33
vd
rH
CM
CM
O
ON
rH
CM
•^
f^
CO
VO
vd
p
•n
Williamsport
-------
•g
•O
"o
.s
§
s
flj
0
O
•|j
O
§
S
j
I
CO
22
S
o
o
5
'E
1
.9 W
s> §2
:i s
.5
o
«>
§
d
O
1
^
O
CO
CO
s
Ov
4J o
•S K
VOrHOt^cOCMCM |-rJ--<*VOOr>OOCM |OOOO
rH OOO\CO'<* •<4-rHCOCOCMrH1OrH'
' rH ' CO ' rH !
i
CM^CMC^t-OiO 'COTtvOCOOOOOCM JiO^fVO
'rHrHt~-O\CMTfrH rHVOTfOOrH1 T'T*
rHVOOCO^O^ "CMCMCMrHCMVOOX IOOCMOO
COCMCOt^CO rH'cMCOrHTfCMCM' ->*COrH
I I I i T-H ' ' rH ' '
C~rHrfcOO\CMCT\ 'COT*COOAOQO'* 'f-iOO
rHrH VOCOCMrHlrHCOCMCN^lOCMrH
VOt^CMOCM^OO |OC~-rHVOrH'^-'^)- |r-VOCM
CM li i3 ;=i ^ .3. £i '^-HrHCMCOCMCOVO '^CMIO
v 1 1 1 1 III r-l 1
TfCTvOOt^OOlO 'VOCMVOVOONOOrH |rHTj-O\
2* :£ ^4 l< C§ ft) rl 'cOrplOvSvOlOCM 'OOrHIO
CM
CO
^
\OCOOOOT-Ht^-lOC**iQ\'^'flr-ITfCSOOt1-VOC^O
0s! CO CQ VO CO ^1 CO VO *O 00 00 VO OK VO VO ^H ^H ON Cj
^-( i— ( C^ ^f *~* i~t
t*^- ON C^- O\ ^3 OQ VO ON ^t" VO VO ^3 ^t" C^^ O3 r~i 00 VO t-H
co oi *o '^ i™^ vo vo *o ^" v^ vo vo vo i"** ON ON *o
1 T-H
1 CNCO T-HTj-'^'CO fs) VO r-l i— 4 VO ^H CN
X 1
S & S 3 s „ S
5 « eow0« ^S?^
0 _ ... a* O ., 'S O-c -a So 'o « S» e
1 1 s 1 1 a H » | * 1 1 1 1 1 1 ! 1
O'5'aSo'-1S&-
co>^ca.[i,PHCHlz;E-iQ
-------
UNITED STATES
Figure 1. Map of the United States and location of the study sites.
-------
US REGIONAL PRODUCTION
WHEAT
SOYBEAN
Figure 2. Regions of wheat, maize and soybean production.
-------
TEMPERATURE (C)
SPOKANE
TEMPERATURE (C)
FAROO
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
TEMPERATURE (C)
DODGE CITY
TEMPERATURE (C)
DES MCHNES
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
TEMPERATURE (C)
ABILENE
TEMPERATURE (C)
MEMPHIS
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
OBSERVED
---oiss •
QFDL
-S-UKMO
Figure 3a. Observed and 2 x CO2 scenario temperature at selected sites in the U.S.
-------
200
PRECIPITATION (MM/MONTH)
SPOKANE
PRECIPITATION (MM/MONTH)
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
PRECIPITATION (MM/MONTH)
DODGE CITY
200
PRECIPITATION (MM/MONTH)
DES KOINES
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
JAN FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
200
PRECIPITATION (MM/MONTH)
PRECIPITATION (MM/MONTH)
MEMPHIS
0
JAN
FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
FEB MAR APR MAY JUN JUL AUQ SEP OCT NOV DEC
OBSERVED
---OISS -
QFDL
-e-UKMO
Figure 3b. Observed and 2 x CO2 scenario precipitation at selected sites in the U.S.
-------
CO
0
CO
LJJ
DC
LLJ
O
CD
CO
03
"^
CD
m
cO
03
CO CO
C\j T-^
-------
0
N
"CO
LLJ
DC
LLJ
O
I I I I
OJ
CO
U)
0)
0)
co co
o o
C/) CO C/J
000
X I
p p
S
;! o> oo
o
II
•o
o
o o
. CO CO
CO CO cO
CQ CQ CQ
O
i
CL
O
CM
o
C\J
0
CM
CL
O
CM
CL
CM
O
CM
CL
o
CL
o
CL
8
tf
I!
<= '-
00
oooooooooo
CMr- T-CMCO^LOCO^
p|9!A
-------
o
cc
o
O
CM
DL
o
CM
Q_
CM
O
CM
QL
O
o
CM
o
CM
CL
CM
O
CM
Q.
O
O
CL
O
Q_
8
w
i
I
.a
t
i
J
.s
§1
>> -5
O S
ca 9
•a o
•a
O
•o
eo
E
O
CM
o
CM
O
CO
O
LO
O
CO
PI9IA
-------
20
15
10
S
0
-5
-10-
-15-
-20-
-25-
-30-
-35-
•40-
-45-
-50-
-55-
-60
Annual US Wheat Production
[ Without C02
WHhCO2
Current Production - 64 mfl MT
GISS
QFDL
UKMO
20
15
10
5
0
-5
-10
-15-
-20-
-25-
-30-
-35-
•40-
-45-
-50-
-55-
-60
Annual US Corn Production
| Without CO2
WHhCOZ
Current production -183 mil MT
GISS
GFDL
UKMO
Annual US Soybean Production
Without CO2
WXhCO2
Currant production - 49 mB MT
GISS
GFDL
UKMO
Figure 5. Change in annual national production of wheat, maize and soybean with the GCM climate
change scenarios.
-------
O
10
O
CM
CO
CO
O
O
CO
O
01
CO
CO
O
03
O
O
CM
CO
CO
CD
CL
Q_
O
CO
CO-
LU
.O
'l-l
a
c
I
W5
I
oo
00
3
CD
I
C5
5s
flS *T3
II
|i
§ a
a S
"" ts
c -S
6 ^
E
CD
-------
CO
C0.0>
8,1
£S
JC CO
O u.
o
T3 ^
05-8
•• BM^BHMH HtanB
CO
c
CD
CD o
.NJ CO
CC CO
~~ CO
O
o
CM
I
o
CO
•^
f **•
to
10
10
•o
o
CO
CO
o
O)
IO
10
IO
•o
a>
o
CO
CO
•o
a>
CC
o
CO
h-
CO
g
Q_
CO
O
CL
-------
POSSIBLE IMPACTS OF CLIMATE CHANGE ON MAIZE YIELDS IN MEXICO
Diana Overman, Max Dilley, and Karen O'Brien
Pennsylvania State University, Pennsylvania, USA
Leticia Menchaca
Centra de Ciencias de la Atmosfera, UNAM, Mexico
MEXICO-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Maize Production
Analysis of Climate Change Scenarios
Study Design
Site Characteristics
METHODS
Base Climate and Climate Change Scenarios
Soils and Management Variables
Crop Model Calibration and Validation
RESULTS AND DISCUSSION
Sensitivity Analysis of Baseline Yields to Inputs
Sensitivity Analysis of Maize Yields to Climate and CO2
Maize Yields with GCM Climate Change Scenarios
Maize Yields with the GISS Transient Scenarios
Adaptive Responses
CONCLUSION
REFERENCES
MEXICO-2
-------
SUMMARY
This study used global climate models and a crop growth model to estimate the potential impacts of
climate change on maize yields in Mexico. Projected climate change caused simulated maize yields to decrease
dramatically in two main Mexican regions, but the magnitude of the decrease varied with climatic scenario and
with the initial set of management variables selected for the simulation. The decreases in yield were caused
primarily by temperature increases, which shortened the duration of the crop's life cycle, particularly the grain-
filling period. The duration of the crop's growth period became shorter, exerting a dramatic negative pressure
on yields. These decreases were slightly counteracted by the beneficial physiological CO2 effects, as simulated
in this study. A sensitivity analysis indicated that decreases in crop yields would be severe under global
warming unless irrigation expands, fertilizer use increases, or new varieties that are more tolerant of heat are
developed.
INTRODUCTION
Background
With production of 25 million metric tons per year, Mexico is one of the world's top fifteen cereal
producers. Although Mexico has exported cereals in the past, net cereal imports have exceeded 5 million
metric tons in recent years. Mexico imported $284 million worth of maize in 1987, $394 million in 1988, and
$441 million in 1989. The country is continually striving to support a growing population with an agricultural
system that relies on relatively low and variable rainfall. For Mexico, an increase in temperature could
aggravate the existing nutritional and economic problems. More than one-third of Mexico's rapidly growing
population works in agriculture, a sector whose prosperity is critical to the nation's debt-burdened economy.
Although only one-fifth of Mexico's crop land is irrigated, this area accounts for half of the value of the
country's agricultural production, including many export crops. Many irrigation districts rely on small reservoirs
or wells that deplete rapidly in dry years. The remaining rainfed crop land supports many subsistence farmers
and provides much of the domestic food supply. Frequent droughts reduce harvest and increase hunger and
poverty in many areas of Mexico.
Maize Production
Maize is the most important crop grown in Mexico. In 1988, production totalled 14.4 million tons
(52% of the total production of major crops), grown on 11.5 million hectares of land (56% of the total crop
land). Almost every region in Mexico produces maize, but the states of Veracruz, Tamaulipas, and Michoacan
tend to be the largest producers. Yields are generally low, ranging from about 350 kg ha"1 in Quintana Roo
in 1986 to 3,670 kg ha"1 in Sonora in 1986. Average yields are 1,250 kg ha"1. Maize is produced in a wide
variety of environments, from highlands to coasts, dry to humid climates, and poor to fertile soils.
Maize was probably domesticated in Mexico, and many of its wild ancestors can still be found. It was
the staple crop of pre-Columbian populations and continued to be an important component of the diet under
Spanish colonialism. In contemporary Mexico, maize is the basis of subsistence and the primary cash crop for
the majority of poorer and smaller farms. It also provides the corn flour for the tortillas eaten with almost
every meal in both urban and rural Mexico. Each Mexican consumes approximately 250 kilograms of maize
per year. Most maize is rainfed. In recent years, 11% of the maize area has been irrigated, providing about
20% of the production. Thus, maize production is extremely vulnerable to climatic variation and drought. The
rapid increase in the Mexican population has contributed to the growth in the demand for maize in recent
MEXICO-3
-------
years. Coupled with environmental and economic problems, the situation has led the Mexican government to
import considerable amounts of maize in recent years.
In 1950, the expansion of the area planted to maize combined with the use of improved seeds and
fertilizer led to a large increase in Mexican maize production. Mexico was one of the leading participants in
the "Green Revolution" and considerable attention has been focused on breeding improved varieties of both
rainfed and irrigated maize. The use of improved seeds and fertilizer varies widely in Mexico. In some regions,
such as the Valley of Puebla, improved seeds, agricultural chemicals, and agricultural extension work has led
to much higher yields in good years. However, in other regions, some farmers cannot afford additional
chemical fertilizer or irrigated land, and yields remain low.
Analysis of Climate Change Scenarios
Liverman and O'Brien (1992) analyzed the possible impacts of global warming on climate and water
availability in Mexico. Their study used five climate models to construct scenarios for temperature,
precipitation, and evaporation at a range of stations in Mexico, and highlighted some of the problems that
arise from using climate models to predict future climate. Their model versions are the same as those used
in this study, and they used the same methodology that we have used to construct the climate scenarios.
Climate models are unable to adequately represent current climate in many parts of Mexico, and climate
change scenarios created with different models vary widely. The principal conclusion of the study is that
Mexico is likely to be warmer and drier under global warming conditions. With all of the climate models
analyzed, potential evaporation increased and moisture availability decreased, even in those cases where the
model projected an increase in precipitation. A sensitivity analysis of our calculations of evaporation and
moisture deficit indicates that water availability could increase only with higher rainfall and relative humidities,
or significantly less solar radiation and windiness.
Study Design
In this study, we used a crop simulation model (CERES-Maize) (Jones and Kiniry 1986; IBSNAT
19S9) and climate change scenarios generated from three General Circulation Models to analyze the possible
impacts of global warming on Mexican maize yields. The CERES-Maize model uses daily weather data to
estimate the effects of climate on yields of rainfed or irrigated maize. The model is physiologically based,
simulating the influence of solar radiation, temperature, nutrients, and water availability on maize development
through major phonological stages.
It has been suggested that crops will benefit from higher levels of atmospheric carbon dioxide,
resulting in higher yields. The CERES-Maize model also includes an option to simulate the direct
physiological effects of elevated CO2 levels on crop growth and the efficiency of water use (Acock and Allen
1985). However, the model does not account for losses from, and changes in, pests and diseases. It may,
therefore, overestimate yields.
Site Characteristics
Our choice of sites for this study was limited by the availability of long time-series of daily climate
data, and by the need to find sites where experimental data were available for calibrating and validating the
CERES model. With the cooperation of CYMMIT, we were able to obtain daily climate data and limited crop
experimental data for Poza Rica, a site on the Gulf Coast of Mexico, (+20°32' N, 97°26' W; 60 m altitude)
and Tlaltizapai, a site in the highlands south of Mexico City (+18°41' N, 99°08' W; 940 m altitude) (Figure
1). These sites are representative of important maize-growing regions in Mexico. The Tlaltizapan site is
MEXICO-4
-------
representative of maize in the central plateau and might be used to approximate conditions in the states of
Mexico and Puebla. Poza Rica is more typical of the Gulf coast states such as Veracruz, Tabasco, and
Tamaulipas.
METHODS
Base Climate and Climate Change Scenarios
We were able to obtain climate data from 1973 to 1989 for Tlaltizapan and Poza Rica, with missing
data for 1986 at the first site. Because observed solar radiation data were only available for a small number
of years, we used sunshine hours to generate the solar radiation for the two sites using WGEN (Richardson
and Wright 1984).
The climate change scenarios for each site were generated from three GCMs: GISS (Goddard Institute
for Space Studies, Hansen et al. 1983); GFDL (Geophysical Fluid Dynamics Laboratory Model, Manabe and
Wetherald 1987); and UKMO (United Kingdom Meteorological Office Model, Wilson and Mitchell 1987).
Changes in temperature, precipitation, and solar radiation projected by the GCMs were applied to the
observed (base) climate variables to create the climate change scenarios for Tlaltizapan and Poza Rica. The
base climate sets were also modified with the GISS transient model to create transient scenarios for the 2010s,
2030s, and 2050s (Hansen et al. 1989). Table 1 presents the seasonal and annual temperature and precipitation
changes projected by the GCMs.
The mean annual observed temperature at Tlaltizapan is 24°C; and the average observed rainfall is
about 96 mm from November to May, and 729 mm from June to October. The mean annual temperature at
Poza Rica is 23°C, and the average rainfall is about 380 mm from November to May, and 834 mm from June
to October.
Soils and Management Variables
The soils in Tlaltizapan are silt-clay soils 1.0 to 1.8 m deep ("isothermic udic pellusterts"). Because
we were unable to obtain detailed soil information, we used an isothermic udic pellusert from Guatemala from
a soil data base. Soils in Mexico tend to be poor in nutrients and extremely dry at the beginning of the growing
season. Therefore, we modified the initial soil conditions to represent low soil nitrogen and water content. We
also reduced the organic content of the soil by reducing the amount of crop residue in the initial conditions.
In Poza Rica, the site with a lower altitude, soils are typically sandy loam of moderate depth. Because
of the lack of detailed soil information, we used a generic medium-depth sandy loam soil from the IBSNAT
data base and modified the initial soil conditions to reflect low soil fertility and water content at the beginning
of the growing season.
In rural Mexico, maize is typically grown by poorer farmers who cannot afford chemical fertilizers.
Rather than use the high levels of fertilization and plant densities practiced at the experimental stations, we
selected agronomic conditions more typical of average farmers in the region. We used 50 kg of nitrogen
fertilizer at planting in the model runs arid soils of relatively low fertility and moisture content to represent
Mexican conditions. In addition, we have simulated maize yields in the absence of nitrogen limitations for
purposes of comparison. We used a planting date of May 20 at Tlaltizapan and June 15 at Poza Rica. In the
cases where we simulated irrigated maize, we assumed a 0.5 m irrigation depth at a rate of 50% efficiency.
Crop Model Calibration and Validation
The CERES-Maize crop model (Ritchie et al. 1989) was used for the simulation of maize yield,
MEXICO-5
-------
season length duration, and crop growing season evapotranspiration. To calibrate the model for the varieties
and conditions of this study, we used several databases obtained from CYMMIT experimental plots between
1987 and 1989. Five genetic coefficients may be modified to calibrate the model phenology and yield with
observed phenology and yield data (Table 2). We began by running the model using the genetic coefficients
of the SUWAN-1 variety that was listed in the IBSNAT database (IBSNAT 1989). This variety has genetic
characteristics similar to some of the Mexican varieties grown at both Tlaltizapan and Poza Rica. We
compared the predicted values for anthesis date, maturity date, yield, kernel weight, grains per square meter,
grains per ear, and total biomass, with experimental data from CYMMIT at both stations. Coefficients P5 and
G3 modify the grain-filling duration and kernel-filling rate, respectively. We repeatedly ran the model and
adjusted P5 (within the ranges suggested by the CERES-Maize User's Guide) to obtain as close a match as
possible between the simulated and predicted phenologies. We reduced the coefficients G2 (maximum number
of kernels per plant) and G3 (kernel-filling rate) to get closer to observed yields. Since we had data for more
than one experimental plot at each station, we validated our calibrated variety with data from a second plot
and made small modifications to the coefficients based on the results. The results of the validations are shown
in Table 2.
RESULTS
Sensitivity Analysis of the Baseline Yield to Inputs
Table 3 shows the influence of management variables on simulated maize yields under the baseline
climate. Simulated yields at the two sites were rather sensitive to our assumptions about soil, fertilizer use,
management practices, sowing date, and initial conditions. Our baseline results for 16 years at Tlaltizapan
produced an average rainfed yield of 4.02 t ha"1 with nutrient stress conditions. Without nitrogen constraints,
rainfed yields are 4.55 t ha"1. At Poza Rica, 17 years of baseline climate produced an average rainfed yield of
3.18 t ha"1 with nutrient stress. Without nitrogen constraints, rainfed yields average 3.911 ha"1. The sensitivity
of yields to our assumptions of nutrient-poor, dry soils and limited fertilizer use is clear. Yields increased at
both sites with increases in fertilizer application and with deeper, moister, more fertile soils.
A delay in the planting date produced lower yields at both sites, while planting 10 days earlier
reduced yields at Tlaltizapan, but slightly increased yields at Poza Rica. The unmodified SUWAN variety
produced slightly higher yields at both sites than the calibrated variety. Yields were lower with a planting
density of two plants per square meter and higher at Tlaltizapan with four plants per square meter.
These variations in responses to current practices suggest that a full set of practices should be tested
for better projections of the impacts of climate change. Due to lack of detailed site information, we made
some rather arbitrary decisions about soils, initial conditions, planting dates, and genetic coefficients.
Sensitivity Analysis of Maize Yields to Climate and CO2 Changes
We also conducted a sensitivity analysis of the CERES-Maize model to changes in climate of +2°C
and +4°C, +20% and -20% precipitation, and 330 ppm and 555 ppm of CO2 (Table 4). At Poza Rica, base
yields decreased with temperature increases and rainfall decreases; base yields increased under the +20%
rainfall scenario. The simulation of the physiological CO2 effects results in a slight increase of scenario yields
in all cases.
Maize Yields under GCM Climate Change Scenarios
The effect of GCM climate change scenarios on simulated maize yields at Tlaltizapan and Poza Rica
MEXICO-6
-------
is presented in Table 5; For1 each climate scenario, yields were simulated under conditions with nutrient stress
(to represent the actual-situation) and under conditions without nutrient limitations. The physiological effects
of elevated atmospheric CO2 (555 ppm) were evaluated in all simulations. In this experiment we assumed that
the planting dates and varieties remain unchanged under the climate change scenarios.
In Tlaltizapan, base rainfed yields decreased under all climate change scenarios, even when the
physiological effects of GO2 were included in the simulation. The decreases in yield ranged from 20% for the
GFDL scenario, to 61% for the UKMO scenario. These low scenario yields occurred because the higher
temperature accelerates the phenology and the crop matures faster with less time for gain-filling. There is also
drought stress early in the growing season under the scenarip conditions.
Baseline rainfed yields of maize at Poza Rica also decreased with global warming, even with the
physiological effects of 555 ppm CO2- The yield decreases at this site were smaller than those at Tlaltizapan,
but nevertheless, they are significant ranging from -6% (GISS) to -26% (UKMO).
Maize Yields under the GISS Transient Scenarios
Yields of rainfed maize have also been simulated at Tlaltizapan and Poza Rica for three GISS
transient scenarios, corresponding to the decades of the 2010s, 2030s, and 2050s (Table 6). At Tlaltizapan,
yields decreased consistently as the climate changes, while at Poza Rica the yield responses varied through
time. The latter result indicates that the trajectory of agricultural response to climate change may be non-
linear.
Adaptive Responses
The sensitivity analyses of the baseline yields to management variables indicate the importance of
some of the current uncertainties in estimating the impacts of future climate changes. In the previous sections
we have presented results assuming no change in the average planting date or in the varieties used. But clearly,
as climate changes, farmers and governments will try to adapt to change. In this section, we examine whether
changes in maize varieties and nutrient levels may offset the negative impacts of a warmer climate.
The amount of fertilizer used is critically important to the maize yields predicted by the CERES-Maize
model (Tables 3 and 5). At present, many Mexican producers can only afford to use small doses of nitrogen
fertilizer at planting. If more fertilizer becomes available to more farmers, some of the yield reductions under
the climate change scenarios might be offset (Table 5). With full fertilization, simulated maize yields increased
under irrigation at Poza Rica, but the increase does not fully offset the negative effects of climate change on
maize yields. However, given the environmental and economic constraints and trends in agricultural inputs in
Mexico, unlimited water and nutrients are extremely unlikely.
We also simulated planting varieties more suited to the new climate conditions, since changing the
variety or the genetic characteristics of crops may also help farmers to adapt to global warming. For example,
planting the unmodified SUWAN or PIONEER varieties, rather than the calibrated varieties, produced slightly
higher yields under the GISS scenario at both sites.
CONCLUSION
In this study we found that rainfed and irrigated maize yields in Mexico decreased with global
warming, in spite of differences among the climate model results. Sensitivity analyses indicated that the
decreases in crop yields will be severe under global warming unless irrigation expands, use of fertilizer
increases, or new varieties of plants are developed.
MEXICO-7
-------
REFERENCES
Acock, B., and L.H. Allen Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.), Direct Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of
Energy. DOE/ER-0238. Washington, D.C. pp. 33-97.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
Review, Vol III, No. 4: 609-662.
Hansen, J., I. Fung, A. Lacis, 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, 93(D8):9341-9364.
IBSNAT (International Benchmark Sites Network for Agrotechnology Transfer Project). 1989. Decision
Support System for Agrotechnology Transfer Version 2.1 (DSSAT v2.1). Dept. Agronomy and Soil Sci.,
College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Jones, C.A., and J.R. Kiniry. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development.
Texas A&M Press. College Station. 194 pp.
Liverman, D., and O'Brien. 1992. Global Warming and Climate Change in Mexico. Global Environmental
Change, in press.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.
Richardson, C.W., and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
MEXICO-8
-------
Table 1. Temperature differences (°C) and precipitation changes (%) between GCM lxCO2 and
2xCO2 climate change scenarios.
Temp. Diff. (°C) Precip. Changes (%)
Site/GCM
Tlaltizapan
GISS
GFDL
UKMO
Poza Rica
GISS
GFDL
UKMO
Spr.
4.1
2.3
3.7
4.1
2.3
4.1
Sum.
4.1
2.5
3.5
4.1
2.5
4.0
Fall
4.3
2.9
2.9
4.3
2.9
4.0
Win.
4.5
2.9
3.0
4.5
2.9
3.6
Annual
4.2
2.7
3.3
4.2
2.7
3.9
Spr.
-6
14
-33
-6
14
-5
Sum.
-6
-19
42
-6
-19
-13
Fall
-2
-16
11
-2
-16
16
Win.
-3
10
-24
-3
10
-29
Annual
-4
-9
5
-4
-9
-6
MEXICO-9
-------
Table 2.
Simulated and observed (1989) yield and season length at Tlaltizapan and Poza Rica.
Site
Experiment
Yield (T/Ha)
Obs. Sim.
Season Length (days)
Obs. Sim.
Tlaltizapan
Tlaltizapan
Poza Rica
Poza Rica
Poza Rica
1
2
1
2
3
6.20
5.83
5.27
6.98
6.56
6.58
5.72
9.12
6.53
7.50
122
122
112
138
138
127
127
109
133
133
Coefficients of SUWAN maize variety and modified coefficients of the varieties used at Tlaltizapan (TLA) and
Poza Rica (PR):
SUWAN: PI (380); P2 (0.6); P5 (780); G2 (750); G3 (7.0)
TLA: PI (320); P2 (0.6); P5 (780); G2 (500); G3 (8.0)
PR: PI (350); P2 (0.6); P5 (780); G2 (550); G3 (8.5)
MEXICO-10
-------
Table 3.
Sensitivity of CERES-Maize yield to changes in management variables.
Management variables
Simulated Yield (T/Ha)
Tlaltizapan Poza Rica
Base scenario*
Planting 10 days later
Planting 10 days earlier
Variety SUWAN
100 kg ha'1 N fertilizer
200 kg ha"! N fertilizer
More productive soil**
V. good initial conditions
Medium initial conditions
2 plants m"2
4 plants m'2
4.02
1.41
3.94
4.75
4.56
4.55
4.05
4.66
4.38
2.86
4.41
3.18
1.12
3.26
3.31
3.81
3.91
4.75
4.53
3.66
2.70
3.16
*Tlaltizapan planting date May 20; Poza Rica planting date June
**Medium silt clay at Tlaltizapan; Deep sand at Poza Rica.
15.
MEXICO-11
-------
Table 4.
Sensitivity of CERES-Maize yield to changes in temperature, precipitation, and atmospheric
CO2.
Scenario
Poza Rica (t ha'1)
BASE
T+2 C, 330 ppm CO2
T+2 C, 555 ppm CO2
T+4 C, 330 ppm CO2
T+4 C, 555 ppm CO2
P+20%, 330 ppm CO2
P+20%, 555 ppm CO2
P-20%, 330 ppm CO2
P-20%, 555 ppm CO2
3.18
2.90
3.10
2.58
2.89
3.26
3.06
3.30
MEXICO-12
-------
Table 5.
Effect of GCM climate change scenarios on CERES-Maize yield.
Simulated Yield (T/Ha)
Nutrient stress* No nutrient limitations
Scenario
BASE
GISS, 330 ppm CO2
GISS, 555 ppm CO2
GFDL, 330 ppm CO2
GFDL, 555 ppm CO2
UKMO, 330 ppm CO2
UKMO, 555 ppm CO2
Tlaltizapan
4.02
3.11
—
2.74
3.20
2.34
—
Poza
Rica
3.18
2.76
2.97
2.26
2.70
1.83
2.35
Tlaltizapan
4.49
3.49
3.77
3.07
3.47
3.92
3.93
Poza
Rica
3.98
2.80
3.30
2.62
3.18
1.98
2.67
*Actual conditions
MEXICO-13
-------
Table 6.
Effect of GISS transient scenarios on CERES-Maize yield.
Scenario
Simulated Yield
(T/Ha)
Tlaltizapan Poza Rica
BASE, 330 ppm CO2
GISS 2010s, 405 ppm CO2
GISS 2030s, 460 ppm CO2
GISS 2050s, 530 ppm CO2
4.02
3.67
3.32
2.95
3.18
3.27
2.48
2.95
MEXICO-14
-------
MEXICO
Figure 1. Map of Mexico and location of the sites selected for the study.
-------
LU
o:
=)
CEO
LLJQ
r
LU
RXXXXXXXXXXXXXXXXXXS^^
o
Z)
Q
LJ_
O
CO
CO
o
LU
S8
CQ
£ * A
0 S33H93Q
-------
o
Is
LJJ
LLI
or
DL
LU
R%gggg?Sggg8gg%888SSg?SSSSggg$^^
i .Q
Q
u_
O
CO
CO
o
LLI
S8
DQ
IAI1AI
-------
LU
DC
LUO
LU
Q
Li-
CD
CO
CO
O
LU
58
00
-------
o
I
9=0
00
s
LJJ
Q
U_
O
CO
CO
o
LLJ
S
CD
a
-------
-------
SECTION 3: SOUTH AMERICA
-------
-------
POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE FOR BRAZILIAN
AGRICULTURE: APPLIED SIMULATION STUDIES FOR WHEAT, MAIZE,
AND SOYBEANS
Otavio Joao Fernandes de Siqueira
National Research Center for Wheat
CPACT/EMBRAPA, Brazil
Jose Renato Boucas Farias
National Research Center for Soybean
CNPS/EMBRAPA, Brazil
Luis Marcelo Aguiar Sans
National Research Center for Maize and Sorghum
CNPMS/EMBRAPA, Brazil
Research partially supported by CNPq (National Council for Technical and Scientific
Development), Contract 040-2941/90.
BRAZIL-1
-------
ACKNOWLEDGEMENTS
The authors address thanks to the following institutions: EMBRAPA (CNPT, CNPMS, CNPS,
CNPMF, CPAA, CPATU, CPATB, CPATSA, NMA), AC-SA/SP, IPAGRO-SAA/RS,
DNEMET/CRMC, CNPq, US EPA, US AID, and collaborators: Dr. Rogerio Remo Alfonsi, Dr. Ivo
Pierozzi Jr., Dr. Aristides CSmara Bueno, Dr. Jaime Ricardo Tavares Maluf, Dr. Fernando Silveira
da Mota, Dr. Moacir Antdnio Berlato, Dr. Delmar Pottker, Dr. Sirio Wietholter, Dr. Gilberto Rocca
da Cunha, Dr. Douglas Godwin, Dr. Cynthia Rosenzweig, Dr. Ron Benioff, Dr. Ana Iglesias. Also
to local support, Adriano de Mello e Silva and Pitagoras Ribeiro, and to family, Marisa, Otavio, and
Fabiana.
BRAZIL-2
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Agroecological Regions and Sites
METHODS
Baseline Climate
Sensitivity Scenarios
Climate. Change Scenarios
GISS Transient Scenarios
Crop Models and Inputs'
Calibration and Validation of the Crop Models
Limitations of the Study
RESULTS AND DISCUSSION
Sensitivity Analysis
Crop Changes under GCM Climate Change Scenarios
National Grain Yields
Transient Results
Adaptation Strategies to Climate Change
CONCLUSIONS
REFERENCES
BRAZIL-3
-------
SUMMARY
Wheat, maize and soybean production were simulated with the CERES-Wheat, CERES-
Maize and SOYGRO crop models to estimate the possible impact of global climate change at 13
sites in Brazil. Climate change scenarios for each site were created by combining observed climate
data with the output of the GISS, GFDL and UKMO General Circulation Models (GCMs).
Historical climate data were used as the base scenarios. The three GCMs project temperature
increases, changes in precipitation, and small variations of solar radiation. The sensitivity of the
simulated crop yield to changes in temperature, precipitation, and atmospheric CO2 was also tested.
Temperature increases resulted in lower grain yield, biomass, and season length for wheat
and maize, but soybean was less affected. The simulated physiological effects of CO2 partially
compensated for the yield decreases under the scenarios of climate change alone in wheat and maize,
and fully compensated in the case of soybean. Wheat yields declined at all sites under the three
GCM climate change scenarios, although the magnitude of the decreases varied with scenario and
region; the largest decreases occurred in the Central region (24 and 46% under the GFDL and
UKMO scenarios respectively) and in the Central South region (43% under the GISS scenario).
Maize yields also declined at all sites. The largest decreases were in the Northern region (26%
under the GISS scenario), while the decreases in the South and Central South region varied but
were less than 20%. In contrast to wheat and maize, simulated soybean yield remained the same (in
the Northern regions) or increased under the GCM climate change scenarios.
Adaptation strategies such as irrigation, changes in planting date, and increased nitrogen
fertilization helped improve yields, but not enough to compensate for all of the losses under climate
change. The crop simulation model was used to design a hypothetical more heat-resistant cultivar
that showed promising results for potential adaptation to warmer climates, but the feasibility of this
strategy still needs to be tested through breeding programs.
INTRODUCTION
The rising concentration of greenhouse gases in the atmosphere may lead to increased
global temperatures (IPCC 1990) and the study of the potential impacts of global climate change
on ecosystems and agriculture is a relatively new area of research (Smith and Tirpak 1989a, 1989b).
In Brazil, very few studies have been conducted in this area. Some results have been presented by
Mota et al (1984) using statistical models. In this study, we analyze the possible impact of global
climate change on wheat, maize, and soybean production in Brazil using crop growth simulation
models and climate change scenarios created with GCM results. These crops are the major
agricultural commodities in Brazil (along with rice), and cover approximately 25 million ha of
cultivated land. In addition, this study analyzes adaptation strategies to minimize the impact of the
projected climate changes on crop production.
Agroecological Regions and Sites
Thirteen sites were selected for the simulation study (Figure 1 and Table 1). The locations
of the sites range from latitude 30 degrees South to close to the equator. Nine sites are concentrated
in the subtropical and tropical/high-elevation regions and the other four sites are distributed among
the tropical, semi-arid, and equatorial/sub-equatorial regions. The sites represent the main
agricultural regions of Brazil, and they were selected based on previous agroclimatic studies (Mota
BRAZIL-4
-------
1989; Mota and Agendes 1986; Alfonsi et al 1981; Queiroz et al. 1979). Nine sites are located in the
most important agricultural region of the country, the South and Central-South regions. Almost
99% of the national wheat production and more than 80% of the maize and soybean production are
concentrated in these regions.
METHODS
Baseline Climate
The baseline climate is represented by the historical data available for each site during the
period 1951-1980 (Table 2) and includes daily maximum and minimum temperatures, precipitation,
and hours of sunshine. Daily solar radiation data were generated using hours of sunshine, and the
results were compared to actual data where available.
The lowest mean annual temperature is found in Vacaria, which has the most significant
seasonal temperature variation. In contrast, Belem shows the highest mean annual temperature with
the smallest variation during the year. Precipitation varies among the sites, with the highest
precipitation occurring in the northern sites and the lowest in the northeast. There is also variability
in seasonal precipitation in some central sites, where the lowest monthly rainfall occurs during the
winter. Seasonal differences in solar radiation are more apparent in the southern sites.
Sensitivity Scenarios
To analyze the sensitivity of the crop models to temperature, precipitation, and CO2 levels,
sensitivity scenarios were created by combining step changes in the climate variables (0, +2°C,
+4 °C temperature changes; 0, +20%, -20% precipitation changes). The physiological effects of 555
ppm CO2 were also considered for each scenario.
Climate Change Scenarios
This study used climate change scenarios generated by three equilibrium general circulation
models: Goddard Institute for Space Studies (GISS) (Hansen et al. 1983), Geophysical Fluid
Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987), and United Kingdom Meteorological
Office (UKMO) (Wilson and Mitchell 1987). These GCMs are three-dimensional models which
incorporate physical knowledge of the processes involved in the transfer of the energy among the
earth, oceans, and atmosphere.
The climate change scenarios for each site were created by applying the changes between
the lxCO2 and 2xCO2 monthly GCM-simulated climate variables (differences were used for
temperatures and ratios for precipitation and solar radiation) to the corresponding daily baseline
climate variables. The scenario changes in mean annual temperature, precipitation and solar
radiation are shown in Table 3 for each site.
The annual temperature increases range from +2°C to +6°C, but there are significant
differences among the scenario projections. In general, the largest increases correspond to the
UKMO scenario. The sites included in the Central-South region show the largest temperature
changes from March to November, which is the growing period of wheat. For the sites in the
Northeast region, the UKMO scenario projects higher temperatures in the winter (June to August).
Under the climate change scenarios, precipitation projections vary greatly, especially for the
more southern sites. For the most part, the projected annual precipitation increased in comparison
BRAZIL-5
-------
with the current (observed) climate. The GFDL scenario projects the largest precipitation increases
for sites in the South region from September to November, and for sites in the Central-South region
from March to May. The precipitation projected by the GFDL and UKMO scenarios for December
for most of the Southern sites is lower than the current precipitation, indicating a higher probability
of drought problems for summer crops. Smaller annual increases in precipitation are projected for
the sites in the Northeast and North regions, where the UKMO scenario projects decreases of about
10% to 15%. The precipitation reductions projected by the UKMO scenario during the summer
(December to February) and winter (June to August) might bring additional stress for the crops in
the northern sites, especially considering the currently low water supply in the winter.
GISS Transient Scenarios
The GISS transient scenarios were used to assess the effect of gradual changes in climate
for the decades of the 2010s, 2030s, and 2050s on crop production (Hansen et al. 1988). The
atmospheric CO2 concentrations considered were 405 ppm, 460 ppm, and 530 ppm, respectively, for
these decades.
Crop Models and Inputs
Crop-growth models developed by IBSNAT (Jones et al. 1990) were selected for this
simulation study: CERES-Wheat version 2.10 (Godwin et al. 1989; Ritchie and Otter 1985), CERES-
Maize (Jones and Kiniry 1986), and SOYGRO (Jones et al. 1988). The IBSNAT crop models
simulate plant development and growth by integrating soil, plant, climate (daily maximum and
minimum temperatures, precipitation, and solar radiation), and .management factors.
The IBSNAT crop models include an option to simulate the direct physiological effects of
CO2 atmospheric concentrations on plant photosynthesis and water use, based on experimental
results (Rose 1989; Curry et al 1990). The photosynthetic enhancement of 555 ppm CO2 is 1.17 for
wheat, 1.21 for soybeans, and 1.06 for maize. Increases in stomatal resistance are also simulated.
Soil data were obtained from regional soil survey studies (Brasil 1971, 1973; Larach et al.
1984; Oliveira et al. 1984; Mothei et al. 1979; Santos et al 1983). Soils were chosen to represent the
local soils of the sites selected. The soil profile for Passo Fundo (crop model calibration site) was
created with local data.
The cultivars, plant population, and planting date used as input for the crop growth
simulation models are shown in Appendix A. The information was obtained from published crop
management reports (QueiiozetaL 1979; Miyasaka and Medina 1981; Vernetti 1983; Reuniao 1987;
CSBPT 1988; CCSBPT 1989; CCBPT 1988).
Although there are water deficits in some regions, almost all maize and soybean crops are
cultivated without irrigation. For winter wheat, irrigation is recommended for some regions,
including Campinas, Campo Grande, and Sete Lagoas. In Campo Grande, wheat could be produced
in a rainfed system, if it is planted at a different planting date.
Calibration and Validation of the Crop Models
CERES-Wheat. The local wheat cultivar BR 14 was chosen for most dryland sites, and its
genetic coefficients were determined using experimental field data. The cultivar ANZA, with the
original genetic coefficients included in the DSSAT database, was selected for the irrigation sites.
The model was validated in Passo Fundo with data from several field experiments, (Figure 2) and
BRAZIL-6
-------
there is a satisfactory agreement between the observed and estimated grain yields (Siqueira 1991).
In Brazil, the CERES model has also been validated for wheat in Sao Paulo (South region)
(Anunciacao and Liu 1991).
CERES-Maize. The genetic coefficients for the local cultivar used in the South and Central-
South regions (PIO 3230) were determined by comparing data from a field experiment conducted
in Taquari involving irrigation, nitrogen, and population levels (Matzenauer et al. 1988). Figure 2
shows selected results of the validation. Cultivar SUWAN-1, with the original genetic coefficients
included in the DSSAT data base, was used for the other regions because the simulated results are
in agreement with the regional observed crop parameters.
SOYGRO. The genetic coefficients of the cultivar DAVIS were calibrated using data from
a field experiment conducted in Passo Fundo in 1989, and the model was validated with data from
several field experiments (Siqueira and Berg 1991) (Figure 2). The cultivars VICOJA and JUPITER
were used for the sites in the Northeast and North regions. In Belem, simulated and observed yields
and anthesis dates show a close relationship.
Limitations of the Study
The crop models have not been validated in all of the regions in this study. Technology and
land use are assumed to be constant, even though it is certain that they will change in the future.
The direct physiological effects of CO2 on crop development and yield may be different than the
simulated effects.
The GCM climate change scenarios do not include changes in climatic variability that might
represent a very important factor for crop production, especially in the more vulnerable regions. The
spatial resolution of the climate change scenarios as created in this study is low.
RESULTS AND DISCUSSION
Sensitivity Analysis
A sensitivity analysis was conducted at each site to evaluate the effect of step changes in
temperature and precipitation on wheat, maize, and soybean (Tables 4, 5, and 6). For wheat, an
increase in temperature resulted in significant reductions in crop season length and grain yields
(Table 4). An increment of +4°C resulted in a shortening of the crop season length by about 15%,
and a 40% to 50% decrease in grain yields. With the physiological effects of 555 ppm CO2 on the
crop included in the +4°C scenario, wheat yield increased in comparison with the +4°C scenario
alone, but the effects did not completely compensate for the negative impact of higher temperature
on yields.
For maize, the effect of warmer temperatures on grain yield varied among the regions
(Table 5), ranging from about -20% in the South and Central-South to -28% in the Northeast
region. Warmer temperatures reduced the maize crop season length by an average of 15%, but the
effect was greater for the more southern latitude sites. There were no significant effects on maize
grain yield from increased precipitation except in the Northeast region. Lower precipitation may be
beneficial for maize in the North region and detrimental in the Northeast. The physiological effects
of 555 ppm of CO2 were smaller on maize grain yields than on wheat yields, as expected from the
lower response of maize, a C4 crop, to higher CO2.
BRAZIL-7
-------
Temperature and precipitation changes had a different effect on soybean yields than on
maize and wheat yields (Table 6). A 2°C temperature increase with the physiological effects of
doubled CO2 resulted in small increases in yield. In addition, the temperature increase did not
significantly affect the length of the crop season. With the physiological effects of CO2, soybean
yields generally increased under the scenarios with higher temperatures.
Crop Changes under GCM Climate Change Scenarios
The results of the simulation of wheat, maize, and soybean growth under the GCM climate
change scenarios are shown in Table 7. Tables 8, 9, and 10 and Figure 3 show the results for each
region.
Wheat. GCM scenarios projected wheat biomass, grain yield, and season length decreases
in comparison with baseline data for all sites (Table 7). Wheat yield reductions were the largest
with the UKMO scenario. All scenarios projected a shorter wheat crop season, especially for the
South region. These changes were driven by the temperature increases of the scenarios, as described
in the sensitivity analysis.
Figure 3 shows the percent change in wheat grain yield for each region under the different
climate scenarios. With the physiological effects of CO2, the negative effect of the climate change
alone was partially diminished (Figure 3). Wheat yields in the Central region were most vulnerable
to future climate changes under the GFDL and the UKMO scenarios: yield reductions were near
24 and 46%. Under the GISS scenario, losses were projected to be about 43% for the Central-South
region. For the most part, the South region appeared to be less vulnerable to the climate changes,
with projected average losses in yield of 22%.
Maize. All GCM scenarios projected reductions in maize biomass production, grain yield,
and season length, when compared to the present climate (Tables 7 and 9 and Figure 3). Decreases
in season length under the climate change scenarios varied in each region, but they averaged about
15%.
The effect of climate change on crop season length and yield was a consequence of the
increases in temperature projected by the GCMs. With the physiological effects of CO2, the
projected reductions in grain yield were diminished in comparison with the yields projected under
the climate change scenarios alone. Under the GFDL and the UKMO scenarios with physiological
CO2 effects, the largest decreases in yield were in the South and the Central-South regions, ranging
from 11% to 20%. Under the GISS scenario, the largest reductions were found in the North region
(24%).
Soybean. In general, reductions in soybean biomass and grain yield were smaller than
reductions in wheat and maize projected by the GCM scenarios (Tables 7 and 10 and Figure 3).
With the GCM scenarios alone, there were soybean yield and biomass reductions in almost all of
the regions. Mean reductions in grain yield varied from 5% to 31% (Table 7). W i t h the
physiological effects of CO2 on yield, the SOYGRO model simulated significant yield increases: gains
averaged 22% (Table 7). The results were consistent in all regions except the Northeast where yields
decreased, even when the physiological effects of CO2 were considered (Figure 4). A slightly shorter
crop season length was simulated for soybean with all GCM scenarios for the sites in the South and
the Central-South regions, but this effect was very small in comparison with the shorter crop season
of wheat and maize under the same conditions.
BRAZIL-8
-------
These results agree with the aforementioned sensitivity analysis and indicate that soybean
production in Brazil might not be as adversely affected by climate change as wheat or maize. The
more positive results projected for soybean, however, were dependent on the beneficial physiological
effects of CO2 concentration.
National Grain Yields
The effects of the climatic changes on national crop production wheat, maize, and soybean
yields were estimated by aggregating the regional results weighted by cultivated area. For example,
the yield changes from a region that represents 50% of the national cultivated wheat area was given
a weight of 50% in the national results (Table 11). All results reported include the beneficial
physiological CO2 effects on crop yield. Crop management, technology, and distribution of cultivated
land were assumed to be constant, although these will change in the future.
Wheat is currently grown on 3.6 million ha in Brazil. Using the changes in regional wheat
yields under the GCM scenarios and the contribution of these regions to the national wheat area,
we estimate that the possible impact of climate change scenarios on national wheat production
would be large (reductions of 33%, 18% and 34% under the GISS, GFDL, and UKMO scenarios,
respectively). Although significant reductions in wheat yield are projected for the Central region,
the impact of these regional reductions on the national yield is not highly significant because of the
small acreage that is cultivated (1%) in that region. Maize is the most widely cultivated crop in
Brazil, now cultivated on about 22 million ha. National maize yields were reduced 11%, 11%, and
16% under the GISS, GFDL, and UKMO scenarios, respectively.
Soybean is the second most widely cultivated crop in Brazil, occupying about 15.5 million
ha. Only in the Northeast region under UKMO were losses projected for simulated soybean yields.
In all other regions, soybean yields increased under the conditions of climate change, as simulated
in this study. The estimated changes in national yields under the GISS, GFDL, and UKMO climate
scenarios are +26%, +23%, and +18%, respectively. The 17% decrease in yield under the UKMO
scenario in the Northeast region did not have a large impact on the national yield estimate since
that region only contributes about 1% of the total soybean cultivation area in the country.
Transient Results
Wheat and maize yields and season lengths were simulated under the GISS transient
scenarios in Passo Fundo (South region) (Figure 4). Linear increases in temperature were projected
from 1990 to the year 2050. Wheat yield and season length decreased under GISS transient
scenarios; the rate of reduction decreased after the year 2030. In contrast to wheat, the simulated
maize grain yield increased until the year 2030 with the physiological effects of CO2 (460 ppm) and
then decreased. The length of the maize crop season decreased linearly over this same period.
Adaptation Strategies to Climate Change
Wheat and maize development and yield were affected by the climate changes projected by
the GCM scenarios, and the changes were largest under the UKMO scenario. This part of the study
aims to define possible alternatives that would compensate for the negative impact of the climate
changes on wheat and maize. The simulation was carried out in Passo Fundo because of the careful
calibration and validation of the CERES model at that site. Additional and preliminary strategies
were also simulated for maize and soybeans in
BRAZIL-9
-------
-------
Petrolina in the Northeast region, a vulnerable area to climate change.
Crop management alternatives (irrigation, new cultivars, and nitrogen management) were
evaluated as possible adaptation strategies for wheat under the climate change conditions (Figure
13). The UKMO scenario was run at Passo Fundo with various combinations of wheat genetic
coefficients and management practices. When tested separately, the adaptation strategies failed to
improve yields significantly. Only when several high-cost adaptation strategies (irrigation and
nitrogen management) were combined were wheat yields restored to 90% of their baseline values.
The changes in maize crop management that were tested (irrigation and changes in planting
date) did not compensate for the decrease in maize yield under the UKMO scenario in Passo Fundo
(South region). The crop growth model can test the performance of a hypothetical new cultivar
under the conditions of climate change. The development of a hypothetical new cultivar with a
different P5 "genetic coefficient" would improve maize yield production under the climate change
scenarios. The P5 coefficient is the growing degree days from flowering to maturity (Ritchie et al
1989). A higher P5 coefficient extends the simulated duration of the grain-filling period under
warmer climate conditions. An increase of 20% in the actual value of the P5 coefficient compensated
for the projected grain yield losses caused by the UKMO scenario at Passo Fundo. The feasibility
of breeding for extended grain-filling periods should be further explored by crop geneticists.
At Petrolina, an improvement in the crop management practices such as irrigation and
increased nitrogen fertilization could compensate for the yield decreases under the UKMO scenario.
However it is important to notice that in this case, base yields also increase substantially. At this
site, irrigation also increases soybean yields under the base case and the UKMO scenario and would
fully compensate for any negative impact of climate change on soybean yields in the vulnerable
Northeast region.
CONCLUSIONS
Projected climate change with the physiological effects of CO2 reduced modeled wheat and
maize yields significantly, while soybean yields increased. Most of the wheat and maize yield losses
were associated with a significant shortening of the growing season due to increased temperatures,
which allowed less time for biomass accumulation. In contrast, soybean growing season was not
reduced significantly, because soybean responds to temperature either positively or negatively
depending on phenological stage. The variable season length response probably contributed to the
more positive soybean yield responses to climate change. Soybean temperature responses at a wide
range of sites require further study.
Adaptation strategies such as irrigation, changes in planting date, and increased nitrogen
fertilization helped improve wheat and maize yields, but not enough to compensate for all of the
losses projected for the climate change scenarios. The development of a hypothetical new, more
heat-resistant cultivar showed promising results, although the feasibility of this strategy still needs
to be tested through breeding programs.
These results imply substantial reduction in the national production of wheat and maize and
increases in the production of soybean under climate change. Differential impacts on the country's
regions are projected, with the Northeast especially vulnerable to maize and soybean yield decreases
and the Central region vulnerable to wheat yield declines.
BRAZIL-10
-------
REFERENCES
Alfonsi, R.R., M.J. Pedro Junior, and AP. de Camargo. 1981. Zoneamento agroclimatico da soja
no Brasil. In Miysaka, S. and Medina, J.C., eds. A soja no Brazil. s.L, Banco do Estado de
Sao Paulo/FINEP-FNDCT/ Ministerio da Agricultura/Ministerio da Industria e
Comercio/Coqp. Agricola de Cotia.
Anunciacao, Y.M.T., and W.T. Liu. 1991. Estimativa da produtividade do trigo em campo
experimental, utilizando o modelo fisiologico CERES-Wheat V.2.10. In Congresso Brasileiro
de Agrometeorologia, 7th. Vicosa. Universidade Federal de Vicosa.
Brasil. 1971. Ministerio da Agricultura. Departamento Nacional de Pesquisa Agropecuaria. Divisao
de Pesquisa Pedologica. Levantamento de reconhecimento dos solos do sul do estado do Mato
Grosso. Rio de Janeiro. (Tech. Bull. 18).
Brasil. 1973. Ministerio da Agricultura. Departamento Nacional de Pesquisa Agropecuaria. Divisao
de Pesquisa Pedologica. Levantamento de reconhecimento dos solos do estado do Rio Grande
do Sul Recife. (Tech. Bull. 30).
CCBPT. 1988. Reuniao da Comissao Centre Brasileira de Pesquisa de Trigo, 4th. Sao Paulo, SP. In
Recomendacoes da Comissao Centra Brasileira de Pesquisa de Trigo para o ano de 1988. Sao
Paulo, CAC-CC.
CCSBPT. 1989. Reuniao da Comissao Centro-Sul Brasileira de Pesquisa de Trigo, 5th. Cornelio
Procopio, PR. In Recomendacoes da Comissao Centro-Sul Brasileira de Pesquisa de Trigo
para 1989. Londrina, PR. CAC-CC.
CSBPT. 1988. Reuniao da Comissao Sul-Brasileira de Pesquisa de Trigo, 20th. Porto Alegre, RS.
In Recomendacoes da Comissao Sul-Brasileira de Pesquisa de Trigo para a cultura do Trigo
em 1988. Passo Fundo, EMBRAPA-CNPT.
Curry, R.B., R.M. Peart, J.W. Jones, K.J. Boote, and L.H. Allen. 1990. Simulation as a tool for
analyzing crop response to climate change. American Society of Agricultural Engineers,
33(3):981-990.
Godwin, D., J. Ritchie, U. Singh, and L. Hunt. 1989. A user's guide to CERES wheat - V2.10.
International Fertilizer Development Center. Muscle Shoals, USA
Hansen, J., I. Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy, G. Russel, and P. Stone. 1988. Global
Climate Changes as Forecast by Goddard Institute for Space Studies Three-Dimensional
Model. /. Geophys. Res. 93:9341-9364.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
Efficient Three-Dimensional Global Models for Climate Studies: Models I and II. Monthly
Weather Review, Vol. Ill, No.4.
BRAZIL-11
-------
IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J.
Ephraums (eds). Intergovernmental Panel on Climate Change. Cambridge.
Jones, C.A., and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and
Development. College Station. Texas A&M University Press.
Jones, J.W., K.J. Boote, S.S. Jagtap, G. Hoogenboom, and G.G. Wilkerson. 1988. SOYGRO v5.41:
Soybean Crop Growth Simulation Model User's Guide. Florida Agr. Exp. Sta. Journal No.
8304, IFAS. Univ. of Florida.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of
DSSAT. In International Benchmark Sites Network for Agrotechnology Transfer. Proceedings
of the IBSNAT Symposium: Decision Support System for Agrotechnology Transfer. University
of Hawaii.
Larach, J.O.I., A. Cardoso, A.P. de Carvalho, D.P. Hochmuller, P.J. Fasolo, and M. de Rauen. 1984.
Levantamento de reconhecimento dos solos do estado do Parana.
EMBRAPA-SNCLS/SUDESUL/IAPAR, Curitiba. 2t. (Tech. Bull, 27).
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Matzenauer, R., AL. Pons, J.R.T. Maluf, and A.C. Bueno. 1988. Efeito da irrigacao e densidade de
plantas na cultura do milho: I. Rendimento de graos e componentes de rendimento. In
Reuniao tecnica anual do milho, 33. Porto Alegre, RS.
Miyasaka, S. and J.C. Medina. 1981. A Soja no BrasiL Banco do Brasil/FINEP/MA/MIC. Secao de
Divulgacao, ITAL. 1062p.
Mota, F.S. da 1989. Clima, technologia e produtividade do trigo no Brasil. In Agrometeorologia do
trigo no BrasiL Campinas, Sociedade Brasileira de Agrometeorologia
Mota, F.S. da and M.O. de O. Agendes. 1986. Clima e agricultura no Brasil. Porto Alegre, SAGRA.
Mota, F.S. da, Z.B. Berny, and M.O. de O. Agendes. 1984. Perspectivas dimaticas da agricultura ate
o ano 2000, no sul do Brasil,sob uma provavel agao do efeito estufa. Pelotas,
EMBRAPA-DDT. 70p. (EMBRAPA-UEPAE de Pelotas. Doc. 11).
Mothei, E.P., J.AM. do Amaral, and R.D. dos Santos. 1979. Levantamento de reconhecimento
detalhado e aptidao agricola dos solos da area do Centra Nacional de Pesquisa de Gado de
Corte, Mato Grosso do Sul. EMBRAPA-SNLCS, Rio de Janeiro, 225p. (Tech. Bull, 59).
Oliveira, J.B. de and J.R.F. Menk. 1984. Latossolos roxos do estado de Sao Paulo. Institute
Agronomico, Campinas. (Tech. Bull. 82).
BRAZIL-12
-------
Queiroz, E.F., N. Neumaier, and E. Torres. 1979. Ecologia e manejo da cultura. In Empresa
Brasileira de pesquisa agropecuaria. Centro Nacional de Pesquisa de Soja, Londrina, PR.
Ecologia, manejo e adubacao da soja. Londrina. (EMBRAPA-CNPSO. Tech. Circ. 2).
Reuniao de Pesquisa de Soja da Regiao Sul, 15th. Cruz Alta, RS. In Recomendacoes da Reuniao de
Pesquisa de Soja da Regiao Sul-1987/88. Cruz Alta, FECOTRIGO-CEP. 75P.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-wheat: A user-oriented
wheat yield model. In Willis, W.O., ed. ARS Wheat Yield Project. Washington, DC: USDA
Agricultural Research Service. ARS-38.
Ritchie, J., U. Singh, G. Godwin, and L. Hunt. 1989. A user's guide to CERES-maize - V2.10 Muscle
Shoals, USA. International Fertilizer Development Center.
Rose, E. 1989. Direct (physiological) effects of increasing CO2 on crop plants and their interactions
with indirect (climatic) events. In: The Potential Effects of Climate Change in the United
States. Smith, J.B. and D.A. Tirpak, eds. EPA. Washington, DC. V2.
Santos, P.CT. dos, L.S. Vieira, M. de N.F. Vieira, and A Cardoso. 1983. Os solos da Faculdade de
Ciencias Agrarias do Para. FCAP. Belem. (FCAP. Infom. 5).
Siqueira, O.J.F. de 1991. Avancos na informatica, aplicados as relacoes solo- planta-clima: Exemplo,
utilizando o sistema de suporte de decisao "DSSAT" - Nitrogenio x Trigo. In Reuniao
national de pesquisa de trigo, 16. Dourados, MS. Resumes. Dourados: EMBRAPA/UEPAE
Siqueira, OJ.F. de and M. van den Berg. 1991. Validacao do modelo de simulacao SOYGRO-Soja
- Passo Fundo, RS. In Reuniao de pesquisa de soja da regiao sul, 19. Pelotas, RS. Soja;
resultados de pesquisa 1990-1991. Passo Fundo: EMBRAPA/CNPT. (Doc. 3).
Smith, J.B. and D.A. Tirpak. 1989a. The potential effects of global climate change on the United States:
Appendix C - Agriculture. EPA. Washington, DC. VI and V2.
Smith, J.B. and D.A. Tirpak. 1989b. The potential effects of global climate change on the United States.
Report to Congress. EPA. Washington, DC. 413p.
Vernetti, F. de Jesus. 1983. Soja. Campinas, Fundacao Cargill.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a
Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
BRAZIL-13
-------
Table 1.
Site
Characteristics of sites selected for the simulation study.
Lat.
Long.
Elevation Weather
(m) Data
Soil
Pelotas, RS
Passo Fundo, RS
Sao Borja, RS
Vacaria, RS
Ponta Grossa, PR
Londrina, PR
Campinas, SP
Campo Grande, MS
Sete Lagoas, MG
Cruz das Almas, BA
Petrolina, PE
Manaus, AM
Belem, PA
31.47S
28.15S
28.39S
28.33S
25.06S
23.19S
22.53S
20.27S
19.28S
12.40S
9.23S
3.08S
1.28S
52.29W
52.24W
56.00W
50.42W
50.10W
51.19W
47.06W
54.37W
44.15W
39.06W
40.30W
60.01W
48.27W
13
667
99
955
868
566
669
530
732
226
366
48
24
52/80
51/80
56/80
51/80
54/80
58/80
51/80
74/80
60/80
71/80
65/80
71/80
67/80
Hapludult
Haplorthox
Paleudalf
Haplohumox
Haplorthox
Haplorthox
Eutrorthox
Haplorthox
Haplusthox
Haplorthox
Eutrusthox
Acrorthox
Haplorthox
BRAZIL-14
-------
Table 2.
Observed baseline climate at selected sites.
TEMPERATURE
Dec-Feb Mar-May Jun-Aug
Sep-Nov
Annual
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
PRECIPITATION
(mm)
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
22.9
22.6
25.0
20.2
22.0
24.3
24.0
25.3
23.4
25.9
26.6
26.6
27.0
Dec-Feb
297
474
348
384
480
627
648
678
738
273
255
708
960
18.7
18.4
20.5
15.9
18.6
21.4
21.6
23.3
21.8
24.9
25.6
26.6
27.0
Mar-May
264
348
372
291
309
324
267
390
201
342
255
903
1101
13.5
14.2
15.5
11.8
15.1
18.0
18.9
21.6
19.1
22.1
24.0
26.6
27.2
Jun-Aug
366
441
306
369
276
213
114
123
33
303
24
348
450
17.5
18.4
20.0
15.9
18.8
21.8
22.1
24.0
22.6
24.2
27.0
27.6
27.4
Sep-Nov
315
501
399
408
381
420
330
507
393
243
69
420
366
18.2
18.4
20.2
15.9
18.6
21.4
21.6
23.6
21.7
24.3
25.8
26.8
27.2
Annual
1,242
1,764
1,425
1,452
1,446
1,584
1,359
1,698
1,365
1,161
603
2,379
2,877
BRAZIL-15
-------
SOLAR RAD. (MJ/m2) Dec-Feb Mar-May Jun-Aug Sep-Nov
Annual
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
18.4
17.7
17.8
17.1
15.1
16.3
16.1
16.2
15.9
15.7
163
11.7
13.1
11.2
11.6
11.5
11.4
11.4
13.3
13.4
13.7
14.5
13.2
14.3
11.2
12.7
8.0
8.6
8.2
8.5
9.6
11.3
11.6
12.1
13.6
11.3
13.4
13.7
15.6
14.6
14.7
15.0
14.4
13.7
15.2
15.3
15.6
15.0
14.3
17.2
14.3
16.2
13.0
13.1
13.1
12.8
12.4
14.0
14.1
14.4
14.7
13.6
15.3
12.7
14.4
BRAZIL-16
-------
Table 3. Annual GCM scenario changes in temperature, precipitation, and solar radiation
at selected sites.
CHANGES IN
TEMPERATURE (°C)
PRECIPITATION
RATIOS
SOLAR RADIATION
RATIOS
SITE
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo
Grande
Sete Lagoas
Cruz das
Almas
Petrolina
Manaus
Belem
GISS
4.0
4.5
4.8
4.5
4.5
4.5
4.5
4.5
4.1
4.3
43
3.5
4.0
GFDL
4.2
3.5
3.5
3.5
3.1
3.1
3.2
3.0
4.2
2.5
2.5
2.7
2.7
UKMO
4.5
6.1
5.6
6.0
6.0
4.8
4.8
5.7
6.0
4.5
4.6
3.9
3.9
GISS
1.00
0.98
1.65
0.98
0.98
1.24
1.24
1.24
1.08
1.03
1.03
1.25
1.09
GKDL
1.00
1.20
1.20
1.20
1.21
1.21
1.25
1.17
1.07
1.08
1.08
1.20
1.00
UKMO
1.14
1.05
1.15
1.05
1.05
1.20
1.20
1.08
1.26
0.83
0.84
0.97
0.80
GISS
1.06
1.02
0.98
1.02
1.02
1.00
1.00
1.00
1.05
1.02
1.02
1.02
1.02
GFDL
1.01
0.99
0.99
0.99
0.99
0.99
0.96
0.98
0.98
1.01
1.01
1.01
1.01
UKMO
1.03
1.04
1.02
1.04
1.04
1.03
1.03
1.03
1.14
1.02
1.04
1.11
1.00
BRAZIL-17
-------
Table 4. Sensitivity of the CERES-Wheat model to changes in temperature, precipitation,
and CO2 levels (330 ppm and 555 ppm CO2). Simulated yields and season length
for different regions.
SIMULATED GRAIN YIELD (t ha'1)
SOUTH
Precip.
Diff.
(%)
0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.
(°C)
+0
+2
+4
0
+2
+4
0
+2
+4
330
ppm
2.30
1.88
1.47
2.26
1.86
1.46
2.32
1.89
1.48
555
ppm
2.66
2.27
1.87
2.61
2.24
1.85
2.71
2.30
1.89
C. SOUTH
330
ppm
2.36
1.63
1.03
2.35
1.64
1.04
2.36
1.63
1.03
555
ppm
2.91
2.22
1.54
2.90
2.21
1.54
2.96
2.24
1.54
CENTRAL
330 ppm
2.93
2.18
1.44
2.93
2.17
1.44
2.94
2.18
1.44
555
ppm
3.36
2.71
2.00
3.34
2.70
2.00
3.32
2.67
1.97
SIMULATED SEASON LENGTH (DAYS)*
Temp.
Diff (°C)
0
+2
+4
SOUTH
330
ppm
124
116
108
C. SOUTH
330
ppm
103
95
89
CENTRAL
330
ppm
100
91
85
*Season length changes (as simulated with the CERES-Wheat model) were affected only by
temperature.
BRAZIL-18
-------
Table 5. Sensitivity of the CERES-Maize model to changes in temperature, precipitation,
and CO2 level (330 ppm and 555 ppm of CO^). Simulated yields and season length
for different regions.
SIMULATED GRAIN YIELD (t ha'1)
Precip.
Diff.
°
0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.
0
+2
+4
0
+2
+4
0
+2
+4
SOUTH
330 555
ppm
7.78
7.14
6.31
7.92
7.26
6.41
7.57
6.92
6.11
ppm
8.49
7.81
6.90
8.51
7.85
6.95
8.37
7.63
6.70
SIMULATED
SOUTH
Temp.
Diff.
0
+2
+4
330
ppm
134
118
103
C.SOUTH
330 555
ppm
6.66
6.15
5.36
6.58
6.10
5.33
6.70
6.17
5.35
SEASON
ppm
7.05
6.56
5.75
6.94
6.47
5.70
7.14
6.62
5.78
LENGTH
C.SOUTH
330
ppm
117
106
98
N.EAST
330 555
ppm
4.87
4.11
3.49
5.04
4.32
3.67
4.54
3.79
3.16
(DAYS)*
ppm
5.44
4.68
4.14
5.52
4.74
4.18
5.30
4.52
3.98
N.EAST
330
ppm
106
98
94
NORTH
330 555
ppm
4.39
3.90
3.42
4.19
3.72
3.28
4.64
4.10
3.55
ppm
4.36
3.91
3.43
3.99
3.60
3.19
4.67
4.15
3.62
NORTH
330
ppm
104
96
92
*Season length changes (as simulated with the CERES-Maize model) were affected only by
temperature.
BRAZIL-19
-------
Table 6. Sensitivity of the SOYGRO model to changes in temperature, precipitation, and
CO2 levels (330 ppm and 555 ppm of COJ. Simulated yields and season length for
different regions.
SIMULATED GRAIN YIELD (t ha'1)
Prec.
Diff.
0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.
0
+2
+4
0
+2
+4
0
+2
+4
SOUTH
330
ppm
2.91
2.99
2.79
3.08
3.18
3.00
2.63
2.67
2.47
555
ppm
3.92
4.14
3.77
4.26
4.17
3.95
3.68
3.72
3.46
SIMULATED
Temp.
Diff.
(°C)
0
+2
+4
SOUTH
330
ppm
143
137
134
555
ppm
C.SOUTH
330
ppm
3.10
2.83
2.35
3.29
2.98
2.46
2.90
2.65
2.16
SEASON
555
ppm
4.60
4.36
3.88
4.71
4.48
4.01
4.43
4.18
3.70
LENGTH
C.SOUTH
330
ppm
125
118
114
555
ppm
N.EAST
330
ppm
3.26
3.06
2.79
3.36
3.18
2.92
3.06
2.84
2.54
(DAYS)*
555
ppm
4.08
3.81
3.45
4.18
3.92
3.60
3.86
3.56
3.19
N.EAST
330
ppm
110
112
116
555
ppm
NORTH
330
ppm
2.17
1.94
1.79
2.21
2.00
1.85
2.10
1.88
1.71
555
ppm
3.08
2.93
2.82
3.10
2.96
2.88
3.05
2.89
2.76
NORTH
330
ppm
91
90
92
555
ppm
*Season length changes (as simulated with the SOYGRO model) were affected only by
temperature.
BRAZIL-20
-------
Table 7. Effects of GCM climate change scenarios on wheat, maize, and soybean yields in
selected sites.
Site
BASE
Simulated Yield (t ha"1)
Climate Scenario
Climate Scenario Alone + Physiological CO2 Effects
GISS GFDL UKMO BASE GISS GFDL UKMO
WHEAT
Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas (*)
Campo Grande (*)
Sete Lagoas (*)
Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Bclcm
Manaus
2.54
1.99
2.66
2.01
2.28
2.02
2.93
2.12
2.93
6.26
8.17
9.20
7.60
7.64
7.36
7.16
6.72
5.41
5.10
4.64
4.00
4.79
1.73
1.21
1.89
0.93
1.50
0.93
1.14
0.59
1.51
5.01
6.83
7.49
5.54
5.88
6.21
6.34
5.34
4.46
3.77
2.59
3.02
3.82
1.52
1.25
1.94
1.21
1.52
1.35
1.57
1.23
1.72
3.93
6.91
7.24
5.24
6.08
5.78
5.93
5.83
4.66
3.94
3.09
3.45
3.92
1.62
0.91
1.69
1.01
1.17
0.95
1.17
0.76
1.10
MAIZE
5.01
5.85
6.93
4.21
5.29
5.76
5.78
4.28
4.67
4.07
2.81
3.64
3.96
2.85
2.32
2.89
2.54
2.70
2.42
3.52
2.79
3.36
7.58
8.93
9.64
8.25
8.07
7.77
7.56
7.08
5.79
5.64
5.25
3.65
5.06
2.14
1.62
2.18
1.33
1.91
1.33
1.72
1.00
2.04
6.06
7.42
7.92
5.99
6.26
6.59
6.72
5.63
4.92
4.49
3.06
2.54
4.10
1.94
1.65
2.22
1.62
1.88
1.78
2.24
1.84
2.24
5.12
7.54
7.68
5.94
6.42
6.10
6.33
6.17
5.12
4.92
3.73
3.94
4.23
2.05
1.32
2.02
1.44
1.57
1.34
1.74
1.23
1.57
6.02
6.64
7.39
5.10
5.72
6.17
6.18
4.56
5.05
4.73
3.36
3.65
4.30
SOYBEAN
Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Belcm
Manaus
2.84
3.03
2.94
2.48
3.28
2.16
3.77
3.43
3.05
3.01
3.52
1.87
1.89
2.20
2.38
3.68
2.36
3.24
1.73
3.44
2.97
2.53
2.57
3.18
2.38
1.49
2.12
2.39
3.52
1.83
3.23
1.60
3.10
2.91
2.62
2.63
2.83
3.17
1.50
3.66
1.71
3.65
1.55
2.76
0.94
3.22
1.76
2.66
2.33
2.51
2.36
1.19
2.95
4.38
3.78
3.52
4.28
4.64
4.80
4.92
4.05
3.74
4.42
3.14
3.01
2.99
3.67
4.64
3.23
4.38
4.12
4.36
4.41
3.62
3.19
3.44
3.16
2.84
3.35
3.76
4.48
2.73
4.37
4.01
4.07
4.29
3.50
3.28
3.99
3.02
2.80
3.30
2.82
4.76
2.36
4.18
3.76
4.20
3.63
3.57
2.93
3.17
3.14
2.55
SOUTH REGION: Pelotas, Passo Fundo, Sao Borja, Vacaria, and Ponta
SOUTH REGION: Londrina, Campinas, Campo Grande, and Sete Lagoas (S.
region for wheat). NORTHEAST REGION: Cruz das Almas and Petrolina.
Manaus and Belem. (*) Irrigated simulation.
Grossa. CENTRAL
L. in the CENTRAL
NORTH REGION:
BRAZIL-21
-------
Table 8.
Effect of the GCM climate change scenarios on simulated wheat.
Climate Scenario Alone
Climate Scenario
+ Physiological CO2 Effects
Region
BIOMASS (t ha'1)
BASE
GISS
GFDL
UKMO
GRAIN YIELD (t ha'1)
BASE
GISS
GFDL
UKMO
SEASON LENGTH (Days)
BASE
GISS
GFDL
UKMO
EVAPOTRANSPIRATION
BASE
GISS
GFDL
UKMO
South
8.82
6.18
6.33
5.50
2.30
1.43
1.49
1.28
124
107
109
104
(mm)
274
247
243
246
CSouth
7.30
3.01
4.51
3.28
2.36
0.89
1.38
0.96
103
87
93
87
264
248
250
252
Central
7.72
4.21
4.74
3.22
2.93
1.51
1.72
1.10
100
85
89
82
303
294
277
305
South
10.30
7.87
7.96
7.27
2.66
1.84
1.86
1.68
124
107
109
104
248
227
222
227
CSouth
9.04
4.55
6.37
4.89
2.91
1.35
1.95
1.43
103
87
93
87
233
219
224
222
Central
8.76
5.63
6.10
4.57
3.36
2.04
2.24
1.57
100
85
89
82
260
253
238
259
BRAZIL-22
-------
Table 9.
Effect of the GCM climate change scenarios on simulated maize.
Region
Climate Scenario Alone
South CSouth NEast North
Climate Scenario
+ Physiological CO2 Effects
South CSouth NEast North
BIOMASS (t ha'1)
BASE
GISS
GFDL
UKMO
GRAIN YIELD
BASE
GISS
GFDL
UKMO
14.03
12.54
12.06
11.60
(tha-1)
7.78
6.15
5.88
5.46
13.32
11.85
12.03
11.42
6.66
5.59
5.55
5.12
10.64
7.62
7.86
8.28
4.87
3.16
3.52
3.44
10.82
8.78
8.52
9.72
4.39
3.42
3.68
3.80
14.73
13.27
12.90
12.63
8.50
6.73
6.54
6.18
13.72
12.32
12.46
11.94
7.05
5.96
5.93
5.49
11.50
8.75
9.34
9.34
5.45
3.78
4.32
4.04
10.70
8.60
9.33
10.17
4.36
3.32
4.08
3.97
SEASON LENGTH (days)
BASE
GISS
GFDL
UKMO
134
105
109
101
EVAPOTRANSPIRATION
BASE
GISS
GFDL
UKMO
448
407
394
405
117
100
101
97
(mm)
414
378
378
400
106
88
94
90
316
258
268
381
104
92
94
92
305
306
294
297
134
107
109
101
377
347
339
353
117
100
101
97
347
318
318
338
106
88
94
90
268
264
233
244
104
92
94
92
254
256
244
249
BRAZIL-23
-------
Table 10.
Effect of the GCM climate change scenarios on simulated soybean.
Region
Climate Scenario Alone
South CSouth NEast North
Climate Scenario
+ Physiological CO2 Effects
South CSouth NEast North
BIOMASS (t ha1)
BASE
GISS
GFDL
UKMO
YIELD (t ha'1)
BASE
GISS
GFDL
UKMO
6.53
5.89
5.44
5.08
2.92
2.83
2.70
2.44
5.69
5.04
4.74
3.92
3.10
2.71
2.55
2.02
6.06
5.18
5.33
4.28
3.20
2.60
2.75
2.01
3.86
3.37
3.33
3.08
2.17
1.94
1.90
1.78
7.77
8.19
7.72
7.41
3.63
3.84
3.74
3.48
7.02
7.80
7.52
7.05
3.70
4.13
3.95
3.63
6.17
6.76
6.91
5.81
3.56
3.32
3.47
2.80
4.82
5.21
5.12
4.95
2.76
3.00
2.91
2.85
SEASON LENGTH (days)
BASE
GISS
GFDL
UKMO
EVAPOTRANSP.
BASE
GISS
GFDL
UKMO
143
133
135
133
(mm)
462
495
453
496
123
119
120
114
432
453
442
462
109
110
109
110
381
433
400
407
91
92
90
92
302
345
324
336
135
133
135
133
454
494
454
502
121
121
122
122
446
462
446
500
110
110
109
110
334
432
396
408
92
92
90
92
324
330
310
322
BRAZIL-24
-------
Table 11.
Aggregated yield changes under GCM climate change scenarios. Results include the
physiological effects of CO2 on yield.
Simulated Yield Change (%)
Production
Region
SOUTH
CENTRAL-SOUTH
CENTRAL
NATIONAL
SOUTH
CENTRAL-SOUTH
CENTRAL
NORTHEAST
NORTH
NATIONAL
SOUTH
CENTRAL-SOUTH
CENTRAL
NORTHEAST
NORTH
NATIONAL
Source of Production data:
Crop
WHEAT
WHEAT
WHEAT
WHEAT
MAIZE
MAIZE
MAIZE
MAIZE
MAIZE
MAIZE
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
IBGE and Bank of Brazil
(t x 1000)
1,573
2,028
24
3,625
6,695
11,131
2,495
1,126
330
21,778
6,408
6,943
2,135
96
0
15,582
GISS
-21
-43
-30
-33
-13
-10
**
-22
-24
-11
30
32
**
9
38
30
GFDL
-19
-17
-24
-18
-16
-11
**
-11
-7
-11
25
28
**
12
34
23
UKMO
-27
-39
-46
-34
-20
-18
**
-17
-10
-16
20
22
**
-6
31
18
(**) not simulated
BRAZIL-25
-------
Table 12.
Adaptation strategies.
(A) Effect of changes in planting date, irrigation, and nitrogen stress on simulated maize and
soybean yields. The climate change scenario simulations include the physiological effects of
CO2 on yield.
Yield
Change
Simulated Yield (t ha'1) from Base
Site
Passo Fundo
Petrolina
Petrolina
Crop Strategy
Maize Oct 15, rainfed(*)
Sep 15, rainfed
Nov 15, rainfed
Dec 15, rainfed
Jan 15, rainfed
Oct 15, irrig.
Maize Rainfed, N stress(*)
Rainfed, N
Irrig., N stress
Irrig., N
Soybean Rainfed
Irrig.
BASE
8.17
8.48
4.64
6.72
5.37
7.05
3.39
3.73
UKMO
6.69
6.54
5.75
5.72
6.62
6.79
3.98
5.34
5.10
6.37
2.64
4.19
(%)
-18
-20
-30
-30
-19
-17
-14
15
10
37
-22
24
(*) current practice
N stress: 80 Kg N/Ha
N: N for maximum yield (N balance off in the model)
(B) Sensitivity of the CERES-Maize model to changes^in the P5 coefficient of the cultivar PIO
3230. Scenario simulations include the physiological effects of CO2 on yield.
Site Crop
Passo Fundo Maize
Strategy
Change in P5
P5=995*
P5=795
P5=1195
P5=1395
Simulated Yield (T/Ha)
Base UKMO
8.18 6.86
5.21
8.51
10.09
Yield
Change from
Base (%)
-16
-36
4
23
(*) calibrated coefficient for the cultivar PIO 3230 in Passo Fundo.
BRAZIL-26
-------
Appendix A. Cultivars and crop management data used to run the crop models.
Crop
WHEAT
MAIZE
SOYBEAN
Region
SOUTH
SOUTH
SOUTH
SOUTH
SOUTH
C.SOUTH
C.SOUTH
C.SOUTH
CENTRAL
SOUTH
SOUTH
SOUTH
SOUTH
SOUTH
C.SOUTH
C.SOUTH
C.SOUTH
C.SOUTH
N.EAST
N.EAST
NORTH
NORTH
SOUTH
SOUTH
SOUTH
SOUTH
SOUTH
Site
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
CRUZ
D.ALMAS
PETROLINA
MANAUS
BELEM
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
Cultivar
BR14
BR14
BR14
BR14
BR14
BR14
ANZA*
BR 14/ANZA*
ANZA*
PIO-3230
PIO-3230
PIO-3230
PIO-3230
PIO-3230
PIO-3230
PIO-3230
PIO-3230
PIO-3230
SUWAN-1
SUWAN-1
SUWAN-1
SUWAN-1
DAVIS
DAVIS
DAVIS
DAVIS
DAVIS
Plant
Pop.
(pl/m2
330
330
330
330
330
330
350
350
350
5
5
5
5
5
5
5
5
5
5
5
5
5
40
40
40
40
40
Planting
Date
JUN. 15
JUN. 15
MAY. 31
JUL. 15
JUN. 15
APR. 15
APR. 30*
APR.30*
APR. 30*
OCT. 15
OCT. 15
OCT. 15
NOV. 15
OCT. 15
OCT. 15
OCT. 15
OCT. 30
OCT. 30
OCT. 15
OCT. 15
NOV. 15
NOV. 15
NOV. 15
OCT. 15
NOV. 15
NOV. 15
NOV. 15
BRAZIL-27
-------
C.SOUTH
C.SOUTH
C.SOUTH
C.SOUTH
N.EAST
N.EAST
NORTH
NORTH
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
CRUZ
D.ALMAS
PETROLINA
MANAUS
BELEM
DAVIS
DAVIS
DAVIS
DAVIS
VICOJA
VICOJA
JUPITER
JUPITER
40
40
40
40
40
40
40
40
NOV. 15
NOV. 15
NOV. 15
OCT. 15
NOV. 30
NOV. 30
NOV. 30
NOV. 30
"Irrigated. Others: rainfed.
Local cultivars: BR 14: PIV=1.9; P1D=1.5; P5=6.0; Gl=3.2; G2=0.6; G3=3.9.
PIO-3230: PI =220; P2=0.85; P5=995; G2=720; G3=5200.
BRAZIL-28
-------
I.BELCM
. MANAUS
3. PCTROUNA
, CRUZ OAS ALMAS
s. sere LAOOAS
«. CAMPO ORANDE
CAMPMAS
• . LONDKINA
• . PONTA WtOSS
10. PAS50 PUNDO
II. VACARIA
SAO •oruA
13. PCUOTAS
EquatorMl and
Figure la. M^> of Brazil; climatic regions and sites selected for the study.
-------
Figure Ib. Wheat agroecological regions; C=Central, CS=Central-South, S=South.
-------
MAIZE - SOYBEAN
Figure Ic. Maize and soybean agroecological regions; N=North, NE=Northeast,
CS=CentraI-South, S=South.
-------
6000
5000 -
4000
Observed
Grain
Yield 3000
(kg/ha)
2000 -
1000
1000
6500
Observed '
Grain
Yield 800°
(kg/ha)
7000
7000
5000
4000
Observed
Grain 3000
Yield
(kg/ha)
2000
2000 3000 4000 5000
Simulated Grain Yield (kg/ha)
6000
CERES-Maize
-K
7600 8000 8600
Simulated Grain Yield (kg/ha)
9000
SOYGRO
1000^—
1000
2000 3000 4000
Simulated Grain Yield (kg/ha)
5000
Figure 2. Observed and simulated grain yields (kg ha'1) using the CERES-Wheat (Siqueira 1991),
CERES-Maize (adapted from Matzenauer et al. 1988), and SOYGRO (Siqueira and
Berg 1991) at Rio Grande do Sul.
-------
WHEAT
Relative Yield Changes (%)
SOUTH
CENTRAL-SOUTH
330 ppm CO2
MAIZE
Relative Yield Changes (%)
SOUTH
C.8OUTH NORTHEAST NORTH
330 ppm CO2
SOYBEAN
Relative Yield Changes (%)
-40
SOUTH C.SOUTH NORTHEAST
330 ppm CO2
NORTH
WHEAT
Relative Yield Changes (%)
SOUTH
CENTRAL-SOUTH
555 ppm CO2
CENTRAL
MAIZE
Relative Yield Changes (%)
C.SOUTH NORTHEAST
555 ppm CO2
NORTH
SOYBEAN
Relative Yield Changes (%)
SOUTH C.SOUTH NORTHEAST
555 ppm CO2
NORTH '
CDGFW. BUKMO
Figure 3. Regional yield changes for wheat, maize, and soybean under GCM climate change
scenarios.
-------
Temperature Differences (Celsius)
Precipitation Ratio (GISS/Base)
1010 (YEAR) 2030
GISS Transient Scenarios
20M
o
mo
2010 (YEAR) 2030
GISS Transient Scenarios
WHEAT
Grain Yield (kg/ha) Season Length (days)
B
-«- *MIN YIELD -»~ (CATCN LCNOTN
MAIZE
Grain Yield (kg/ha) Season Length (days)
ioooo,e — — —1130
rioei i4soi
GISS Transient Scenarios
eooo
,100 5000
2060 yeAB w
* *EA*ON LIN*TH
2010 2030 2050
_1_J<£S' , _ (*«» IS30-55SS
GISS Transient Scenarios
Figure 4. (A) Temperature and precipitation changes with the GISS transient
scenarios at Passo
-------
Relative Yield (%)
~BRl4
P1V.2.6JBfWoH
P1V-2.6.IBR
P1V-2.6
P1V-19
Season Length (days)
8 • 10
P-5 Genetic Coefficient
P1V-2.6.IRR.No(f
P1V-J.6.IRR
P1V-2.6
P1V-1.9
« « • 10
P-5 Genetic Coefficient
Grain Yield (t/ha)
P1V-2.5JRR.Noff
2JJSH
Grain Yield (t/ha)
< • • 10
P-5 Genetic Coefficient
< • a 10
P-5 Genetic Coefficient
P1V-1.8
Figure 5. Adaptation studies for two cultivars of wheat (BR 14 and BR 23) in Passo Fundo to
the UKMO climate change scenario. The results include the physiological CO2 effects.
Changes in the genetic coefficients and crop management. IRR: Irrigation; Noff: full
nitrogen fertilization.
-------
IMPACTS OF GLOBAL CLIMATE CHANGE ON MAIZE PRODUCTION IN
ARGENTINA
O. E. Sala and J.M. Paruelo
Universidad de Buenos Aires
Buenos Aires, Argentina
ARGENTINA-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Agricultural Systems in Argentina
Objectives
METHODS
Baseline Climate Data and Climate Change Scenarios
Crop Model Calibration
Soils and Management Variables
RESULTS AND DISCUSSION
The Effects of Climate Change
Sensitivity Analysis
Adaptation Strategies
Simulations with Hypothetical New Hybrids
IMPLICATIONS OF THE RESULTS
REFERENCES
ARGENTINA-2
-------
SUMMARY
Climate change is predicted by three GCMs to increase temperature and precipitation in the Rolling
Pampa. These climatic changes may decrease maize yields, if all other variables remain constant. The
simulation results suggest that decreases in maize yields may be offset by (1) changing the sowing date of
current maize varieties, or (2) using hybrids more adapted to the climate change conditions. However, at the
warmer boundaries of the Pampa, it is unlikely that decreases in maize yields could be completely offset by
changes in management practices. Under climate change conditions, maize production may also be constrained
in the northern edge of its current distribution due to extreme high temperatures. At the same time, maize
production could expand into areas to the south that are currently limited by frost.
INTRODUCTION
Agricultural Systems in Argentina
The Pampa region (Figure 1) covers approximately 34 million hectares of agricultural land (Hall et
al. 1991). One-third of the area is used to grow grain crops; the rest, which is comprised of meadows and
natural grasslands, is used to fatten steers and conduct cow/calf operations. The Pampa region has been
subdivided according to ecological criteria (Le6n et al. 1991). Of these subregions, the Rolling Pampa is the
most productive. Here, 50 - 75% of the region is devoted to grain crops.
The Pampa region has a temperate, humid climate and lacks a fully developed dry season. There is
a gradient of rainfall from 1,000 mm in the NE of the region to 600 mm in the SW. Mean annual temperatures
range from 17 °C in the northern part of the region to 14 °C in the SE. Similarly, there is a gradient of the
frost-free period which ranges from 180 to 260 days.
Soils in the region are mainly Mollisols developed on a deep mass of Pampean loess (Frenguelli 1925).
This is a quaternary sediment which was originally transported by wind and later redistributed by water. There
is a clear gradient of soil texture, from coarse soils in the West to fine-textured soils in the East.
Maize is one of the most important crops of the Pampa, occupying 2.3 millon hectares (Hall et al.
1991), with production concentrated in the Rolling Pampa in the NE portion of the region. Maize is alternated
with wheat and soybean crops. Sunflower and grain sorghum are also important (1.5 and 1.1 millon hectares,
respectively, from 1983 to 1985). The average yield of maize in the region from 1980 to 1986 was 4.2 t ha'1.
Objectives
The objective of this work is to assess the possible effect of climate change on maize yields in the
Pampa of Argentina. Climate change driven by an increase of greenhouse gases, mainly CO2, will affect crops
in two ways. One, an elevated CO2 level may enhance net photosynthesis and change stomatal conductance
(making the water use per unit leaf area more efficient). These responses are called the "direct CO2 effects".
Two, the expected temperature increases and changes in precipitation patterns would alter the growth and
development of the crop. In this study, we explore the effects of climate change alone and climate change with
the direct effects of increasing CO2 on maize. We use climate change scenarios developed from three General
Circulation Models (GCMs) and apply them to a dynamic process crop simulation model to determine the
potential yield changes under future climate conditions.
METHODS
ARGENTINA-3
-------
Baseline Climate Data and Climate Change Scenarios
The climatic data used for the modeling experiments are from the town of Pergamino in the Rolling
Pampa (lat. 33 °S, long. 60 °W) (Figure 1). We used daily precipitation and maximum and minimum
temperature data for the period 1960-1984. Daily radiation was estimated using sunshine hours data.
Climate change scenarios were generated from three GCMs: GISS (Goddard Institute for Space
Studies, Hansen et aL 1983); GFDL (Geophysical Fluid Dynamics Laboratory Model, Manabe and Wetherald
1987); and UKMO (United Kingdom Meteorological Office Model, Wilson and Mitchell 1987). Changes in
temperature, precipitation, and solar radiation projected by the GCMs were applied to the observed (baseline)
climate to create the climate change scenarios for Pergamino.
The annual and seasonal changes in temperature and precipitation projected by the GCMs at
Pergamino are shown in Table 1. The three scenarios predict significant increases in annual temperatures
ranging from 4.5 °C to 5.2 °C; the temperature increases are larger during the summer months. The projected
precipitation changes vary. The UKMO and GISS scenarios predict precipitation increases (29% and 10%,
respectively), while the GFDL scenario predicts an annual decrease in precipitation (-5%).
Crop Model Calibration
The CERES-Maize crop model (Ritchie et al. 1989; Jones and Kiniry, 1986) was chosen for the
simulations. This model has been widely validated in different agroecological conditions (Hodges et al. 1986).
We further calibrated and validated the CERES-Maize model for the conditions of the Pampa to determine
its suitability as a simulating tool. The calibration was performed for two single flint type hybrids, DAF11 and
DAF12, from Dekalb Argentina. Table 2 presents the derived genetic coefficients that define the maize
varieties. A summary of the simulated and observed validation results is presented in Table 3.
Soils and Management Variables
The main soil type of the Rolling Pampa is the Pergamino series. It is a typical argiudol with no
serious constraints for agriculture. It is comparable to soils of the mid-western U.S. or the Ukraine, but it does
not freeze in winter; thus tillage is feasible year-round (Hall et al 1991) (Appendix 1).
Maize in the Rolling Pampa is mostly rainfed, and long-cycle hybrids account for 80% of the cultivated
area. There are some medium-cycle hybrids grown. Maize is sown in early September and harvested in early
February the following year. For this simulation, the sowing date of long-cycle hybrids was the 277th day of
the Julian calendar and for medium-cycle hybrids, the 267th day. The duration of the fallow is directly
correlated with maize yields in the region. A fallow period longer than 120 days is reported for 37% of the
crops.
RESULTS AND DISCUSSION
The Effects of Climate Change
The changes in temperature and precipitation predicted for the doubled CO2 scenarios produced
decreases in maize yields (Table 4). Under the GISS scenario, yields simulated with nitrogen stress were
reduced about 19% in comparison with base yields. Under the GFDL and UKMO scenarios, the yield
reductions were larger (about 36% in both cases). These simulated yield reductions are mainly a result of a
shorter growing period. The higher temperatures under the climate change scenarios trigger the onset of
maturity stages earlier than under the present climate.
ARGENTINA-4
-------
The direct effect of the increased atmospheric CO2 on the crop resulted in very small yield increases.
These did not compensate for the yield decreases simulated under the GCM scenarios alone.
Yearly variability of the yields decreased in the doubled-CO2 runs. Since precipitation variability was
not modified in this modeling exercise we suggest that the observed reduction in yield variability is a result
of the shortened growing season, with the crop being constrained to a portion of the year where precipitation
variability was lower. Results simulated without nitrogen stress are also shown in Table 4.
Sensitivity Analysis
The purpose of the sensitivity analysis is to assess the maize model responses to changes in
temperature, precipitation, and CO2. The experiment consisted of running the model for three temperature
conditions (control, +2°C, and +4°C); three precipitation conditions (control, +20%, -20%); and two CO2
conditions (330 ppm and 555 ppm). The response variables analyzed are yield, season length, and
evapotranspiration (Table 5).
Yields decrease as a result of increases in temperature, due to a decrease in the season length which
particularly affects the critical grain-filling period. The simulated increases in precipitation do not produce
increases in simulated yields as we had expected. A possible explanation may be that the modeled crop was
under more nitrogen stress under the higher precipitation scenario, due to nutrient leaching. Alternatively,
with seasonal rainfall, percentage increases of precipitation may still be quite small. Further analysis of the
nitrogen balance in the CERES-Maize model and the simulation conditions are needed.
The direct effects of 555 ppm CO2 result in a small increase in simulated yield, but this yield increase
did not compensate for the negative effects of a +2° C temperature increase. The direct effects of CO2 on yield
are larger under the lower precipitation conditions because the beneficial effect of CO2 on simulated water
use is more apparent under low precipitation conditions.
Adaptation Strategies
The major effect of climate change predicted by this modeling exercise is that temperature increases
result in yield reductions due to a shortening of the growing period. We suggest two possible adaptive
strategies to climate change: (1) a shift in the sowing date; and (2) a change in the hybrid.
Changes in the sowing date had large effects on simulated maize yields (Figure 3) and on the length
of the growing period (Figure 2). Under the baseline climate, the optimum simulated sowing date is close to
the date most frequently used in the region. However, under the GISS climate change scenario, the curve
relating sowing dates and yields shows a bimodal response, with maxima occurring with very early or very late
sowing dates. Both dates avoid the high-stress months of midsummer (January and February) and consequently
decrease the water stress.
Therefore, one possible adaptive strategy to climate change conditions would be to sow maize very
early so that the growing period occurs mostly during the cooler part of the year. Planting two months earlier
than the present sowing date can fully compensate for the yield decreases under climate change conditions.
An alternative strategy is to sow very late and avoid the hot months of midsummer.
These strategies have some risk associated with them. The yield variability increases markedly as the
sowing date moves away from October/November (Figure 2), because of extremely low temperatures in some
years. Although these strategies simulate an optimum yield under climate change conditions, they may not
represent a practical optimum because of the biological uncertainties and the economic risks associated with
them. Each of these alternatives implies major changes in the agricultural system of the region.
Simulations with Hypothetical New Hybrids
ARGENTINA-5
-------
A possible strategy for adaptation to climate change could be to replace the currently available hybrid
with a hypothetical new hybrid better adapted to climate change conditions. To analyze this alternative,
hypothetical new hybrids were created under the GISS scenario by increasing the genetic coefficients, PI} P2,
and P5, from 10% to 50% (Figure 4). Pj represents the time period during which the plant is not responsive
to changes in the photoperiod. P2 defines the photoperiod sensitivity of the cultivar. P5 is the number of degree
days above a base of 8°C from silking to physiological maturity.
The figure shows that an increase in the genetic coefficients of the potential new hybrids of between
10% and 20% is enough to restore the GISS climate change yields to their baseline level. This simulation
exercise suggests that hypothetical new hybrids may be able to take advantage of a prolonged growing season
and higher precipitation amounts. The extent to which this could be possible in practice needs to be further
investigated with maize breeders.
IMPLICATIONS OF THE RESULTS
This study suggests that projected climate changes resulting from an increase in greenhouse gases may
result in decreases in maize yields in the Rolling Pampa. The yield decreases may be compensated for by
changing the planting date of the hybrids presently used. But in the warmer boundaries of the rolling Pampa,
it is unlikely that maize yields could be fully restored to their previous level. Nevertheless, maize production
could expand into areas that are currently limited by the length of the frost-free period.
Using the same tools that we used in this exercise, we plan to conduct further model simulations for
different locations in order to estimate the geographical location of the potential new maize-producing region
in Argentina under climate change conditions. A qualitative analysis suggests that a large portion of the maize-
producing area will be located on soils of poorer quality than present soils. Therefore, the average soil fertility
of the maize area may decrease, possibly resulting in decreased yields, unless additional fertilizer is added to
the soil. These hypotheses can be tested further by using these same modeling tools on a larger number of
sites.
ARGENTINA-6
-------
REFERENCES
Frenguelli, J. 1925. Loess y limos pampeanos. Anales de la Sociedad Argentina de Estudios Geogrdficos GAEA
1:7-91.
Hall, A.J., CM. Rebella, CM. Ghersa, and J.Ph. Culot. 1993. Field crops systems of the pampas. In C.J.
Pearson (ed) Ecosystems of the World, Field Crop Ecosystems of the World. Elsevier Scientific
Publishing Company, Amsterdam, The Netherlands.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient three-
dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-662.
Hodges, T., D. Botner, C Sakamoto, and J. Hays-Haug. 1986. Using the CERES Maize model to estimate
production for the US corn belt. Agricultural and Forest Meteorology 40:293-303.
Jones, C.A. and J.R. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development. College
Station. Texas A&M Press.
Le6n, R.J.C. 1991. Natural Grasslands of South America. In R.T.Coupland (ed) Ecosystems of the World,
Volume 8, Natural Grasslands. Elsevier Scientific Publishing Company, Amsterdam, The Netherlands.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Ritchie, J., U. Singh, D. Godwin, and L. Hunt. 1989. A user's guide to Ceres Maize- v2.10. International
Fertilizer Development Center. pp86.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
ARGENTINA-7
-------
Table 1. Average seasonal changes of temperature and precipitation for the GISS, GFDL, and UKMO
GCM climate change scenarios in Pergamino.
Scenario/variable
SPRING SUMMER
FALL WINTER ANNUAL
GISS
Temperature °C
Precip. (%)
GFDL
Temperature °C
Precip. (%)
UKMO
Temperature °C
Precip. (%)
3.8
-3
4.0
0
4.8
51
5.4
30
5.5
20
5.9
14
5.5
10
4.3
-34
5.1
40
4.5
15
4.4
-25
5.0
33
4.8
10
4.5
-5
5.2
29
ARGENTINA-8
-------
Table 2.
Genetic coefficients derived for single flint type hybrids.
Genetic coefficient
DAF11 hybrid DAF12 hybrid
PI (Juvenile)
P2 (Photoperiodism)
P5 (Grain filling duration)
G2 (Kernel number)
G3 (Kernel weight)
260
1.0
700
625
9.6
260
1.0
700
710
8.6
ARGENTINA-9
-------
Table 3.
Validation of the CERES-Maize model for Pergamino.
Crop variable
DAF11 hybrid
Sim. Obs.
DAF12 hybrid
Sim. Obs.
Emergence
Anthesis
Beginning grain filling
End grain filling
Yield (t ha'1)
Kernel weight (g)
Grain/ear
Biomass (t ha'1)
3-11
10-1
20-1
20-2
10.12
0.275
383
20.57
3-11
13-1
-
-
9.96
0.281
381
22.68
3-11
10-1
20-1
20-2
10.15
0.246
429
20.58
3-11
9-1
~
~
10.25
0.251
436
20.25
ARGENTINA-10
-------
Table 4. Effects of climate change on simulated rainfed maize yield, season length, season
precipitation, and evapotranspiration with and without nitrogen stress.
Climate Scenario With Physiological
Climate Scenario Alone
CO2 Effects
Management
Rainfed
Nitrogen
Stress
Rainfed
Fertilized
Simulated
Variable
Yield
(t ha'1)
SD
SXength
(d)
SD
S. PP
(mm)
SD
ET (mm)
SD
Yield
(t ha -1)
SD
ET (mm)
SD
BASE
3.69
0.85
126
8
611
142
487
20
9.42
1.44
494
22
GISS
3.00
0.70
102
4
498
102
460
21
7.89
0.95
471
21
GFDL
2.52
0.61
101
4
530
134
458
25
7.39
1.13
458
25
UKMO
2.40
0.48
99
3
594
115
448
22
8.20
0.95
475
22
BASE
3.77
0.82
126
8
611
142
398
16
9.95
1.30
419
19
GISS
3.04
0.51
102
4
498
102
387
24
8.52
0.85
414
23
GFDL
2.52
0.53
101
4
530
134
408
18
8.29
0.91
408
19
UKMO
2.54
0.42
99
3
594
115
367
27
8.88
0.85
417
22
ARGENTINA-11
-------
Table 5.
Sensitivity analysis of CERES-Maize to climate and CO2 changes.
Simulated variable*
Changes
Precip.
(%)
0%
20%
-20%
Temp.
(°C)
0
2
4
0
2
4
0
2
4
Yield
(t ha'1)
3.69
3.35
3.31
3.52
3.19
3.18
3.76
3.49
3.35
330 ppm CO2
Season L.
(days)
126
113
104
126
113
104
126
113
104
ET
(mm)
486
457
437
487
456
437
482
453
433
Yield
(t ha'1)
3.77
3.38
3.35
3.49
3.18
3.12
3.92
3.59
3.52
555 ppm CO2
Season L.
(days)
126
113
104
126
113
104
126
113
104
ET
(mm)
397
376
364
395
373
360
399
378
366
* Simulations with nitrogen stress.
ARGENTINA-12
-------
ARGENTINA
Figure 1. Map of Argentina and location of Pergamino
-------
150
140
*"••>
cn
<9 130
tn 120
o
(A
CO
cu
en
110
100
90
Baseline
CLIMATE CHANGE
ISAug ISSep 150ct IBNov
Sowing date
IBDec
15Jan
Figure 2. Effect of sowing date on maize yield for the baseline and the climate change conditions.
Vertical bars represent standard deviations.
-------
tti
ID
•H
Baseline
CLIMATE CHANGE
15Aug
IBSep
150ct IBNov
Sowing date
ISDec
Figure 3. Effects of sowing date on the season length of maize for the baseline and climate change
conditions.
-------
6 -
cd
EH
5 -
ID
:H 4 H
Climate Change condition
DAF12 (baseline condition)
Control Hyb 1 Hyb 2 Hyb 3 Hyb 4 Hyb 5
Figure 4. Simulated season length of hybrids with genetic coefficients PI, P2, and P5 ranging from
10% to +50% (Hybl to Hyb5) higher than the control hybrid.
-------
IMPACT OF CLIMATE CHANGE ON BARLEY IN URUGUAY:
YIELD CHANGES AND ANALYSIS OF NITROGEN MANAGEMENT SYSTEMS
Walter E. Baethgen
International Fertilizer Development Center (IFDC)
USA
URUGUAY-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Objectives
Representative climate and soils
Managements practices and nitrogen available for the crop
METHODS
Baseline climate data
Climate change scenarios for the region
Crop model and simulation strategies
Validation and calibration of the crop model
RESULTS AND DISCUSSION
Effects of 2xCO2 GCM climate change on barley production
Sensitivity analysis
Transient analysis
Adaptation strategies to climate change
CONCLUSIONS
REFERENCES
URUGUAY-2
-------
SUMMARY
This study utilized global climate models (GCMs) and dynamic crop growth models to estimate the
potential agricultural effects of climate change on barley production and nitrogen management systems in
Uruguay. The barley crop simulation model was calibrated and validated in great detail in the study,
considering seven parameters of growth and development of the crop. Since barley management practices vary
widely in Uruguay, several crop management scenarios were considered combining three planting dates and
two levels of available nitrogen. Projected climate change caused simulated barley yields to decrease in all
strategies considered. The decrease in modeled grain yields were caused primarily by temperature increases
which shorten the duration of the crop life cycle, particularly the grain-filling period. These decreases were
partially counteracted by the beneficial physiological CO2 effects on crop growth and water use as simulated
in this study. The negative effects of GCM climate change were worse when no nitrogen fertilizer was applied,
and/or when planting was delayed. The variability of grain yields was larger under the GCM scenarios and
when planting dates were delayed, but it was not affected by nitrogen fertilization. A possible adaptation of
barley management systems in Uruguay to climate change conditions is an optimization of the soil nitrogen
available for the crop, but even considering this adaptation strategy, significant production losses were still
associated with the climatic conditions projected by the UKMO scenario.
INTRODUCTION
Background
The climatic characteristics of Uruguay (Table 1) allow the production of subtropical and temperate
species. The most important crops in Uruguay are wheat, barley, rice, maize, sorghum and sunflower. The
country is self-sufficient in food, and 30-40% of the country's total exports are textiles, beef, hides and cereals.
Most of the wheat, maize, sorghum and sunflower are consumed in the country; barley and rice are primarily
export crops. Due to the economic importance of barley as an export, it was selected for this study.
The enhanced greenhouse effect of increased atmospheric concentration of CO2 and other trace gases
could lead to higher global surface temperatures and changed hydrological cycles (IPCC 1990). Global climate
change may affect crop production in some areas of the world, as previous impact studies have shown (Parry
et al. 1988). A change in climatic conditions could affect crop production in Uruguay.
Barley crops in Uruguay are sometimes damaged by excess soil water that causes leaching and
denitrification, with consequent increased requirements for nitrogen fertilization, lower economic returns, and
greater potential for nitrate contamination of groundwater. A possible change in climatic conditions may have
an important impact on barley production and on the environmental consequences of the use of nitrogen
fertilizers.
Objectives
This study had two main objectives. One goal was to study the potential impact of global climate
change on barley production in Uruguay under a wide range of management practices, in particular available
nitrogen levels and planting dates. Another goal was to evaluate soil and crop management practices (nitrogen
fertilizer use, cultivar characteristics, and planting dates) that may be better adapted to the possible climate
change conditions.
Representative Climate and Soils
URUGUAY-3
-------
The crop-growing area of Uruguay is small (about 700,000 hectares) and the major agricultural regions
are located in the west and southwest of the country; the barley production is concentrated in those regions.
Given the relatively small size of the barley growing area in Uruguay, the climate variables recorded in the
different meteorological stations within the area are very similar; therefore, one site was selected for the study:
"La Estanzuela" Experimental Station (34° 27' S, 57° 46' W) (Figure 1).
The soils are heterogeneous and important variability exists with respect to soil water-holding capacity,
the ability to supply nitrogen through mineralization, and the ease with which the soils can be tilled. The
dominant soils in the area are Mollisols and Vertisols. Typical ranges of the most important chemical
characteristics of unfertilized soils of the area are: pH 5.5-6.5; organic mater 25-60 g/kg; 2.0-4.0 mg/kg available
phosphorus (Bray 1); 0.25-0.60 cmol(+)/kg of potassium. Most Mollisols in the area have more than 25% of
clay in the superficial horizon and usually a heavy textured B horizon (more than 40% clay). The Vertisols
are less differentiated and usually present more than 30% of clay through all the profile. Although soil
variability in the region is important, for this simulation study we selected one representative soil profile that
accurately represents the textures and water capacities of the main agricultural soils in the region. The
available nitrogen content in the soil was varied to create different nitrogen conditions that represent the
agricultural soils and management practices in the area.
Management Practices and Nitrogen Available for the Crop
Typical farms in the area produce crops and raise livestock. Consequently, the grain crops and
pastures are rotated. For example, three years of crops (such as wheat, barley, sunflower and sorghum)
alternate with three years of pasture (typically a mixture of white clover, red clover, birdsfoot trefoil and tall
fescue). As a result of this system, soils have a variable ability to supply nitrogen to the crops, depending on:
(a) time since the last pasture was plowed, (b) soil tillage, and (c) soil type. Soil tillage is especially important
for barley production because the planting time of the recommended cultivars coincides with the time of the
year when soil tillage is the most difficult (Figure 2). Thus, years with low rainfall during summer and fall
usually result in good soil tillage, high nitrogen levels, and high yields. Years with heavy rainfall during soil
preparation often result in inadequate seedbeds, low plant populations, poor natural soil nitrogen supply, and
low grain yields.
Winter crops in the west and southwest can be planted as early as April (cultivars with photoperiod
sensitivity) and as late as August. With the exception of a few newly developed cultivars, barley cannot be
planted earlier due to lodging problems. This simulation study considered three possible planting dates and
two levels of available nitrogen to construct simulations that represent the wide variety of conditions used for
barley production in Uruguay.
METHODS
Baseline Climate Data
The baseline climate data were obtained from Mr. Ricardo Romero, Agrometeorologist of the La
Estanzuela Experimental Station of INIA (National Agricultural Research Institute of Uruguay). The period
included in the dataset was 1 January, 1966 to 31 December 1989. The solar radiation values were obtained
from the sunshine hours based on the total possible hours and the observed hours using the WGEN program
(Richardson and Wright 1984). Mean monthly temperature, precipitation and solar radiation for the period
are shown in Table 1. The temperature regime is seasonal and the largest monthly precipitation corresponds
to the late summer.
URUGUAY-4
-------
Climate Change Scenarios for the Region
Global Climate Models (GCMs) were used to derive climate scenarios. Runs of these models were
used to estimate change in climate variables (temperature, precipitation, and solar radiation). These changes
were then used to modify the baseline climate and produce the climate change scenarios at each site. The
equilibrium GCMs used were: Goddard Institute for Space Studies (GISS, Hansen et al. 1983); Geophysical
Fluid Dynamics Laboratory (GFDL, Manabe and Wetherald 1987); and United Kingdom Meteorological
Office (UKMO, Wilson and Mitchell 1987). Transient scenarios were created from the GISS transient model
simulations (Hansen et al. 1988).
The following climate scenarios were run: (a) baseline; (b) GCMs (GISS, GFDL, and UKMO)
doubled CO2 climate change; (both (a) and (b) with and without the direct effects of CO2 on photosynthesis);
(c) sensitivity studies with combinations of 0, +2°C, +4°C, and 0, +20%, -20% precipitation, over the
baseline, each with and without direct CO2 effects; (d) transient climate change for the 2010s, 2030s and 2050s
using the GISS model; and (e) adaptive responses, e.g. the use of soil and crop management practices that are
better adapted to the possible new climatic conditions.
The three GCMs used to predict climate changes under doubled CO2 conditions show similar trends
with respect to the average mean temperature (Figure 3a). All three models predict an increase in the monthly
average of about 5°C. The UKMO model predict the largest increase (about 6°C); GFDL produces the
smallest change (approximately 4°C).
The GCMs show contrasting trends for precipitation (Figure 4a). Both the GISS and UKMO models
predict a general increase in total precipitation. Increases in rainfall are predicted for spring and fall, although
the GISS model also predicts a considerable increase in rainfall for the winter months. The GISS and UKMO
models predict higher means for annual precipitation than the baseline scenario - approximately 150 and 300
mm higher, respectively. In contrast, the GFDL model predicts a small decrease in total precipitation for most
of the fall and winter months except for September (Figure 4). Excluding this month, the mean annual
precipitation predicted with the GFDL model is about 50 mm lower than the baseline.
All three models predict very small changes in solar radiation (Figure 5a). The UKMO model projects
a slightly larger solar radiation value for January and February, approximately 2 MJ/m2.
Because the UKMO model predicts the most unfavorable climate changes (largest temperature and
precipitation increases) for barley production in the area it was used for the adaptive response studies
discussed earlier. It was also the only GCM included in this study in which a model gridbox included the entire
land area of Uruguay.
The GISS model was also used to predict transient climate changes for the 2010s, 2030s, and 2050s.
Figures 3b, 4b and 5b show the results of these predictions for temperature, rainfall, and solar radiation,
respectively. The results indicate a gradual temperature increase through the 2010s, 2030s, and 2050s, reaching
the maximum at the doubled CO2 scenario. The variation in the total precipitation was less systematic and
showed a very high rainfall value for March in the 2030s (GCM regional precipitation projections are highly
uncertain). As expected, the predicted solar radiation values showed almost no change from the baseline data
over the three decades.
Crop Model and Simulation Strategies
An International Fertilizer Development Center (IFDC) - United Nations Development Programme
(UNDP) global project was started in 1990, which will include the validation and regional adaptation of the
wheat, rice, maize, soybean, sunflower, and barley CERES models for further studies. The CERES-Barley
model, based on the CERES generic model, was used for all the simulation activities reported here (IBSNAT
1989). The soil used for the modeling activities, (fine, messic, typic, Argiudoll) was the same soil that was used
URUGUAY-5
-------
for the model validation. A complete description of the chemical and physical properties for 10 cm layers
throughout the soil profile is shown in Table 2.
The simulations are described in Table 3. Three possible fertilization strategies were used: (a) no N
fertilizer; (b) 60 kg N/ha at planting + 60 kg/ha at the end of tillering (a common practice in the region); and
(c) optimal N fertilization. This last strategy simulates the situation of having a diagnostic tool available to
ensure optimal N fertilizer application, one of the key objectives of the barley research mentioned above. Also,
given the long planting season for winter crops, three different planting dates were used to run the model: 1
June (early), 21 July (normal), and 28 August (late).
Frequent rains in summer and in fall result in problems with the soil tillage for winter crops. Common
consequences are inadequate seedbeds and a very compacted layer at 20-25 cm. The contraction is caused by
numerous tillage operations performed in a short period of time. An attempt was made to simulate this
situation by setting the value of the weighting factor to determine new root growth (WR in the CERES
models) equal to zero for all soil layers below 30 cm. However, the model was insensitive to the simulated
poor soil preparation. Setting the value of the weighing factor to determine new root growth (WR in the
CERES model) equal to 0.0 for all soil layers below 30 cm did not affect the results in crop production (data
not shown). Therefore, the model was not used to perform any further simulations changing soil preparation.
Since the climate change scenarios are associated with higher levels of CO2 (designated as 555 ppm
in the figures, representing an equivalent doubling of CO2) than the current climate GCM simulations (330
ppm in the figures), the study includes simulations of the physiological effects of CO2 on barley. Higher levels
of atmospheric CO2 have been found to increase photosynthesis and water use efficiency, resulting in yield
increases in experimental settings (Acock and Allen 1985; Cure 1985).
Validation and Calibration of the Crop Model
A complete dataset is available in Uruguay that is suitable for calibrating and validating the CERES-
Barley model. All the crop production variables (dry matter production, grain yield, kernel weight, grains/m^
grains/ear, etc.) of the CERES model were validated with this dataset. However, the phenology variables
(anthesis dates and maturity dates) were also tested with results from three years of cultivar trials conducted
by the Department of Plant Breeding of INLA, The phonological variables were calibrated and validated for
planting dates similar to the early, normal and late dates used in the simulation studies.
The CERES-Barley model validation procedures for this study were performed in two stages: (a)
determination of the genetic coefficients and validation of the phonological variables of the model; and (b)
validation of the crop production variables. Data from several barley planting date trials were used for the first
stage. The genetic coefficients, P1V and P1D, were adjusted for the cultivar CLE-116, a newly developed
cultivar from INIA, using three years of data. The model's accuracy in predicting anthesis and maturity dates
was excellent. The difference between observed and simulated dates for all years and planting dates included
in the study was always less than three days. The CERES model also accurately predicted the end of tillering
in most situations.
The validation of the crop production variables was performed using data from 1991 research. The
most relevant results of this validation are shown in Figures 2a-2g. For example, there was a high correlation
(r>0.9, P<0.01) between the simulated and observed results for grain yield, total biomass production, straw
production, grains/m2, and kernel weight. However, the model did not simulate grains/ear or for nitrogen
content of the grain as well. Considering the grains/ear, the principal problem was that the model did not
correctly predict the number of spikes per m2 (data not shown). In all treatments, the predicted number of
spikes/m2 was always lower than the observed data, indicating a problem with the model routines that calculate
tiller production, or the number of tillers that produce spikes. Observation of the predicted number of tillers
showed that, in most cases, the simulated values were similar to the ones observed, and in some cases, the
URUGUAY-6
-------
predicted values were higher than those observed. This suggests that a problem may exist predicting the
proportion of tillers that will produce spikes, a topic to be studied in further validation research.
The reason for the poor performance of the model in predicting the nitrogen content of the grain is
not clear. Problems may exist during the grain-filling stage because the length of this stage greatly affects the
N content of the grain. Finally, the kernel weight and the N content of the grain simulated by the model
showed a smaller sensitivity to N fertilizer than that observed in the field research. The kernel weight
depended only on the value of G2 (kernel weight genetic coefficient) and was constant for all the N
treatments. Similarly, the N content of the grain observed in the field varied more than the values predicted
by the model. Although these limitations of the model justify further work on the routines and in the
validation procedures, the general performance of the CERES-Barley model was very good. This was
particularly true for simulating grain yield, biomass production, and dates of anthesis and maturity.
RESULTS AND DISCUSSION
Effects of 2xCO2 GCM Climate Change on Barley Production
Yield. Figures 6a and 6b show the effects of the climate change predicted with the three GCMs on
barley production. Without nitrogen fertilization, the GCMs predicted a 50% mean decrease in grain yield.
When nitrogen fertilizer was applied at rates and timings that are similar to Uruguayan farmers' practices (60
kg N/Ha at planting + 60 kg N/ha at tillering), the yield was reduced by 40%. In all of the climate scenarios
tested, N fertilizer affected the grain yields, especially in the normal (NP) and late (LP) planting dates. Figure
6 indicates that normal planting date yields doubled and that late planting yields tripled with N fertilizer.
The interaction of planting date and nitrogen fertilizer for the baseline and the three GCMs resulted
in a substantial difference between early planted (EP) and NP barley yields when no fertilizer was applied. In
contrast, the yields of EP and NP were veiy similar when N fertilizer was used. A possible explanation of these
results is that early planted barley remains in the vegetative growth stage for a longer period of time than NP
barley. This gives the soil more time to supply larger quantities of N (through mineralization) at a stage when
N requirements are relatively low. When barley is planted at a normal date, the crop is much more dependent
on fertilizer N.
Direct CO2 effects. Figure 6b shows the physiological effects of CO2 on crop growth. The results
indicate that the baseline yields increased by 20% and the GCM climate change scenario yields increased by
30%. The yields simulated with the GCM scenarios that included the direct effects of CO2 were 30% lower
than the yields of the baseline climate at 330 ppm of CO2 concentration.
Yield variability. Baseline yields were less variable than yields simulated with three GCM scenarios
(Figures 7a and 7b). However, the main source of variability in grain yields was planting date. Figure 7 shows
that the CV increased from 20% to almost 50% when the planting time was delayed to a later date. Also, the
CVs for the GCMs were much larger those for the baseline at late planting dates. Finally, N fertilization did
not affect variability of the grain yield: the CVs with or without fertilizer were similar.
Biomass and season length. Decreases in barley grain yield can be caused by a reduction in the total
biomass production (which is highly correlated with grain yields), and/or by a reduction in the grain filling
period. To explain the observed reductions of grain yield, we studied the effect of the GCMs on the total
production of barley biomass (Figures 8a and 8b). The experiments showed that the biomass production
changed in proportion to grain yields, and as a result, the barley harvest index (grain yield/total biomass,
expressed as a percent) did not change. Therefore, reductions in grain yield were mainly due to reductions in
the production of total dry matter. The decrease in total biomass production was caused by a shorter growing
season (Figure 9). The figure indicates that: (a) the period from emergence to maturity was shortened by a
delay in the planting date; and (b) for any given planting date, all of the GCM scenarios caused a reduction
URUGUAY-7
-------
of the period equal to 30 days for EP, 20 days for NP, and 15 days for LP - about 20% in all cases.
Sensitivity Analysis
The results in Figures 3 and 4 indicate that the GCMs predicted important changes in the temperature
and total precipitation. A sensitivity analysis was performed to try to identify which of these variables was
responsible for the negative effect on the grain yields (Figures lOa and lOb). The analysis indicates that the
barley grain yields were much more sensitive to temperature changes than to precipitation changes. This is
expected since Uruguay does not have periods of water shortages in the winter. As mentioned above, one
factor causing reductions in winter crop yields is the excess of water in the soil. In all of the simulation runs,
there were no years with periods of water stress in the vegetative stage and very few years with periods of mild
water stress during the reproductive stage.
The results of the sensitivity analysis indicate that the mean grain yields were reduced by 0.45 t ha"1
for each 1°C of mean temperature increase. This reduction is similar for the three tested precipitation
scenarios. Grain yields decreased 0.101 ha"1 for each 10% decrease in total precipitation and did not decrease
with the potential period of excess of water in the soil.
Figure 11 shows the results of the sensitivity analysis on total biomass production. The effects of
temperature and precipitation changes were similar to the effects on the grain yields, suggesting that the
reduction in grain yield - caused by an increase in the mean temperature - is a consequence of a reduction in
the total biomass produced. Figure 12 shows the effect a of temperature increase on the length of the period
from emergence to maturity. The results indicate that the crop growing season was shortened by five days for
each 1°C increase in the mean temperature.
Transient Analysis
The analysis of the transient climate change using the GISS model indicates a gradual increase in the
mean temperature through the 2010s, 2030s and 2050s. As expected, the transient analysis of barley production
indicated a reduction in grain yields and an increase in yield variability (Figures 13a and 13b). Barley grain
yields were reduced by 1.7 t ha"1 from the baseline for the 2010s, and by an additional 0.35 t ha"1 for each of
the following decades. There is no apparent reason for this differential reduction since the temperature change
predicted for the barley-growing season was gradual.
Adaptation Strategies to Climate Change
Adaptive responses were tested to mitigate the negative effects of the predicted climate changes of
theUKMO model, including the improvement of nitrogen fertilizer management and the creation of a cultivar
with a longer grain-filling period. An additional cultivar which is more adapted to earlier planting dates was
also simulated. The results of these runs indicate the importance of the these alternatives (Figures 14a and
14b). The use of a more adequate fertilizer regime produced an increase in yield of 1 t ha"1 over the
unfertilized crop. An additional 11 ha"1 increase was achieved without N limitations. Finally, another 11 ha"1
increase was gained by using the optimum N management with an improved "hypothetical cultivar" designed
with a higher number of kernels/m2, and a longer grain filling period.
The adaptive response analysis shows that using improved crop management practices would increase
currently attainable grain yields of 2.5-3.0 T/Ha (baseline with 60 kg N/ha at planting + 60 kg N/ha at tillering,
in Figure 14a), and results in a mean grain yield of 5.0 T/Ha, even under the unfavorable weather conditions
predicted by the UKMO model. However, Figure 14a also shows that the mean grain yield achieved with these
same crop management practices would be 6 t ha"1 with 330 ppm CO2, or 7.5 T/ha with direct effects of
URUGUAY-8
-------
doubling the CO2 concentration - indicating a potential loss of 1.0-2.5 t ha"1 under the predicted climate
changes.
A second set of simulation runs to compare the barley response to nitrogen fertilizer under baseline
and UKMO conditions. For the baseline runs, CLE-116 was used at the normal planting date. The same
cultivar was used for the UKMO runs at the early planting date. The planting date was changed for the
UKMO runs because the model had predicted an increase in the mean temperature, which in turn resulted
in a shorter growing season. The probable response of farmers to this change would be to plant at earlier dates
and avoid high temperatures during the grain-filling period. The results presented for the climate change
scenario included the physiological CO2 effects.
Figure 15 shows the results of testing four nitrogen fertilizer rates. The response curve was adjusted
with linear regression analysis. Because the baseline response curve was much steeper than the UKMO curve,
the baseline maximum yield was more than 11 ha"1 higher than the UKMO maximum yield. Also, the amount
of N fertilizer needed to attain the maximum grain yield under UKMO was 2.6 times more than the amount
required to attain the same yield under current climatic conditions.
CONCLUSIONS
The climate changes predicted by the three GCMs are increased mean temperature, unchanged or
increased precipitation, and unchanged solar radiation. These climate conditions resulted in reduced crop
growing seasons and lower total biomass production and grain yields for barley in Uruguay. The negative
effects of the GCMs were worse when no nitrogen fertilizer was applied, and/or when the planting date was
delayed. The variability of grain yields was greater under the GCM conditions and when the planting dates
were delayed, but it was not affected by N fertilization.
The sensitivity analysis indicated that the decrease in grain yields is related to the reduction of the
barley-growing season. This reduction results from the temperature increase predicted by the GCM climate
change scenarios. In the major barley-growing region of Uruguay, the estimated yield effect of increasing the
mean temperature by 1°C is equivalent to the estimated effect of a 40% reduction of the total precipitation.
The adaptive response studies indicated that although there is a good potential for developing adaptive
management practices for the predicted climate changes, significant losses in crop production can still be
expected.
URUGUAY^
-------
REFERENCES
Acock, B. and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C.
Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure (eds.). Direct
Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S. Department of Energy.
Washington, D.C.
Hansen, J., I. Fung, A. Lascis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1989. Global Climate Changes
as Forecasted by Goddard Institute for Space Studies Three-Dimensional Model. Journal of
Geophysical Research, 93: 9341-9364.
Hansen, X, G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy and L. Travis. 1983. Efficient Three-
Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather Review, Vol
III, No. 4: 609-662.
IBSNAT. 1989. International Benchmark Sites Network for Agrotechnology Transfer Project. Decision Support
System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil Sci., College
of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
IPCC 1990. Intergovernmental Panel on Climate Change. First Assessment Report.
Manabe, S. and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.
Parry, M.L., T.R. Carter and N.T. Konijn. 1988. The impact of climatic variations on agriculture. Vol 1
Assessments in cool temperate and cold regions. Vol 2 Assessments in semi-arid region. Kluwer,
Dordecht, Netherlands. 876 pp. and 764 pp.
Richardson, CW. and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Wilson, C.A. and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
URUGUAY-10
-------
Table 1. Mean temperature, precipitation, and solar radiation for baseline data in La Estanzuela,
Uruguay (1966-1989).
Mean Total
Month Temperature Precipitation
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
°C
21.1
20.3
18.5
15.7
12.6
9.6
9.8
10.5
12.2
14.4
16.6
19.8
mm
77.6
85.8
107.0
58.7
59.5
62.3
67.2
73.0
76.6
96.7
80.7
78.1
Solar
Radiation
MJ/m2
23.0
20.8
16.9
13.0
9.4
7.3
8.0
10.7
14.4
18.5
21.6
23.3
URUGUAY-11
-------
Table 2.
Description of the soil used for modeling activities.
Layer
cm
0-5
5-10
10-20
20-30
30-40
40-50
50-60
60-70
70-80
80-90
90-100
LL
DUL
SAT
?D
OC
SW
LL
0.124
0.124
0.119
0.127
0.186
0.200
0.199
0.192
0.193
0.185
0.186
DUL
cm3/cm
0.320
0.320
0.290
0.300
0.290
0.310
0.310
0.300
0.290
0.280
0.270
= Lower limit of soil
=
Drained
SAT
3
0.350
0.350
0.320
0.350
0.350
0.360
0.360
0.350
0.340
0.320
0.310
SW
0.320
0.320
0.290
0.300
0.290
0.310
0.310
0.300
0.290
0.280
0.270
BD
g/cm3
1.28
1.28
1.32
1.32
1.36
1.40
1.40
1.42
1.42
1.42
1.42
OC
%
2.08
2.08
2.08
2.08
.78
.78
.95
.95
.14
.14
.14
NH4
5.0
5.0
5.0
2.0
2.0
1.0
1.0
1.0
1.0
1.0
1.0
NO3
mg/kg
20.0
15.0
10.0
5.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
PH
5.6
5.6
5.6
5.6
6.1
6.1
6.6
6.6
6.6
6.8
6.8
extractable water
upper limit of soil water content
= Saturated water content
= Bulk density
=
Organic
Carbon
= Initial soil water content
URUGUAY-12
-------
Table 3.
Management practices used in the simulations.
a) Baseline and GCMs (x2 CO2) with and without direct effects of doubled CO2 concentration
TOTAL: 48 runs
Planting Date
Early (1-June)
Normal (21-July)
Late (28-August)
Early (1-June)
Normal (21-July)
Late (28-August)
N fertilizer
No N Fertilizer
No N Fertilizer
No N Fertilizer
60 + 60 Kg N/ha (Plant. + Till.)
60 + 60 Kg N/ha (Plant. + Till.)
60 + 60 Kg N/ha (Plant. + Till.)
b) Sensitivity analysis (0,+2,+4 °C X 0,-20,+20% mm) with and without 2xCO2 TOTAL: 18 runs
Planting Date N fertilizer
Normal (21-July) 60 + 60 Kg N/ha
c) Transient climate change with and without 2xCO2 TOTAL: 6 runs
Planting Date N fertilizer
Normal (21-July) 60 + 60 Kg N/ha
d) Adaptive Response with and without 2xCO2 TOTAL 12 runs
Planting Date
Normal (21-July)
Normal
Early
Normal
Normal
Early
N fertilizer
60 + 60 Kg N/ha
Optimal
Optimal
Optimal
Optimal
Optimal
Cultivar
CLE-116
CLE-116
CLE-116
Improved #!(*)
Improved #2
Improved #3
Improved cultivar #1 = CLE-116 with increased grains/m2
Improved cultivar #2 = Improved #1 with longer grain filling period
Improved cultivar #3 = Improved #2 better adapted to early planting
URUGUAY-13
-------
LLJ
O
OC
111 CO
O O
~Z. LL.
O
D)
L.
E
£
O
LLJ
Q
O
O
O
CL
LJJ
CO
O
DC
Q_
<
DC
CO
LLJ
LL
CO
CO
LJJ
O
X
LJJ
CO
1
LJJ
CO
I
O
LJJ
DC
*«^
CO
1
CL
Z> O
^ LJJ
5 DC
CO
H
LU
DC
1
1
CL
L
L
f
L
«
^
L
LI
~\
-1
>
>
ooooooooo
CO'^CMOCOCO'sJ-CM
-------
O
> c
«3S
•go
CO
ra w »•
I
rt
-C
1 I
f
»
3
o
a
•V
v-
*
\
•O
J)
O
C
.g
i *
cd
33
15
"o
•a
o
m
CO
LU
DC
LJLJ
O
di
s
1
°a
si
a-3
« e
_ w S
- ^^
o
I
•5
I
-------
II
"
I
TJ «-
«H 0)
^2 O.
_ to
•§i
^6
10
a
m
10
at
•t
g
2
sl
S
»«
R
O
O
CM
D)
LL
I
O)
£
K> O.
-------
LJJ
o
I
o
UJ O
fe =
< •£
*sr co
*>. v_
~"! 0
UJ
CO
DC
H
"3
111
O
X 0
Ob
Hi 'ra
II
-J ,0
o «-
o
o
"co
o
§
Q.
I
C
CO
o
Q_
UJ
CO
_J X
Z> H
-^ z
o
-- <
o
CO
ID
CM
O ID O
CM T- T-
O
S
o>
Q.
I
C
CO
o
o
•z.
Q.
CD
CO
_. 3
C
-- CO
o
CM
_j CO O
LU 52 o
Q O CM
O
2
\ \
LJJ O
CO CO
< O
m CM
C
0)
"co
C
cc
"O
C
03
O
-------
LLJ
O
z
<
o
LU
H
UJ
CO
<
DC
c
g
*s
•t?
'g.
'o
Q.
Ill
CO
_J I
13 V-
cc
o
10
CM
o
o
CM
o
o
o
ID
o
CM
Q O CM
o
UJ O
CO CO
< O
m CM
c
0)
'co
C
05
C
03
O
o
c
o
•+-»
03
LJLJ
O
<
X c
0.9
< .9-
O
O
O
'co
E
v^>
c
g
*-•
CO
o
o
O
Q.
UJ
CO
_J I
Z) H
DC
CO
CO
o
O ID
H ! I
o
^ Q
CO '-'
CO U.
m O
g.
"o
0)
ol
0
O)
C
03
6
D)
LL
o o o o o
ID O ID O IO
CM CM i- -i-
-------
LLJ
O
LJLJ .2
DC
v_
h- -^
5 °
dgco
CO
<
DC
H
JQ
LLI
O
< C
I .2
0 «
HI TJ
H «3
< DC
2 V.
- co
O
O
CM
C
O
CO
DC
O
CO
O
CO
10
CM
O U> O
CM T- T-
IO
CM
C
O
u->
CO
O
CO
|
DL
LU
CO
_J X
13 I-
-> Z
O
Is
DC
<
t <
O
Q.
LU
CO
_l I
3 H
~3 Z
O
-- <
_
LU
Q
O
LU
Q
O
O
CM
CO
CO
o
o
ID
o
CM
I !
LU O
CO CO
< O
CO CM
CO 5
CO 6
I !
o
I o
CO U-
m O
co
C
03
O
c
O
TJ
03
QC
O
CO
LJL
O
co
10
CM
o 10 o
CM T- T-
-------
Fig. 6a: BASELINE AND GCMs
MEAN GRAIN YIELDS
GRAIN YIELD (T/ha)
EP-NO NP-NO LP-NO EP-N NP-N LP-N
STRATEGY
MODEL
BASE
GFDL
GISS
UKMO
Fig. 6b: BASELINE AND GCMS
MEAN GRAIN YIELDS
GRAIN YIELD (T/ha)
6-
5-
-40%
EP-NO NP-NO LP-NO EP-N NP-N LP-N
STRATEGY
MODEL (2 X CO2)
EZ1 BASX 8SS3 GISX
Hi! GFDLX • UKMX
-------
Fig. 7a: BASELINE AND GCMs
CV % GRAIN YIELDS
YIELD C.V. (%)
EP-NO NP-NO LP-NO EP-N
STRATEGY
NP-N
LP-N
MODEL
BASE
GFDL
GISS
UKMO
Fig. 7b: BASELINE AND GCMS
CV % GRAIN YIELDS
YIELD c.v. (%)
so -ri
EP-NO NP-NO LP-NO EP-N NP-N LP-N
STRATEGY
MODEL (2 X CO2)
BASX K\\\\N GISX
GFDLX HH UKMX
-------
Fig. 8a: BASELINE AND GCMS
MEAN BIOMASS PRODUCTION
BIOMASS (T/ha)
16 -d
EP-NO NP-NO LP-NO EP-N NP-N LP-N
STRATEGY
cm
™
BASE
GFDL
MODEL
l\\\\\l GISS
Wf& UKMO
Fig. 8b: BASELINE AND GCMS
MEAN BIOMASS PRODUCTION
BIOMASS (T/ha)
16 -d
14-
EP-NO NP-NO LP-NO EP-N NP-N LP-N
STRATEGY
MODEL (2 X CO2)
BASEX ESS3 GISSX
GFDLX W% UKMOX
-------
CO
<
CD
LU
LU
O
DC
LU
E LU
LU
LU
0
§5
QC
LU
o o o o o o
CD ^f CM O 00 CD
CO
CO
o
LU
Q
O
LU
CO
<
CD
-------
Fig. 10a: SENSITIVITY ANALYSIS
MEAN GRAIN YIELDS
GRAIN YIELD (T/ha)
+2
+4 0+2+40
TEMPERATURE CHANGE
+2 +4
CO2 CONCENTRATION
M YSENS SSSI YSENX
Fig. 10b: SENSITIVITY ANALYSIS
C.V. GRAIN YIELDS
C.V. YIELD (%)
+4 0+2+40
TEMPERATURE CHANGE
+2 +4
CO2 CONCENTRATION
^ SSOppm ^^ 555ppm
-------
O
CO h-
< O
Q_
>- CO
\- CO
=: <
— o
UJ
CO <
.. LU
D)
LL
CO
CO
g
CQ
C02 CONCENTRATION
555ppm
I
£
Q.
Q.
O
CO
CO
1
\ i I I \ I \
CD ^t CV! O CO CD -sf CM O
-------
Fig. 12: SENSITIVITY ANALYSIS
EMERGENCE - MATURITY LENGTH
DAYS EM - MAT
0 +2 +4
TEMPERATURE CHANGE
-------
Fig. I3a: TRANSIENT ANALYSIS (GISS)
MEAN GRAIN YIELDS
GRAIN YIELD (T/ha)
5-
4-
3-
2-
1
0
1
BASELINE 2010'S
2030'S
2050's
CO2 CONCENTRATION
S! 330ppm 555ppm
Fig. I3b: TRANSIENT ANALYSIS (GISS)
C.V. GRAIN YIELDS
c.v. (%)
BASELINE 2010'S
2030's
2050's
CO2 CONCENTRATION
S 330ppm 555ppm
-------
Fig. I4a: ADAPTIVE RESPONSE (UKMO)
MEAN GRAIN YIELDS
8
7-
6-
5-
4-
3-
2-
1-
GRAIN YIELD (T/HA)
NO GC
IMPR. N + CULTIVAR
IMPROVE is!
V-O N 0 V-O 60+60 V-O Nopt V-1 Nopt V-1 Nopt
CO2 CONCENTRATION
H 330ppm K\\\\M 555ppm
Fig. I4b: ADAPTIVE RESPONSE (UKMO)
C.V. GRAIN YIELDS
c.v. (%)
NOGC
IMPROVE N PERT.
IMPR. N + etlLTIVAR
I
V-O N 0 V-O 60+60 V-O Nopt V-1 Nopt V-1 Nopt
CO2 CONCENTRATION
H 330ppm K\\\\N 555ppm
-------
Fig. 15
Grain Yield (T/ha)
4 - 3.48 T/ha
1
0
96 kg N/ha
37 kg N/ha
0
40 80
Fertilizer (kg N/ha)
120
BASELINE -a- UKMO (2xCO2)
NP £P
-------
-------
SECTION 4: EUROPE
-------
-------
POSSIBLE EFFECTS OF INCREASING C02 CONCENTRATION
ON WHEAT AND MAIZE CROPS IN NORTH AND SOUTHEAST FRANCE
R. Delecolle, D. Ripoche
Station de Bioclimatologie INRA, France
F. Ruget, G. Gosse
Station de Bioclimatologie INRA, France
FRANCE-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Agricultural regions and systems
METHODS
Climate
Crops Models and Management Variables
Calibration of the crop models
Climate change scenarios
RESULTS WITHOUT ADAPTATION
Wheat
Maize
RESULTS WITH ADAPTATION
Wheat
Maize
DISCUSSION
REFERENCES
FRANCE-2
-------
ACKNOWLEDGMENTS
The authors are grateful to Claude Varlet-Grancher and Pierre Pluchard (both from INRA) for
providing wheat data.
FRANCE-3
-------
SUMMARY
This study evaluated the potential effects of climate change due to increased trace gas concentrations
in the atmosphere on wheat and maize yields in France. Sensitivity studies showed that yields generally
decreased with increasing temperature, probably due to a shorter crop-growing season. When the direct effects
of CO2 on the crop were included, yield decreases were somewhat counteracted.
Under both temperate and mediterranean climates in France, winter cereal yields did not decrease
under future climate conditions (provided that irrigation supply was not limiting under dry conditions). Under
temperate climate, maize yields increased, while under mediterranean conditions, the reduction of phase
duration induced a drastic yield decrease, even under optimal irrigation. Adaptation simulations (change in
planting date) produced slightly higher yields. If these results are realized with global warming, maize may be
farmed in France more extensively in the future.
INTRODUCTION
This study was part of an international endeavor to assess the possible effects of global climate change
on world food production in the coming century. Any study of this type must account for the climate
modifications induced by increasing trace gas concentrations in the atmosphere, which are simulated by general
circulation models (GCMs), and the direct effects of increasing carbon dioxide concentration on plant
physiology. These studies are particularly interesting in northern and middle Europe, where cropping systems
will likely shift, within a few decades, from high-input production to more extensive practices, due to an
increasing concern for pollution costs and over-production. It is therefore necessary to estimate whether
climate change will create extra costs (irrigation, pesticides), or if increasing CO2 will promote crop production
at steady costs. In this study, we used the CERES plant process models to estimate the response of wheat and
maize crops in France and to address the effects of climate change on yield and water demand.
Agricultural Regions and Systems
The research was conducted in two different agricultural areas in France (Figure 1). These regions
were chosen because they represent two extremes of French agroclimatologic conditions. The northern plain
is characterized by intensive cropping practices, semi-oceanic climate, and little water stress. The soils are deep
silts with high field capacities. The most frequently grown crops are wheat and sugarbeet, with several
secondary crops such as peas, potatoes, and beans. Maize was introduced in the northernmost areas 15 years
ago.
The southeast has diverse crops. Fruits and vegetables, maize, sunflower, and sorghum are grown in
irrigated areas; wheat (soft and durum) and barley, are usually rainfed except on very shallow soils. The climate
is mediterranean, with drought periods during the summer, characterized by a large interannual variability. Soil
types are highly variable and generally shallow with low field capacities. In river valleys, however, alluvial
deposits and water tables insure better water supplies.
METHODS
Climate
For northern France, we used a timeseries of climate data from 1951 to 1980 (1984 for wheat) that
was available from Versailles-INRA (lat. 49°N). This weather station, albeit located in the southern part of
FRANCE-4
-------
the study area, provides data that are representative of the area. For the South, the Avignon-INRA weather
station (lat. 44.9°N) provided data from 1968 to 1984.
Crop Models and Management Variables
Wheat and maize were the sample crops for this study. These crops provide a physiological comparison
between C3 and C4 crop responses to the direct effects of increased COV and a comparison between winter
and spring crops for testing the influence of indirect climatic effects on season length and water requirements.
Both crops were tested at each site.
The choice of genotypes was guided by the availability of calibration data. For wheat, we chose one
genotype per location to account for the specific conditions and farmers' practices at each site. We chose the
genotype Arminda (winter type, long cycle) for northern France, and the genotype Fidel (half-spring,
medium-length cycle) for the South. For maize, we used only one genotype, Dea, which can be grown in both
climates, with special regard to its insensitivity to photoperiod. It is often used as a control in productivity
tests.
Crop models. The CERES-Wheat (Godwin et al. 1989) and CERES-Maize (Ritchie et al. 1989) models
provided by the DSSAT package were used to simulate wheat and maize crops. Modified versions accounted
for higher-than-ambient CO2 concentration changes. Photosynthesis enhancement factors for various CO2
concentrations came from Allen et al. (1987).
Soils. For northern France, we used a standard "Deep Silt Loam" soil (IBSNAT 1989) tp represent
the soils found in this area. For the Southeast, we created a specific soil file from the soil properties that exist
in a typical alluvial area near Avignon-although farmers would grow mostly vegetables and other high-return
crops on this soil.
Irrigation. For all of the simulations, the initial soil water was set to 100% capacity for each layer.
This is a realistic assumption for both locations under present climatic conditions. When irrigation was applied,
the "automatic" option was chosen to provide the crop with a non-limiting situation.
Calibration of the Crop Models
Wheat. The CERES-Wheat genetic coefficients for both wheat genotypes were estimated from a
dataset from the Mons-en-Chaussee-INRA experiment station (49.8 °N). The data were representative of the
wheat area in the North. We used a sowing date experiment with three years of data and with eight
independent treatments. Experimental data were split into two sets-one for calibration and the other for
validation. The calibration was carried out in two phases. First, each parameter was calibrated directly from
observed results (phase durations, biometric ratios, growth rates) to obtain rough parameter values. Second,
these values were used in model runs and arbitrarily adjusted to attain a pseudo-best fit of observed data
(Table 1). The validation procedure showed acceptable results for Fidel, but underestimated yields for Arminda
by 13%, due to underestimation of final kernel weights.
Durations from sowing to anthesis were simulated adequately most of the time, but simulated anthesis-
to-maturity periods were generally too short in comparison to reality. These results were attained with values
of Gx coefficients which are different from what is suggested in DSSATs CERES-Wheat guide for European
wheats (Godwin et al. 1989). All growth coefficients are highly correlated, thus leading to potentially large
instabilities in estimated values-a crucial point for study in the future. Figure 2 displays the sensitivity of
simulated yields to values of P1D, Gl, and G2.
Maize. Two sets of data were used to study maize. The results from a planting-density trial in
Mons-en-Chauss6e were used for the calibration of genetic parameters and initial values (sowing depth). The
results from another density trial in Grignon (near Versailles) allowed us to test the consistency of estimated
FRANCE-5
-------
parameters. PI and sowing depth bave been adjusted to attain satisfactory anthesis dates and maximum leaf
area indices, and P2 was set to zero because Dea has been shown to be a non-photoperiodic genotype.
The calibration of parameters linked to grain number and grain filling was problematic. In France,
maize kernels rarely reach "physiological maturity" (P5). Also, the potential grain number was dramatically
increased to insure an adequate simulation of the final number. Finally, the grain-filling rate (G3) was set to
far below the potential value. Therefore, the set of parameters obtained (Table 2) is different from the values
indicated for our latitude (Ritchie et al. 1989).
Climate Change Scenarios
Climate change ratios from three GCMs have been applied to both weather series. Three equilibrium
GCMs were used: GISS (Goddard Institute for Space Studies; Hansen et al. 1983); GFDL (Geophysical Fluid
Dynamics Laboratory Model; Manabe and Wetherald 1987); and UKMO (United Kingdom Meteorological
Office Model; Wilson and Mitchell 1987). Transient climate scenarios were also created with the GISS
transient model (Hansen et al. 1988). Figure 3 shows the monthly averages of changes in required weather
variables for the three doubled CO2 and transient (2010s, 2030s, 2050s) scenarios.
Temperature. For both regions, temperatures simulated by GISS and GFDL lie 3-4 °C above the
present mean values, whereas UKMO predicts more drastic changes-more than 10°C during the summer in
Avignon, and more than 6°C during the winter in Versailles.
Precipitation. According to all scenarios, precipitation increases during the winter in Avignon and
throughout the year in Versailles. UKMO simulates a dramatic precipitation decrease during the summer in
the South. Individual monthly values for precipitation change are very chaotic in the Southeast for the GISS
and UKMO simulations, and in the North for GFDL.
Solar radiation. For UKMO and GISS, the simulations of solar radiation suggest an increase of about
10% over the year for both sites. GFDL simulations are very close to baseline values.
RESULTS WITHOUT ADAPTATION
Wheat
The following results correspond to current management practices: sowing on November 1 and plant
population of 250 plants m"2 for both sites.
Yield and season length. To synthesize the effects of global change on crop behavior (yield, days to
maturity), Figures 4-9 show scenarios arranged by the corresponding yearly mean increase in temperature. We
simulated both irrigated and rainfed production at each location. For the North (Versailles), the effect of
irrigation is negligible, so irrigated and rainfed treatments are represented by one set of points in Figures 4a
and 4b. The effects of climate change alone tended to decrease yields up to almost 30% (GFDL), probably
due to the shortened season length of up to one month (Figure 4b). When CO2 physiological effects were
added, this effect was counteracted; the yield changes were almost nonexistent (at least for the transient and
equilibrium GISS scenarios).
Figures 5a and 5b show the results for irrigated and rainfed wheat in Avignon. The irrigated results
were similar to those obtained from Versailles. Rainfed wheat was less adversely affected by climate change,
with the exception of the UKMO climate change scenario. When the direct effects of CO2 were included, all
scenarios produced increased yields except UKMO. Table 3 summarizes results of water consumption by the
crops as a percent variation from the baseline for Versailles (irrigated) and for Avignon (irrigated and rainfed).
Water consumption. The simulation results show the difference between the effects of climate change
alone, due to a shortening of the crop season, and the effects of climate change and increased CO2, due to
FRANCE-6
-------
increasing stomatal resistance. In both cases, the major reduction in evapotranspiration (ET) (Table 3) was
mostly attributable to climate effects (up to about two thirds), with a magnitude that varies according to
scenario. See also the sensitivity analysis (Appendix A). In the North, transient scenarios induced a small
reduction in ET, whereas equilibrium scenarios had a greater influence (up to a 22% reduction). For irrigated
wheat in the South, the overall effect on evapotranspiration could range from -10% to -20%. Rainfed crop
ET showed less dramatic reductions.
Maize
The following results correspond to sowings on May 1 in Versailles and April 15 in Avignon. The
plant population was 9 plants m'2 in both cases.
Yield and season length. Figures 6a and 6b summarize the results for yield and season length for maize
in Versailles, with the same representation as for wheat. Irrigation did not modify season length, which showed
a significant decline when the mean temperature roses, especially in the extreme case of UKMO. Surprisingly,
the yield response to climate change alone shows a marked optimum for conditions in 2010. This could be
interpreted as a consequence of the shift in development phases, which positioned the grain-filling period more
favorably in terms of available solar radiation. Even if season length was reduced, the yield was not
substantially decreased below the baseline level even in the least favorable conditions (equilibrium scenarios).
This trend was confirmed by sensitivity analysis results (Appendix B). Adding the CO2 physiological effect
increased yields up to 15-20% above the baseline.
Figure 7 shows the results from irrigated and rainfed maize at Versailles. Irrigation had only a slight
influence on the yields. (Irrigated baseline was 8.521 ha"1 compared with 8.431 ha"1 under rainfed conditions.)
Although changing the climate reduced the length of the development phases, it also increased the yields in
the rainfed case (Figure 8a), probably due to the same phenomenon that occurred in Versailles (better timing
of development to radiation availability). Adding direct CO2 effects led to a large increase in grain dry-matter
production. Rainfed baseline yields were very low (simulated 2.3 t ha"1). Yet if the simulation results are
reliable, a high CO2-high temperature environment would lead to high yields under rainfed conditions
(simulated 5.91 ha"1 for UKMO). With irrigation, maize yields showed a decline from 10 to 30% with climate
change alone, and adding CO2 effects did not consistently modify the results (Figure 9). The worst yields in
this case were still higher than the best rainfed simulations.
Water consumption. Table 4 summarizes simulated ET for maize. In Versailles, decreases in water use
largely depended on the assumed scenario. When direct CO2 effects were added, a 20% to 30% decrease in
water consumption occurred in the equilibrium and 2050 scenarios for irrigated and rainfed crops. These
results were generally similar to results for irrigated crops in Avignon.
RESULTS WITH ADAPTATION
Possible adaptations to future climate conditions were tested with a range of sowing dates to adjust
the timing of crop phonological events for photothermal time and radiation availability.
Wheat
For wheat, three sowing dates were tested for both locations: two of them earlier (September 1 and
October 1) than the regular date (November 1) and one later (December 1). A four-point response curve for
grain yield appears in Figure 10. Figure 10 also illustrates the grain dry-matter response to the sowing date,
representing (in percent of lxCO2 baseline value for November 1 sowing) the transient scenarios including
direct CO2 effects. For Versailles, the sowing date ded not help to counteract the effect of season length in
FRANCE-7
-------
the most pessimistic case (2050). Whatever the scenario, the yield difference between the first and last sowing
was about the largest that can be expected in an agronomic experiment at potential production level. In
Avignon, even earlier sowing dates improved the situation in the near future projections (compare September
1, October 1, November 1,2010s, and 2030s), but the initial soil water conditions would probably hinder this
solution.
Maize
Figure 11 illustrates the effects of two more sowing dates, 15 days earlier and 15 days later than the
regular date. For the North, it appears that an earlier sowing date would not modify the yield changes, but
delaying the planting date clearly reduced the benefit of climatic change. Earlier sowing dates present a
problem for spring crops that are limited by soil temperature. Additional studies on the possible warming of
upper soil layers under changing weather conditions are needed to study this adaptation.
In the South, postponing the planting dates would counterbalance the negative influence of climate
change, at least in the near future. This effect may be due to a better synchrony between phasic development
and available energy, as was suggested for Versailles. For the 2050 scenario, further study of an earlier sowing
(April 1) is necessary, assuming again that more information on soil temperature will be available.
DISCUSSION
The overall trends that can be deduced from this study are: (1) Season lengths will be shortened by
climate change; (2) Yields will decrease in proportion to the magnitude of warming, but may be partially
mitigated by direct CO2 effects (up to an increase of 5°C in temperature); and (3) Water use will decrease.
Two major questions remain: (1) Do the multiplicative coefficients used to represent high-CO2
photosynthetic response account for long-term plant behavior especially with possible acclimation? and (2)
How realistic is the simulation of the water budget in the crop models?
The calibration problems we encountered preclude straightforward interpretation of absolute values,
but the relative or differential results probably deserve more credit. Assuming that complete water repletion
of all soil layers at sowing may strongly temper the following statements, it may be concluded that: (1) Under
both temperate and mediterranean climates in France, winter cereal yields will not be decreased by future
conditions (provided that irrigation supply is not limiting under dry conditions); and (2) Under temperate
climate, maize could take advantage of development-phase shrinkage and improve its radiation-use efficiency.
The traditional assumption that C3 crops would take precedence over C4 crops because they use CO2 more
efficiently would become questionable in this case. Inversely, under mediterranean conditions, the reduction
of the phase duration may induce a drastic decrease in yields, even under optimal irrigation. Could maize
become a potential crop for extensive fanning in the two regions, as the simulations would suggest? The
answer depends in part on the ability of CERES-Maize to account for the effect of water stress on ear fertility
and the potential grain number.
Finally, the diversity of French climates and soils prohibits generalizing these results for the entire
country. The two sites chosen for this study represent extremes for intensive crops in France; more continental
situations should be studied to investigate the behavior of crops in the eastern agricultural plains regions under
future climate.
FRANCE-8
-------
REFERENCES
Allen L.H., KJ. 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 Biogeochimical Cycles, 1:1-14.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy and L. Travis. 1983. Efficient Three-
Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather Review, Vol
III, No. 4:609-662.
Hansen, J., I. Fung, A Lascis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1988. Global Climate Changes
as Forecasted by Goddard Institute for Space Studies Three-Dimensional Model. Journal of
Geophysical Research, 93:9341-9364.
IBSNAT 1989. DSSAT (Decision Support System for Agrotechnology Transfer) Version 2.1 User's Guide.
Honolulu,: Univ. of Hawaii.
Godwin D., J. Ritchie, U. Singh, and L. Hunt. 1989. A user's guide to CERES Wheat V. 2.10. Michigan State
University-IFDC-IBSNAT, Muscle Shoals, AL.
Manabe, S. and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.
Ritchie J., U. Singh, D. Godwin, and L. Hunt. 1989. A user's guide to CERES Maize V. 2.10. Michigan State
University-IFDC-IBSNAT, Muscle Shoals, AL.
Wilson, C.A. and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
FRANCE-9
-------
FIGURES CAPTIONS
Figure 1. Location of test sites (Versailles and Avignon) and calibration station
(Mons-en-Chauss6e) used for France.
Figure 2. Sensitivity analysis of wheat yield to parameters P1D, Gl and G2. Mean values (solid
line) and confidence intervals (dotted lines) are computed for all available years in
each weather series.
Figure 3. Changes in weather variables as simulated by GCMs (a,b: temperature; c,d: rainfall;
e,f: radiation).
Figure 4. Yield variation (in percent of lxCO2 baseline) and season length for wheat in
Versailles (squares = climate change alone; circles = climate change with
physiological CO2 effects).
Figure 5. Yield variation (in percent of lxCO2 baseline) and season length for wheat in
Avignon (squares = climate change alone; circles = climate change with
physiological CO2 effects).
Figure 6. Yield variation (in percent of lxCO2 baseline) and season length for rainfed maize
in Versailles (squares = climate change alone; circles = climate change with
physiological CO2 effects).
Figure 7. Yield variation (in percent of lxCO2 baseline) for irrigated maize in Versailles
(squares = climate change alone; circles = climate change with physiological CO2
effects).
Figure 8. Yield variation (in percent of lxCO2 baseline) and season length for irrigated maize
in Avignon (squares = climate change alone; circles = climate change with
physiological CO2 effects).
Figure 9. Yield variation (in percent of lxCO2 baseline) for irrigated maize in Avignon
(squares = climate change alone; circles = climate change with physiological CO2
effects).
Figure 10. Wheat-yield response to sowing date (in percent of lxCO2 baseline) for GISS transient
scenarios in Versailles and Avignon.
Figure 11. Maize-yield response to sowing date (in percent of lxCO2 baseline) for GISS transient
scenarios in Versailles and Avignon.
FRANCE-10
-------
Table 1.
Values of genetic coefficients used for wheat.
Genotype P1V P1D
Fidel
4.5 3.0
(0.5) (3.5)
Arminda 6.0 4.5
P1V
P1D
P5
Gl
G2
G3
(6.0) (3.5)
Vernalization coefficient
Photoperiodism coefficient
Grain filling duration coefficient
Kernel number coefficient
Kernel weight coefficient
Spike number coefficient
(Values recommended in DSSAT are
P5
4.5
(2.5)
4.5
(4.0)
shown between
Gl G2
4.5 1.5
(4.0) (3.0)
4.6 1.2
(4.0) (3.0)
brackets)
G3
2.5
(2.0)
1.7
(2.0)
FRANCE-11
-------
Table 2.
Values of genetic coefficients used for maize.
Genotype
PI
P2
P5
Gl
G2
Dea
PI
P2
P5
Gl
G2
180
(130)
Juvenile phase coefficient
Photoperiodism coefficient
0.0
(0.2)
640 800
(680) (780)
7
(8)
Grain-filling duration coefficient
Kernel number coefficient
Kernel weight coefficient
(Values recommended in DSSAT are
shown between brackets)
FRANCE-12
-------
Table 3.
Simulated total evapotranspiration for wheat (in percent base
Scenario
Climate Change Alone
Climate Change and
CO2 Physiol. Effects
BASE
GISS
GFDL
UKMO
2010
2030
2050
BASE
GISS
GFDL
UKMO
2010
2030
2050
BASE
GISS
GFDL
UKMO
0
-5
-16
-16
4
3
-1
0
-12
-12
-10
-6
-5
-10
0
-7
-9
-7
Versailles (irrigated)
-6
-12
-22
-16
1
-1
-7
Avignon (irrigated)
-8
-8
-18
-16
-8
-10
-16
Avignon (rainfed)
-4
-11
-13
-10
FRANCE-13
-------
Table 4.
Simulated total evapotranspiration for maize (in percent of base
Scenario
Climate Change Alone
Climate Change and
CO2 Physiol. Effects
BASE
GISS
6FDL
&KMO
2010
2030
2050
BASE
GISS
GFDL
UKMO
BASE
GISS
GEDL
UKMO
2010
2030
2050
0
-11
-16
-11
0
-4
-6
0
-10
-16
-11
0
-8
-9
-7
-6
-8
-9
Versailles (rainfed)
-16
-24
-29
-25
-5
-12
-19
Versailles (irrigated)
-17
-25
-29
-26
Avignon (irrigated)
-16
-23
-23
-20
-11
-16
-21
FRANCE-14
-------
Appendix A. Sensitivity analysis of GERES-Wheat to climate and CO2 changes.
CO,
Changes
Precip. Temp.
VERSAILLES
Yield Seas.L. ET
Yield
AVIGNON
Seas. L. ET
(ppm) (%) (°C)
330 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
555 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
(tfaa-1)
6.4
5.9
5.3
6.5
6.0
5.3
6.3
5.9
5.2
7.7
7.0
6.2
7.7
7.0
6.2
7.6
6.9
6.2
(days)
252
241
231
252
241
231
252
241
231
252
241
231
252
241
231
252
241
231
(mm)
483
460
437
491
466
442
470
449
429
454
431
410
458
435
412
445
424
405
6.32
6.3
5.8
6.8
6.6
5.9
5.5
5.8
5.5
7.9
7.7
6.9
8.2
7.9
7.0
7.1
7.3
6.7
216
205
194
216
205
194
216
205
194
216
205
194
216
205
194
216
205
194
421
396
375
440
410
387
394
375
357
403
376
357
418
387
366
380
361
343
FRANCE-15
-------
Appendix B. Sensitivity of CERES-Maize to climate and CO2 changes.
CO,
Changes
Precip. Temp.
VERSAILLES
Yield Seas. L.
ET
AVIGNON
Yield Seas. L. ET
(ppm) (%) (°C)
330 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
555 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
(t ha'1)
8.4
9.3
8.6
8.5
9.3
8.6
9.4
9.1
8.5
10.0
9.9
9.1
9.0
9.9
9.1
9.0
9.9
9.1
(days)
159
128
108
159
128
108
159
128
108
159
128
108
159
128
108
159
128
108
(mm)
442
416
384
450
423
389
432
405
376
369
349
323
375
354
328
361
342
318
10.9
9.5
8.5
10.9
9.5
8.5
10.8
9.5
8.5
11.6
10.2
9.1
11.6
10.1
9.0
11.5
10.2
9.1
109
97
89
109
97
89
108
97
89
109
97
89
109
97
89
108
97
89
458
428
408
461
432
411
453
424
405
385
361
345
389
363
346
381
356
341
FRANCE-16
-------
OMons
Versailles
Figure 1.
Location of test sites (Versailles and Avignon) and calibration station
(Mons-en-Chauss£e) used for France.
-------
AVIGNON cv. FIDEL
VERSAILLES cv. ARMINDA
o
JC
10
4
P1D
101-
0
2 34 5
P1 D
I
o
10
4
G1
10
234
G1
I
a
-------
Versailles
10 12
- G1SS — GFDL -a- UKMO
- Trans 2010 -*- Trans 2030 -*- Trans 2050
— GISS -— GFDL -*- UKMO
-s- Trans 2010 -K- Trans 2030 -*r Trans 2050
-GISS --GFDL -x-UKMO I
- Trans 2010 -*- Trans 2030 -*- Trans 2050
Avignon
- QSS —- GFDL -«- UKMO
• Trans2010 -K- Trans2030 -*- Trars2050
- GISS —- GFDL -«- UKMO
- Trans2010-*- Trans2030-*- Trans2050
- GISS —- GFDL -*- UKtVO
- Trans 2010 -x- Trans 2030 -*- Trans 2050
Rgure3.
Changes in weather variables as simulated by GCMs (a,b: temperature; c,d: rainfall;
e,f: radiation).
-------
Versailles wheat (rainfed or irrigated)
-5
I3'
I'I
0 2030
4 O \ 3"° I
•30 -25 -20 -15 -10 -5 0 5 10 15
Yield change 06)
V(
6
ff
|5
& 4
1
I3
c «
s2
!'
jrsailles wheat (rainfed or irrigated)
• 1 UK
• [5
" [Ja
. nr
* |2K
_ _ Jlo
220 225 230 235 240 245 250 L^
Season length (days)
JW3[
a]
SJ
*]
o]
o]
E]
Rgure4.
Yield variation (in percent of lxCO2 baseline) and season length for wheat in
Versailles (squares = climate change alone; circles = climate change with
physiological CO2 effects).
-------
Avignon wheat (irrigated)
7-i—
I
IUKMO
GFDL
GISS
2050
2030|
•~W 12010 I
-30 -25 -20 -15 -10 -5 0 5 10 15
Yield change (%)
Avignon wheat (rainfed)
arly
^
04—
-25
-20 -15 -10 -50 5
Yield change (%)
10 15 20
o
o
CD
5
i.
S
0
f
Avignon wheat (rainfed or irrigated)
'! " [U
5-,
! • [I
3'
i ' —
. 1 m 1 —
11 [a
i
185 190 195 200 205 210 215 4=
Season length (days)
-------
Versailles maize (rainfed)
J
•" — Trim.. A
«,....... ...... , ,^,
»-»» --O-
K. ., ....^~
i ~ r-m . . r
UKMO!
5 10
rieU change {%)
6
&
• 5
s
j«
o 3-
a°
S
i 2
E
ff
>•
10
Versailles maize (rainfed)
• • IUKMO |
" •• IGFDL |
IGISS
J2U50 1
" "" ' ' •"" |2030|
• J2010
0 110 120 130 140 150 160 I°^EI
Season length (days)
Figure 6.
Yield variation (in percent of lxCO2 baseline) and season length for rainfed maize
in Versailles (squares = climate change alone; circles s climate change with
physiological CO2 effects).
-------
Versailles maize (irrigated)
_ 0
Jh
i 2
0 2 4 6 8 10 12 14 16 18 -
Yield change (%)
Figure 7.
Yield variation (in percent of lxCO2 baseline) for irrigated maize in Versailles
(squares = climate change alone; circles = climate change with physiological CO2
effects).
-------
Avignon maize (rainfed)
20 40 60 80 100 120 140 160
YieU change fr)
S6
s
i
s.
i3-
Avignon maize (rainfed)
^UKMO
GFDLj
6ISS!
80
SS
90 95
Season length (days)
100
RgureS.
Yield variation '«n percent of lxCO2 baseline) and season length for irrigated maize
in Avignon (squares = climate change alone; circles = climate change with
physiological CO2 effects).
-------
7
§6
I3
Avignon maize (irrigated)
» •
, • :
« «
« o "••
1
* r^j
GFDL|
GISSl
-30 -25 -20 -15 -10 -5
Yeti change <%)
Figure 9.
Yield variation (in percent of lxCO2 baseline) for irrigated maize in Avignon
(squares = climate change alone; circles = climate change with physiological CO2
effects).
-------
Versailles wheat (irn'gated)
I 5
I
I *
i
I 2
i
1-
0-
2010
2030
2050
Septl
Oct1 Ncw1
Sowing da!e
Dec1
1 0-
t
I'21
o
-4-
-&
Avignon wheat (irrigated)
2010
2030
-*~
2050
Septl
OcM Nov1
Sowing date
Dec)
Figure 10. Wheat-yield response to sowing date (in percent of lxCO2 baseline) for GISS transient
scenarios in Versailles and Avignon.
-------
1&
18
2 17
o
£ 16
I15'
1+
I 13
-------
-------
POTENTIAL EFFECTS OF GLOBAL WARMING AND
CARBON DIOXIDE ON WHEAT PRODUCTION IN THE
FORMER SOVIET UNION
Gennadiy V. Menzhulin and Larisa A. Koval
Department of Climatic Changes
State Hydrological Institute, Russia
Alexander L. Badenko
Laboratory of Plant Biophysics
Agrophysical Institute, Russia
FSU-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Crop Production
Objectives
METHODS
Crop Model
GCM Climate Change Scenarios
Transient Scenarios
INPUT INFORMATION
Climate Data
Soil Data
Initial Soil Conditions
Management Variables
Calibration
Validation
RESULTS AND DISCUSSION
Sensitivity Analysis
Description of the GCM Climate Change Scenarios
Wheat Crop Changes
Projection of National Wheat Yield
ADAPTATION
SOURCES OF UNCERTAINTY
Models
Input Parameters
Technological Adjustments
Variability
FUTURE RESEARCH NEEDS
REFERENCES
FSU-2
-------
SUMMARY
This study combines the CERES-Wheat crop growth model and climate change scenarios
to estimate the possible impacts of climate change on wheat production (the main food crop of the
country) in 19 sites in Russia and the former Soviet Republics. This is the first crop modeling study
in the former Soviet Union (FSU) that uses future climate change scenarios generated by General
Circulation Models (GCMs). The results will make a significant contribution to the understanding
of possible yield changes under future climate.
1. The sensitivity of the crop model to arbitrary incremental temperature and precipitation
changes was estimated in two regions. A 2°C increase in temperature decreased yields by
an average of approximately 20%, and a temperature increase of 4°C resulted in yield
decreases of more than 30%. These yield decreases probably result from a shortening of the
wheat season under a higher temperature. Changes in the precipitation by ±20% had a
smaller impact on simulated wheat yields than changes in temperature.
2. GCM climate change alone resulted in considerable regional differences in simulated
changes in wheat yields compared to baseline yields. The simulation study considered the
beneficial physiological effects of CO2 on wheat yields. Under the GISS scenario, spring and
winter wheat yield increases exceeding 50% were obtained in the northwestern, central, and
eastern regions of Kazakhastan. These regions are presently fairly arid, but with the increase
in precipitation in the climate change scenarios they are predicted to become very favorable
to wheat production. In contrast, under the GFDL scenario, wheat yields were not favored
in these regions. In general, the climate change scenarios used were more favorable for
winter than for spring wheat growth and production.
4. The GCMs forecast significant changes in temperature. These changes would result in the
contraction of the crop growing season and the cold period of the year, and could be of
great importance for agriculture in Russia and the republics of the FSU. In some regions
these changes would raise the possibility of new cropping techniques, especially the ability
to grow a second crop during the calendar year. The potential decrease in duration of the
cold period is particularly significant in the case of vegetable crops which currently have a
short growing season.
5. The results of the simulated yield changes for each site were aggregated to estimate the
possible impact of climate change on national wheat yield. Under the GISS scenario and
considering the direct physiological effects of CO2 on crop growth and water use, winter and
spring wheat yields increased by 41% and 21% respectively. The GFDL and UKMO
scenarios were significantly less favorable for the simulated wheat production. The largest
yield decreases—19% -were simulated under the UKMO scenario.
6. Changes in sowing date and irrigation were studied as possible adaptation strategies to
climate change in two regions. While small changes in the sowing date did not appear to
have a large impact on yields, irrigation had a dramatic effect on spring wheat yields.
FSU-3
-------
INTRODUCTION
Background
Russia and the former Soviet Republics have the world's largest area of arable land.
Agriculture has been predominant throughout all periods of the country's history, and there is
important ongoing research on the impacts of climate on crop yields (Golzberg 1967; Mishchenko
1984; Shashko 1985). Scientists in Russia and the former Soviet Republics have developed several
empirical statistical models for forecasting crop yields in different regions of the country.
Publications of the Agrometeorological Forecast Department of the Moscow Hydrometeorological
Centre (Ulanova 1984) illustrate the developments in this field. Recently, scientists have developed
prognostic models with a physical and physiological basis. For example, the St. Petersburg
Agrophysical Institute has developed a special Supporting Agrotechnological Decisions System for
the study of prognostic crop models (The Modeling...., 1982).
In the last fifteen years, a number of important studies have been done in the former Soviet
Union related to the possible impacts of climate change on agricultural systems and crop production
(Zhukovskiy and Belchenko 1988, Zhukovskiy et al. 1992, Sirotenko et al. 1984). These studies did
not include the physiological (direct) effects of increased CO2 on crop yield.
Menzhulin (1976 a,b, 1984), Koval and Sawateyev (1982), and Menzhulin and Nikolayev
(1987) have developed crop models that consider both the climate impacts and the statistical analysis
of annual variation, and can be combined with different climate and technological scenarios to
evaluate the CO2 physiological effects on yield. These crop models have been used to study the
impact of climate on wheat yield in Russia and the former Soviet Republics, North America, and
Europe (Menzhulin and Sawateyev 1980, 1981; Menzhulin et al. 1987 a,b; Anthropogenic... 1987;
and Prospects... 1990; Rosenzweig et al. 1993).
Historically, Russian scientists and their colleagues have not used climate change scenarios
generated by the General Circulation Models (GCMs) to study the possible impact of climate change
on agriculture. The climate change scenarios were paleoclimatic analogues of global warming: up
to 1°C (year 2000) used the Holocene Optimum; up to 2°C (year 2025) used the Eemian
Interglaciar; and up to 4°C (year 2050) used the Pliocene Optimum. The carbon dioxide
concentrations corresponding to these three periods would be 380 ppm, 420 ppm, and 560 ppm,
respectively.
Crop Production
Today, agriculture presents a complex set of challenges in the former Soviet Union.
Agricultural and economic policies and the monopoly of the agricultural industry by the State have
led agriculture to its current crisis. These past policies have resulted in a country that cannot provide
for itself with adequate agricultural production. The low effectiveness of Soviet agriculture is evident
by the following statistics. In 1991, more than 22 million people were involved in the agricultural
sector of the former USSR, which is more than in any country in the Organization for Economic
Co-operation and Development (OECD) comprised of the countries of Europe, North America,
Australia, New Zealand, and Japan. Yet agricultural production in the former Soviet Union is five
times smaller than that of the OECD as a whole. The efficiency of Soviet agriculture is ten times
lower than that of the USA, the Netherlands, Canada, and Belgium.
The general agricultural crisis is particularly manifested in grain production. Average yields
of winter and spring wheat, the main components of national grain production, increased slightly
FSU-4
-------
before the 1970s, but leveled off during the last two decades. Although Russia holds a leading
position in the total amount of wheat production, it ranks 35th in yield level among the 45 wheat-
producing countries.
Objectives
The actual crisis in Soviet agriculture is sufficient cause for alarm, but the alarm is greater
if the problem has to be solved during a period of possible climate change. This is the first crop-
modeling study in the FSU that uses future climate change scenarios generated by GCMs. The
results will make a significant contribution to the understanding of possible yield changes under
future climate.
An analysis of Soviet agricultural production (Appendix 1) suggests that the best crop for
the simulation study is wheat. Wheat (winter and spring) is the main food crop in Russia and the
other republics. Its mean annual production is 100 million t. Other crops such as sugarbeet, potato,
and barley are also major components of national agricultural production, and subsequent studies
of the impacts of climate change on their production would be valuable. The production of wheat
in Russia and the Soviet Republics has been described and analyzed in more detail than that of any
other crop, providing empirical information for the validation of the simulation model.
Simulations of wheat yields and production changes require meteorological, soil, and crop
management information about the zone of production. We chose 113 administrative regions
(oblasts) in the main grain-producing area of the country (Figure 1 and Appendix 4) that include
65 regions of Russia, 25 regions of Ukraine, 6 regions of Belarus, 13 regions of Kazakhastan, the
Baltic Republics, and Moldova. The Republics of the Caucasus and Middle Asia, and the regions
of East Siberia and Russian Far East were not included because their wheat production does not
significantly contribute to the total national wheat production.
METHODS
Crop Model
The simulation model used was the CERES-Wheat v2.10 model (Ritchie and Otter 1985)
developed by IBSNAT (International Benchmark Sites Agrotechnology Transfer). The model has
been calibrated and validated over a wide range of geographical locations, and thus the results of
the simulation are comparable to previous and current studies in other regions of the world.
GCM Climate Change Scenarios
We used the scenarios of future climate generated by three equilibrium GCMs: the Goddard
Institute for Space Studies (GISS) (Hansen et al. 1983), the Geophysical Fluid Dynamics Laboratory
of Princeton University (GFDL) (Manabe and Wetherald 1987), and the United Kingdom
Meteorological Office (UKMO) (Wilson and Mitchell 1987). The climate scenarios provided by
these GCMs correspond to the global warming resulting from an increase of the greenhouse gas
content in the atmosphere equivalent to the doubling of CO2 concentration. It is projected that CO2
would account for 83.3% (550 ppm) of the equivalent concentration (660 ppm) and that other trace
gases (methane, nitrogen oxides, etc.) would account for 16.7%.
FSU-5
-------
The local climate scenarios were developed from the GCM output (temperature,
precipitation, and solar radiation) that is associated with a particular grid network for each GCM.
We developed the local climate change scenarios by two methods.
Simplified method. Local scenarios for each site were created by the direct extrapolation
(attribution) of the changes in climate variables at the nearest gridpoint. This method has been used
in previous climate change impact studies (Smith and Tirpak 1989), and therefore the results of
simulation under such scenarios can be compared to other studies. We created scenarios for 19
selected sites with this method (Appendix 2).
Multi-point interpolation. We developed this more complex method to create climate change
scenarios that simulate local climate more accurately (Appendix 3). This method used the GCM
output of changes in temperature, precipitation, and solar radiation occurring in several of the
nearest gridpoints to each site. This method was applied to generate scenarios in all of the 113
selected regions. With this complete set of scenarios we estimated the national changes in spring and
winter wheat production.
Transient scenarios
In addition to the equilibrium GCM scenarios, we used scenarios generated by the GISS
transient model for the decades of the 2010s, 2030s, and 2050s (Hansen et al. 1988). We calculated
the local changes of climate variables in 113 selected regions (as described above) and applied them
to the crop model.
INPUT INFORMATION
Climate Data
Ten years (1970-79) of daily maximum and minimum temperatures and precipitation data
were obtained from the Union Research Institute of Meteorological Information (World Data
Center "B" in Obninsk, Moscow Region) at 51 locations. When necessary, extrapolation of data
based on geographical proximity and environmental and climatic conditions was done to create data
for the entire set of 113 selected locations included in this study.
Daily solar radiation values were generated using long-term monthly radiation normals
(obtained from the Solar Radiation Data Bank in the Geophysical Observatory (St. Petersburg) and
the solar radiation generator included in the CERES-Wheat v2.10 model. Appendix 2 shows the
stations selected to create scenarios for the 19 representative sites discussed in this report.
Soil Data
Soil data were obtained from the largest soil information archive at the Dokuchayev Soil
Scientific Museum, with the help of the Soil Departments at the St. Petersburg and Moscow
Universities. All the information in these archives is uniformly systematized (Ivanova and Rozov
1967, Glazovskaya 1966). The large territory included in this study and the wide variety of soils
within each region required careful study for the selection of the representative soils and the
determination of the input parameters for the simulation (Aderikhin 1964; Kovda and Lobova 1964;
Liverovskiy 1965; Dobrovolskiy 1968; Glazovskaya and Friedland 1980; and Kovda 1973). We
selected 26 soil types in the wheat-producing region. For a complete representation of grain
production in each region, it was necessary to include more than one soil in each region (Table 1).
FSU-6
-------
Initial Soil Conditions
Initial soil moisture in the profile, acidity, and the content of nitrates and ammonium must
be defined for the simulation model. Unfortunately, there is not enough reliable annual information
on initial soil moisture and nitrogen content for each region. Therefore, we used the following
standard values. For arid regions, the soil water content at the end of the previous season was
assumed to be equal to the lower limit (LOL) of water availability to the plant. In the regions where
precipitation is always sufficient to fill the soil water store to the drained upper limit (DUL), the
initial value was taken as equal to DUL just before sowing (Ritchie et al. 1989). Although there is
nitrogen stress in some areas, we assumed that the nitrogen was not limiting.
Management Variables
For spring wheat, the simulated planting date occurs when the average temperature is about
5°C. For winter wheat, the simulated planting date occurs in autumn, about 60 days before the mean
temperature reaches 0°C.
Calibration
The CERES-Wheat model has been calibrated and validated extensively with a number of
wheat cultivars (Ritchie and Otter 1985). In this study, however, we used specific cultivars that were
representative of production in Russia and the other republics, and that were not previously
included in the CERES model. Therefore, we had to determine the genetic coefficients and calibrate
the model for the cultivars used in the FSU. (The Manual... 1982).
According to specifications developed by the National Seed Inspection, all varieties of wheat
cultivars recommended for use in national grain production are divided into eight adaptation groups:
East European Forest; East European Forest-Steppe; East European Southern Steppe; Volga
Steppe; West-Siberian Forest Steppe; East Siberian Forest Steppe; East Siberian Forest; and Middle
Asian Rainfed.
The genetic coefficients were determined based on experimental data provided by the
Agrophysical Research Institute. We calibrated the model with Russian cultivars so that the
simulated values correspond effectively to the experimental observations on duration and
phenophases, weight and number of grains in the ear, growth of phytomass and leaf area and yield.
Table 1 shows the adaptation group number (1 to 8) and the wheat cultivars selected for
the 19 regions. The actual wheat production in any region of the FSU involves more than one
cultivar. The number of cultivars used and their relative contribution to the total regional
production is subject to yearly changes and the introduction of new cultivars. We used a set of
nonvariable genetic parameters that represents only one cultivar and therefore we do not completely
represent current practices. This assumption could introduce possible error into the results that
could be significant because we have simulated only ten years of wheat production.
Validation
The model was validated for the eight adaptation groups by comparing observed
experimental values of yields from the National Bureau of Agricultural Statistics to simulated values.
FSU-7
-------
RESULTS AND DISCUSSION
Sensitivity Analysis
We estimated the sensitivity of the CERES-Wheat model to step changes in temperature
and precipitation. The mean daily temperature was increased by 2°C and 4°C, and the precipitation
was changed by +20% and by -20%. All possible combinations were evaluated. The physiological
effects of increased CO2 were also considered in each scenario. The sensitivity study was simulated
for two important wheat producing regions, Zhitomir and North Bashkiria (for both spring and
winter wheat).
Table 2 shows the results of simulations conducted with climate change alone and with
climate change and physiological effects of CO2. The results show large decreases in simulated yields
due to changes in temperature. In the case of climate change alone, a 2°C increase in temperature
decreased yields by approximately 20%; a temperature increase of 4°C resulted in yield decreases
of more than 30% (the largest decrease was 45% for winter wheat in North Bashkiria). The
physiological effects of CO2 partially compensated for the adverse impact of temperature increases
in simulated yields. For example, for spring wheat in North Bashkiria, a 4°C increase produced a
40% decrease in yield for temperature change alone (330 ppm) and a 17% decrease for temperature
change with physiological effects of CO2 (550 ppm). The impact of changes in precipitation on
winter and spring wheat yields was noticeably smaller than the impact of changes in temperature.
The simulated yield increases due to the beneficial direct effects of CO2 were close in
magnitude to the yield decreases produced by a +2°C temperature increase. In all cases, when
considering the physiological CO2 effects, the simulated yields of winter and spring wheats increased
more than 20% compared to the baseline yields with a maximum of 35% in the case of winter wheat
in North Bashkiria.
In the CERES-Wheat model, the phenological stages are influenced by temperature, and
our results show that the length of the growing season was not affected by changes in precipitation.
In addition, the physiological CO2 effects did not have any influence on the length of the growing
period in this case. A temperature increase of 2°C resulted in a decrease in the length of the spring
wheat growth period by 8 days in both Zhitomir and North Bashkiria. With temperature increases
of 4°C, the decrease was 14 and 15 days for spring wheat and 10 and 18 days for winter wheat in
Zhitomir and North Bashkiria respectively.
The most significant evapotranspiration (ET) change (-18%) was for spring wheat in
Zhitomir under the warmest and driest scenario (temperature +4°C and precipitation -20%). ET
was not significantly affected by the simulated physiological effects of CO2. Under the baseline
climate, ET was reduced only 6% due to the physiological CO2 effects. It is important to note that
this estimate is dramatically smaller than estimates from experimental data (Rose 1989).
Description of the GCM Climate Scenarios
Tables 3,4, and 5 show the seasonal and annual changes in temperature, precipitation, and
solar radiation under the three GCM scenarios for the selected sites. The temperature changes
(Table 3) simulated by the GISS and GFDL scenarios are comparable in all sites and average 4 -
5°C higher than current temperatures. The changes under the UKMO scenario are almost twice as
large as those simulated by the other two scenarios. In most cases, the differences between scenario
and baseline temperatures are more noticeable in the cold season.
FSU-8
-------
The average annual and seasonal precipitation under the climate change scenarios is
significantly greater than the baseline precipitation in most cases (Table 4). In general, the largest
amounts of precipitation are projected by the GISS and UKMO scenarios.
Table 5 shows changes in solar radiation under the three scenarios. There is a general
tendency for increased solar radiation under the GFDL and UKMO scenarios. Under the GISS
scenario, solar radiation decreases slightly in most sites.
Wheat Crop Changes
The simulated changes in yield, season length, and evapotranspiration under GCM climate
change scenarios are given in Tables 6, 7, and 8 for the 19 representative sites of the grain
production area of Russia, the Ukraine, Belarus, and Kazakhstan.
Yield. Climate change alone resulted in considerable regional differences in simulated
changes in wheat yield under the GCM scenarios. The maximum increases in yield were projected
under the GISS scenario. The following results are for climate change including the
physiological effects of CO2.
Under the GISS scenario, increases in spring wheat yields exceeding 50% were obtained in
the northwestern, central, and eastern regions of Kazakhastan (Aktyubinsk, Zelinograd, Karaganda).
In the same regions, the changes in winter wheat yields were also characteristically high. These
regions are fairly arid, but with the increase in precipitation under the climate change scenarios,
were very favorable to wheat production. The GFDL scenario was not as favorable for wheat
production in the regions listed above. Under this scenario, the decreases in yields were
characteristic of the sites in Kazakhstan, as in most of the other sites.
A systematized interpretation of the meaning of climate change for wheat production is
complicated because of the considerable differences in yield changes estimated under the three GCM
scenarios. Yet we can conclude that, on average, the climate change scenarios used were more
favorable for winter than for spring wheat yields.
Length of the growing season. To analyze the possible impacts of global climate change
on agriculture in the FSU, it is important to consider the changes in the lengths of the growing
period and the warm seasons under the climate scenarios (Table 7). All three GCM scenarios
resulted in a shortening of the growing season for both winter and spring wheat at all sites. Under
the UKMO scenario, which is characterized by the highest temperature increases, the growing season
in the Baltic regions for winter wheat decreased by more than two months. For spring wheat the
decrease was about one month. The shortening of the simulated growing season for winter and
spring wheats occurred in all regions of the FSU.
The GCMs forecast significant changes in the temperature regime, which would result in
the contraction of the crop-growing season and of the cold period of the year, and could be of great
importance for agriculture in Russia and the republics of the former Soviet Union. In some regions
these changes would raise the possibility of new cropping techniques, especially the growth of a
second crop during the calendar year. The decrease in the duration of the cold period is especially
significant in the case of vegetable crops, which have currently have a short growing season in many
regions.
Evapotranspiration. The evapotranspiration changes under the GCM scenarios were generally
insignificant (Table 8). In most cases, the total crop ET compared to baseline data. We believe that
the CERES-Wheat model needs to be improved in order to simulate the crop water regime
accurately. In the simulation of evapotranspiration of the wheat crop, it is necessary to account for
FSU-9
-------
the interactions among leaf temperature, moisture level, and the physiological effects of CO2 that
increase stomatal resistance (Rose 1989).
Projections of National Wheat Yields
The national production changes for equilibrium and transient scenarios were estimated
using the 16-point interpolation scheme for all 113 sites, not the nearest gridpoint extrapolation
method.
Aggregation of the GCM-scenario results. The possible impact of climate change on wheat
production in Russia, the Ukraine, Belarus, and Kazakhstan can be estimated from the simulated
changes in yield under the climate change scenarios. Table 9 shows estimates in yield changes
calculated from the results of 113 subregions of the grain production zone. Under the GISS
scenarios and considering the physiological effects of CO2, winter wheat production increased by
41% and spring wheat production increased by 21%. The GFDL and UKMO scenarios were
significantly less favorable for the simulated wheat production. Under the UKMO scenario and
considering the physiological effects of CO2, the simulated wheat production decreased by 19%.
Aggregation of the transient scenario results. Wheat yield changes were also simulated under
the GISS transient scenarios. The first two transient scenarios correspond to the intermediate levels
of climate change, which are simulated for the decades of the 2010s and 2030s, with a corresponding
increase in greenhouse gas concentration in the atmosphere equivalent to 405ppm and 430ppm of
CO2. The third GISS transient scenario can be used as another alternative scenario of climatic
change for the decade of the 2050s.
Winter and spring wheat yields were simulated under the three GISS transient scenarios for
the 113 selected regions. Table 9 shows the results of the aggregation of the yield changes. The
results suggest that the production of both winter and spring wheat would gradually increase due
to climate change.
ADAPTATION
A realistic approach to the problem of climate change impacts on agriculture requires
studying possible measures to avoid unfavorable consequences. A detailed investigation of this
question was beyond the scope of this study, but we analyzed some possible adaptation responses,
such as changes in the sowing date and in irrigation on spring wheat yields at Zhitomir and Saratov
regions (Table 10). The adaptation responses were analyzed under the UKMO scenario because it
is the most unfavorable scenario. This small adaptation study shows approaches and methods that
can be and should be applied to a more detailed study of adaptation responses to unfavorable
climate changes in agriculture in Russia and in the former Soviet Republics.
The most significant result of the analysis was that changes in the sowing date (both later
and earlier) by 15 days and even by 30 days did not noticeably affect crop yields. Under the UKMO
scenario, there is a considerable precipitation increase in the winter and spring seasons at these two
sites (Table 4). However, these precipitation changes did not fully compensate for the moisture
limitation of wheat in the summer season due to a temperature increase. In addition, in the Saratov
region, the UKMO scenario predicts a 24% decrease in the summer precipitation, which is already
limited. Irrigation could have a large impact on spring wheat yields in these two regions, especially
in Saratov.
SOURCES OF UNCERTAINTY
FSU-10
-------
Models
The use of the simulation crop model (CERES-Wheat) and the climate change scenarios
(GISS, GFDL, and UKMO) introduce uncertainties. In principle, we could quantify the uncertainty
associated with the use of GCMs in the agricultural zones, applying the methods proposed in Global
Comparison of Selected GCM: Control Runs and Observed Data (Kalkstein 1991).
Input Parameters
Here we include uncertainties derived from the limited availability of current information
on input parameters of the models. As discussed above, the possible error in the calculations of the
climate scenarios could be smaller when we use the procedure for daily climate data generation
described in Appendix 3. The standard procedure used to develop climate scenarios for important
agricultural sites that are situated far from the GCM gridpoints could be incorrect. With the
multipoint interpolation system we can simulate more accurately the current climate than when we
use the GCM output from the nearest gridpoint. Therefore, additional uncertainty may be generated
when using the conventional system to create local scenarios.
Another potential error is associated with the large size and climatic diversity of some of
the study regions that would require data from more than one station to represent their climate
more accurately.
Technological Adjustments
There are unavoidable uncertainties in estimating the future changes of agricultural
technology and management in response to climatic change. These are significant factors limiting
the reliability of the results. Climate change can alter the geography of the current regions and
influence soil processes, and will eventually create demand for newly developed technologies.
Variability
Another major factor that limits the reliabiity of these results is the direct extrapolation to
the future of the current variation oif daily temperatures and precipitation. In order to obtain more
reliable estimates of future changes in wheat yields, it would be necessary to consider potential
differences in climate variability.
FUTURE RESEARCH NEEDS
The present study is a contribution to the understanding of the possible impacts of climate
change on the FSU's agricultural systems. However, future research is still needed. With improved
models and a more accurate set of input parameters, some of the uncertainties could be eliminated.
An adaptation study with a larger scope-one that would consider the complex
agrotechnological decisions that might decrease the negative consequences of climate change on
wheat production—is essential for continued progress in this field. In addition, further research
should consider the impact of climate change on the production of other major crops. Finally, the
frequency of extreme climate events, especially the occurrence of drought periods, needs to be
investigated.
The impact of drought variability on agriculture was addressed during the preparation of
the Intergovernmental Panel on Climate Change (IPCC) Working Group II report. However,
FSU-11
-------
methods for estimating such variability were not clearly established within the framework of the
IPCC or other projects. It will require considerable work to compile and analyze GCM output and
to prepare appropriate input data, but such an effort is justified by the importance of potential
impacts of future climate variability in the FSU and other agriculturally important regions of the
world.
FSU-12
-------
REFERENCES
Anthropogenic Climatic Changes. 1987. Eds: M.I. Budyko and Yu.A. Izrael. Leningrad,
Gidrometeoizdat. 406 p. (Russian). English Translation: Arizona University Press, 1990.
Aderikhin, P.G. 1964. Chernozems of the USSR. In: Genesis, Classification and Cartography of Soviet
Soils: Proceeding of the 8th International Congress of the Soil Science. Moscow, Nauka Publ.
House. (Russian).
Dobrovolskiy, V.V. 1968. Soil Geography. Moscow, High Education Publ. House. 350 p. (Russian).
Glazovskaya, M.A. 1966. The Principles of the Soil Classification. Soviet Soil Science Journal, Ne 8.
p. 14-25. (Russian).
Glazovskaya, M.A., and V.M. Friedland 1980. The Map of the USSR Soils (1:4000000). Moscow,
Geodesic and Geographic Publ. House. (Russian).
Golzberg, LA. 1967. Agroclimatological and Microclimatological Research in the Soviet Union.
Transactions of the Main Geophysical Observatory, vol. 218. p.75-88. (Russian).
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
Efficient Three-Dimensional Global Models for Climate Studies: Model 1 and 2. April
Monthly Weather Review, Vol. Ill, Na 4: 609-662.
Hansen J., I. Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy, and G. Russell. 1988. Global Climate
Changes as Forecasted by Goddard Institute for Space Studies Three- Dimensional Model.
Journal of Geophysics Research, 92: 9341-9364.
Ivanova, E.N. and N.N. Rozov 1967. The Classification of the Soils in the USSR. Soviet Soil Science
Journal, N2 2. p. 34-43.
Kalkstein. L.A. (ed.) 1991. Global Comparisons of Selected GCM: Control Runs and Observed
Climatic Data. Centre of Climatic Research University of Delaware. Report Prepared for
US EPA, Office of Policy, Planning, and Evolution. Climate Change Division. 251 p.
Koval, L.A., and C.P. Sawateyev. 1982. The Using of Crop Productivity Models for Estimations of
Climatic Changes Influence on the Crops Productivity. In: The Problems of Land Hydrology,
eds. I.V. Popov and S.A. Kondratyev. Leningrad, Hydrology and Meteorology Publ. House.
p. 215-233. (Russian).
Kovda, V.A., 1973. The Fundamentals of Soil Science. Moscow, Nauka Publishing Co., Book 1: 380
p., Book 2: 467 p. (Russian).
Kovda, V.A., and E.V. Lobova. 1964. The Soils Map of Asia. Moscow, Soviet Academy of Science
Publ. House. (Russian).
FSU-13
-------
Liverovskiy, Yu.A. Soils of the Soviet Union. Transactions of Moscow State University, Vol. 1, N2 2.
p. 51-121. (Russian).
Manabe, S., and R.T. Wetherald. 1987. Large-Scale Changes in Soil Wetness Induced by an Increase
in Carbon Dioxide. Journal of Atmospheric Sciences, 44: 1211-1235.
The Manual on Selection of Wheat Cultivars for USSR Cereal Regions. 1982. Moscow, Agricultural
Ministry Publ. House. 86 p.
Menzhulin, G.V. 1976a. Biosphere Aspects of Climate Modeling: Climate Variations Effect on the
Crops Productivity. Papers ofSov. Amer. Symp., Tashkent. 12 p.
Menzhulin, G.V. 1976b. The Influence of Climatic Changes on Crop Plant Productivity. Transactions
of the State Hydrological Institute, Vol. 365. p. 41-48. (Russian).
Menzhulin, G.V. 1984. The Influence of Contemporary Climate Changes and CO2 Content
Evaluation on the Crop Productivity. Soviet Meteorology and Hydrology, Nfi 4. p. 95-101.
Menzhulin, G.V., and M.V. Nikolayev. 1987. Methods for Estimating the Year-to-year Variability
of Cereals Crop Yields. Transactions of the State Hydrological Institute, Vol. 327. p. 113-131.
(Russian).
Menzhulin, G.V., and S.P. Sawateyev. 1980. The Impact of Expected Climate Changes on Crop
Productivity. In: The Problems of Atmospheric Carbon Dioxide, ed. M.I. Budyko. Transactions
of Sov. Amer. Symp., Leningrad, Gidrometeoizdat. p. 186-197. (Russian).
Menzhulin, G.V., and S.P. Sawateyev. 1981. The Present-day Climatic Changes and the Problems
of Crop Productivity. Transactions of the State Hydrological Institute. Vol. 271. p. 90-103.
(Russian).
Menzhulin, G.V., L.A. Koval, M.V. Nikolayev, and S.P. Sawateyev. 1987a.The Estimating of
Climatic Changes Agroclimatic Consequences: Scenario for Northern America. Transactions
of the State Hydrological Institute, Vol. 327. p. 132-146. (Russian).
Menzhulin, G.V., L.A. Koval, and M.V. Nikolayev. 1987b. The Agroclimatic Consequences of
Present Day Global Climate Changes. In: The Problems of Agroclimatic Providing for the
USSR Foodstuff Programme, ed. I.G. Gringhof. Leningrad, Gidrometeoizdat. p. 72-81.
(Russian).
Mishchenko, Z-A. 1984. The Bioclimate of the Day and the Night. Leningrad, Gidrometeoizdat. 280
p. (Russian).
The Modeling of Agroecosystems Productivity. 1982. Eds. N.F. Bondarenko, N.G. Zhukovskiy, N.G.
Mushkin, S.V. Nerpin, R.A. Poluektov, and I.B. Uskov. Leningrad, Gidrometeoizdat. 264
p. (Russian).
FSU-14
-------
Prospects of Future Climate. 1990. Eds. M. MacCracken, AD. Hecht, M.I. Budyko, Yu.A Izrael. A
Special US/USSR Report on Climate and Climate Change, Lewis Publishers. 270 p.
Rose, E. 1989. Direct (Physiological) Effects of Increasing CO2 on Crop Plants and their
Interactions with Indirect (Climatic) Effects. In: The Potential Effects of Global Climate
Change on the United States. Appendix C, Agriculture, Vol. 2, eds. J.B. Smith, D.A. Tirpak.
Office of Policy, Planning and Evaluation US EPA, Washington, DC.
Ritchie, J.T., and S. Otter. 1985. Description and Performance of CERES-Wheat: A User-oriented
Wheat Yield Model. In: Willis W.O., ed. ARS Wheat Yield Project. Washington D.C.: US
DOA, Agricultural Research Service. Ars-38. p. 159-175.
Ritchie, J.T., D.C. Godwin and U. Singh. 1989. Soil and Weather Inputs for the IBSNAT Crop
Models. In: IBSNAT Symposium Proceedings: 81st Annual Meeting of the American Society
of Agronomy, Las Vegas, Nevada, October 1989. p. 31-45.
Rosenzweig, C, G. Menzhulin, and L Koval. 1993. Climate Change and Agriculture in the former
Soviet Union. In: Science, Agriculture and Environment in the FOrmer Soviet Union, M.S.
Strauss and S.M.B. Thomson, eds. AAAS, Washington, DC.
Shashko, D.I. 1985. Agroclimatic Resources of the USSR. Leningrad, Gidrometeoizdat. 247 p.
(Russian).
Sirotenko, O.D., E.V. Abashina, and V.N. Pavlova. 1984. The Estimates of Climate Variations of
Crops Productivity. The Proceedings of the USSR Academy of Science: Physics of Atmosphere
and Ocean, Vol. 20, N 11. p. 1104-1110. (Russian).
Smith, J.B. and D.A. Tirpak. 1989. The Potential Effects of Global Climate Change on the United
States. US Environmental Protection Agency, Washington, DC.
Ulanova, E.S. 1984. Meteorological Factors and the Cereal Crops Productivity, p. 95-100. (Russian).
Wilson, C.A., and J.RB. Mitchell. 1987. A Doubled-CO2 Climate Sensitivity Experiment with a
Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
Zhukovskiy, E.E., and G.G. Belchenko. 1988. The Stochastic System of Crop Yields Forecasting.
Transactions of the Agrophysical Institute, Vol. 70. p. 2-7. (Russian).
Zhukovskiy, E.E., G.G. Belchenko, and T.N. Brunova. 1992. The Stochastic Analysis the Climate
Change Influence on The Crop Productivity Potential. Meteorology and Hydrology Journal,
Vol.1, (in press) (Russian).
FSU-15
-------
APPENDIX 1. Production (x 100,000 t)/Area (x 100,000 ha) of the principal food crops.
Crops
W.Wheat
S. Wheat
Sugar Beet
Potatoes
Barley
Maize
Vegetables
Oats
Rye
Beans
Sunflower
Millet
Buckwheat
1940
145/143
172/260
180/12
759/77
121/105
51/36
137/15
168/202
210/231
15/24
26/35
44/60
13/21
1950
114/125
197/260
208/13
886/85
64/81
66/48
93/13
130/162
180/236
12/20
18/36
17/38
13/30
Years
1960
182/121
461/483
577/30
844/91
160/110
98/51
166/15
120/128
182/162
27/33
40/
32/38
6/14
1970
427/185
576/467
789/34
968/81
383/213
94/34
212/15
142/92
130/100
76/51
61/48
21/27
11/19
1987
462/153
371/314
904/34
760/62
584/306
323/46
292/17
219/118
181/97
100/64
61/42
39/28
13/16
FSU-16
-------
APPENDIX 2.
Meteorological and solar radiation stations used for the 19 representative
sites.
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Region
W Russia
C Chernozem of
Russia
Central Russia
Middle Volga
South Volga
Southern Russia
East Euro
Russia
Western
Ukraine
Northwestern
Ukraine
Eastern Ukraine
Southern
Ukraine
Moldova
Western Belarus
Central Belarus
Baltic
NW Kazakhstan
N Kazakhstan
C Kazakhstan
E Kazakhstan
Site
(Lat/Long)
Bryansk
52.9/33.4
Kursk
51.7/36.0
Yaroslavl
57.6/38.9
Saratov
51.6/46.6
Astrakhan
46.6/48.0
Rostov-Don
47.4/40.8
N Bashkiria
55.0/55.6
LVov
49.5/23.6
Zhitomir
50.0/28.4
Kharkov
49.5/36.3
Kherson
46.7/33.5
Kishinev
47.0/28.9
Brest
52.0/24.2
Minsk
53.3/26.0
Lithuania
55.0/23.0
Aktyubinsk
49.9/57.9
N Kazakhstan
54.6/68.3
Zelinograd
51.6/70.1
Karaganda
50.0/72.4
Meteorological
Station
(Lat/Long)
Smolensk
54.7/32.0
Kursk
51.7/36.0
Kostroma
57.8/41.0
Saratov
51.3/46.0
Astrakhan
48.2/46.2
Rostov-Don
47.0/40.0
Ufa
54.9/55.9
L'vov
49.5/23.6
Vinnitsa
49.3/28.3
Kharkov
50.1/36.3
Askania-Nova
46.3/33.5
Kishininev
47.0/28.9
Brest
52.2/23.9
Minsk
54.0/27.3
Kaunas
54.5/24.0
Aktyubinsk
50.2/57.1
Petropavlovsk
54.5/69.1
Zelinograd
51.5/68.2
Karaganda
49.1/73.1
Distance +
(km)
290
0
125
75
270
100
35
0
80
75
45
0
40
165
125
95
90
210
125
Solar
Station
(Lat/Long)
Br.forest
53.1/34.5
Ostrogozhsk
50.9/39.0
Vologda
59.2/40.0
Saratov
51.3/46.0
Or. Promysel
45.9/47.5
Rostov-Don
47.0/40.0
Ufa
54.9/55.9
L'vov
49.5/23.6
Kiev
50.6/30.3
Kharkov
50.1/36.3
Kherson
46.4/32.4
Kishinev
47.0/28.4
Vasilyevichi
52.2/30.0
Minsk
54.0/27.3
Zarasay
55.7/26.3
Orenburg
51.4/55.1
Petropavlovsk
54.5/69.1
Atbasar
51.1/71.2
Karkaralinsk
49.5/75.5
+ Distance from the agricultural region to the meteorological station
FSU-17
-------
APPENDIX 3.
Method for calculating local climate change scenarios.
The procedure for developing the local climate change scenarios using the nearest GCM
gridpoint and the local climate data set may not always be the best representation of the local future
climate. In certain cases there may be discrepancies between the observed climate variables and the
simulated baseline climate for a site. Problems may arise when the climate variables are
nonuniformly distributed on the region, and when two grain-producing sites are closely situated or
within the same GCM gridbox. This is a particular problem in the creation of the scenarios in the
113 grain-producing regions selected for this study, because the gridpoint resolution of the GCMs
used (GISS, GFDL, and UKMO) is lower than ideal. The distance between gridpoints in the
latitudinal direction for any of the GCMs used is two times the size of the average region selected
for the study; Furthermore, when we want to develop a map of the spatial distribution of the
changes of climatic variables and the productivity in the agricultural territory.
These problems arise because of the large agricultural region used in this study. To obtain
potential representative estimates of changes of national agricultural production under the climate
change scenarios it is important to study the responses of crops yields in relatively small regions. In
addition, it would be best to represent the simulation results as territorial distributions of output
parameters in the forms of maps with isolines.
We developed a method called Multipoint Interpolation for developing local climate
scenarios based on a modification of the standard method. The method is based on the interpolation
of climate variables from several gridpoints for each GCM to each of the centers of the wheat-
producing regions (113). For reasons of symmetry, the 16 nearest surrounding gridpoints were used
in this modified procedure. To calculate the values of the three climate variables (temperature,
precipitation, and solar radiation) for each agricultural site we used the formula:
X(in the site) = Sum[p(i) * X(i)]; i=l,2,...16
where X(i) are the climatic variables in each of the 16 gridpoints, and p(i) are their relative weights.
To estimate the relative weight values (p(i)), we assumed that the weight of each of the 16
gridpoints that surround the agricultural site for each variable must be proportional to the value of
the correlation coefficient between its magnitudes in the gridpoint and of the site. The coefficient
of correlation for temperature (K(T,i)), precipitation (K(P,i)), and solar radiation (K(S,i)) have been
plotted as a function of distance by the formulas:
= exp[:r(i)/R(T)]
= exp[-r(i)/R(P)]
= exp[-r(i)/R(S)]
i= 1,2,.... 16
where R(T), R(P), and R(S) are so-called radii of correlation, (e.g., the distances for which
correlation has decreased e = 2.7183). The final formulas for weights after normalization can be
represented as follows:
= K(T,i)/Sum[K(T,i)];
= K(P,i)/Sum[K(P,i)];
= K(S,i)/Sum[K(S,i)].
FSU-18
-------
To carry out the calculations in accordance with this procedure, the values of these three
parameters must be known: radii of temperature, precipitation, and solar radiation correlation. In
this particular study we considered these values to be:
R(T) = 1500 km; R(P) = 600 km; R(S) = 700 km.
The use of this procedure establishes a better correlation between the local climate and the
climate scenarios generated by the GCMs. But in spite of the use of this more accurate technique
of interpolation, there is still a degree of uncertainty associated with the local climatic scenarios.
Specifically, there is also uncertainty associated with the extrapolation of present spatial
meteorological fields into the future. For more accurate use of this multipoint interpolation
procedure, it is necessary to have information on the changes of correlation radii R(T), R(P), and
R(S) during climate change.
FSU-19
-------
APPENDIX 4.
A complete list of sites in the FSU used in the simulation study. The
numbers below each site indicate its latitude/longitude. An asterisk (*)
indicates the sites selected as examples for this report. In these sites the
climate change scenarios were created by the simplified method.
1.
3.
*5.
7.
9.
11.
13.
*15.
17.
*19.
*21.
*23.
25.
*27.
29.
*31.
33.
35.
37.
39.
Vinnitsa
48.7/28.6
Dnepropetrovsk
48.4/34.7
Zhitomir
50.0/28.4
Kiev
50.1/31.0
Lugansk
49.0/38.8
Odessa
46.6/30.4
Rovno
50.4/26.0
Kharkov
49.5/36.3
Cherkassy
49.0/31.3
Moldova
47.1/28.6
Bryansk
52.9/33.4
Kursk
51.7/36.0
Oryol
52.8/36.2
Brest
52.0/24.2
Gomel
52.0/29.4
Minsk
53.3/26.0
Vladimir
56.0/40.9
Ivanovo
57.2/41.4
Kostroma
58.2/43.2
Mordovia
54.4/44.3
2. Volyn
50.8/24.4
4. Donetsk
47.9/37.5
6. Zaporozhye
47.3/35.7
8. Kirovograd
48.3/32.2
10. Nikolaev
47.2/32.0
12. Poltava
49.6/33.8
14. Sumy
51.0/33.9
*16. Kherson
46.7/33.5
18. Chernigov
51.1/32.0
20. Belgorod
50.7/37.7
22. Voronezh
51.1/40.0
24. Lipetsk
52.6/38.8
26. Tambov
52.6/41.5
28. Vitebsk
55.2/28.3
30. Grodno
53.0/24.2
32. Mogilyev
53.6/30.4
34. Nizhniy Novgorod
55.9/44.2
36. Kaluga
54.3/35.0
38. Mary
56.4/47.8
40. Ryazan
54.2/40.7
FSU-20
-------
41.
43.
45.
47.
49.
51.
53.
55.
57.
*59.
61.
63.
65.
67.
*69.
*71.
73.
*75.
77.
79.
81.
83.
85.
Tula
53.9/37.4
Chuvashia
55.4/47.0
Moscow
55.4/38.1
Volgograd
49.9/44.1
Krasnodar
45.2/39.2
Stavropol
44.8/43.2
West Orenburg
51.8/54.3
Penza
53.1/44.7
Tataria
55.3/51.0
North Bashkiria
55.0/55.6
Kurgan
55.4/65.0
South Yekaterinburg
57.2/61.9
Region "B" of Mid.Ural
54.8/58.9
South Chelyabinsk
53.1/60.2
North Aktyubinsk
49.9/57.9
Karaganda
50.0/72.4
Kustanaiy
52.4/63.5
North Kazakhstan
54.6/68.3
C'trl Semipalatinsk
48.7/80.0
North Ural
50.2/52.9
Tver
57.1/34.5
St. Petersburg
59.2/29.0
Novgorod
58.2/32.0
42.
*44.
*46.
48.
*50.
52.
54.
*56.
58.
60.
62.
64.
66.
68.
70.
72.
74.
76.
*78.
80.
82.
84.
86.
Udmurtia
56.6/52.2
Yaroslavl
57.6/38.9
Astrakhan
46.6/48.0
Kalmykia
45.9/44.4
Rostov
47.4/40.8
Samara
53.0/50.6
East Orenburg
51.3/59.6
Saratov
51.6/46.6
Ulyanovsk
53.9/48.0
South Bashkiria
52.9/57.3
Perm
57.5/55.9
Region "A" of Mid.Ural
57.8/59.9
North Chelyabinsk
55.0/60.5
South Urals Region
53.2/59.1
North Guryev
57.5/53.4
Kokchetav
53.4/70.3
Pavlodar
52.3/76.0
North Semipalatinsk
50.3/79.6
Zelinograd
51.6/70.1
South Ural
49.3/49.4
Smolensk
54.6/32.3
Pskov
57.1/28.5
Vologda
59.5/40.4
FSU-21
-------
87.
89.
91.
93.
95.
97.
99.
101.
103.
105.
*107.
109.
111.
113.
Latvia
56.2/23.8
Estonia
58.2/25.5
Kaliningrad
54.2/20.6
South Tomsk
58.0/83.9
South- West Novosibirsk
55.0/77.9
South-East Kemerovo
54.1/87.0
North-East Altai
53.6/84.4
North Omsk
57.2/73.4
North Tyumen
58.9/68.6
Vyatka
57.9/49.8
Lvov
49.5/23.6
Carpathians
47.7/22.3
Ternopol
49.0/25.1
Crimea
45.4/34.0
*88.
90.
92.
94.
96.
98.
100.
102.
104.
106.
108.
110.
112.
Lithuania
55.0/23.0
Karelia
61.9/32.4
North Tomsk
59.5/80.0
NE Novosibirsk
55.6/82.1
North-West Kemerovo
56.0/86.0
North-West Altai
52.1/79.8
South-East Altai
51.9/83.0
South Omsk
55.1/73.0
South Tyumen
56.9/68.4
East Kazakhstan
50.7/82.3
Ivano-Frankovsk
48.4/24.0
Chernovtsy
48.0/26.1
Khmelnitsky
49.1/26.0
FSU-22
-------
Table 1. Soil types and wheat cultivars used in the simulation.
SITE
SOIL TYPE
WHEAT CULTIVAR (Spring,
Winter)
Bryansk Gray Wooded!
Kursk Chernozem:
Leached, Modal
Yaroslavl Dernopodzolic
Saratov Chernozem: Steppe, Modal,
Southern; Deep Chestnut
Astrakhan Light Chestnut, Brown Steppe,
Desert
Rostov-Don Deep Chestnut Chernozem:
Southern, Deep
N. Bashkiria Chernozem: Modal,
Leached
Lvov Leached Chernozem,
Dernopodzolic
Zhitomir Modal Chernozem
Dernopodzolic
Kharkov Modal Chernozem
Kherson Chernozem:
Deep, Steppe, Southern
Kishinev Gray Wooded,
Modal Chernozem
Brest Podzolic, Ferruginous, Humic
Minsk Dernopodzolic Podzolic
Lithuania Glee Wooded
Aktyubinsk Deep Chestnut, Brown Steppe
Desert
N. Kazakhstan Brown Meadow Steppe
Zelinograd Chernozem: Steppe, Southern;
Deep Chestnut
Karaganda Chestnut: Light, Deep
Palmira, Bezostaya-1
Albidum-43, Mironovskaya-808
Palmira, Bezostaya-1
Albidum-43, Mironovskaya-808
Albidum-43, Mironovskaya-808
Lutescens-62, Krymskaya
Albidum-43, Mironovskaya-808
Ostka, Lutescens-17
Ostka, Lutescens-17
Albidum-43, Mironovskaya-808
Lutescens-62, Krymskaya
Lutescens-62, Krymskaya
Palmira, Bezostaya-1
Palmira, Bezostaya-1
Palmira, Bezostaya-1
Albidum-43, Mironovskaya-808
Albidum-43, - NO -
Albidum-43, - NO -
Albidum-43, Mironovskaya-808
FSU-23
-------
Table 2. Sensitivity analysis of CERES-Wheat to climate and CO2 changes in Zhitomir and N.
Bashkiria.
Changes from baseline
CO2 level Precip.
(ppm) (%)
330 0%
20%
-20%
555 0%
20%
-20%
330 0%
20%
-20%
555 0%
20%
-20%
Temp. Yield
(°C) (%)
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
-16
-33
5
-12
-27
-8
-26
-44
26
8
-9
30
12
-4
20
0
-19
N. Bashkiria
0
-22
-40
2
-18
-36
-4
-26
-44
21
1
-17
27
3
-14
13
-2
-21
Season ET
L.(days) (mm)
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
Spring Wheat
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-4
-9
3
0
-4
-6
-11
-18
-6
-9
-11
-4
-6
-8
-9
-13
-18
0
-4
-9
4
-2
-4
-8
-12
-16
-6
-6
-10
-4
-4
-6
-10
-12
-16
Yield Season
(%) L.(days)
0
-23
-38
8
-13
-27
-18
-36
-52
27
0
-17
36
14
-3
16
-13
-34
N. Bashkiria
0
-25
-45
13
-12
-34
-20
-40
-56
35
10
-15
48
22
-3
17
-9
-32
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
Winter
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
ET
(mm)
0
-3
-6
4
2
0
-8
-12
-13
-6
-7
-9
-2
-3
-5
-10
-14
-10
Wheat
0
-3
-5
5
2
1
-7
-10
-11
-5
-5
-7
1
-1
-3
-9
-11
-13
FSU-24
-------
Table 3. Temperature change (°C) predicted by GISS, GFDL, and UKMO climate change
scenarios.
Site
Spring Summer Autumn Winter Annual
Bryansk GISS
GFDL
UKMO
Kursk GISS
GFDL
UKMO
Yaroslavl GISS
GFDL
UKMO
Saratov GISS
GFDL
UKMO
Astrakhan GISS
GFDL
UKMO
Rostov-on-Don GISS
GFDL
UKMO
N. Bashkiria GISS
GFDL
UKMO
Lvov GISS
GFDL
UKMO
Zhitomir GISS
GFDL
UKMO
Kharkov GISS
GFDL
UKMO
Kherson GISS
GFDL
UKMO
Kishinev GISS
GFDL
UKMO
4.3
4.6
11.0
4.6
4.6
11.0
4.4
5.3
9.6
4.7
4.5
10.1
3.6
4.0
10.2
4.6
4.0
9.0
5.0
4.4
9.8
3.7
4.6
10.4
4.3
4.6
11.4
4.6
4.6
10.6
4.8
4.1
10.6
4.8
4.1
10.6
1.0
4.8
8.6
1.4
4.1
8.6
0.7
4.1
5.8
3.0
3.6
8.3
4.3
3.6
8.1
1.4
3.8
8.9
3.9
3.8
8.5
2.3
5.2
8.2
1.0
4.8
7.1
1.4
4.1
8.5
3.5
4.5
8.5
3.5
4.5
8.5
4.3
4.7
7.5
4.6
4.2
7.5
4.4
4.8
6.5
5.0
3.7
7.1
4.8
3.6
7.3
4.6
4.0
7.5
4.6
4.3
7.0
4.2
4.8
7.7
4.3
4.7
7.1
4.6
4.2
7.2
4.7
4.3
7.2
4.7
4.3
7.2
6.3
4.2
8.5
6.6
4.1
8.5
6.8
4.6
9.9
6.9
4.1
8.4
4.6
4.2
8.2
6.6
3.9
9.5
6.4
4.2
8.0
5.8
4.6
8.7
6.3
4.2
8.7
6.6
4.1
9.7
4.7
3.9
9.7
4.7
3.9
9.7
4.0
4.6
8.9
4.3
4.2
8.9
4.1
4.7
7.9
4.9
4.0
8.9
4.3
3.8
8.7
4.3
3.9
8.7
5.0
4.2
8.3
4.0
4.8
8.8
4.0
4.6
8.6
4.3
4.2
8.0
4.4
4.2
8.0
4.4
4.2
8.0
FSU-25
-------
Brest
Minsk
Lithuania
Aktyubinsk
N. Kazakhstan
Zelinograd
Karaganda
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
3.7
4.6
11.4
4.3
4.6
11.4
4.0
5.3
11.5
5.0
4.8
10.1
5.0
5.2
10.5
5.0
5.3
10.5
5.0
5.4
10.5
2.3
5.2
7.1
1.0
5.2
7.1
1.6
4.7
6.2
3.9
4.0
9.1
4.1
4.0
8.3
4.1
4.1
8.3
4.1
4.5
8.3
4.2
4.8
7.1
4.3
4.8
7.1
3.5
5.0
6.0
4.6
5.8
7.0
4.1
6.1
7.7
4.1
6.4
7.7
4.1
5.9
7.7
5.8
4.6
8.7
6.3
4.6
8.7
5.3
5.3
9.5
6.4
5.0
9.0
7.5
6.0
6.9
7.5
6.0
6.9
7.5
6.1
6.9
4.0
4.8
8.6
4.0
4.8
8.6
3.6
5.1
8.3
5.0
4.9
8.8
5.2
5.3
8.3
5.2
5.4
8.3
5.2
5.5
8.3
FSU-26
-------
Table 4. Precipitation change (%) predicted by GISS, GFDL, and UKMO climate change
scenarios.
Site
Spring Summer Autumn Winter Annual
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-on-Don
N. Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
+24
+36
+71
+28
+15
+72
+34
+8
+79
+16
+28
+70
+22
+10
+68
+30
+3
+44
+40
+18
+77
+27
+25
+49
+24
+36
+60
+28
+15
+52
+15
-1
+52
+15
-1
+52
+35
-3
-24
+37
+20
-24
+74
+2
-12
+38
+57
-24
-98
+31
-20
+37
+8
-32
+45
+18
-31
+9
-16
-16
+35
-3
+6
+37
+20
-35
-11
+9
-35
-11
+9
-35
+33
+8
+25
+18
-3
+25
+35
+23
+33
-6
+6
+18
-8
-5
+14
+18
-27
+21
-22
+44
+21
+6
+2
+19
+33
-3
+6
+18
-3
+28
+6
+15
+28
+6
+15
+28
+16
+6
+34
+38
+18
+34
+36
+5
+47
+41
+26
+30
+19
+19
+22
+38
+38
+37
+31
+39
+28
+14
-6
+26
+16
+6
+47
+37
+18
+23
+5
+29
+23
+5
+29
+23
+27 '
+12
+27
+30
+12
+27
+45
+9
+37
+22
+29
+23
+33
+13
+21
+30
+5
+17
+24
+29
+23
+14
+2
+19
+27
+12
+38
+30
+12
+ 16
+4
+12
+16
+4
+12
+16
FSU-27
-------
Site
Spring Summer Autumn Winter Annual
Brest
Minsk
Lithuania
Aktyubinsk
N. Kazakhstan
Zelinograd
Karaganda
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
+27
+25
+70
+24
+25
+70
+29
+2
+84
+40
+12
+59
+51
+33
+84
+51
+16
+84
+51
+47
+84
+9
-16
+6
+35
-16
+6
+49
-9
+11
+45
+7
-22
+99
+37
-26
+99
+ 11
-26
+99
+24
-26
+6
+2
+33
+33
+2
+33
+28
+24
+29
-22
+94
+24
+52
+38
+2
+52
+33
+2
+52
-4
+2
+ 14
-6
+47
+16
-6
+47
+ 14
+23
+50
+31
+46
+32
+78
+76
+30
+78
+22
+30
+78
+23
+30
+14
+2
+38
+27
+2
+38
+32
+ 10
+42
+24
+40
+23
+70
+46
+22
+70
+20
+22
+70
+22
+22
FSU-28
-------
Table 5. Solar rdiation change (%) predicted by GISS, GFDL, and UKMO climate change
scenarios.
Site
Spring Summer Autumn Winter Annual
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-on-Don
N. Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
0
+10
+22
-1
+14
+22
-5
+47
+8
-3
+11
+31
+1
-2
+29
-1
+1
+18
-4
+10
+29
-2
+2
+26
0
+10
+23
-1
+14
+21
0
-1
+21
0
-1
+21
-3
+6
+9
-3
-2
+9
-11
+1
+5
-2
-3
+15
0
-1
+16
-3
0
+9
-1
-1
+1
+1
+7
+6
-3
+6
0
-3
-2
+4
0
0
+4
0
0
+4
-6
+6
+10
-4
-9
+10
-9
-2
+6
0
+2
+15
+5
+6
+14
-4
+11
+6
0
+9
+11
-4
+7
+12
-6
+6
+6
-4
+9
+5
+7
+4
+5
+7
+4
+5
-16
+83
+4
-20
+62
+4
-18
+37
-1
-20
+24
+6
+1
+25
+3
-20
+14
0
-20
+10
+ 10
-12
+51
+7
-16
+83
+ 1
-20
+62
-6
+1
+16
-6
+1
+16
-6
-6
+26
+11
-7
+20
-11
-8
+20
+4
-6
+8
+16
+2
+7
+17
-7
+4
+7
-6
+ 15
+ 15
-4
+ 17
+ 12
-6
+26
+8
-7
+20
+6
+2
+5
+6
+2
+5
+6
FSU-29
-------
Site
Brest
Minsk
Lithuania
Aktyubinsk
N.Kazakhstan
Zelinograd
Karaganda
Spring Summer Autumn
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
-2
-2
+23
0
+2
+23
-2
+9
+14
-5
+21
+13
-5
+6
+1
-5
+10
+11
-5
+10
+11
+1
+7
0
-3
+7
0
-2
+3
-3
-1
+2
+8
0
0
+14
0
-1
+14
+0
-1
+14
-4
+7
+6
-6
+7
+6
-4
-1
+14
0
+12
+3
-2
+12
+15
-2,
+6
+15
-2
+9
+15
Winter Annual
-12
+51
+1
-16
+51
+1
-1
+37
-2
-20
+7
-2
-21
+1
+1
-21
+21
+1
-21
+35
+ 1
-4
+17
+8
-6
+17
+8
-4
+12
+6
-6
+10
+5
-7
+5
+10
-7
+9
+10
-7
+12
+ 10
FSU-30
-------
Table 6. Simulated change in wheat yield under projected GCM climate scenarios for selected
sites. The physiological effects of 555 ppm CO2 were considered in each climate change
scenario.
Base Yields
(t ha'1)
Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N. Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N. Kazakhstan
Zelinograd
Karaganda
Spr
5.0
3.0
3.5
2.1
0.7
3.2
2.8
6.7
5.4
2.9
1.9
2.8
4.2
4.0
4.3
1.6
4.6
2.4
2.0
Win
6.2
6.0
5.7
3.2
1.0
3.7
4.6
5.8
4.0
5.2
2.1
2.4
3.9
4.6
4.7
1.3
-
-
1.7
Simulated yield changes
under GCM scenarios %
GISS GFDL UKMO
Spr
+12
+25
+18
+ 4
+ 9
+12
-5
+ 6
+16
+ 4
+30
+31
+ 5
+10
+17
+61
-5
+55
+59
Win
+23
+23
-28
+45
+13
+51
+25
+18
+13
+34
+52
+95
+22
+23
+42
+68
-
-
+102
Spr
-31
+ 1
+ 1
+ 6
+51
-9
-3
-20
-18
-11
-3
+ 3
-25
-32
-26
+16
-1
-12
+ 6
Win
0
+ 9
-34
+45
+28
+30
+11
-8
-4
+12
+25
+71
+ 6
+ 2
-38
+ 12
-
-
+ 5
Spr
+34
-31
-1
-48
-7
-30
-38
-19
-33
-44
+ 1
+11
-35
-32
-27
-36
-38
-28
-18
Win
-12
-16
0
-44
+32
+30
-53
-16
-33
-17
+91
+ 130
-33
0
+47
- 1
-
-
+ 9
Base yields are simulated considering a CO2 concentrationof 330 ppm.
FSU-31
-------
Table 7. Simulated changes in wheat season length for GCM climate change scenarios at selected
sites. The physiological effects of 555 ppm CO2 were considered in each climate change
scenario.
Season Length Changes (days)
GISS GFDL UKMO
Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N-Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N.Kazakhstan
Zeli-grad
Karaganda
Spr
-8
-9
-1
-8
-9
-8
-13
-10
-8
-8
-12
-11
-10
-8
-8
-8
-14
-10
-9
Win
-23
-22
-72
-20
-14
-20
-20
-25
-23
-24
-22
-22
-20
-18
-33
-16
-
-
-16
Spr
-19
-11
-6
-6
-7
-8
-11
-19
-15
-10
-10
-11
-18
-20
-18
-8
-14
-11
-12
Win
-25
-20
-20
-15
-15
-19
-16
-25
-20
-20
-18
-18
-23
-23
-38
-16
-
-
-19
Spr
-27
-24
0
-16
-14
-13
-22
-23
-26
-21
-19
-19
-26
-25
-25
-13
-22
-18
-17
Win
-55
-49
-16
-35
-39
-34
-37
-52
-44
-44
-50
-48
-52
-50
-70
-32
-
-
+32
FSU-32
-------
Table 8. Simulated changes in wheat evapotranspiration for projected GCM climate scenarios at
selected sites. The physiological effects of 555 ppm CO2 were considered in each climate
change scenario.
Change in ET (mm)
Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N.Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N.Kazakhstan
Zeli-grad
Karaganda
Spr
-9
-5
-2
0
-4
-2
-4
-7 •
-8
-7
-4
-8
-10
-8
-4
+7
+2
+15
+15
GISS
Win
-6
-3
-30
+1
0
-2
-1
-5
-8
-4
-3
-2
-12
-10
-2
+8
-
-
+22
Spr
-12
-4
-10
-1
-4
-10
-8
-13
-9
-6
-8
-10
-15
-12
-16
-4
-5
-12
-2
GFDL
Win
+4
+4
+8
+7
+2
-3
-1
-1
+6
0
-3
-2
-9
-3
-12
+9
-
-
+12
UKMO
Spr
-14
-10
+20
-10
-5
-10
-13
-9
-8
-11
0
-2
-11
-9
-4
-6
-13
-10
-6
Win
0
+2
+27
+8
0
+1
+4
-7
0
+3
-4
-1
-24
-3
-6
+11
-
-
+18
FSU-33
-------
Table 9.
Aggregated national wheat yield change (%) for GCM equilibrium and the GISS
transient climate change scenarios.
Scenario Carbon Dioxide Yield changes (%)
(PPm) Spring Wheat Winter Wheat
GISS-Trans-2010 405
GISS-Trans-2030 460
GISS-Trans-2050 530
GISS 555
GFDL 555
UKMO 555
+8
+12
+18
+21
-4
-18
+16
+24
+38
+41
+12
+9
FSU-34
-------
Table 10. Simulated wheat yield response to changes in sowing date and irrigation under the
UKMO climate change scenario. The physiological effects of 555 ppm CO2 were
considered in the simulation.
Site
Management Sowing date
Saratov
Rainfed
Rainfed
Irrigated
Irrigated
15 days early
30 days early
15 days early
30 days early
% Change from
Baseline yield
Zhitomir
Rainfed
Rainfed
Irrigated
Irrigated
15 days late
30 days late
15 days late
30 days late
-1
-3
+50
+40
+1
-1
+70
+100
FSU-35
-------
-------
SECTION 5: AFRICA
-------
-------
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT AND MAIZE YIELDS
IN EGYPT
H.M. Bid
Soils & Water Resources Institute,
ARC, Ministry of Agriculture, Egypt
EGYPT-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
METHODS
Climate Data and Climate Change Scenarios
Crop Models
Validation of the Crop Models
RESULTS AND DISCUSSION
Crop Sensitivity to Temperature Increases
Yields under GCM Climate Change Scenarios
Yields under GISS Transient Scenarios
Adaptation Strategies for Wheat and Maize to Climate Change
Future Research Needs
REFERENCES
EGYPT-2
-------
SUMMARY
The potential impact of climate change on maize and wheat production in Egypt was evaluated by
simulating crop production under different climatic scenarios and by analyzing crop sensitivity to temperature
increases in the two major agricultural regions of Egypt. Increases of 2°C and 4°C reduced wheat and maize
yields in the Delta and Middle Egypt regions. Under GCM climate change scenarios, yields decreased in
comparison to current climate conditions, even when the beneficial effects of CO2 were taken into account.
The UKMO scenario produced the largest yield decreases. According to this simulation study, the impact of
climate change on national wheat and maize production would be severe. Future adaptation strategies to
climate change may involve the development of new, more heat-tolerant cultivars.
INTRODUCTION
Major crops in Egypt include wheat (used as a staple food crop), maize (used primarily as coarse grain
for animal feed), clover, cotton, rice, sugar cane, fava bean, and soybean. The national wheat and maize
production do not meet the current demand for these crops, and each year additional amounts have to be
imported—up to 60% of total consumption in the case of wheat. The rapid growth of the country's population,
the economic stress of reliance on food imports, and the limited area for agriculture (most of the country is
a desert) require Egyptians to find new ways to increase agricultural productivity. If climate change as
projected by atmospheric scientists (IPCC 1990) adversely affects crop production, Egypt would have to
increase its reliance on costly food imports.
The purpose of this study was to investigate the potential effects of climate change on wheat and
maize yields and irrigation demands in Egypt. Wheat and maize are irrigated under the flood irrigation system,
using water from the Nile River. Two important agricultural regions were selected for the study (Figure 1).
The Sakha region, at the north of the Nile delta, is represented by the site Sakha (Khafr El-Sheik governorate)
(31.07°N; 30.57°E). This region of the delta is the most fertile area in Egypt and produces about 60% of the
national wheat and 75% of the total maize crop (Table 1). The Giza region, near Cairo, is represented by the
site Giza (30.03°N; 31.13°E).
METHODS
Climate Data and Climate Change Scenarios
Daily maximum and minimum temperatures, precipitation, and solar radiation data were obtained
for Sakha from 1975 to 1989 and for Giza from 1960 to 1989 (Table 2).
The climate change scenarios used in this simulation study were created using three equilibrium
General Circulation Models (GCMs) combined with the observed daily climate data for each site (Table 2).
Three GCMs were used: Goddard Institute for Space Studies, (Hansen et al. 1983), Geophysical Fluid
Dynamics Laboratory, (Manabe and Wetherald 1987), and United Kingdom Meteorological Office, (Wilson
and Mitchell 1987). Table 3 shows the projected seasonal and annual GCM climate changes for these sites.
In Giza, the GISS and UKMO climate change scenarios presented consistent, unexplained errors in the years
1968 (GISS scenario) and 1974 and 1989 (UKMO scenario). These years were not included in the study.
The study also includes: (a) a sensitivity study to arbitrary changes in temperature (+2° C and +4° C);
and (b) a transient scenario study, using the GISS transient run A for the years 2010, 2030, and 2050 (Hansen
et al. 1988). Atmospheric CO2 concentrations of 405 ppm, 460 ppm, and 530 ppm were used for the years 2010,
2030, and 2050, respectively.
EGYPT-3
-------
Crop Models
Crop yields and demand for irrigation water were estimated with the CERES-Wheat (Ritchie and
Otter 1985) and CERES-Maize (Jones and Kiniry 1986) models using the version developed by the
International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT 1989). The IBSNAT crop
models can simulate the physiological effects of CO2, i.e., increase in rate of photosynthesis and change in
evapotranspiration due to increases in stomatal resistance (Acock and Allen 1985). Simulations for the GCM
scenarios and sensitivity analyses included climate change alone and climate change with the physiological
effects of CO2 at both sites.
Management Variables for the Crop Models
Typical soils at Giza and Sakha are montmorillonitic, thermic, slightly calcareous, and deep (Abdel
Wahed 1983). The texture, albedo, and water-related specific characteristics of these soils are adequately
represented by the generic soil (Medium Silty Clay) provided for the study (Jones et al. 1990).
Wheat and maize are grown using flood irrigation (Gad-El Rab et al. 1988; Eid 1977) . For the
simulations, the automatic irrigation option was chosen to provide the crops with nonlimiting water; the
models do not include an option that simulates flooding. Maize and wheat are fertilized in the regions in this
study, and therefore the simulations did not consider nitrogen stress.
Crop Model Validation
The CERES-Wheat and CERES-Maize models were validated by comparing observed data on biomass,
yields, and maturity dates to simulated values (Table 4). The results of the validation experiment indicate that
the CERES crop models can be used at the selected sites in Egypt. The observed data on grain yield and
season length were very close to the corresponding simulated values. The observed total biomass was slightly
smaller than the simulated one. According to these results, the models were considered validated for the
conditions of the study.
RESULTS AND DISCUSSION
Crop Sensitivity to Temperature and CO2 Increases
Table 5 shows simulated grain yield, season length, evapotranspiration (ET), and irrigation demand
in response to arbitrary changes in temperature. The physiological effects of CO2 were considered in each
scenario (shown as 555 ppm in the tables). Increases in temperature resulted in lower grain yields for both
crops at the two sites; yield reductions showed a linear relation to temperature increases. Yield decreases were
larger for maize, a summer crop, than for wheat, a winter crop. The temperature-induced reductions in maize
and wheat may be due to a shortening of the grain-filling periods. Although not simulated in CERES-Maize,
high temperatures as projected by the GCM scenarios may also cause pollinization failure. When the direct
effects of CO2 were considered, simulated grain yields increased in all scenarios compared with the results in
the scenarios of climate change alone. These increases, however, did not compensate for the yield decreases
under the higher temperature scenarios compared to base yields.
Total crop evapotranspiration and irrigation demand increased with temperature increases in the case
of maize, but decreased in the case of wheat, due to the shortening of the crop growing season. When the
physiological effects of CO2 were considered, ET and irrigation water demand were reduced for both crops
compared to the case of climate change alone. These reductions are the consequence of the CO2-induced
EGYPT-4
-------
decreases in the transpiration rate per unit leaf area (Acock and Allen 1985) and the shortening of the crop
growing season included in the crop models.
Yields under GCM Climate Change Scenarios
All climate change scenarios considered resulted in simulated decreases in maize and wheat yields at
both sites (Table 6 and Figures 2 and 3). The largest decreases in yield for both wheat and maize were under
the UKMO scenario in Giza. The physiological effects of CO2 caused simulated maize yields to increase
slightly and wheat yields to increase substantially compared to yields under the climate change scenario alone.
The relative responses of wheat and maize are due to the greater CO2 photosynthetic response in C3 crops
(wheat, barley, soybean, cotton, rice, and many others) than in C4 crops (maize, sugarcane, and sorghum)
(Cure 1985).
Considering the simulated negative impacts of climate change on simulated wheat and maize yields
in the Delta and Middle Egypt regions, it is possible to conclude that climate change may bring about
substantial reductions in the national grain production.
Transient Scenarios
The crop responses to the transient GISS scenarios are shown in Table 7. Simulated grain yield
decreases were not linear with time; the largest reductions correspond to the 2030s.
Adaptation Strategies for Wheat and Maize to Climate Change
Changes of maize and wheat cultivars were considered as possible adaptation strategies to climate
change (Table 6). Nevertheless, for the two locations, all cultivars tested had similar losses in yield under the
climate change scenarios in comparison to baseline yields. It will be important for Egypt to develop new
cultivars that are more adapted to higher temperatures. Since the crop models can be used to identify
appropriate crops, varieties, and management strategies to maximize benefits and minimize risks associated
with future climatic change, further simulation studies would be valuable in assessing the risks associated with
given production strategies.
EGYPT-5
-------
REFERENCES
Abdel Wahed, A. 1983. Research report on soil survey. Soil Survey EMCIP Res. Extension Centers.
Publication No.62. Consortium for International Development. ARC, Cairo, May 1983.
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure (eds.), Direct
Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of Energy. DOE/ER-0238.
Washington, D.C. pp. 33-97.
Bid, H.M. 1977. Effect of some climatic factors on evapotranspiration, growth, and yield of wheat. Ph.D.
Thesis, Fac. of Agric., Al-Azhar Univ., Egypt.
Gad-El Rab, G.M., N.G. Ainer, and H.M. Bid. 1988. Water stress in relation to yield of wheat and some water
relations in wheat. Egypt. J. Soil. Sci. 28, No.4, pp.433-445.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient three-
dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-622.
Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.
IBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project. 1989. Decision
Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil
Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Jones, J.W., S.S. Jagtap, G.Hoogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer.
Las Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.
Jones, C.A, and J.R. Kiniry. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development.
Texas A&M University Press. College Station, TX. 194 pp.
IPCC 1990. The IPCC Scientific Assessment, eds. J.T. Houghton, G.J. Jenkins, and JJ. Ephraums. Cambridge
University Press.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
model. In W.O. Willis (ed.). ARS Wheat Yield Project. Department of Agriculture, Agricultural
Research Service. ARS-38. Washington, D.C. pp. 159-175.
EGYPT-6
-------
Wilson, C.A., and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
EGYPT-7
-------
Table 1.
Wheat and maize production in Egypt.
Area (ha)
Production (t)
Yield (t ha'1)
WHEAT
Giza Region (1984)
Sakha Region (1984)
Egypt total
(avg. 1980-1988)
MAIZE
Giza Region (1984)
Sakha Region (1984)
Egypt total
(avg. 1980-1988)
26,111
39,395
572,000
54,269
36,264
754,000
107,055
162,110
2,166,000
309,290
204,557
3,386,000
4.15
4.12
3.79
5.66
6.64
4.49
EGYPT-8
-------
Table 2. Observed average temperature, total precipitation, and solar radiation during the crop
growing season at Giza and Sakha.
Maize at Giza
Temperature
Precipitation
Solar radiation
Maize at Sakha
Temperature
Precipitation
Solar radiation
Wheat at Giza
Temperature
Precipitation
Solar radiation
Wheat at Sakha
Temperature
Precipitation
Solar radiation
25.6 °C
0.0mm
549 MJ m'2
23.9° C
2.0mm
546 MJ m'2
16.5 °C
17.0 mm
408 MJ m'2
15.4 °C
42'.0 mm
399 MJ m'2
EGYPT-9
-------
Table 3. Seasonal and annual changes in temperature, precipitation, and solar radiation under GCM
climate change scenarios at Giza and Sakha.
GCM CLIMATE CHANGES AT SAKHA AND GIZA
Temperature Change (°C)
Spring
Summer
Autumn
Winter
Annual
Spring
Summer
Autumn
Winter
Annual
GISS
5.1
3.2
4.4
4.0
4.2
Precipitation
GISS
-7.1
350.0
27.3
5.9
55.7
GFDL
4.5
4.4
4.1
3.7
4.2
Change (%)
GFDL
-19.2
0.0
-20.0
-10.0
-15.3
UKMO
4.7
4.1
4.5
4.5
4.4
UKMO
-12.5
-37.0
1.2
-8.9
-13.8
Solar Radiation Change (%)
Spring
Summer
Autumn
Winter
Annual
GISS
-0.3
-4.2
-1.2
0.0
1.7
GFDL
2.0
-0.6
0.6
0.8
0.6
UKMO
6.2
6.1
1.3
8.7
5.5
EGYPT-10
-------
Table 4. Simulated and observed wheat and maize yields, biomass, and season length at Giza and
Sakha.
Yield (t ha1)
Cultivar
Giza-2
H.204
Cairo-1
Pio.514
D.C.202
Year
1985
1985
1958
1985
1984
Sim
9.95
10.70
9.77
9.92
9.20
Obs
Maize
9.90
10.54
9.61
9.83
Maize
8.93
Biomass
Sim
(Sakha)
23.47
24.77
24.75
24.11
(Giza)
31.59
(tha1)
Obs
26.73
28.45
25.94
26.53
24.44
S. length
(days)
Sim
120
122
122
122
128
Obs
121
121
121
121
124
Sakha 8
1983
Giza-156 1975
Mexipak65 1975
Wheat (Sakha)
4.30 4.35 18.72 14.41.
Wheat (Giza)
6.22 6.14 14.77 18.27
6.24 6.27 15.10 16.73
153 156
132 133
132 133
Maize experiments: Sakha Clayey soil; planting June 1; row spacing 70 cm; 5.7 plants m"2; initial soil water
(depth, water content) (5,.246) (10..246) (15,.208) (15,.208) (15..208) (15,.318) (15,318) (15,.318); irrigation
dates and amounts (Julian day, mm) (169,158) (192,37) (199,39) (204,35) (210,41) (216,39) (222,36) (228,37)
(234,37) (240,35) (246,36) (253,37) (260,39) (267,37) (274,38) (282,36) (290.35) = 751 for H.204;
N-fertilization dates and amounts (Julian day, kg ha'1} (169,69) (190,133).
Wheat experiments: Sakha Clayey soil; planting December 24; row spacing 10 cm; 350 plants"2; irrigation dates
and amounts (Julian day, mm) (385,146) (29,35) (42,34) (73,35) (84,34) (95,37) (103,34) (111,36) (119,34)
(126,40) (132,36) (137,35) (142,38) (147,37); N-Fertilization dates and amounts (Julian day, kg ha'1) (18,63)
(46,63).
EGYPT-11
-------
Table 5. Sensitivity analysis of yield, season length (SL), total evapotranspiration (ET), and irrigation
water (Irrig.) simulated with CERES-Wheat and CERES-Maize.
WHEAT (SAKHA)
WHEAT (GIZA)
CO2
level
(ppm)
330
555
CO2
level
(ppm)
330
555
Temp
Increase
(°C)
0
2
4
0
2
4
Temp.
Incr.
(°C)
0
2
4
0
2
4
Yield
(t ha'1)
3.80
3.30
3.20
4.40
3.89
3.46
Yield
(T/Ha)
10.25
8.73
7.54
10.33
9.11
7.96
SL
(days)
141
133
125
141
133
125
MAIZE
SL
(days)
115
107
104
115
107
104
ET
(mm)
552
547
511
517
498
498
(SAKHA)
ET
(mm)
569
565
609
480
481
521
Irrig.
(mm)
475
456
430
432
418
404
Irrig.
(mm)
524
527
566
433
438
475
Yield
(t ha'1)
5.60
4.40
3.20
6.00
4.80
3.58
Yield
(T/Ha)
9.80
8.70
7.50
10.40
9.20
8.00
SL
(days)
118
106
96
118
106
96
MAIZE
SL
(days)
120
114
114
120
114
114
ET
(mm)
288
246
206
260
219
181
(GIZA)
ET
(mm)
583
608
665
500
525
577
Irrig.
(mm)
250
209
174
220
183
151
Irrig.
(mm)
543
567
624
462
482
534
The cultivar used at Sakha was "Giza-2" and at Giza, "D.C. 202." Similar results were obtained with
cultivars "H-204," "Cairo 1," and "Pioneer 514."
The cultivar used at Sakha was "Sakha-8" and at Giza, "Giza-156." Similar results were obtained with
the cultivar "Mexipak-65" at Giza.
EGYPT-12
-------
Table 6. Simulated maize and wheat yields under baseline climate and GCM climate change scenarios
at Sakha and Giza.
Simulated maize yield (t ha'1}
Site
Sakha
Cultivar
Giza-2
H.204
Cairo-1
Pio. 514
BASE
330
10.25
10.60
7.61
10.15
GISS
330
7.96
8.16
7.61
7.79
GISS
555
8.41
8.62
8.04
8.23
GFDL
330
7.56
7.85
7.24
7.44
GFD
L
555
7.98
8.28
7.65
7.86
UKMO
330
8.38
8.63
8.01
8.21
UKMO
555
8.46
8.71
8.07
8.28
Giza D.C.202
9.83
6.96 7.35 7.73 8.16
3.20
3.38
Simulated wheat yield (t ha"1)
Sakha Sakha-8
4.16
3.03 3.20 3.21 3.41
3.27
3.46
Giza
Giza 156
Mexipak65
5.57
5.42
3.22 3.59
3.17 3.52
3.44
3.38
3.82
3.74
1.32
1.30
1.49
1.46
330: Effects of climate alone on crop yield
555: Physiological effects of CO2on crop yield
EGYPT-13
-------
Table 7. Simulated wheat and maize yields under the GISS transient scenarios. The physiological
effects of CO2 on crop yield were included.
Crop
Site
Scenario
Yield (t ha'1)
MAIZE Sakha* BASE
2010
2030
2050
Giza** BASE
2010
2030
2050
WHEAT Sakha* BASE
2010
2030
2050
Giza** BASE
2010
2030
2050
10.25
8.76
7.93
8.15
9.80
8.99
8.29
7.74
4.40
4.79
4.47
3.89
5.60
5.48
4.31
3.58
Maize cultivar shown at Sakha "Giza-2". Similar results were obtained with "H-204," "Cairo-1," and
"Pioneer-514." Wheat cultivar simulated at Sakha "Sakha 8."
Maize cultivar simulated at Giza "D.C.202." Wheat cultivar simulated at Giza "Giza-156"; similar
results were obtained with "Mexipak-65."
EGYPT-14
-------
EGYPT
Figure 1. Map of Egypt and location of the crop modeling sites.
-------
YIELD (T/HA)
SAKHA
SITE
GIZA
CLIMATE SCENARIOS
IBASE330 ^GISS330 E^GISS555 ^GFDL330
IGFDL555 OUKMOSSO ^UKMO555
Irrigated simulation
Figure 2. Simulated maize yield under GCM 2xCO2 climate change scenarios.
-------
YIELD (T/HA)
0
SAKHA
SITE
GIZA
CLIMATE SCENARIOS
IBASE330 I^GISS330 ^GISS 555 MGFDL330
IGFDL555 OUKMO330 ^UKMO 555
Irrigated simulation
Figure 3. Simulated wheat yield under GCM 2xCO2 climate change scenarios.
-------
-------
IMPLICATIONS OF CLIMATE CHANGE FOR MAIZE YIELDS IN ZIMBABWE
Paul Muchena
Plant Protection Research Institute, Zimbabwe
ZIMBABWE-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Description of the Agricultural Regions
Farming Sectors, Management Systems and Crops in Zimbabwe
Food Trade and Vulnerabilities
Objectives
METHODS
Baseline climate data
Climate Change Scenarios
Crop Model, Inputs and Simulations
Validation of the crop model
RESULTS
Sensitivity analysis
Maize yields under GCM climate change scenarios
Adaptations to climate change
CONCLUSIONS
REFERENCES
ZIMBABWE-2
-------
SUMMARY
The potential impact of climate change on maize production in Zimbabwe was evaluated by simulating
crop production under climate scenarios generated by General Circulation Models (GCMs). The baseline
climate data for each site was also modified by increasing daily air temperatures by 2°C and 4°C for a
sensitivity analysis. Maize yields decreased under the GCM climate change scenarios, even when the direct
beneficial effects of CO2 were included in the simulation. Temperature increases of 2°C and 4°C reduced
maize yields at all sites.
If these results are realized, the impact on national maize production in Zimbabwe would be severe.
Many farmers would not find maize economical to produce. Adaptation results suggest that major changes in
the farming system might ameliorate some losses, but the costs of additional fertilizer, seed supplies, and
irrigation could be high.
INTRODUCTION
Description of the Agricultural Regions
Zimbabwe is situated in southern Africa between the latitude of 15° and 23° S and the longitude of
25° and 33 °E, and occupies a land area of 390,759 square kilometers. Although Zimbabwe lies within the
tropics, only one-fifth of the country's area experiences typical tropical climate. The rest of the country
includes areas of altitude between 600 and 1,200 meters above sea level (three-fifths of the total area) and
areas above 1,200 meters (one-fifth of the total area). Agricultural land constitutes 85% of the land resources
in the country. The land is divided into five natural agroecological regions on the basis of rainfall and
temperature (Figure 1).
Farming Sectors, Management Systems and Crops in Zimbabwe
The agricultural land is used by large- and small-scale commercial sectors, and by communal and
resettlement sectors. The large-scale commercial sector includes private and state enterprises (39% of the total
agricultural land). Private enterprises (about 4,500 large-scale farmers) hold title deeds to the land. The large-
scale cropping system is intensive and mechanized, using modern production technologies over large areas of
monoculture. Summer crops are irrigated when necessary, and winter crops are always irrigated. The small-
scale commercial sector consists of small private farms in which the farmers hold title deeds to their farms (4%
of the total agricultural land).
The communal sector comprises about 900,000 farming families (4.5 million people; 49% of the total
agricultural land). Individual farmers hold family rights to limited areas for cultivation (2.5 ha/family).
Monoculture of maize, the staple crop, generally dominates the system. The cropping system is similar to the
small-scale sector, although the farms are smaller and the production capacity is comparatively less. About
40,000 families have been resettled on about 8% of the total agricultural land. As in the case of the communal
sector, the resettlement sector has socioeconomic constraints due to poor planning and management.
Maize is the primary food crop food in Zimbabwe and occupies about half of the total agricultural
crop land. It is grown by all of the farming sectors, although the productivity varies largely among them (mean
yields of 3.21 ha'1 and 1.11 ha'1 in the commercial and communal sectors, respectively). Eighty percent of the
area sown with maize is communal. Other important crops are small grain cereals (sorghum, finger millet, and
pearl millet); oil-seed crops (soybeans, groundnut, and sunflowers); and important cash crops (tobacco, cotton,
coffee, tea, and horticultural crops). The small grains use about 20% of the total crop land and are generally
ZIMBABWE-3
-------
grown by communal farmers.
Food Trade and Vulnerabilities
Crop production in Zimbabwe usually exceeds demand, especially in years when rainfall is not limiting.
However, when rainfall is below average, production in the marginal regions is often not enough to meet local
demand, and the population can experience food deficits.
Objectives
During the last 10 years, the frequency of drought has increased in Zimbabwe, causing crop yields to
be below average. If climate change brings about warmer and dryer conditions in the region, crop production
may be seriously damaged. For an adequate impact study, it is important to select sites representing the
different agroecological regions of Zimbabwe. For this study, these three sites were selected: Banket, Gweru,
and Chisumbanje. The sites represent high, low, and marginal productivity regions, respectively. They were also
chosen because the long-term, daily weather data sets (solar radiation, maximum and minimum air
temperatures, and precipitation) were available to run the crop simulation model.
The specific objectives of the study are to evaluate: (1) potential climate changes that might occur
from doubled CO2; (2) the possible impact of climate change on maize yields; and (3) the sensitivity of maize
yields to temperature increases (+2°C or +4°C).
METHODS
Baseline Climate Data
Figure 2 shows the daily observed climate data (precipitation, solar radiation, maximum and minimum
air temperatures) for Banket (1968 to 1988), Gweru (1960 to 1987), and Chisumbanje (1962 to 1988).
Chisumbanje has the warmest average annual temperature. The seasonal precipitation regime is similar at all
sites (wet in the summer and dry in the winter), but there are large differences in the total precipitation among
the sites. Chisumbanje is the driest site with average annual precipitation of about 560 mm.
Climate Change Scenarios
Using the GCMs, the observed climate data were modified to create climate change scenarios for each
site. The GCMs used were developed by the Goddard Institute for Space Studies (GISS) (Hansen et al. 1983),
the Geophysical Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987), and the United Kingdom
Meteorological Office (UKMO) ((Wilson and Mitchell 1987). Daily observed climate values were modified
with the monthly outputs of the GCMs for the particular gridbox where the site is located.
The climate change scenarios created for each site show major changes from the current climate
(Table 1). Temperature increases are very similar in the three sites. The GFDL scenario produces the lowest
temperature increase (3.6°C), while the UKMO scenario produces the largest (5°C). Precipitation changes
vary among the sites: the GFDL scenario is the driest. Considering the temperature increases projected by the
GCMs and their possible effects on evapotranspiration, the climate change scenarios could imply water
shortages, especially in the sites that currently have low precipitation, such as Chisumbanje.
The baseline climate data for each site was also modified by uniformly increasing the values of the
daily air temperatures by +2°C and +4°C.
ZIMBABWE-4
-------
Crop Model, Inputs and Simulations
The maize simulation model used was the CERES-Maize model (Jones and Kiniry 1986). The model
simulates crop responses to changes in climate, management variables, soils, and different levels of CO2 in the
atmosphere. The software used to run the programs was developed by the Decision Support System for
Agrotechnology Transfer (DSSAT) and includes database management, crop models, and application programs
(Jones et al. 1990).
Management variables. Maize in Zimbabwe is mainly rainfed and partially fertilized. For the
simulations, maize was not irrigated and nitrogen was applied four and eight weeks after planting (60 kg ha'1
each time). Additional simulations to evaluate potential adaptive strategies considered possible changes in
management variables. For the irrigation simulation, the water demand was calculated assuming 100%
efficiency of the automatic irrigation system; a 1-meter irrigation management depth; automatic irrigation when
the available soil water is 50% or less of capacity; and the soil water for each layer is re-initialized to 100%
capacity at the start of each growing season. The plant population was higher in the irrigated simulations (7
plants m"2) than in the rainfed simulations (5 plants m"2).
Soils. Zimbabwe's soils are predominantly derived from granite and are often sandy and light-textured,
with low agricultural potential due to low nutrient content (especially nitrogen and phosphorus). Nevertheless,
in all regions, there is a significant portion of soils with a heavier clay content that is more suitable for crop
growth. The representative soils in Banket and Gweru are medium sandy loam soil and in the smaller region
of Chisumbanje are medium silt loam (as described in Jones et al. 1990).
Physiological CO2 effects. Since the scenario climate change has higher levels of CO2 than the current
climate, the CERES-Maize model includes an option to simulate the physiological effects of CO2 on
photosynthesis and water-use efficiency that results in higher yields (Acock and Allen 1985).
Simulations. For all climate scenarios included in this study (GCMs, sensitivity, GISS transient, and
UKMO scenarios for adaptation), maize growth was simulated under the conditions of climate change alone
and under the conditions of climate change with the physiological effects of CO2 on crop growth and yield.
Validation of the Crop Model
The CERES-maize model was validated at Banket and Gweru using local experimental and climate
data. The experimental data included soil fertility before planting, cultivar, planting date, phonological growth-
stage components and growth analysis, harvesting date, and final yield components. Experimental crop data
and climate were used for the 1987-88 season in Banket, and for the 1985-86 and 1986-87 seasons in Gweru.
In Banket, the observed yield was 13.6% lower than the simulated yield, and the observed season length was
1.6% shorter than the simulated season length. In Gweru, the mean observed yield was 2% lower than the
simulated yield and the observed season length was 1% longer than the simulated season length. The results
indicate that CERES-Maize is adequate to simulate maize growth under the conditions of this study, especially
to evaluate changes in phenology and relative changes in crop yields.
RESULTS
Sensitivity Analysis
A temperature increase of 2°C over the baseline climate reduced maize yields by 8% to 14% and by
24% to 27% when the baseline temperature was increased 4°C (Table 2). The largest yield decreases were in
Chisumbanje, suggesting that, at this site, maize is already close to the upper temperature limit for growth.
The increase in temperature reduced maize yields because the growing season was shortened and the crop had
ZIMBABWE-5
-------
less time for grain-filling and biomass accumulation. When the direct effects of CO2 were added, maize yields
increased slightly, but not enough to compensate for the adverse effects of the temperature increase.
Maize Yields under GCM Climate Change Scenarios
In Banket, simulated maize yields under the GISS, GFDL, and UKMO scenarios were reduced by
21%, 14%, and 28%, respectively. In Gweru, the reductions were 28%, 26%, and 50%, and in Chisumbanje,
they were 27%, 25%, and 39% (Figure 3). These yield reductions were a consequence of the shortening of the
growing season induced by higher temperatures. The physiological effects of CO2 caused maize yields to
increase at all sites, due to the higher photosynthetic rate of the crop with the elevated CO2. In Chisumbanje,
the probability of obtaining an acceptable yield (e.g., 2.51 ha"1) decreased with an increase of +2°C and under
the UKMO scenario (Figure 4).
Adaptation to Climate Change
We tested possible adaptation strategies to climate change at Gweru because, under the UKMO
scenario, it projects the largest yield decreases. Two high-cost adaptation strategies were tested: (1) increased
fertilization (double the amount of nitrogen) and (2) a combination of increased fertilization and irrigation
(simulated under the automatic irrigation option of the CERES model) (Figure 5). Increased fertilization
alone had a small, positive effect on the. yield under the UKMO scenario. However, even if increased
fertilization was combined with irrigation (nonlimiting water), the combined positive effect was not enough
to fully compensate for the yield losses under the UKMO scenario.
CONCLUSIONS
The results show that even when the physiological effects of CO2 were included, modeled maize yields
decreased significantly at all sites, as a result of climate change. The effect of climate change on maize
'production may force some farmers to switch to other crops with higher thermal requirements and drought
tolerance, such as sorghum, pearl millet, and finger millet. Although the results suggest that farmers may be
able to offset some of the yield losses by adaptation, this solution may not be a very realistic one. This is
because the costs of additional fertilizer, seed supplies, and irrigation may be high in a country with very
limited financial resgurces.
ZIMBABWE-6
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient tree-
dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-622.
Jones, J.W., S.S. Jagtap, G.Hogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer.
Las Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.
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. 194 pp.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Wilson, C.A, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate sensitivity Experiment with a Global Climate
Model Including a Simple Ocean. JGR 92:13315-13343.
ZIMBABWE-7
-------
Table 1. Annual temperature differences and precipitation ratios under GCM scenarios at selected
sites in Zimbabwe.
Temperature Changes
BANKET
GWERU
CHISUMBANJE
GISS
4.5
4.6
4.5
GFDL
3.7
3.7
3.6
UKMO
4.8
5.0
3.9
Precipitation Ratios
BANKET
GWERU
CHISUMBANJE
GISS
1.18
1.22
1.24
GFDL
1.04
0.93
1.05
UKMO
1.10
1.59
1.09
ZIMBABWE-8
-------
Table 2.
Sensitivity analysis of CERES-Maize to temperature and CO2 changes in Zimbabwe.
Simulated Yield (T/Ha)
Simulated Season Length
(days)
CO2 level
(ppm)
330
555
Temp.
(°Q
0
2
4
0
2
4
Banket
4.93
4.40
3.68
5.25
4.70
3.95
Gweru
3.72
3.41
2.83
4.41
3.99
3.29
Chisumb.
2.86
2.47
2.09
3.09
2.63
2.22
Banket
120
107
96
120
107
96
Gweru
135
115
101
135
115
101
Chisumb.
97
90
87
95
88
85
ZIMBABWE-9
-------
ZIMBABWE
-------
BANKET
TEMPERATURE (C)
PRECIP (MM/MONTH)
200
150
100
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
TEMPERATURE ESi PRECIPITATION
GWERU
TEMPERATURE (C)
PRECIP (MM/MONTH)
200
- 160
• 100
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
• TEMPERATURE
3 PRECIPITATION
CHISUMBANJE
TEMPERATURE (C)
PRECIP (MM/MONTH)
200
150
• 100
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
TEMPERATURE ES3 PRECIPITATION
Figure 2. Baseline climate for Banket, Gweru, and Chisumbanje.
-------
BANKET, ZIMBABWE
YIELD (T/HA)
GISS
GFDL UKMO
I BASE CD CC CD CC'OE
GWERU. ZIMBABWE
YIELD (T/HA)
GISS
GFDL UKMO
I BASE CUCC CD CODE
CHISUMBANJE, ZIMBABWE
YIELD (T/HA)
GISS
GFDL UKMO
I BASE CD CC CD CC'DE
Figure 3. Simulated maize yields under base climate and GCM climate change scenarios. CC indicates
simulations under climate change alone; CC+DE indicates simulations under climate change
including the direct effects of CO2 on maize yield.
-------
m
O
cc
a.
o
DISTRIBUTION OF MAIZE YIELDS
FOR CHISUMBANJE, ZIMBABWE
i i i i i i i i i i i i i i i i i i i i i i t i
BASEUNE
*;
+2degC
X
+2 deg C, -20% P
0.00 0.40 0.80
YIELD, T/HA
Figure 4. Cumulative probability of maize yields under baseline, +2°C and -20% precipitation, and the
UKMO GCM scenario at Chisumbanje. Acceptable yield equals 2.5 T/ha.
-------
CHANGE IN MAIZE YIELD
WITH HIGH COST ADAPTATIONS
% CHANGE (BASE-3.72 MT/HA)
GWERU
UKMO
+FERT
+FERT+IRRIG
Figure 5. Maize yield changes from baseline under the UKMO scenario considering possible adaptation
strategies at Gweru. The direct effects of CO2 on yield were included in the simulation.
+FERT indicates simulations with double the amount of nitrogen fertilizer (120 kg/ha
applied at 4 and 8 weeks after planting). +FERT +IRRIG indicates simulations with double
the amount of nitrogen fertilizer combined with full irrigation.
-------
SECTION 6: ASIA
-------
-------
IMPLICATIONS OF GLOBAL CLIMATE CHANGE FOR AGRICULTURE IN
PAKISTAN: IMPACTS ON SIMULATED WHEAT PRODUCTION
Ata Qureshi
Climate Institute
Washington, DC, USA
and
Ana Iglesias
Inst. Nacional de Investigaciones Agrarias (INIA)
Madrid, Spain
PAKISTAN-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Description of the Regions and Sites
Aims of the Study
METHODS
Observed Climate
Climate Change Scenarios
Wheat Growth Simulation
Limitations of the Study
RESULTS
Sensitivity Analysis
Climate Change Scenarios
Continuous vs. Noncontinuous Soil Water Balance
Changes in Management under Climate Change
DISCUSSION
REFERENCES
PAKJSTAN-2
-------
SUMMARY
This study used global climate models (GCMs) and dynamic crop growth models to estimate the
potential agricultural effects of climate change in Pakistan. Under present climate conditions, wheat is
currently under stress due to high temperatures and arid conditions. Projected climate change caused simulated
wheat yields to decrease dramatically in the major regions of agricultural production, even under fully irrigated
conditions. Decreases in modeled grain yields were caused primarily by temperature increases that shortened
the duration of the life cycle of the crop, particularly the grain-filling period, exerting a strong negative
pressure on yields. These decreases were somewhat counteracted by the beneficial physiological CO2 effects
on crop growth, as simulated in this study. Some adaptation strategies were tested that partially offset the
negative impacts of climate change on wheat yields.
INTRODUCTION
Background
Pakistan is located between the Tropic of Cancer (23 °N) and latitude 38 °N and longitudes 61 °E and
77 °E (Figure 1). The total area of the country is 803,943 km2. The main river, the Indus, flows across the
regions of the Northwestern frontier (NWFP), Punjab, and Sind; the Baluchistan region is traversed by the
Zhob river. The Punjab region (in the great Indogangetic plain) accounts for the major part of agricultural
production (Wescoat and Leichenko 1990). Pakistan's climate is generally hot and arid, except some locations
have a distinct southwest monsoon rainy season.
Agriculture is a main contributor to the Pakistani economy. Agricultural systems in Pakistan (arable
crops, rangelands, forestry, and fisheries) employ 52% of the labor force and account for 32% of the Gross
National Product (GNP). Crop production accounts for about 70% of agricultural revenues and contributes
to export earnings. Cotton textile output is the most important industry. Food processing is also a major
industry, and wheat is the staple food crop in Pakistan. The principal crops grown in Pakistan are cotton, rice,
wheat, sugarcane, fruits, and vegetables. Cotton and rice are the main export agricultural commodities. Wheat
is of very high economic value and represents a large proportion of the food in the region. Therefore, wheat
was selected for this climat change simulation study. Wheat production occurs over a large area of Pakistan
and under very different climatic and management conditions. Temperature, precipitation, and soil moisture,
as well as frequency of heat waves and droughts, are significant factors in crop productivity.
Crop production in Pakistan is heavily dependent on the irrigation systems that have been installed
over the past four decades. Pakistan boasts the largest system of integrated irrigation in the world, covering
75% (195,000 km2) of the cultivated land (Wescoat and Leichenko 1990). Irrigated cropland accounts for about
90% of Pakistan's agricultural production. The irrigated areas in the Indus basin and plains (154,000 km2)
account for the bulk of the national harvest. Another 10 million hectares are cultivated under rainfed
conditions, also known as the "barani" method of crop production. Irrigation management practices in Pakistan
vary, but in most cases the system involved is "partial irrigation", whereby water is provided only to ensure
limited production. The potential need for increased irrigation is a concern for some areas of the region under
present circumstances. Overapplication and inadequate drainage has aggravated waterlogging in other areas.
In 1985, Pakistan suffered critical water shortages, leading to inadequate irrigation water supplies for the main
crop growing season and causing power blackouts (Wescoat and Leichenko 1992).
In addition to irrigation, crop production in Pakistan is also increasingly dependent on chemicals, such
as fertilizers and biocides. The environmental damage resulting from this dependence is one of Pakistan's most
severe problems. The inadequate attention to water drainage has resulted in soils that suffer from waterlogging
PAKISTAN-3
-------
and salinization, reducing soil fertility on about 100,000 km2 of land. At least 400 km2 of irrigated land is lost
to agriculture each year. In addition, more than 12,000 km2 of agricultural lands suffer from soil erosion.
Uncertain water supplies, soil problems, and floods along the entire Indogangetic plain also cause crop
production losses. Natural disasters such as earthquakes and landslides also account for destruction of crops
and damage to the country's agricultural infrastructure. In the first half of the 1980s, floods caused $3 billion
of damage to crops, buildings, roads, and other installations. Sedimentation from upper watersheds could
completely fill several reservoirs in the not-too-distant future.
At present, Pakistan is self-sufficient in grain production and is a net grain exporter. The decline of
the agricultural resource base could change the net balance. In addition, as the current 2.9% population
growth rate continues, the amount of cultivated land per rural habitant is expected to decrease from 0.32
ha/person (1983) to a projected 0.14 ha/person in the year 2010. The 1990 population was 122.7 million, and
the projected population for 2025 is 267.1 million (Wescoat and Leichenko 1990).
Climate change could have a critical impact in a country that already has limitations on agricultural
and natural resources (Myers 1989; Qureshi 1989). The enhanced greenhouse effect, due to an increased
atmospheric concentration of CO2 and other trace gases, is projected to lead to higher global surface
temperatures and changed hydrological cycles (IPCC 1990). Evidence of the potential impacts of climate
change in the Indian subcontinent has been mostly limited to studies in northern India, where a temperature
increase of 0.5°C is estimated to reduce wheat yields by about 10%; similar increases in central India would
probably lead to larger percentage reductions from a lower base yield (IPCC 1990; Gadgil et al 1988). Few
studies have projected potential agricultural impacts in Pakistan.
Description of the Regions and Sites
To assess the potential impacts of climate change, four sites were selected to represent the broad
differences in the agroecological zones of Pakistan (Figure 1).
The upper Indus plain in the Punjab region varies from 150-300 meters in altitude and is a fertile
agricultural area that accounts for about two-thirds of the total national wheat production (Table 1).
The site of Jhelum represents wheat production in the northern plateau of Punjab's agroecological
region. A large portion of the production is dryland, but irrigation is also important.
The lower Indus plain corresponds to the Sind province, is lower in altitude than the Punjab, and
declines to sea level at the coast. There are large tracts of deserts and marshes to the southeast, but
there are also agricultural areas that account for about 20% of the national wheat production. The
site of Khanpur represents wheat production in the region, which is conducted mainly under
irrigated conditions.
The northern area of Pakistan is mountainous, and wheat production is limited. It is represented by
the site of Gilgit in this study.
The central west area of Pakistan is represented by the site of D.I. Khan, which is also sandy and
mountainous. The wheat production is dryland.
The region of Baluchistan falls in the great plateau (914-1220 meters in altitude). Relatively small
amounts of wheat are produced in this region, which is unimportant from an international trade
perspective. Therefore, a site from this region was not included in this study.
Since Jhelum and Khanpur together account for nearly 90% of the country's wheat production, some
of the discussion that follows will pertain to these two sites only.
Aims of the Study
PAKISTAN-4
-------
This study analyzed the results of the simulation of wheat production, crop water use, and irrigation
demand at four contrasting locations under current climate and under climate change scenarios generated with
GCMs. The study also included a sensitivity analysis of arbitrary temperature and precipitation changes. The
physiological effects of elevated CO2 on wheat yields were simulated in all climate change scenarios.
METHODS
Observed Climate
Climate data for the four sites were obtained from the Pakistan Meteorology Department and include
monthly temperatures and precipitation. Daily climate variables were generated from monthly temperature and
precipitation data using the weather generator supplied by IBSNAT (IBSNAT 1989). Daily solar radiation
values were generated from observed monthly sunshine values using the WGEN program (Richardson and
Wright 1984).
Pakistan experiences extreme temperature fluctuations both seasonally and daily, and its climate is
generally arid. The three seasons are: November to February (cool and dry), March to May (hot and dry), and
June to October (hot and humid). The latter is the southwest monsoon season. In the Indus valley, the
climate becomes progressively more arid from north to south. In Sind, temperatures sometimes rise to 50 °C
in the summer, and precipitation can drop to below 100 mm per year.
Figure 2 shows the observed climate for the selected sites. The temperature regime shows strong
seasonally at all four of the sites, with the summers being very hot. Khanpur has the warmest average annual
temperatures. Gilgit is the coolest of the sites. There are large differences in the amounts of precipitation.
Jhelum, in the southwest monsoon region, has the highest precipitation and Khanpur has the lowest, with less
than 10 mm during 10 months of the year. Gilgit is also very dry, but is slightly wetter than Khanpur, with
maximum precipitation during the spring. D.I. Khan has a similar precipitation pattern to Jhelum, but it has
less rainfall.
Climate Change Scenarios
GCM Climate Change Scenarios. The GCMs compute climatic variables for different longitude and
latitude gridboxes; the predicted variables do not reflect variations within the gridbox. Mean annual changes
in climate variables from doubled CO2 simulations of three GCMs-Goddard Institute for Space Studies
(GISS, Hansen et al 1983), Geophysical Fluid Dynamics Laboratory (GFDL, Manabe and Wetherald 1987),
and United Kingdom Meteorological Office (UKMO, Wilson and Mitchell 1987)-were applied to the
generated daily climate to create climate change scenarios for each site. GCMs were used to create climate
change scenarios because they produce climate variables which are internally consistent; thus they allow for
comparisons between or among regions. The method used the differences between lxCO2 and 2xCO2 monthly
GCM temperatures, and the ratio between 2xCO2 and lxCO2 monthly precipitation and solar radiation
amounts (lxCO2 refers to current climate conditions and 2xCO2 refers to the climate that would occur with
an equivalent radiative forcing of doubled CO2 in the atmosphere).
The annual averages of the climate change scenarios show some major differences when compared
with the observed climate (Table 2). The temperature increases are similar at the four sites. The GISS and
GFDL scenarios produce comparable temperature increases (3 ° C to 5 ° C) while the UKMO scenario produces
more drastic changes (5°C to 7°C). The precipitation changes are more variable among the sites. In general,
the GCMs predict increases in global precipitation associated with warming because warmer air can hold more
water vapor. Khanpur, the more humid site under current conditions, shows the largest increase in
precipitation (UKMO scenario). In Gilgit, all scenarios show very small precipitation increases. In Jhelum
PAKISTAN-5
-------
(Punjab region), the GISS and GFDL scenarios show very small increases in precipitation amounts, while the
UKMO scenario shows decreases. The climate change scenarios in D.I. Khan are comparable to those in
Jhelum.
Because the water supply in the Indus basin depends entirely upon climate conditions within the basin,
any decrease in precipitation could affect the quantity of irrigation water available for the crops. Wescoat and
Leichenko (1992) used a river-basin model to show that flows in the Indus delta are sensitive to changes in
precipitation. When the GISS (+3.2°C, +30% precipitation) and the GFDL (+3.6°C, +20% precipitation)
climate changes were put into their model, water flow increased almost 30%. However, a 2°C temperature
increase combined with a 20% decrease in precipitation decreased flows into the delta by close to 50%, which
would result in economic and ecological devastation.
Sensitivity Scenarios. An alternative approach to analyze the possible impacts of different climate on
crop yield is to specify incremental changes to temperature and precipitation and to apply these changes
uniformly to the baseline climate. We used combinations of temperature increases of 0°C, +2°C, and +4°C
and precipitation changes of-20%, 0%, and +20%. The scenario, +2°C with a precipitation decrease of-20%,
is interesting because it assumes a possible temperature increase combined with a large decrease in monsoon
rain, which is not simulated under the GCM climate scenarios.
Wheat Growth Simulation
CEKES-Wheat Model. Potential changes in wheat physiological responses (yield, season length, ET,
and irrigation demand) were estimated with the CERES-Wheat model (Ritchie and Otter 1985) under
different climate scenarios. The CERES crop model has been validated with experimental data from many
locations that encompass a wide range of environments (Ritchie and Otter 1985). The model simulates
physiological crop responses (water balance, phenology, and growth throughout the season) on a daily basis
to the major factors of climate (daily solar radiation, maximum and minimum temperature, and precipitation),
soils (albedo and a variety of measures relating to water in the profile), and management (cultivar, planting
date, plant population, row spacing, and sowing depth). The choice of the CERES-Wheat model allowed us
to compare the simulation results with other studies in different regions of the world, and therefore, view
Pakistan's results from a global perspective.
Simulation of the Direct COZ Effects. Higher levels of atmospheric CO2 have been found to increase
photosynthesis and water-use efficiency, resulting in yield increases in experimental settings (Acock and Allen
1985). Because the climate change scenario has higher levels of CO2 than the current climate GCM simulations
(330 ppm), the CERES-Wheat model includes an option to simulate the physiological effects of CO2 on
photosynthesis and water-use efficiency (Peart et al. 1989) that result an increase in yield. A level of 555 ppm
CO2 was used in the crop models. The physiological effects of CO2 were considered in all scenarios and at all
sites.
Management Variables. The cultivar selected for the study in the four sites was Mexipak (Table 3)
because it accurately represents the most common cultivars in the area. Mexipak is a classic wheat variety
introduced in Pakistan during the Green Revolution. Mexipak (or a cultivar with similar characteristics) is a
variety that performs well almost anywhere in Pakistan, especially in the irrigated areas that produce the bulk
of the national wheat production. Mexipak has been calibrated and validated for many different climate
conditions. The "genetic coefficients'1 of Mexipak were taken from the DSSAT data base (IBSNAT 1989).
Season length and yield were compared with experimental data from the Barani Agricultural Research Institute
for the site of Jhelum. A number of new wheat varieties are currently grown in Pakistan: Chakwal-86, Rewal-
87, and Pak-81.
The management variables used for the CERES-Wheat model were determined for each location
according to information on current practices. Wheat in Pakistan is grown under partial or deficit irrigation,
PAKISTAN-6
-------
in which farmers provide water to the crop but do not seek to optimize yield per hectare. This subirrigation
phenomenon is difficult to simulate exactly with the CERES model, since the input irrigation has to be pre-
scheduled before starting the simulation. Therefore, we simulated wheat production under dryland and
irrigated conditions to provide a range of possible scenarios and to analyze the production changes. We
simulated irrigation under the "automatic option" to provide the crop with a hypothetical, nonlimiting
situation. The amount of irrigation water used and consequent yields obtained are overestimates, but this
approach permits us to compare the relative changes in each site and in each scenario. For the irrigation
simulation, the water demand was calculated assuming: 100% efficiency of the automatic irrigation system; a
1-meter irrigation management depth; automatic irrigation when the available soil water is 50% or less of
capacity; and the soil water for each layer is reinitialized to 100% capacity at the start of each growing season.
To account for the supplementary type of irrigation practiced in D.I. Khan and Khanpur, these sites
are best represented by irrigated simulations. In Jhelum, simulation of both irrigated and dryland production
is appropriate. Here, most wheat is grown under irrigation, but certain marginal areas are dryland. In Gilgit,
wheat production is marginal and basically dryland. For dryland multi-year simulations, it is desirable to
establish the initial soil-moisture levels at planting because soil moisture is a major factor in determining final
yields. Unfortunately, we had no data available on initial soil-moisture conditions, so we tested the sensitivity
of simulated wheat yields to initial conditions. In all areas of wheat production in Pakistan, the level of
fertilization are relatively low. In Jhelum and Khanpur the soils are deeper with higher fertility and better
management practices than at the other two sites.
Simulations. The CERES-Wheat model was run for 29 years of baseline climate and the GISS, GFDL,
and UKMO climate change scenarios under dryland and irrigated conditions. It was also run for the scenarios
created by superimposing arbitrary temperature and precipitation changes on the baseline climate (sensitivity
scenarios). Simulations for all scenarios included climate change alone and climate change with the
physiological effects of CO2 at all sites. The mean and standard deviation of the yield, evapotranspiration,
water applied for irrigation, and crop-maturity date are simulated in the study.
Limitations of the Study
Climate Change Scenarios. While GCMs are useful for climate change studies, current climate models
oversimplify many aspects of the climate system, especially ocean dynamics, cloud physics, and land-surface
hydrology. GCMs do not accurately simulate the current climate at regional scales, and they were not designed
for predictive regional studies. Therefore, the climate change scenarios created from GCM output must not
be viewed as predictions, but as examples of possible future climates for the regions under study. GCM output
analyzed by the U.S. Environmental Protection Agency indicated that the GCMs perform least well in the
Indus Valley and in four other regions of the world. Given the seriousness of these deficiencies, a detailed
examination of the GCM output was undertaken in Pakistan. Different gridboxes of the GISS, GFDL, and
UKMO GCMs were compared with historic temperatures and precipitation levels of sites in the region, and
recommendations were made for the use of climate scenarios in the area (Wescoat and Leichenko 1990). Mean
annual changes of each variable were used in this study, instead of the usual monthly changes. The justification
for the use of the climate change scenarios generated by GCMs is to have comparability among agricultural
production in different countries, and among river basins and climate change impacts projects.
As configured, the climate change scenarios do not alter the patterns of events in the base climate.
Therefore, they do not simulate changes in the underlying variability (e.g., extended periods of high
temperature, droughts, etc.) that can be vital for crops. Dryland yields may be considerably different depending
on whether a change in precipitation results from a change in mean, frequency, or intensity. However, the
scenarios created for this study do result in an increased frequency of temperatures above certain thresholds.
PAKISTAN-7
-------
Also, the weather generator used in this study does not simulate observed daily weather with complete
accuracy.
Crop Models. For this study, we assumed that nutrients are nonlimiting, pests are controlled, and that
there are no catastrophic weather events. These assumptions tend to overestimate the simulated yields. We
also assumed that technology and climatic tolerance of cultivars do not change under the conditions of climate
change, although this is not a realistic assumption (see discussion on adaptation). The physiological effects
of CO2 in the crop model may be overestimated because experimental results from controlled environments,
used to calibrate the model, may not be accurate under windy and pest-infected field conditions.
RESULTS
The specific objectives of this study were to measure the change in simulated crop yields under the
three GCM scenarios and to use the data as part of a world trade economic model.
Sensitivity Analysis
A sensitivity approach to climate change impact analysis is removed from the processes which
influence climate. However, it has the advantage of simulating a controlled experiment and therefore provides
a good understanding of the factors affecting responses; furthermore, it can identify temperature and
precipitation thresholds for crop production. We used combinations of temperature increases of 0°C, +2°C,
and +4°Cand precipitation changes of-20%, 0%, and +20% and superimposed them on the baseline climate.
For each scenario, the wheat model was run under current (330 ppm) and elevated (555 ppm) CO2 levels
(Table 4).
Table 4 shows that the decrease in yields associated with a 2°C temperature increase could be
compensated for when the direct effects of CO2 were taken into account (this result occurred at three of the
four sites). With a 4°C temperature increase, however, wheat yields were significantly lower (particularly at
Jhelum), even with the direct effects of CO2. These yield decreases were a consequence of a shorter season
length, and in some cases, water stress to the crop. A temperature of 2°C above the baseline implied that the
season length would be reduced by 8-14 days at all sites tested; a 4°C increase reduced it by about 20 days at
all sites.
An increase in the temperature also increased the moisture stress of the wheat crop (i.e., the
difference between precipitation and evapotranspiration). The simulated evapotranspiration (ET) decreased
in most cases, despite large increases in temperature and in potential evaporation. This was due to the shorter
growing season, which reduced the total amount of ET, and by decreased demand by the crop because it was
not growing as well. Irrigation demand generally decreased for the same reasons. Most of the sites' modeled
wheat yields were not very sensitive to precipitation changes (see section on continuous vs. noncontinuous
water balance).
Climate Change Scenarios
Currently, high temperatures and low precipitation limit present-day wheat production in Pakistan.
The sensitivity study showed that temperature increases of 2°C and higher resulted in a reduction of crop
yields throughout the country, especially in the main wheat-producing region of Punjab. In this section we
consider the possible impact of the climate change scenarios generated by the three GCMs on wheat
production in four sites distributed throughout Pakistan. The results show the degree to which the impacts
may vary within the regions. Table 5 shows the yield changes under doubled CO2 climate change scenarios for
all of the sites for both dryland and irrigated conditions with and without the direct effects of CO2. At Jhelum
PAKISTAN-8
_
-------
and Khanpur, the two major sites, irrigated and dryland results were very similar, showing major reductions-up
to 80%—in yields. Reductions in yields were projected with all three scenarios, but the reductions were more
significant with the UKMO scenario. Even when the direct effects of CO2 are included, reductions in yields
were still generally significant. At D.I. Khan and Gr/gzY, results were mixed, with some increases in yields.
Yield decreases were driven primarily by the increase in temperature, which causes the duration of
the crop-growth stages (particularly the grain-filling period) to be shortened. The season length was greatly
reduced by about four weeks under the three scenarios (Table 6). Tables 7 and 8 show that simulated ET and
irrigation demand generally decreased under the climate change scenarios.
Continuous vs. Noncontinuous Soil Water Balance
The assumption that the soil-moisture profile was full at the beginning of each growing season is one
that may be acceptable under moist conditions, but it is unrealistic in dry or marginal areas, leading to an
overestimation of the available soil water. To determine how sensitive the yield results are to soil-water
conditions, a series of tests was run using a continuous water balance in addition to the normal method of
resetting the soil moisture to full every growing season. Under the continuous water balance method, the soil
moisture is never reset and is allowed to continue, as computed, from one year to the next. These experiments
were carried out at Jhelum and D.I. Khan for the 29 years under both the current climate and the GISS
climate change scenario.
At Jhelum, the continuous water balance made very little difference on crop yields. This is because
Jhelum receives copious amounts of monsoon rainfall just prior to planting. At D.I. Khan, where precipitation
amounts are lower, the continuous water balance resulted in decreased yields under current climate conditions.
However, the GISS scenario yields also decreased, so the relative changes from base yields remained similar
(+16.3% for the normal reset run; +21.6% for the continuous soil-moisture run). Only relative yield changes
were contributed to the world food trade study.
Changes in Management under Climate Change
This study determined the possible causes of losses in wheat yields under different climatic conditions
(GCM climate change scenarios and sensitivity analysis) a$ the first step in the analysis of adaptation to
climate change. Adaptation strategies included changes in the management practices of the wheat cultivars
currently used (planting date and irrigation) and changes in the cultivar. All adaptation strategies were
simulated in Jhelum under UKMO, the most unfavorable climate change scenario (Table 9). The results of
the simulation show that there was no significant benefit from full irrigation at this site under the UKMO
scenario (a -61% yield decrease in dryland conditions and a -56% yield decrease under full irrigation). This
result was not very surprising, since Jhelum was found to be insensitive to changes in precipitation.
A likely response of farmers to warmer temperatures would be to shift the planting dates to avoid high
temperatures during the grain-filling period and increase the crop-growing season. In Jhelum, a simulated shift
in the planting date to 10 days earlier (from Oct. 20 to Oct. 10) increased the yield losses under the UKMO
scenario (Table 9), probably because the new planting date falls in a very hot period. Nevertheless, a delay in
the planting date to Nov. 1 or Nov. 15 was beneficial for wheat yields in Jhelum (a -61% yield loss under
present management conditions and a -30% yield loss if the planting is postponed to Nov. 15).
Two cultivars were tested as possible alternatives to Mexipak: Sonalika (widely used in India) and
Anza. Sonalika, planted on Nov. 15, resulted in larger yield losses than Mexipak planted on Nov. 15 (a -42%
yield decrease with Sonalika and a -30% with Mexipak). Anza, planted on Nov. 15, was the most favorable
strategy tested. In this case, yields decreased under the UKMO scenario by -20%.
PAKISTAN-9
-------
DISCUSSION
Conclusions. Projected climate change caused simulated wheat yields to decrease dramatically both in
dryland and irrigated production, due to a shorter season length caused by temperature increases. Yield
decreases were somewhat counteracted by the physiological effects of CO2, as simulated in this study, but the
effect was not enough to offset yield decreases under the climate scenarios. The overwhelming impact of
temperature increases, especially when coupled with precipitation decreases, would reduce the time required
for crop maturation and thus reduce wheat yields. If global climate change brings warmer and drier conditions
to Pakistan, additional problems would arise for water supplies for irrigation in the region.
In comparison to simulated climate change agricultural impacts in other hot and dry regions, the
Pakistan study found more severe yield decreases in response to the same climate change scenarios tested
elsewhere (Adams et aL 1990; Smith and Tirpak 1989; Rosenzweig 1990). It is possible that Pakistan may be
one of the regions more severely affected by climate change, and national wheat production may decrease
substantially in the future if climate change occurs as predicted by GCMs. This study suggests that the
development of new cultivars that are more heat-resistant than the present ones is essential to improve the
resiliency of Pakistani wheat production to possible increases in temperature.
The adaptation strategies tested in Jhelum under the UKMO scenario did not compensate fully for
the yield reductions, but a shift in cultivar combined with a later planting date decreased yield losses under
irrigated conditions. With the known pool of cultivars and current resources (water and fertilizer), full
adaptation seems unlikely in Pakistan since the temperature changes suggested by some GCM scenarios would
exceed the temperature the wheat plant can tolerate physiologically.
Further research needs. Possible future adaptations need to be analyzed so that farmers will be able
to adapt their practices in response to altered climate conditions. These studies should include detailed
changes in planting date, fertilizer, and irrigation. Studies should also focus on the effect of the altered
calendar crop on the agricultural system in the region, particularly where more than one crop is harvested each
year.
The possible increase in climatic variability under future climate change scenarios may be an important
factor in determining production losses. Since crop yields exhibit a nonlinear response to heat and cold stress,
any change in the probability of extreme temperature events can be significant. Therefore, a variability study
on possible future yields is important. Because insect pests and diseases are also an important component of
production losses under the present climate, any possible change in pest and disease patterns under climate
change conditions should be analyzed.
Finally, it is important to analyze the potential extension of, or shifts, in the areas of crop production,
the soil constraints in these potential agricultural regions, economic tendencies, and other environmental
factors. Based on the results of the present study, the effects of potential climate change in Pakistan could have
dramatic consequences for agricultural production and it is important to analyze the impact in a larger number
of sites and for other major crops, such as rice. Finally, further research should be done on simulating the
annual water balance more accurately in order to estimate future water availability for irrigation.
PAKISTAN-10
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Adams, R.M., C. Rosenzweig, R.M. Peart, J.T. Ritchie, B.A McCarl, J.D. Glyer, R.B. Curry, J.W. Jones, KJ.
Boote, and L.H. Allen, Jr. 1990. Global climate change and U.S. agriculture. Nature 345(6272):219-
224.
Gadgil, S., AK.S. Huda, N.S. Jodha, R.P. Singh, and S.M. Virmani. 1988. The effects of climatic variations on
agriculture in dry tropical regions of India. M.L. Parry, T.R. Carter and N.T. Ronijn (eds.) The Impact
of Climatic Variations on Agriculture. Chapter 5. pp. 497-576.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient three-
dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-622.
IBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project. 1989. Decision
Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil
Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums
(eds). Intergovernmental Panel on Climate Change. Cambridge.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Myers, N. 1989. Environmental Security: The Case of South Asia. International Environmental Affairs, 1:138-
154.
Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, and L.H. Allen, Jr. 1989. Impact of climate change on crop
yield in the southeastern U.S.A. In J.B. Smith, and D. Tirpak (eds.). The Potential Effects of Global
Climate Change on the United States. Report to Congress. EPA-230-05-89-050. U.S. Environmental
Protection Agency. Washington, DC. Appendix C-l, pp. 2-1 to 2-54.
Qureshi, A 1989. Mitigating climate change: strategies to finance retention of tropical forests. In Coping with
Climate Change-Proceedings of the Second North American Conference on Preparing for Climate
Change: A Cooperative Approach. J.C. Topping, Jr., ed. The Climate Institute, Washington, D.C.
Richardson, C.W., and D.A Wright. 1984. WGEN: A Model for Generating Daity Weather Variables. ARS-8.
US Dept. of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CEKES-Wheat: A user-oriented wheat yield
model. In W.O. Willis (ed..). ARS Wheat Yield Project. ARS-38. U.S. Department of Agriculture,
Agricultural Research Service. Washington, DC. pp. 159-175.
PAKISTAN-11
-------
Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study.
Professional Geographer 42(l):20-37.
Smith, J.B., and D. Tirpak (eds.). 1989. The Potential Effects of Global Climate Change on the United States.
Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency. Washington, DC.
423pp.
Wescoat, J.L., and R. Leichenko. 1990. Climate change scenarios for the Indus basin, Pakistan: Modification
of Scenario Research Design. Report to "Complex River Basin Project Participants".
Wescoat, J.L., and R. Leichenko. 1992. Complex River Basin Management in a Changing Global Climate: The
Indus Basin in Pakistan, A National Assessment. Collaborative Paper No. 5. Center for Advanced
Decision Support for Water and Environmental Systems and the Institute of Behavioral Science. The
University of Colorado. 143 pp.
Wilson, C.A., and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
PAKISTAN-12
-------
Table 1.
Wheat production (1981-89).
Region
Site
Production
Area
Gilgit
Yield
Punjab
Sind
NWFP
Jhelum
Khanpur
D.I. Khan
txlOOO
9,345
2,189
939
ha x 1000
5,404
1,032
791
tha-1
1.73
2.12
1.19
Baluchistan
445
252
1.79
Pakistan
12,918
7,478
1.73
PAKISTAN-13
-------
Table 2. Average annual changes in temperature, precipitation, and solar radiation in the GCM
climate change scenarios.
Temp, changes (°C)
Precip. changes (%)
SR changes (%)
Site
Jhelum
Khanpur
D.LKhan
Gilgit
GISS
4.7
3.2
4.7
4.7
GFDL
4.5
4.5
4.9
4.8
UKM
O
7.0
6.0
7.0
5.7
GISS
13
25
13
13
GFDL
16
19
11
10
UKM
O
-18
40
-18
13
GISS
-1
-2
-1
-1
GFDL
8
0
8
17
UKM
0
5
-1
5
20
PAKISTAN-14
-------
Table 3 Test sites, soils, and management variables.
Site
Jhelum
Khanpur
Gilgit
D.LKhan
Location
+32.97 N
+73.75 E
+28.63 N
+70.67 E
+35.90 N
+74.33 E
+31.82 N
+70.93 E
Representative
Soil
medium
sandy loam
deep
sandy loam
medium
sand
medium
sandy loam
Cultivar
mexipak
mexipak
mexipak
mexipak
Practice
dryland/
irrigation
irrigation
mainly
dryland
irrigation
Planting
date
Oct. 20
Nov. 12
Oct. 30
Nov. 30
Plant population for irrigation practice = 250 plants m":
Plant population for dryland practice = 125 plants m'2
PAKISTAN-15
-------
Table 4.
Sensitivity of irrigated CERES-Wheat yields to temperature and CO2 changes.
Simulated Yields (t ha"1)
CO2 level
(ppm)
330
555
Temp.
Change (°C)
0
2
4
0
2
4
Jhelum
5.28
3.55
2.02
6.60
4.89
3.22
Khanpur
4.79
4.19
3.23
5.80
5.28
4.29
D.I. Khan
3.73
3.50
3.16
4.41
4.19
3.86
Gilgit
4.54
3.97
3.94
5.38
4.85
4.42
PAKISTAN-16
-------
Table 5.
Effects of climate change on wheat yields.
Site
Rainfed/ Base
Irrig. Yield
Percent changes in simulated yields from base
GCM Scenario Alone GCM Scenario + DE
GISS GFDL UKMO GISS GFDL UKMO
Jhelum
Khanpur
D.LKhan
Gilgit
Jhelum
Khanpur
D.LKhan
Gilgit
rainfed
rainfed
rainfed
rainfed
irrigated
irrigated
irrigated
irrigated
tha'1
4.81
4.19
1.94
0.6
5.28
4.79
3.73
4.75
-67
-45
16
20
-60
-43
-20
-17
-29
-54
-55
8
-24
-54
33
-5
-80
-61
-11
-47
-74
-57
-17
-12
-41
-15
61
67
-36
-19
-2
-6
-5
-33
-45
50
-5
-37
60
9
-61
-35
29
-20
-56
-37
6
5
DE = direct effects of CO2 on wheat yield included.
PAKISTAN-17
-------
Table 6.
Effects of climate change on the season length of wheat.
Simulated season lengths (days)
Site
Jhelum
Khanpur
D.I. Khan
Gilgit
Base (days)
128
122
122
150
GISS
100
99
104
132
GFDL
105
96
98
131
UKMO
94
93
97
133
PAKISTAN-18
-------
Table 7.
Effects of climate change on wheat evapotranspiration (ET).
Site
Percent changes in simulated ET from base
GCM Scenario Alone GCM Scenario +DE
Rainfed/ Base ET
Irrig. (mm)
GISS GFDL UKMO
GISS GFDL UKMO
Jhelum
Khanpur
D.LKhan
Gilgit
Jhelum
Khanpur
D.I. Khan
Gilgit
rainfed
rainfed
rainfed
rainfed
irrigated
irrigated
irrigated
irrigated
207
226
226
166
232
267
331
474
-39
-41
-9
-5
-31
-31
-22
-17
1
-36
-10
-7
25
-29
21
-9
-50
-47
-20
-5
-42
-36
-19
13
-34
-35
-11
-5
-31
-32
-27
-23
2
-33
-10
-7
18
-31
14
-17
-44
-40
-19
-4
-38
-34
-24
3
DE = direct effects of CO2 on wheat ET included.
PAKISTAN-19
-------
Table 8.
Effects of climate change on amounts of irrigation.
Site
Base
Percent changes in simulated irrig. amounts from
base
GCM Scenario Alone GCM Scenario +DE
GISS GFDL UKMO GISS GFDL UKMO
Jhelum
Khanpur
D.LKhan
Gilgit
(mm)
128
199
215
337
-45
-31
-28
-8
57
-32
-43
4
-56
-36
-16
33
-43
-33
-33
-16
43
-33
-33
-7
-47
-34
-20
20
DE = direct effects of CO2 on wheat irrig. water demand included.
PAKISTAN-20
-------
Table 9. Effect of changes in wheat cultivar and management on simulated yield under the UKMO
climate change scenario in Jhelum. The physiological effects of CO2 were included in all
UKMO scenario simulations.
Scenario
BASE*
BASE*
UKMO**
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
Cultivar
Mexipak
Mexipac
Mexipak
Mexipak
Mexipak
Mexipak
Mexipak
Sonalika
Anza
Management
rainfed
irrigation
rainfed
irrigation
irrigation
irrigation
irrigation
irrigation
irrigation
Planting Date
Oct. 20
Oct. 20
Oct. 20
Oct. 20
Oct. 10
Nov. 1
Nov. 15
Nov. 15
Nov. 15
Yield
T/Ha
4.81
5.28
1.88
2.35
0.64
2.56
3.71
3.08
4.25
% Change
from Base
-61
-56
-88
-52
-30
-42
-20
*Base conditions using current climate as scenario.
**Simulations under UKMO scenario without adaptation.
PAKISTAN-21
-------
-------
Location of Test Sites for Pakistan
NORTH-WEST
FRONTIER
BALUCHISTAN
PUNJAB
SIND
Arabian Sea
Figure 1. Map of Pakistan with sites selected for this study. The denominations used and the
boundaries shown do not imply any judgement on the legal status of any territory or any
endorsement or acceptance of such boundaries.
-------
CC
CL
o
10
CM
o o
o 10 _
CM T- —
§
0.
o
UI
en
o.
f
LU
IT
111
O IO O IO O IO O
CD CM CM »- 1-
o.
o
I
o
E
3
C3
E
O
o.
"8
H
§
1
Bk
o o o o
10 o 10 o
CM CM T- •*-
O
IO
o o o o
10 O 10 O
CM CM *- 1-
Q.
O
_i
UJ
o
ui
§
E
a.
o
s
a
o
CL
<
s
a.
O
a
i
f
-------
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT
PRODUCTION IN INDIA
D.Gangadhar Rao
GRID A, Hyderabad, India
S.K.Sinha
IARI, New Delhi, India
INDIA-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
METHODOLOGY
Baseline Climate Data
GCM Climate Change Scenarios
The CERES-Wheat Model, Crop Inputs and Soil Data
Model Validation
Simulations
RESULTS AND DISCUSSION
Sensitivity Analysis of the CERES-Wheat model
Wheat Responses to Climate Change at Delhi
Wheat Responses to Climate Changes at Hyderabad
Adaptation Strategies
Implications of Climate Change for Wheat Production in India
REFERENCES
INDIA-2
-------
SUMMARY
This study assessed the impact of climate change on wheat production at two contrasting locations
in India, Delhi and Hyderabad, using the CERES-Wheat simulation model under rainfed and irrigated
conditions. Climate change scenarios for each location were created from three equilibrium General
Circulation Models (GISS, GFDL, and UKMO) and from the transient GISS model (for the 2010s, 2030s, and
2050s). The simulation study considered the physiological effects of increased CO2 levels on the crop in all
scenarios. In all climate change simulations, wheat yields were smaller than those in the current climate, even
with the direct beneficial effects of CO2 on crop yield considered. Yield reductions are primarily due to a
shortening of the wheat-growing season, resulting from the scenario temperature increases.
INTRODUCTION
Agriculture remains an important economic sector in India, despite the development of industry and
services in the post-independence nation. Seventy percent of the population is rural and depends directly or
indirectly on agriculture. Agricultural failures due to natural disasters have been an important factor in the
migration of the population from rural to urban areas. Today, agriculture represents 35% of India's Gross
National Product (GNP). This percentage is likely to decline due to further industrialization, but agriculture
will continue to occupy an important place in the Indian economy. Therefore, global climate change that could
influence agriculture in the Indian subcontinent should be a cause for concern (IPCC 1990).
The Indian agricultural system includes the production of crops, animal products, fisheries, and forest
products. It is impossible to present a holistic picture of the effect of climate change on the agricultural system
because of the many interdependent and interactive factors. Therefore, for the purpose of the present analysis,
we considered the potential effects of climate change on the primary sector of the agricultural system-crops.
Because of the different agroclimatic environments in India, it is possible to grow almost all of the
important cereal, leguminous, oilseed, horticultural, and plantation crops, and it would be interesting to
simulate their potential production under climate change. A simulation study, however, requires reliable input
data and a reasonable model. Among the available crop models, CERES-Wheat has been widely used over
different environments, and therefore is probably suitable for the agroclimatic conditions in India. Wheat is
important in India because it constitutes the basis of its food security system. Although rice simulation models
are also available, ecological niches for rice production in India are highly variable and would require a more
elaborate study. Therefore, the present study was confined to wheat as a test system.
There is a need to have a long-term weather record and high-quality experimental data for validating
any model. The two locations selected for this study (Delhi and Hyderabad), representing contrasting
agroclimates, may be a good starting point (Figure 1). Delhi represents the highly productive regions of
Haryana, Punjab, and Uttar Pradesh (contributing approximately 70% of the national production), and
Hyderabad represents the less productive regions of Andhra Pradesh, Karnataka, and Maharashtra
(contributing approximately 2.6% of the national production) (Table 1).
METHODOLOGY
Baseline Climate Data
For the simulation study daily minimum and maximum temperature, solar radiation, and rainfall data
were used for twenty-eight years (1953-80) for Delhi (28°N;78°E) and for seventeen years (1964-80) for
Hyderabad (17 °N; 78 °E). The monthly averages for these parameters are given in Figure 2.
INDIA-3
-------
Solar radiation was unavailable, but was computed using sunshine hours calculated from the following
formula:
Rs = Ra (a + b * n / N) where
Rs = Solar Radiation in Megajoules m"2
Ra = Extraterrestrial Radiation
n/N = % Sunshine Hours
a,b = Location-specific Constants
The constants a and b were established by the India Meteorology Department.
GCM Climate Change Scenarios
Climate change scenarios for Delhi and Hyderabad were created by combining daily observed climate
data with the monthly output of three General Circulation Models: the Goddard Institute for Space Studies
(GISS) (Hansen et al. 1983), the Geophysics Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald
1987), and the United Kingdom Meteorological Office (UKMO) (Wilson and Mitchell 1987) (Figure 2).
Similarly, transient climate change scenarios were created for the two locations for the decades of 2010s, 2030s,
and 2050s with monthly output from the GISS transient run A (Hansen et al. 1988) (Figure 2).
The CERES-Wheat Model, Crop Inputs and Soils
Growth and development of wheat were simulated using the CERES-Wheat model (Ritchie and Otter
1985). A common wheat genotype (Sonalika) was chosen to represent the most common cultivars used at both
Delhi and Hyderabad. In all the simulations the crop was sown on November 1, and a plant population of 200
plants per square meter was maintained. The irrigated simulations considered three irrigation treatments of
40 mm at the time of sowing, 30 days after sowing (vegetative stage), and 60 days after sowing (reproductive
stage).
The soil characteristics used for the simulations are listed in Appendix A. For Delhi, the soils include
deep, well-drained loamy soils that exhibit a regular decline in the organic carbon with depth, indicating no
disturbance in the profile development. The soil characteristics used for Hyderabad included deep, mildly
alkaline, moderately well-drained clayey soils.
Model Validation
The CERES-Wheat model has been validated using datasets from various wheat-growing areas of the
world (Otter and Ritchie 1985). For the sites tested in India, the simulated phenology did not always replicate
the observed. The average yield of the variety "Kalyansona" is about 4.3 t ha"1, with adequate irrigation at New
Delhi (Sinha and Swaminathan 1991); with the same management conditions, the averaged simulated yield was
5.2 t ha"1. At Hyderabad, the average wheat yield is 2.6 t ha"1 and the average simulated yield was 2.2 t ha"1.
Simulations
The simulations were conducted using the CERES-Wheat model with DSSAT (Decision Support
System for Agrotechnology Transfer) (IBSNAT1989). Both rainfed and irrigated simulations were performed
under all of the climate scenarios described above. The physiological effects of CO2 on the crop were
INDIA-4
-------
considered in each simulation. In addition a sensitivity analysis was performed with arbitrary changes in
precipitation (+20% and -20%) and in temperature (+2°C and +4°C) from the baseline climate.
The CO2 levels and the photosynthetic factors used for the climate change scenarios are:
Climate Change Scenario CO2 Level Photosynthetic Factor
GISS, GFDL, UKMO 555 1.17
2010 405 1.05
2030 460 1.10
2050 530 1.15
RESULTS AND DISCUSSION
Sensitivity Analysis
This section presents the sensitivity of the CERES-Wheat model to changes in precipitation (+20%
and -20%) and temperature (+2°C and +4°C). Figure 3 shows that temperature is the main factor
responsible for yield changes. Both grain yields and biomass decline with increasing temperatures, although
there is a small yield increase when the rainfall is increased by 20%. Field experiments (at the Indian
Agricultural Research Institute and other sites) indicate that wheat yields are very sensitive to elevated
temperatures, particularly during the grain-filling period. The simulation analysis shows similar results.
Wheat Responses to Climate Change at Delhi
At Delhi, the GISS and GFDL climate change scenarios project an increase in the amount of rainfall
from June to October (Figure 2). The air temperature increases under all GCM scenarios throughout the year
and particularly during the post-rainy season. No appreciable change in the solar radiation is observed under
climate change conditions. The GISS transient scenarios show gradual increases in monthly temperature of
at least 5°C by 2050 and sharp increases in rainfall during July (Figure 2). Solar radiation increases slightly
in January and February during the 2030s and 2050s.
Under the GCM equilibrium scenarios alone (not considering the direct CO2 effects), simulated wheat
grain yields were largely reduced in comparison to the base values. Under the transient scenarios, yield
reductions were smaller (Figure 4). When the physiological effects of CO2 are included in the yield simulations,
they partially offset the yield decreases (Figure 4). In fact, under the GISS scenario and the 2010 transient
climate, there was a marginal increase in the simulated grain yields under rainfed conditions. Nevertheless,
under irrigated conditions, yield decreases from the base were apparent even when the direct CO2 effects were
included. The transient scenarios predicted a greater reduction in the response of the crop to irrigation as we
pass from the 2010s to the 2050s. Simulated biomass followed a similar trend, but in this case the marginal
increase in the dryland simulations under the GISS scenario with CO2 effects were not observed (Figure 4).
Temperature increases under the scenarios decreased the length of the growing period and total
evapotranspiration in the crop model simulations (Figure 4). The length of the growing season was not
affected by the physiological effects of CO2. In general, the climate change scenarios predict a greater percent
reduction in evapotranspiration under irrigated conditions.
INDIA-5
-------
Wheat Responses to Climate Changes at Hyderabad
All GCMs show an increase in the average monthly air temperature at Hyderabad. The UKMO model
projects the largest increases-up to 5°C from January to March (Figure 2). The rainfall increases during the
rainy season (May to November) under all scenarios. The solar radiation shows considerable reduction in the
GFDL scenario from June to December (Figure 2) because this scenario predicts higher precipitation during
those months. The transient scenarios show an increase in the mean average temperatures, which reach an
average of 35 °C (Figure 2) during August. Rainfall shows no particular trend and solar radiation shows small
increases from June to December in 2050.
As in the case of Delhi, simulated wheat yields and season length decreased under all GCM scenarios
of climate change (Figure 5). The largest reductions were under the GISS and UKMO scenarios. The transient
scenarios projected a 20% decrease in the 2010s and a 60% decrease in the 2050s. The same trend was
observed with respect to biomass production, except under the GFDL scenario. As in the case of Delhi, the
evapotranspiration was lower.
Adaptation Strategies
Possible strategies to increase wheat yields under climate change conditions are: increasing irrigation,
using more fertilizer, or increasing the plant population. These strategies were tested at Delhi under the
equilibrium GCM scenarios, but only slightly reduced the negative impacts of climate change on simulated
yield (data not shown).
To further evaluate possible adaptive strategies to climate change, two approaches could be tried:
(a) Evaluate the impacts of the climate change scenarios on other Indian wheat cultivars. However, the
tremendous variability in germoplasm requires accurate estimates of the genetic coefficients that describe each
cultivar in the CERES-Wheat model, and these coefficients are not readily available for many of the Indian
genotypes.
(b) Design hypothetical new cultivars that will adapt to climate change conditions, using the CERES-Wheat
model as a tool. Possible breeding objectives could be established by analyzing the genetic coefficients in order
to define a theoretical cultivar that is adapted to climate change.
Implications of Climate Change for Wheat Production in India
The results presented in this report suggest that wheat yields would decrease under equilibrium and
transient climate change conditions at Delhi and Hyderabad. These yield decreases could have a serious
impact on food security, especially in view of the increasing population and its demand of food grains. Since
most of the wheat production comes from the northern plains, where it is almost impossible to increase the
present area of wheat under irrigation, climate change could lead to shortages in food supplies. Furthermore,
the simulated changes in wheat yields could be indicative of the possible impacts of climate change on other
similar crops grown during the same season.
study:
Before considering the above scenario, it is important to evaluate some important limitations of this
(a) Poor simulation of the present climate by the GCMs in this region.
(b) Poor validation of the CERES-Wheat model for some of the Indian wheat cultivars. For example,
the cultivar "Kalyansona", under irrigation at Delhi, has an average growing period of 150-160 days,
INDIA-6
-------
but the simulated growing period was always under 116 days in the present study. In addition, the yield
components are not simulated adequately.
Because of these uncertainties, it would be premature to draw final conclusions on the impact of
climate change on India's food security system. However, the results of this simulation study suggest possible
negative consequences of climate change for the crops in India, and thus, further research needs to be
encouraged through both national and international efforts. It is essential to conduct further research using
the wheat model, and efforts are needed to develop models for crops such as chickpeas, rapeseed, mustard,
and barley. This would require extensive field experiments to generate crop phenology data.
INDIA-7
-------
REFERENCES
Hansen, J., G. Russel, D. Rind, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient Three
Dimensional Global Models for Climate Studies: Models I and II. Monthly Review, Vol III, No. 4:609-
662.
Hansen, J., L Fung, D. Rind, S. Lebedeff, R. Ruedy, and G. Russel. 1988. Global Climate Changes as
Forecasted by Goddard Institute For Space Studies Three Dimensional Model. Journal of Geophysical
Research, 93:9341-9364.
BBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project (IBSNAT). 1989.
Decision Support System for Agrotechnology Transfer Version 2.1 (DSSAT V 2.1). Dept. Agronomy
and Soil Science., College of Trop. Agr. and Human Resources, University of Hawaii, Honolulu,
Hawaii 96822.
IPCC. 1990. Intergovernmental Panel on Climate Change. First Assessment Report.
Manabe, S., and R.T. Wetherald. 1987. Large-Scale Changes in Soil Wetness Induced by an Increase in Carbon
Dioxide. Journal of Atmospheric Sciences, 44:1211-1235.
Otter, S., and J.T. Ritchie. 1985. Validation of the CERES-Wheat Model in Diverse Environments. In: Wheat
Growth and Modeling, eds. W. Day and R.K. Atkins. Plenum Press. New York, pp. 307-310.
Ritchie, J.T., and S. Otter. 1985. Description and Performance of CERES-Wheat: A User-Oriented Wheat Yield
Model In: W.O. Wills, ed. ARS Wheat Yield Project. Washington, DC: U.S. Dept. of Agriculture,
Agricultural Research Service. ARS-38.pp. 159-175.
Sinha, S.K., and M.S. Swaminathan. 1991. Deforestation, Climate Change and Sustainable Nutrition Security:
A Case Study of India. Climate Change, 19:201-209.
Wilson, CA, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
INDIA-8
-------
Table 1.
Area and production of wheat in various states of India.
State
Andhra Pradesh
Arunachal Pradesh
Assam
Bihar
Gujarat
Haryana
Himachal Pradesh
Jammu&Kashmir
Karnataka
Madhya Pradesh
Maharastra
Meghalaya
Orissa
Punjab
Rajastan
Sikkim
Tamilnadu
Tripura
Uttar Pradesh
West Bengal
Delhi, Dadra &
Nagar Haveli
TOTALS
86-87
Area
10.60
2.10
120.00
1839.70
315.10
1782.00
377.10
222.40
306.50
3251.30
735.50
4.80
46.80
3189.00
1843.20
10.20
0.40
2.70
8311.50
397.70
41.20
22820.30
Prodn.
6.90
6.90
125.80
2863.10
661.70
5055.00
492.00
272.10
148.80
3865.00
536.40
6.70
83.30
9458.00
3401.60
16.50
0.20
4.30
16078.50
682.70
119.60
45576.50
87-88
Area
11.80
2.10
98.30
1934.90
192.40
1731.00
374.90
238.40
262.40
3549.10
732.70
4.80
45.70
3126.00
1533.80
11.20
0.10
3.20
8339.80
374.20
37.30
22604.20
Prodn.
4.20
6.90
105.80
2776.60
351.20
,4861.00
351.20
212.00
133.90
4328.80
633.40
6.70
80.90
11066.00
2909.80
17.70
0.10
2.70
16462.90
673.90
107.20
45095.50
Area (x 1000 ha)
Prodn. (million t)
INDIA-9
-------
Appendix A Soil physical properties used for CERES-Wheat simulations at Delhi and Hyderabad.
SOIL AT
DLAYR
10.
15.
15.
15.
30.
30.
30.
30.
30.
-1.
SOIL AT
DLAYR
10.
15.
15.
15.
15.
30.
30.
-1.
DLAYR
LL
DUL
SAT
SW
DB
DELHI
LL
.156
.166
.167
.157
.148
.110
.111
.112
.112
.00
DUL SAT SW
.262 .362 .262
.262 .362 .262
.262 .362 .262
.262 .362 .262
.261 .361 .261
.260 .360 .260
.259 .359 .259
.258 .358 .258
.258 .358 .258
.00 .00 .00
BD
1.37
1.37
1.37
1.37
1.38
1.38
1.39
1.39
1.39
.00
HYDERABAD
LL
.313
.313
.314
.316
.316
.318
.320
.00
DUL SAT SW
.480 .560 .480
.479 .559 .479
.479 .559 .479
.477 .557 .477
.477 .557 .477
.476 .556 .476
.474 .554 .474
.00 .00 .00
BD
1.35
1.36
1.36
1.37
1.37
1.37
1.38
.00
= Thickness of the soil layer
= Lower limit plant extractable
water
= Drained upper limit soil water content
= Saturated water content
= Defaulted soil water content
= Moist bulk density of soil
INDIA-10
-------
INDIA
Figure 1. Map of India and location of the sites selected for the study.
-------
o
a
u.
o
t
CO
CO
5
t
IU
5
m
o o o o o-
o o o o
X CT CM t-
DO
(E
u>
CM
O O O O
O IO O IO
CM T- T-
-------
o
10
o
CM
O
CO
o
CM
O
en
\
111
CO
CO
CO
S
-------
Grain Yield (t/ha)
T*2
T»4
Biomass (t/ha)
T*2
Avg.Rain «330 ppm
»20% Rain »555 ppm
T*4
Avg.Rain »555 ppm f&k-*i «20% Rain *330 ppm
-20% Rain *330 ppm d) -20% Rain +655 ppm
Figure 3.
Sensitivity analysis of CERES-Wheat to temperature, precipitation, and CO2 at Delhi.
-------
40
(% change over ttae value)
CLIMATE CHANGE ALONE
G/SS GFDL UKMO 2010 2030 2050
S
40
20
(% change over base value)
CLIMATE CHANGE + P.E. OF CO2
-40
-60
-80
40
20
G/SS GFDL UKMO 2010 2030 2050
* change over baa. va,u.) CLIMATE CHANGE + P.E. OF CO2
ce
S,
2
M -
-40
-eo
6/SS GFDL UKMO 2010 2030 2050
Painted
Irrigated
Figure 4a. Effect of climate change on simulated yield and biomass at Delhi.
-------
(% ch*ng» over b»»» vmlut)
% SEASON LENGTH
+v
20
0
-20
-40
-fin
Ti 1 1 "" u I
i i i i i i
G/SS GFDZ. i/KMO 2070 2030 2050
40
20
-20
-40
(% chango ovtr t>*t» valut) CLIMATED Cifl&NGE + P.E. OF CO
i
-60 ' «-
G/SS GFDL UKMO 2O10 2030 2050
RAINFED ^^ IRRIGATED
Figure 4b. Effect of climate change on season length and total crop evapotranspiration
(ET) at Delhi. F
-------
40
20
-20
-
-60
-80
-100
(% change over baa* value)
CLIMATE CHANGE ALONE
GISS GFDL. UKMO 2010 2030 2050
40
20
0
-20
-40
-60
-80
(% ching* over i>«« value) CLIMATE CHANGE + P.E. OF CO2
1
GISS GFDL UKMO 2010 2030 2050
40
20
(K change over b»»» value)
-20
-40
ii
GISS GFDL UKMO 2010 2030 2050
RAINFED
i IRRIGATED
Figure 5. The effect of climate change on simulated yield and season length at
Hyderabad.
-------
-------
IMPACT OF CLIMATE CHANGE ON THE PRODUCTION
OF MODERN RICE IN BANGLADESH
Z. Karim, M. Ahmed, S.G. Hussain, and Kh.B. Rashid
Bangladesh Agricultural Research Council
Dhaka, Bangladesh
BANGLADESH-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Background
Climate Vulnerabilities in Bangladesh
Aims and Design of the Study
METHODS
Climate Data and Climate Change Scenarios
Crop Models, Cultivars and Management Variables
RESULTS
Sensitivity of Rice Yields to Temperature Increases
Rice Yields under GCM Climate Change Scenarios
IMPLICATIONS OF THE'RESULTS
REFERENCES
BANGLADESH-2
-------
SUMMARY
Bangladesh is located in a region that is vulnerable to environmental hazards, frequent floods, droughts,
cyclones, and storm surges that damage life, property, and agricultural production. This study uses climate
models combined with crop simulation models to determine the possible effects of climate change on rice
production in major agricultural regions of the country.
Sensitivity simulations showed that rice yields decreased significantly with temperature increases in the
two sites considered. The rice yields under the GCM climate scenarios alone decreased at both sites. When the
physiological CO2 effects were considered, the yield decreases under the climate change scenarios were offset.
If the physiological CO2 effects are not as positive as simulated in this study, rice production in Bangladesh
could be damaged under climate change conditions. A decrease in rice production, combined with the rapidly
increasing population, would threaten the country's food security.
INTRODUCTION
Background
Bangladesh occupies an area of about 55,598 square miles near the Tropic of Cancer between latitudes
20.25°N-26.38°N and longitudes 88.01°E-92.40°E (Figure 1). For its relatively small size, Bangladesh
encompasses a wide range of environmental conditions. Land, soil, hydrology, and climatic variability occur at
the national level and also in small areas of the country. The country is divided into 30 agroecological regions
(FAO 1988) according to physiography, characteristics and length of seasonal flooding, length of the cool
winter period, and frequency of extremely high (>40°C) summer temperatures.
A variety of crops are grown in the country (Statistical Yearbook 1990), but local and modern varieties
of rice represent the dominant crop. Rice is the staple food in the country, currently grown on about 10.3
million hectares of land with production of about 18 million tons. Other important crops are jute and wheat.
Wheat is becoming increasingly significant especially in the northern regions. This study considers the possible
impacts of climate change on modern rice production in two contrasting locations in Bangladesh (Figure 1)
that represent major rice-growing regions.
Climate Vulnerabilities in Bangladesh
Monsoon rainfall occurs everywhere in the country, although the amount and intensity of the rain varies
greatly. About 90% of the precipitation occurs during four months (June-September), but the distribution
within these months differs at different sites (Figure 2)., Annual rainfall ranges from 1,400 mm in the dry
Rajshahi (Northwest) region to more than 5,000 mm in the wet Sylhet (Northeast) region. Regional cropping
patterns are primarily determined by the seasonal monsoon flooding regime. There are three seasons with
different hydrological regimes: the wet season from June to November (called "kharif"), the dry season (called
"rabi"), and a transitional season (called "prekharif"). The winter temperatures are moderate, but the summer
temperatures are extremely .high and may exceed 40°C.
The country is highly vulnerable to climatic events; floods and droughts are very frequent and damaging.
The recent severe floods of 1987 and 1988 produced colossal damages to crops and property. In 1988, about
2.5 million tons of food grain were damaged by floods. Due to the uncertain and uneven distribution of rainfall,
droughts of different intensities may occur and cause large crop losses! During the wet season, the crop most
affected by drought is rice. In the areas that are most severely affected (about 2.3 million ha), yield losses can
BANGLADESH-3
-------
represent a 45% decrease from the achievable yield (Karim et al. 1990; Karim and Akhand 1982). During the
dry season, wheat, potato, and mustard are most affected by drought. Possible climate changes (IPCC 1990)
may bring more unfavorable conditions for agriculture in Bangladesh, as has been previously suggested (Mahtab
1990).
Aims and Design of the Study
Since climatic variability is the most important factor determining agricultural production in Bangladesh,
it is important to study the potential impacts of climate changes on rice production. This study uses climate
change scenarios generated by General Circulation Models (GCMs) and is part of a global research project
funded by the U.S. Environmental Protection Agency.
Two contrasting locations, Mymensingh (Lat. 24.46°N and Long. 90.23°E) and Barisal (Lat. 22.42°N and
Long. 90.23°E), were selected for the study (Table 1). Mymensingh represents the old Brahmaputra, young
Brahmaputra, and Jamuna floodplain regions. This location includes a relatively large proportion of high and
medium highlands ideally suitable for transplanted modern rice cultivars. Barisal represents 2.85 million
hectares of the coastal areas of Bangladesh and constitutes about 30% of the net cultivable area of the country.
The Ganges tidal floodplain is the major agroecological region represented by Barisal, and part of this region
is seasonally inundated by tides.
METHODS
Climate Data and Climate Change Scenarios
Daily maximum and minimum temperatures, precipitation, and hours of sunshine data were obtained
for the two sites. Daily solar radiation values were calculated from sunshine hours using the weather-generating
program included in DSSAT (Jones et al. 1990). Figure 2 shows the monthly average temperatures and
precipitation for Mymensingh and Barisal.
The climate change scenarios used in this simulation study were created from three GCMs: Goddard
Institute for Space Studies (GISS) (Hansen et al. 1983), Geophysical Fluid Dynamics Laboratory (GFDL)
(Manabe and Wetherald 1987), and United Kingdom Meteorological Office (UKMO) (Wilson and Mitchell
1987). The UKMO scenario projects the largest annual temperature increases (4.3°C) (Table 2), while GISS
and GFDL predict smaller annual temperature changes. Annual precipitation is projected to increase
considerably at the two sites (up to 41%) except at Barisal, where annual precipitation is projected to decrease
slightly under the GISS scenario. The study also includes a sensitivity study to changes in temperature (+2°C
and +4°C).
Crop Model, Cultivars and Management Variables
The simulation of rice growth and yield was performed using the CERES-Rice model (Godwin et al.
1992). Since the climate change scenarios have higher levels of CO2 than the current climate GCM simulations,
the CERES-Rice model includes an option to simulate the physiological effects of CO2 on photosynthesis and
water-use efficiency that results in an increase in yield, as documented experimentally (Acock and Allen 1985).
The physiological effects of CO2 were also considered at both sites.
The Bangladesh Rice Research Institute has developed a number of crop varieties; BR11 and BR3 are
the two most extensively used. The cultivar BR11 is grown during the wet season (June-November) and is
transplanted (called "transplanted Aman"). The cultivar BR3 is the main cultivar grown during the dry season
(called "Boro" rice) and is generally irrigated. Occasionally the cultivar BR3 is grown during the wet season
BANGLADESH-4
-------
as transplanted Aman.
The management variables for the simulation model were determined according to current practices in
the two regions considered in this study. Rice growth was simulated under the "automatic irrigation" option
of the model to provide the crop with a nonlimiting water situation. A limitation of this choice of simulation
is that the yield losses due to insufficient water for the crop are not simulated, and therefore the yield could
be overestimated. Nevertheless, relative yield changes under climate change conditions compared to baseline
climate could still be analyzed.
RESULTS
Sensitivity of the Rice Yields to Temperature Increases
Table 3 shows that simulated rice yields decreased significantly with increases in temperature at the two
sites considered. During the transplanted Aman season, rice yields were reduced about 21 ha"1 by a temperature
increase of 4°C at both locations. Similar effects were also observed with the cultivar BR3 in the Boro rice
season. In all cases, the high temperatures result in a reduced season length that partially accounts for the
reduced yields. The crop may also experience heat stress due to the very high temperatures considered in these
sensitivity scenarios, especially during the transplanted Aman season (warmest season). During the Boro season,
base temperatures are lower and rice yields were somewhat less reduced under the high-temperature scenario.
When the direct effects of 555 ppm CO2 were included in the yield simulations, they compensated for the
negative effects of a 2°C temperature increase but not for the effects of a 4°C temperature increase.
Rice Yields Under GCM Climate Change Scenarios
Rice yields decreased significantly at both sites under the three climate change scenarios compared with
the base yields under current climate (Table 4). During the Boro season, the UKMO scenario projected the
largest decreases in yields at the two sites because of the higher temperature projections. In the transplanted
Aman season, simulated yield losses under the GCM scenarios were larger than in the Boro season. Simulated
season length decreased under the three GCM scenarios considered due to the acceleration of phonological
growth stages by high accumulated temperature. When the direct effects of CO2 (555 ppm) were included in
the yield simulation, they compensated for the yield decreases under the climate change scenarios in most cases.
IMPLICATIONS OF THE RESULTS
The current rice production could be damaged under climate change conditions as projected by the
GCMs. Although the simulated direct effects of CO2 may compensate for yield reductions in general, the full
extent of these effects has not been clearly established under field conditions. Another cause for concern in
regard to climate change is the potential for increases in sea-level leading to flooding of low-lying coastal
regions. These regions support high populations and extensive agricultural production. Any possible decrease
in rice production in Bangladesh would have severe consequences on the food resources available to the
country and thus threaten food security. The Bangladesh Planning Commission estimates that the country's
population will increase to 137 million by the year 2000 and to 155 million by the year 2010, leading to further
reduction of the land-person ratio from the low current level of 0.13 ha per person.
In view of the results of this study, serious consideration should be given to reduce population growth,
to develop research strategies to increase rice production, and to limit greenhouse gas emissions. The
development of rice cultivars that are more tolerant to heat and require a shorter growing season is one of the
possible strategies for adaptation to climate change. Policymakers should seriously consider population growth
BANGLADESH-5
-------
regulations as well as qualitative improvements in agronomic research and technology to sustain the country's
economic development.
BANGLADESH-6
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
FAO. 1988. Land Resources and Appraisal of Bangladesh. Technical Report 3, Vol. 1. (BGD/81/035).
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A User's Guide to Ceres-Rice. International
Fertilizer Development Center. Muscle Shoals, AL.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient Three-
Dimensional Global Models for Climate Studies: Models I and II. Monthly Weather Review, Vol III, No.
4:609-662.
IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums
(eds). Intergovernmental Panel on Climate Change. Cambridge.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
14. In: Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer, Las
Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Karim, Z., and N.A. Akhand. 1982. Net irrigation Requirement of Rice and Evapotranspiration of Wheat and
Potato for Different Locations of Bangladesh. BARC Publication No.ll. Dhaka.
Karim, Z., A.M. Ibrahim, A. Iqbal, and M. Ahmed. 1990. Drought in Bangladesh Agriculture and irrigation
Schedules for major crops. BARC publication N°34.
Mahtab, F.U. 1990. Implication of Global Change for Bangladesh. Paper presented at the Workshop on the
Environmental Implication of Global Change, 18th IUCN General Assembly, Perth, Australia,
November 1990.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Statistical Yearbook 1990. 1990. Bureau of Statistics. Government of Bangladesh.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 climate sensitivity experiment with a global model
including a simple ocean. Journal of Geophysical Research, 92: 13315-13343.
BANGLADESH-7
-------
Table 1. Characteristics of the land resources and inundation land type in Bangladesh.
Land Type (ha)
Bangladesh Total
Mymensingh Region
Barisal Region
Highland
4,199,952
156,022
12,117
Medium
Highland
5,039,724
138,303
135,580
Medium
Lowland
1,771,102
63,243
23,593
Lowland
1,101,560
7,821
4,566
Very Lowland
193,243
-
-
Highland: Land above normal inundation level.
Medium Highland: Land normally inundated up to about 90 cm deep.
Medium Lowland: Land normally inundated up to 90-180 cm deep.
Lowland: Land normally inundated up to 180-300 cm deep.
Very Lowland: Land normally inundated deeper than 300 cm.
BANGLADESH-8
-------
Table 2. Annual temperature and precipitation changes from base under GCM scenarios.
Site
Temp, Changes (°C)
Lat/Long GISS GFDL UKMO
Precip. Changes (%)
GISS GFDL UKMO
Mymensingh
Barisal
24.75 N
90.38 E
22.68 N
90.33 E
2.8
4.0
2.8
2.8
4.3
4.3
27
-7
33
33
41
41
BANGLADESH-9
-------
Table 3. Sensitivity of CERES-Rice to temperature and CO2 changes at selected sites.
CO2 level Temp.
(ppm) (°C)
330 0
2
4
555 0
2
4
330 0
2
4
555 0
2
4
Yield Season L.
(t ha'1) (days)
Rice Cultivar BR11, transplanted Aman
Mymensingh
6.28 139
5.14 128
4.39 127
7.69 140
6.51 131
5.74 129
Rice cultivar BR3, Boro Rice
Mymensingh
6.39 156
5.77 150
4.94 147
7.61 156
6.98 150
6.09 148
Yield Season L.
(t ha-1)
Barisal
7.74
6.59
5.71
9.43
8.17
7.29
Barisal
6.81
6.31
5.65
8.13
7.61
6.91
(days)
152
140
138
153
141
139
138
131
129
138
131
129
BANGLADESH-10
-------
Table 4. Simulated rice yield (a) and season length (b) under current climate and yield changes under climate
change scenarios in Bangladesh.
(a) Yield (t ha'1)
Site
Type of rice BASE GISS
% change from base
GISS GFDL GFDL UKMO UKMO
Mymensingh
Barisal
BR 11 Aman
BR 3 Aman
BR 3 Boro
BR 11 Aman
BR 3 Aman
BR 3 Boro
tha'1 330 555 330 555
6.28 -22 0 -27 -6
4.61 -23 0 -27 -6
6.37 -14 5 -18 1
7.74 -19 2 -24 -3
5.7 -18 4 -24 -3
6.81 -11 8 -14 5
330 555
-22 0
-21 1
-25 -6
-20 1
-19 2
-23 -4
(b) Season Length (days)
Site
Mymensingh
Barisal
Type of Rice
BR 11 Aman
BR 3 Aman
BR 3 Boro
BR 11 Aman
BR 3 Aman
BR 3 Boro
BASE GISS GFDL
139 130 127
138 130 127
156 147 148
152 138 139
152 138 138
138 131 129
UKMO
127
127
145
137
138
125
330 - Climate change alone
555 - Climate change with physiological CO2 effects.
BANGLADESH-11
-------
-------
BANGLADESH
Figure 1. Map of Bangladesh and location of the sites used in the simulation study.
-------
BARISAL
TEMPERATURE (C)
PRECIP (mm)
500
10
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
MONTH
Temp (C)
I Precip (mm)
MYMENSINGH
TEMPERATURE (C)
PRECIP (mm)
600
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
MONTH
Temp (C)
I Precip (mm)
Figure 2. Observed monthly averaged temperature and precipitation for selected sites.
-------
IMPACT OF CLIMATE CHANGE ON SIMULATED RICE
PRODUCTION IN THAILAND
C. Tongyai
Soil Science Division, Dept. of Agriculture
Ministry of Agriculture and Cooperatives, Thailand
TrtAILAND-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
METHODS
Observed Climate
Climate Change Scenarios
Crop Model and Management Variables
Validation of the Crop Model
RESULTS AND DISCUSSION
Sensitivity Analysis
Effect of GCM Climate Change Scenarios on Rice Growth
REFERENCES
THAILAND-2
_
-------
SUMMARY
This study used global climate models and crop dynamic models to estimate the potential effects of
climate change on the rice production of Thailand. Projected climate change caused simulated rice yields (both
upland and paddy rice) to decrease dramatically, due to temperature increases. Yield decreases were partially
counteracted by the physiological effects of CO2, as simulated in this study. The high temperatures of the
climate change scenarios had a negative effect on crop growth even when an adequate water supply was
available. We found that in spite of the differences among the GCM scenarios, the locations simulated, and
the agricultural practices of the regions, rice yields under climate change scenarios dramatically decreased in
comparison with base-line yields. The results may be indicative of the potential impacts of climate change on
the agricultural systems in Thailand.
INTRODUCTION
Thailand, located in Southeast Asia, extends over a large latitudinal range (latitude 5°40'N to 20°30'N)
and from longitudes 97°20'E to 105°45'E (Figure 1). The total area of the country comprises 513,115 km2. Six
regions are defined and characterized by their landforms: the Northern Ranges and Valleys, the Northeast
Plateau or Khorat Plain, the Central Plain, the Western mountains, the Southeast coast, and the Peninsular
South. Agriculture is a main contributor to the Thailand economy. Agricultural systems (arable crops,
rangelands, forestry, and fisheries) employ more than 60% of the labor force during the cropping season and
account for 20% of GNP. Crop production accounts for more than 50% of the agricultural revenues. Rice is
the main food crop and the main agricultural export commodity. It is grown in all regions of the country
during both the wet and dry seasons. In. some regions additional water for irrigation is provided when
necessary. Although both long-grain (Thai) and glutinous rice are grown throughout the country, the long-
grain rice is the main crop in most regions and accounts for all rice exports (Table 1).
Floods are an important cause of crop production losses. Natural disasters (tropical or depression
storms and landslides) also account for the destruction of crops and damage to agricultural infrastructure. The
greenhouse effect could lead to higher global surface temperatures and changed hydrological cycles (IPCC
1990). Previous research has suggested that the possible impacts of climate change in Southeast Asia could
be severe (Parry et al 1988). In Thailand a change in the rainfall pattern or an increase in temperature could
lead to soil moisture deficits for the rainfed rice crop, and result in limited irrigation water supply for the
second crop. This study analyzed the simulation of rice production with scenarios of climate change, generated
by General Circulation Models (GCMs), and compared the results to simulations under the current climate.
The study also evaluated the physiological effects of CO2 on rice yields.
To represent the varied range of agricultural practices and regions of the country, simulations were
carried out at four contrasting locations and included a combination of different management practices and
soils. The regions selected for the simulation study are the upper and lower Northern regions and the Central
Plain. The lower Northern region encompasses broad differences in climatic conditions, therefore two sites
were selected in this region (Phitsanulok and Nankhon Sawan). The upper Northern region is represented by
Chiang Mai, and the Central Plain is represented by Bangkok (Figure 1). These sites are representative of the
major rice production areas in Thailand.
METHODS
Observed Climate
THAILAND-3
-------
Climate data for the four sites included in this study were obtained from the Thailand Meteorological
Observatory (1956-1989), and included daily maximum and minimum temperatures and daily precipitation.
Daily solar radiation was estimated from sunshine hours by the WGEN program (Richardson and Wright
1984).
Thailand experiences extreme seasonal climate fluctuations, although its climate is generally humid. Figure
2 shows the monthly temperature and precipitation averages for the 1956-89 period for the site of Bangkok.
Two seasons can be distinguished: wet (May to October) and dry. The mean temperature is always over 25°C
except in the coldest regions (represented by Chiang Mai), and the maximum temperature often exceeds 30°C.
GCM Climate Change Scenarios
The General Circulation Models compute climatic variables for spatial gridboxes across the earth's
surface, but they do not account for the variations within the gridbox. Climate change scenarios for each site
were created based on the GCM gridbox output of climate variables applied to the daily observed climate at
the study sites. The method employs the differences between lxCO2 and 2xCO2 monthly temperatures, and the
ratios between 2xCO2 and lxCO2 monthly precipitation and solar radiation amounts. (lxCO2 refers to current
climate conditions, and 2xCO2 refers to the climate which would occur with an equivalent doubling of CO2.)
Three equilibrium GCMs were used in this study to create the climate change scenarios for each site: the
Goddard Institute for Space Studies model (GISS) (Hansen et al. 1983), the Geophysical Fluid Dynamics
Laboratory model (GFDL) (Manabe and Wetherald 1987), and the United Kingdom Meteorological Office
model (UKMO) (Wilson and Mitchell 1987). Scenarios were also generated with output from a transient
climate change scenario (GISS transient run A, Hansen et al. 1988).
The climate change scenarios show some major differences when compared with the observed climate.
Figure 3 shows monthly differences between observed climate and climate change scenarios for each site. The
temperature increases are very similar at the four sites and in general larger temperature increases are
simulated by the UKMO climate change scenario. With all climate change scenarios, the largest changes in
the seasonal temperatures occurred during the dry season. The scenario precipitation changes vary among the
sites and the distribution of the projected seasonal rainfall was very different from site to site. The largest
increases in projected precipitation usually correspond to the early wet season.
Crop Model and Management Variables
Potential changes in rice physiological responses (yield, season length, evapotranspiration, and
irrigation water demand) were estimated with the CERES-Rice model (Godwin et al 1992) under different
climate scenarios. The software for the climate change simulations of the rice crop was developed by IBSNAT
(IBSNAT1989, Jones et al 1990). The model was run using 30 years of baseline climate and GISS, GFDL, and
UKMO climate change scenarios under dryland and irrigated conditions.
Cultivars. The cultivar RD23 was selected for the study because it represents the most common long-
grain cultivar in Thailand (Table 2). The genetic coefficients of RD23 were taken from the DSSAT database
(Jones et al 1990). In the upper North region (Chiang Mai), the glutinous cultivar NSPT was also used in this
study (Table 2). These cultivars have been calibrated for many different climate conditions, including Thailand.
Management Variables. Rice was planted in early to mid- August, and the plant population was 25
plants m"2. Both upland and paddy rice practices were simulated. For the irrigated simulations, the automatic
option of the CERES-Rice model was used to provide the crop with a nonlimiting water situation. The water
demand was calculated assuming: 100% efficiency of the automatic irrigation system; 1 m irrigation
management depth; and automatic irrigation when the available soil water is 50% or less of capacity. The
initial soil water content for each layer was determined for each type of soil (see below).
THAILAND-4
-------
Soils and Initial Soil Water Conditions. The input soil parameters, nutrient levels, and initial soil
conditions (determined for each layer of soil and for all soils employed in this study) were obtained from a
soil data survey from the Soil Physics Research Group at the Department of Agriculture (Table 3).
Physiological Effects ofCO2. Many crops may benefit from the higher levels of atmospheric CO2 (Acock
and Allen 1985). Because the climate change scenarios have higher amounts of CO2 than the present climate,
the CERES-Rice model includes an option to simulate the physiological effects of CO2 on the rice (i.e., an
increase in photosynthesis and water use efficiency). The physiological effects of CO2 were considered in all
scenarios and at all sites.
Validation of the CERES-Rice Model
The CERES-rice model was validated at Bangkok using observed crop and weather datasets for the
years 1985,1986 and 1987. The rice cultivar used was RD23 (long-grain Thai). The initial soil water conditions
were set to realistic values for the years of the simulation. Observed and simulated yield, biomass, and season
length were compared. Table 3 shows the close correspondence between the simulated and the observed
parameters during the years of the validation experiments.
RESULTS
Sensitivity Analysis
This section characterizes the sensitivity of the CERES-Rice model to changes in temperature at the
four sites selected for the study in Thailand. The direct beneficial CO2 effects on rice growth were simulated
for each set of conditions. Table 4 shows the sensitivity results under irrigated conditions. The temperature
increases resulted in decreases in crop yield and season length at all sites. In the rainfed simulations (not
shown) the relative changes in yields and season length produced by the temperature increases were
comparable to the irrigated simulation.
Effect of GCM Climate Change Scenarios on Rice Growth
This section analyzes changes in simulated crop yield for the GCM climate change scenarios, using
the current climate and management variables as the base reference. Each climate scenario included crop
simulations with 330 ppm CO2 (representing the effects of climate change alone on the crop) and with 555
ppm of CO2 (representing the effects of climate change with the direct beneficial effects of CO2 on yield and
crop water use).
Simulations were conducted to represent different crop management conditions, soils, and initial water
conditions of the soils. There were large decreases in simulated yields under the climate change scenarios alone
compared to the baseline yield (Table 5). These were comparable at all sites and with the different crop
conditions, probably because the negative effect of increased temperature on season length (as shown in the
sensitivity analysis) is the dominant factor driving yield changes in each case. The performance of the rice
cultivar RD23 was compared with the cultivar NSPT, which is a typical photoperiod-sensitive, glutinous indica
cultivar planted in the upper northern region. Generally, the yield of the RD23 cultivar decreased more than
that of the NSPT under the conditions that simulate rice growth in paddies (Table 5).
When the direct CO2 effects were included in the climate change simulations, rice yields increased at
all sites in comparison to the climate change scenarios alone. The direct CO2 effects offset the yield decreases
simulated under the climate change scenarios in some but not all cases (Table 6 and Figure 4). The reaite
of these simulations imply the potential for significant changes for Thailand rice production with climate
THAILAND-5
-------
change. Yield decreases for two of the GCM scenarios tested ranged from -20% to -40%; further studies are
planned to test possible adaptation strategies to yield decreases of such magnitude. The third GCM scenario
implied little change. Better GCM simulations of both current climate and of regional predictions of climate
change are needed to narrow the uncertainties of crucial importance to rice production in Thailand.
THAILAND-6
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocija. 1992. A User's Guide to CERES-Rice. International
Fertilizer Development Center. Muscle Shoals, AL.
Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
Review, Vol III, No. 4: 609-662.
IPCC. 1990. Intergovernmental Panel on Climate Change. 1990. First Assessment Report. Working Group
II. Tegart, W.L. McG., G.W. Sheldon, and D.C. Griffiths (eds.).
IBSNAT. 1989. International Benchmark Sites Network for Agrotechnology Transfer Project. Decision
Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil
Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
14. In Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer,
Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Parry, M., T.R. Carter, and N.T. Konijn (eds.) 1988. The Impact of Climatic Variations on Agriculture.
Kluwer, Dordecht, Netherlands. 876 pp. and 764 pp.
Richardson, C.W., and D.A Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Wilson, C.A, and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
THAILAND-7
-------
Table 1.
Regional rice production (1984-87).
Region
THAI RICE
North Upper
North Lower
Northeast Upper
Northeast Lower
Central East
Central Mid
Central West
South Upper
South Lower
GLUTINOUS RICE
North
Northeast Upper
Northeast Lower
Others
Season
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Area (Ha)
55,200
12,544
1,484,224
65,456
304,704
16,800
611,408
13,872
510,144
62,528
1,184,256
447,824
224,480
15,120
475,040
27,152
111,248
3,840
576,656
196,752
910,976
24,944
Yield (T/Ha)
2.96
3.11
2.14
3.69
1.56
3.02
1.61
2.16
1.84
3.25
2.29
3.95
2.22
3.43
1.55
2.53
1.61
3.37
2.96
1.54
1.39
1.85
Representative Sites:
Chiang Mai (N. Upper); Phitsanulok and Nankhon Sawan
(N. Lower); Bangkok (C. East).
THAILAND-8
-------
Table 2.
Sites, cultiVars, and soil types used in the simulations.
Site
Lat/Long
Rice
Cultivar
Soil Type
Chiang Mai
Phitsanulpk
Nankhon Sawan
Bangkok
+18.5 N
+98.6 E
+16.5 N
+100.2 E
+15.5 N
+100.1 E
+13.4 N
+100.3 E
NSPT
(glutenous)
RD23
(long grain)
RD23
RD23
RD23
RD23
RD23
RD23
RD23
RD23
Hd Clay
Hd Clay
Wangtong
Phimai
Hd Clay
Rb Clay
Np Clay
Suphan
Rangsit
Bangkok
THAILAND-9
-------
Table 3.
Validation of the CERES-Rice model for Bangkok.
Yr.
Yield (t ha'1) Biomass (t ha'1)
Obs. Sim. Obs. Sim.
Season Length (d)
Obs. Sim.
1985
1986
1987
4.23
4.53
4.45
4.40
4.85
4.77
8.93
9.42
8.57
7.89
8.58
8.60
93
90
91
105
106
99
Locations: Klong Luang, Rangsit, Bangkok
Crop Management Conditions:
Soil type; Rangsit Lowland fine, mixed, acid sulfic Tropaquepts
Planting date: August 15
Start of simulation: July 15
Soil water initial condition: 0.456, 0.456, 0.447, 0.414, 0.414, 0.386
Plant population: 25 plants/m2
THAILAND-10
-------
Table 4. Sensitivity analysis of CERES-Rice to climate and CO2 changes. Simulations are for irrigated
production of long-grain Thai rice (BR23).
CHANGES
CO2 level Temp.
(ppm)
330
555
(oQ
0
2
4
0
2
4
Yield
(tha1)
Chiang
(soil: Hd
5.09
3.83
2.74
6.58
5.31
4.29
SIMULATED VARIABLES
Season L. ET Yield Season L.
(days)
Mai
clay)
115
105
102
115
105
102
(mm)
587
574
574
577
569
580
Phitsanulok
(soil: Hd clay)
330
555
0
2
4
0
2
4
4.39
3.32
2.44
5.85
4.84
3.95
104
100
103
104
100
103
591
594
620
582
591
623
(tha-1)
Nankhon
(soil: Rb
4.71
3.62
2.64
6.16
5.07
4.11
(days)
Sawan
clay)
105
101
103
106
101
103
ET
(mm)
572
574
587
561
569
595
Bangkok
(soil: Np clay)
4.63
3.36
2.44
6.14
4.82
3.91
103
97
99
103
98
99
635
625
642
624
621
651
THAILAND-11
-------
Table 5. Effects of climate change on rice yield under GCM climate change scenarios at four sites with
different cultivars and soils. Initial soil water was set to characteristic observed values. (DE
= direct effects of CO2 on rice yield included.)
% CHANGE IN SIMULATED YIELD
GCM Scenario Alone
GCM Scenario +DE of
CO,
Site:
Site:
Site:
Site:
Cult.
Soil
Rainfed/
Irrig.
Chiang Mai
RD23
RD23
NSPT
NSPT
Hd clay
Hd clay
Hd clay
Hd clay
Rainfed
Irrig
Rainfed
Irrig
Base
Yield
tha'1
2.28
5.09
2.24
4.93
GISS
-36
-41
-34
-29
GFDL
-11
-25
-11
-12
UKMO
-25
-39
-25
-28
GISS
-6
-10
-7
-2
GFDL
25
6
18
12
UKMO
10
-10
4
-4
Phitsanulok
RD23
RD23
RD23
RD23
RD23
RD23
Nankhon
RD23
RD23
Bangkok
RD23
RD23
RD23
RD23
RD23
RD23
RD23
RD23
Wangtong
Wangtong
Phimai
Phimai
Hd clay
Hd clay
Sawan
Rb clay
Rb clay
Np clay
Np clay
Suphan
Suphan
Rangsit
Rangsit
Bangkok
Bangkok
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
1.98
4.38
2.61
4.55
2.28
4.39
3.48
4.71
2.58
4.63
2.08
4.68
1.83
4.93
2.82
4.76
-35
-41
-39
-35
-38
-40
-40
-40
-42
-44
-42
-43
-44
-37
-44
-40
-15
-26
-16
-24
-16
-27
-20
-26
-16
-31
-16
-33
-17
-29
-17
-31
-30
-36
-31
-31
-31
-35
-36
-38
-47
-48
-48
-47
-48
-43
-48
-46
-6
-6
-11
-3
-11
-4
-14
-8
-14
-12
-12
-7
-10
-11
-16
-9
23
7
18
5
22
7
11
4
20
0
23
-1
25
-2
18
-1
1
-4
-2
-1
0
-1
-9
-7
-23
-18
-21
-15
-19
-18
-23
-17
THAILAND-12
-------
Table 6. Effects of climate change on rice yields under GCM climate change scenarios in four selected
sites. The water soil initial conditions were set to characteristic observed values. The results
presented are an average of simulations with different soils under irrigated conditions using
cultivar RD23 (long gram Thai rice).
Percent Changes in Simulated Yields from Base
GCM Climate Scenario Alone
GCM Climate Scenario
Including the Physiological
CO2 Effects
Site
Chiang Mai
Phitsanulok
Nankhon Sawan
Bangkok
GISS
-41
-39
-40
-41
GFDL
-25
-26
-26
-31
UKMO
-39
-34
-38
-46
GISS
-10
-4
-8
-10
GFDL
6
6
4
-1
LKMO
-10
-2
-7
-17
THAILAND-13
-------
-------
THAILAND
Figure 1. Map of Thailand showing the sites selected for the study.
-------
BASELINE CLIMATE
BANGKOK, THAILAND
TEMPERATURE (C)
PRECIPITATION (MM/MONTH)
400
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
TEMPERATURE (C)
PRECIP (MM/MONTH)
Figure 2. Observed mean monthly temperature and precipitation in Bangkok.
-------
CLIMATE IMPACT ASSESSMENT FOR AGRICULTURE IN THE PHILIPPINES:
SIMULATION OF RICE YIELD UNDER CLIMATE CHANGE SCENARIOS
Crisanto R. Escafto and Leandro V. Buendia
Philippine Council for Agriculture, Forestry and Natural Resources Res. and Dev.
(PCARRD), Los Banos, Laguna, Philippines
PHILIPPINES-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Rice Production in the Philippines
Climate Restrictions to Crop Production in the Philippines
METHODS
Baseline Climate Data
Climate Change Scenarios
Crop Model and Management Variables
Validation of the Crop Model
Limitations of the Study
RESULTS AND DISCUSSION
Sensitivity of Simulated Rice Yields to Temperature Increases and CO2
Rice Yields Under GCM Equilibrium and Transient Climate Change Scenarios
Adaptation Strategies to Climate Change
REFERENCES
PHILIPPINES-2
-------
SUMMARY
This study used General Circulation Models (GCMs) and a crop growth model to estimate the
potential impact of climate change on rice yields in two main agricultural regions of the Philippines,
represented by the sites of Batac and Los Banos. Projected climate change caused simulated rice yields to
decrease at both locations, but the decreases were larger at Los Banos, which has a lower latitude. The
decrease in simulated rice yield at these tropical sites was caused primarily by the GCM-projected temperature
increases which shorten the crop-growing cycle. When the direct CO2 effects on crop growth were included
in the simulation, the yield losses under the climate change scenarios were completely offset in Batac, but only
partially offset in Los Banos. If climate change brings conditions similar to those projected by the GCM
scenarios, rice production in Los Banos, an important agricultural region, could be significantly reduced.
Testing of possible adaptation strategies to climate change suggested dramatic changes in the cropping patterns
of rice and in the agricultural systems of the Philippines.
INTRODUCTION
Rice Production in the Philippines
The Philippines is an archipelago of about 7,000 islands scattered between latitudes 4° 30' and 21° 30'
north and longitudes 116° and 127° east (Figure 1). The islands are located in the humid tropical belt with
twelve of the 7,000 islands having individual areas of more than 1,000 km2. About 50% of the country's 29
million hectares are cultivated lands. Agriculture is considered the economic lifeline of the Philippines. More
than 50% of the working population is engaged in agriculture and more than 70% of the foreign exchange
earnings are derived from exports of agricultural products. Rice is the staple food of the Filipino people (in
1980 the consumption of rice was 93 kg per capita), and the current population (more than 61 million people)
depends on the rice supply from the 3.4 million ha of cultivated rice each year.
The major areas of rice production (Table 1) are in the regions of Ilocos (represented in this study
by the site of Batac; 18°03'N and 120°32'E), Cagayan Valley, Central Luzon, and Southern Tagalog
(represented in this study by the site of Los Banos; 14°11'N and 121°15'E). Rice management practices vary:
24% of the total area of rice is irrigated during the wet season; 19% is irrigated during the dry season; 38%
is rainfed; and 19% is upland rice. The first three practices are, included in the "wetland" system and the fourth
practice takes place on dry land. Wetland rice production in the Philippines takes place mainly in low-elevation
areas (50 m or lower), in a few areas of low-terraced paddies on intermediate upland slopes, and in isolated
spots on terraces on mountain slopes. Wetland areas are typically the coastal plain regions, the river basins,
and the broad depositional plains. Rice areas on dry land are located on steeper slopes and higher topographic
positions.
Fully irrigated wetland rice areas produce two or more crops of rice annually. Partially irrigated
wetland rice regions produce one crop of rice a year, and are frequently seeded with dryland crops during the
drier part of the year. Dryland rice regions generally produce one rice crop a year, and are seeded with other
crops during the drier part of the year. A shift from rice to other dryland crops, or vice versa, is dictated* by
the relative economic advantages of growing each crop.
Climate Restrictions to Crop Production in the Philippines
Mean annual rainfall ranges from 1,200 mm to 4,600 mm and mean annual temperature at middle-
to low-elevation ranges from 25.8 °C to 27.9.° C. Climatic differences at middle and low elevations are primarily
PHILIPPINES-3
-------
due to rainfall distribution from typhoons and monsoon rains. Agriculture is very vulnerable to climate
hazards, such as the occurrence of tropical cyclones and floods, and to the variability in the dates of the rainy
season that may cause drought during some periods. A large proportion of the potential rice production is lost
each year due to tropical cyclones, floods, and droughts. For example, during the severe drought in 1973, 0.5
million t of potential rice harvest was lost. In 1979 about 1 million t of rice was lost due to floods and
typhoons. In January of 1991, the Philippine weather bureau (PAGASA) confirmed that several provinces were
hit by a drought related to the "El Nino" oscillation. Only 10 weeks after the onset of the El Nino drought,
rice and corn crops suffered estimated damages of $752.8 million.
Scientists predict global climate changes from an increase of atmospheric CO2 and other gases (IPCC
1990). The objective of this study was to evaluate potential climate change impacts on rice production in the
Philippines.
METHODS
Baseline Climate Data
Daily maximum and minimum temperature and precipitation data for the period 1980-89 for Batac
and Los Bafios were used to estimate 30 years of daily weather data for each location using a weather
generator program (Richardson and Wright 1984, Jones et al. 1990).
Climate Change Scenarios
Climate change scenarios were created with the output of three equilibrium GCMs combined with the
daily climate data for each site. The GCMs used were: Goddard Institute for Space Studies (Hansen et al.
1983), Geophysical Fluid Dynamics Laboratory (Manabe and Wetherald 1987), and United Kingdom
Meteorological Office (Wilson and Mitchell 1987). The mean seasonal temperature and precipitation changes
projected by the GCM scenarios are presented in Table 2. The mean annual temperatures are projected to
increase from 2.1 °C to 3.7 °C, while annual precipitation is projected to decrease by as much as 14% under
the GISS and UKMO scenarios, and to increase slightly under the GFDL scenario.
Transient scenarios were created with the GISS transient model for the decades of the 2010s, 2030s,
and 2050s (Hansen et al. 1988). The atmospheric CO2 concentrations considered were 405 ppm, 460 ppm, and
530 ppm, for the respective decades. This study also included a sensitivity test of simulated rice yield to
arbitrary changes in temperature (+2°C and +4°C).
Crop Model and Management Variables
Crop model The simulation of rice growth was performed with the CERES-Rice model developed by
the International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT) (Jones et al. 1990, Godwin
et al. 1992). The CERES-Rice model can simulate increases in photosynthesis and changes in
evapotranspiration driven by increases in atmospheric CO2; these responses are known as the "direct" effects
of CO2 (Acock and Allen 1985).
Cultivars and planting dates. The rice cultivar IR64 was used for the simulation at both sites with an
observed growing cycle of 110 to 120 days. The population density was 25 plants m'2; the row spacing was 0.2
m; and the planting dates were May 15 for Batac, and September 15 for Los Banos.
Soils. The parent materials of wetland rice soils are mainly recent alluvium and Pleistocene marine
sediments. In the higher topographic positions, the soils are composed of Miocene and Pliocene sediments.
The typical soil in Batac is classified as fine, montmorillonitic, isohyperthermic Udorthentic Pellustert; at Los
PHILIPPINES-4
-------
Baftos the typical soil is fine, mixed, isohyperthermic Typic Haplaquoll. The initial soil conditions for the
simulation are: (a) soil water initial condition = full; (b) soil pH = 6.8 for all layers; and (c) soil nitrogen
amendment = 50 kg N ha"1.
Simulations. Simulations for the GCM scenarios and sensitivity analysis included climate change alone
and climate change with the physiological effects of CO2 (Ackok and Allen 1985) at the two sites. Simulations
of possible adaptive strategies were performed under the most unfavorable climate scenario in each
location-GFDL for Batac, and UKMO for Los Baftos.
Validation of the Crop Model
The CERES-Rice model was validated for Los Baftos, comparing simulated parameters to the
observed experimental data from the Los Bafios Irrigation and Nitrogen Study conducted in 1980 at the
International Rice Research Institute. The rice cultivar was IR36, and the observed and simulated yields were
closely related (6.57 t ha"1 simulated and 6.90 t ha"1 observed).
Limitations of the Study
The 30 years of daily weather data used in this simulation study were generated from 10 years of
available data. Generated weather data, although statistically comparable with observed 10-year data, may not
project the actual daily occurrence of weather events over 30 years. Simulations under irrigated conditions used
the automatic irrigation option of the model, but may not represent the actual practice in the Philippines and
may lead to overestimated rice yields. However, because all of the simulations were done under the same
conditions, the relative changes in yield under the different climate scenarios compared to the base yields may
still be considered significant. The rice yields obtained in the simulation are "potential" yields because the
simulation considered that nutrients are nonlimiting; weeds, insect pests, and diseases are controlled; and there
are no typhoons over the simulated period.
RESULTS AND DISCUSSION
Sensitivity of Simulated Rice Yields to Temperature Increases and CO2
A temperature increase of 2°C caused decreases in yield at both locations, a 15% reduction in Batac,
and a 27% reduction in Los Baftos (Table 3). An increase of 4° C above the base temperatures resulted in yield
decreases of 27% in Batac and 53% in Los Baftos. The physiological effects of CO2 (555 ppm in the Table)
in the +2°C scenario compensated for the yield losses due to the increase in temperatures at both locations.
Nevertheless, in the +4°C scenario, the physiological effects of CO2 only partially compensated for the yield
losses, and significant yield decreases were still evident, especially in Batac (-25%).
Rice Yields Under GCM Equilibrium and Transient Climate Change Scenarios
Simulated rice yields decreased under all climate change scenarios in comparison to baseline yields
(Table 4 and Figure 2). The largest yield decreases were simulated in Los Baftos under the GISS and UKMO
scenarios. These yield decreases may be partially a consequence of a shortened crop season due to higher
temperatures. In Los Baftos, the length of the rice season decreased more than two weeks under the GISS and
UKMO scenarios. The shortening of the rice-growing season in Los Bafios was also responsible for a decrease
in simulated total crop evapotranspiration.
When the direct effects of increased CO2 (555 ppm) were considered in the simulations with the
PHILIPPINES-5
-------
climate change scenarios, the negative impact of climate change on rice yield was offset in Batac. In Los Banos,
the rice yield decreased significantly under the GISS and UKMO scenarios, even when the direcfeffects of CO2
were considered.
A linear decrease in yields occurred under the GISS transient scenarios (2010s, 2030s, and 2050s)
(Table 5). In Batac, the physiological CO2 effects offeet the yield decreases, but not in Los Banos.
Adaptation Strategies to Climate Change
Because our results indicated taht climate change may cause decreases in rice yields in some regions
in the Philippines, we considered possible adaptation strategies. The strategies were evaluated under the most
severe climate change scenario (the one that resulted in the largest yield decreases) for each site: GFDL in
Batac, and UKMO in Los Banos. We tested a range of rice cultivars and various planting dates. Table 6 shows
the results of the adaptation simulation strategies; all simulations include the direct CO2 effects on rice yields.
In Batac, planting rice one month earlier increased simulated rice yield of all cultivars tested under
the GCM climate change scenario. In Los Banos, the optimum planting date of the original cultivar (IR64)
under climate change conditions was one or two months earlier than the current planting date. The highest
simulated yields under climate change conditions for Los Banos were obtained by planting the cultivar IR43
one or two months before the current planting date.
September is currently the recommended planting month for rice in Los Banos for the dry season.
September planting dated allow a farmer to harvest rice 4-5 times in two years. If climate conditions force
farmers to plant earlier, the rice cropping pattern could be dramatically affected. Additional problems may
arise when shifting the crop calendar. For example, in Batac, farmers report that a July planting date may not
be appropriate for rice because strong winds in October and November affect the grain-filling stage. Thus,
additional climate factors such as wind need to be considered in the development of adaptation strategies.
These adaptation experiments suggest that the negative impact of climate change on rice yields can
be overcome by planting earlier and changing cultivars, but they also imply major changes to the current
farming system of the Philippines. Such changes may lead to widespread and varied alterations in rural
regions.
PHILIPPINES-6
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A User's Guide to CERES-Rice. International
Fertilizer Development Center. Muscle Shoals, AL. (in press).
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lascis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
three-dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-
622.
Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.
Jones, J.W., S.S. Jagtap, G.Hogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer. Las
Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.
IPCC1990. The IPCC Scientific Assessment, eds. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums. Cambridge
University Press.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Richardson, C.W., and D.A. Wright. 1984. WGEN:A model for generating daify weather variables. ARS-8. U.S.
Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
Wilson, C.A, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
PHILIPPINES-7
-------
Table 1.
National and regional rice production in the Philippines (1980-89).
Region
CAR
Ilocos * BATAC
Cagayan Valley
Central Luzon
S Tagalog * LOS BANGS
Bicol
W. Visayas
C. Visayas
E. Visayas
W. Mindanao
N. Mindanao
S. Mindanao
C Mindanao
NATIONAL
Area
haxlOOO
78
299
324
472
378
310
464
110
198
136
120
186
243
3,317
Production
txlOOO
170
725
826
1,461
869
646
1,112
158
370
321
325
558
706
8,246
Yield
T/Ha
2.21
2.42
2.64
3.09
2.30
2.08
2.40
1.45
1.86
2.36
2.71
3.03
2.90
2.49
*Sites modeled
PHILIPPINES-8
-------
Table 2. Seasonal and annual temperature and precipitation changes in the GCM climate change
scenarios.
SITE/
Season
Temperature changes (°C)
Precipitation changes (%)
GISS
GFDL UKMO
GISS
GFDL
UKMO
BATAC
Winter
Spring
Summer
Fall
Annual
2.3
3.0
2.7
3.5
2.9
4.3
2.4
2.5
2.3
2.9
3.0
0.9
2.3
3.1
2.3
-33
-17
-25
20
-14
-12
3
37
-13
31
-49
-9
12
-2
LOS BANGS
Winter
Spring
Summer
Fall
Annual
3.3
3.2
3.2
3.8
3.4
2.5
2.1
2.1
1.7
2.1
4.2
4.0
2.9
3.5
3.7
13
11
-47
10
-5
-40
13
27
25
-13
-27
-5
7
-5
PHILIPPINES-9
-------
Table 3.
Sensitivity of simulated rice yields to temperature increases and CO2
SIMULATED YIELD (t ha'1)
SITE
Los Banos
Batac
TEMPERATURE
CHANGE
0 (base)
+2°C
+4°C
0 (base)
+2°C
+4°C
Climate Scenario
Alone
(330 ppm CO2)
5.92
5.03
4.34
4.03
2.92
1.89
Climate Scenario +
P.E. of CO2 (555
ppm CO2)
7.01
6.23
5.59
5.34
4.12
3.00
P.E. - Physiological effects of CO2: the direct beneficial effects of 555 ppm CO2 on rice yield are included in
the simulation.
PHILIPPINES-10
-------
Table 4.
Simulated yield, season length, and evapotranspiration (ET) under GCM climate changes
scenarios.
SIMULATED YIELD (t
ha'1)
Scenario Scenario +
Alone (330 PE of CO2
Simulated
Season
Simulated
SITE
Batac
SCENARI
O
BASE
GISS
GFDL
UKMO
ppm CO2)
5.91
4.99
4.93
5.51
ppm CO2
7.19
6.19
6.11
6.77
(days)
121
118
118
118
(mm)
1,659
1,263
2,151
1,409
(mm)
586
637
631
343
Los Banos
BASE
GISS
GFDL
UKMO
3.86
2.09
2.75
2.04
5.15
3.22
3.91
3.14
120
103
109
103
928
1,057
982
983
421
402
409
403
P.E. - Physiological effects of CO2.
PHILIPPINES-11
-------
Table 5.
Simulated rice yields under the GISS transient scenarios and the GISS equilibrium scenario.
SIMULATED YIELD (t ha'1)
SITE
Batac
SCENARIO
BASE
GISS 2010s
GISS 2030s
GISS 2050s
GISS
Scenario
Alone
5.91
5.49
5.31
4.81
4.99
Scenario
+ P.E. of CO2*
~
5.86
6.03
5.86
6.19
Los Banos BASE 3.86
GISS 2010s 3.21
GISS 2030s 2.82
GISS 2050s 2.12
GISS 2.09
* See METHODS for the CO, levels under each scenario.
3.58
3.54
3.20
3.22
PHILIPPINES-12
-------
Table 6. Effect of changes in rice cultivar and planting dates on rice yields simulated with the
CERES-Rice model under base climate and GCM climate change scenarios.
Site Scenario
Batac BASE
GFDL
GFDL
GFDL
GFDL
GFDL
GFDL
GFDL
Los Banos BASE
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
Cultivar
IR64
IR64
IR64
IR64
IR43
IR43
IR43
UPLR 15
IR64
IR64
IR64
IR64
IR64
IR43
IR43
IR43
IR43
UPLR 15
Planting Date
May 15
May 15
Apr 15
June 15
May 25
Apr 15
June 15
May 15
Sept 15
Sept 15
Aug 15
July 15
Oct 15
Sept 15
Aug 15
July 15
Oct 15
Sept 15
SIMULATED
YIELD t ha'1
5.91
6.11
6.48
5.75
6.42
6.83
6.19
5.01
3.86
3.14
3.93
4.87
2.69
3.39
4.17
5.03
2.79
2.02
PHILIPPINES-13
-------
-------
PHILIPPINES
Figure 1. Map of the Philippines and location of the sites selected for the simulations.
-------
% CHANGE FROM BASE YIELD
BATAC
LOS BANOS
GISS.330
GFDL.555
GISS.555 HH GFDL.330
UKMO.330 CHI UKMO.5S5
Figure 2. Simulated rice yield changes under GCM climate change scenarios.
-------
EFFECTS OF CLIMATE CHANGE ON RICE PRODUCTIONAND STRATEGIES FOR
ADAPTATION IN SOUTHERN CHINA
Zhiqing Jin, Daokou Ge, Hua Chen, and Juan Fang
Department of Agrometeorology
Jiangsu Academy of Agricultural Sciences
People's Republic of China
CHINA-1
-------
ACKNOWLEDGEMENTS
The authors would like to express their profound appreciation to Dr. C. Rosenzweig, Prof. L.Z. Gao, Dr. J.T.
Ritchie, and Dr. T.Y. Chou for their suggestions, encouragement, assistance, and advice in this work.
CHINA-2
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Agroecological System
Literature Review
METHODS
Climate Data
Climate Change Scenarios
CERES-Rice Crop Model
Management Variables and Soils
Calibration and Validation
Simulation Scenarios
Crop Index
RESULTS AND DISCUSSION
Impacts on Rice Yields
Impacts on Season Length
Impacts on Irrigation
Impacts on Cropping Systems
Adaptation Strategies
REFERENCES
CHINA-3
-------
SUMMARY
The CERES-Rice model was calibrated and validated for nine sites in Southern China to examine its
suitability to model rice production in the region, using agronomic data from three or more successive years.
After determining the genetic coefficients for the cultivars, the CERES-Rice model was run a second time for
the same locations for a time period of 20-30 years. The model used local climate data (1958-86) and doubled-
COZ climate change scenarios generated from the GISS, GFDL, and UKMO General Circulation Models
(GCMs), with and without supplemental irrigation (to model paddy and upland rice, respectively). This study
assessed the direct physiological effects of CO2 on rice growth for each scenario. Finally, the study examined
several strategies for adapting rice production to climate change.
The results of the study are listed below. They should not be regarded as predictions, but as plausible
assessments of the potential effects of climate change on rice production in Southern China.
(1) Climate change alone
Simulated rainfed rice yields decreased under climate change alone due to increases in temperature
that shorten the growing season for rice. For some sites, however, sharp decreases in precipitation
were also an important factor in the decreased yield of rainfed rice.
Rice yields simulated under "automatic" irrigation also decreased. Although irrigation did not fully
compensate for the negative effects of climate change, it significantly improved simulated rice yields,
especially in regions where precipitation decreased under climate change conditions.
(2) Climate change with the direct effects of CO2.
In rainfed rice, the direct effects of CO2 compensated for the negative effects of climate change alone
in most sites, except in sites where rainfall sharply decreased in the climate change scenarios.
In irrigated rice, the three GCM scenarios produced increases in modeled rice yields in comparison
with the baseline yields in the northern sites, but decreases in the central and southern sites. These
findings suggest that there is less compensation by the physiological effects of CO2 in areas with high
temperature.
(3) General results.
Simulated irrigated rice yields are higher and have less year-to-year variability than rainfed yields.
Under all climate change scenarios studied, the amount of water needed for automatic irrigation
greatly increased in areas where precipitation sharply decreased.
Evapotranspiration (ET) for rainfed rice was usually less than that for automatic irrigated rice.
Therefore, cultivation of upland rice may be extended into areas where irrigation water is not
available.
An increase in temperature would increase China's rice-based cropping system. The northern limits
for double-rice and triple-rice cropping systems could be moved northward about 5-10 degrees of
latitude, depending on the climate scenario.
CHINA-4
-------
Introducing upland rice cultivars to areas where precipitation sharply decreased under the climate
change scenarios significantly improved rainfed rice yields at some sites, but not at others.
For paddy rice, adjusting planting dates ameliorated the negative effects of climate change on modeled
yields in the northern part of the study region, but not in the southern part.
INTRODUCTION
Since 1949, China has made great progress in food production. With only 9% of the world's cultivated
land, China supports more than one-fifth of the world's population. However, with increasing population, a
sharp decrease in cultivated land, lack of water resources, environmental pollution, and frequent natural
disasters, it is difficult for China to continue to increase its food production.
Recently, scientists have suggested that the addition of greenhouse gases to the atmosphere will alter
global climate, increasing temperatures and changing rainfall and other climate patterns (IPCC 1990). The
combined effects of climate change and physiological effects of CO2 on crops may result in a net increase in
crop yields in some cases, especially in the high latitudinal regions. In low latitudes, agriculture could be
negatively affected (Parry et al. 1988).
The objective of this study is to characterize the direction, magnitude, and degree of uncertainty of
the potential impacts of global climate change on rice production in Southern China. This study is important
because: (1) China is the largest rice producer and consumer in the world; (2) Southern China plays an
important role in the nation's rice production; and (3) Southern China appears to be more vulnerable to
climate change than the more northern regions of China (Jin et al. 1990).
The study is based on simulations with the CERES-Rice model. The model used actual daily weather
data (about 30 years) from nine sites in seven provinces of Southern China (Figure 1) and regional climate
change scenarios generated from General Circulation Models (GCMs). Rice growth is simulated for both
rainfed and automatic irrigation conditions to represent some mountain rice areas without irrigation and the
plains rice areas with an extensive irrigation system. Changes in rice yield, growth duration, evapotranspiration,
and irrigation are estimated under baseline and climate change conditions. The physiological effects of CO2
are also analyzed under each scenario.
The results of this study should be considered as indicators of the possible impacts of global warming.
Agroecological System
Southern China is subject to the influence of the monsoonal wind system of eastern Asia, which makes
the climate wet and hot in the summer, and cold and dry in the winter. Ample rainfall combined with high
temperatures and high intensity of solar radiation during the rice-growing season are generally favorable
conditions for rice production.
About 1,000 rivers and more than 200 large lakes provide a copious irrigation system, and rich soil
and intensive farming have made this region one of the world's greatest centers of rice production.
Nevertheless, some unfavorable climate events constrain rice production. Summer and autumn droughts are
rather common, and typhoons, floods, and heat waves occur frequently.
The topography in Southern China is varied and complex. Altitude drops from west to east, leaving
only about 15% of the land for agricultural purposes. Consequently, the farmers must adopt a multiple-
cropping system.
Southern China is comprised of 16 provinces: Anhui, Fugian, Guangdong, Guangxi, Guizhou, Hainan,
Hubei, Hunan, Jiangsu, Jiangxi, Sichan, Taiwan, Tibet, Yunnan, Zhejian, and Shangai. Rice is the most
CHINA-5
-------
important crop in all of them, except in Tibet.
Southern China can be divided into four rice subregions: the Middle and Lower reaches of the Yanzi
River; the Southwest; South China; and the Yellow-Hui Rivers valleys. Table 1 shows information on land use
and rice production for each subregion.
Literature Review
The impacts of 2xCO2 climate change on agricultural crops have been studied mainly by scientists from
developed countries. In North America, Robertson et al. (1987) analyzed the impacts of climate change on
yields and soil erosion for selected crops in the Southern U.S., Central Prairie, and Northern Plains using the
Erosion Productivity Impact Calculator (EPIC); Allen et al. (1987) reported that the soybean yield could
increase about 32% with a doubling of CO2; Smit (1989) studied how crop yields in Ontario (Canada) would
be affected by climate change (GFDL and GISS scenarios); and Jones et al. (1986) used the Soybean Integrated
Crop Management (SICM) model to estimate potential changes in earworm damage to soybean under climate
change scenarios, concluding that earworm damage may increase in the Grain Belt region under a warmer
climate. Using an agroclimatic approach, Rosenzweig (1990) suggested that the wheat cultivars in the Southern
Great Plains would change from winter to spring wheats due to the lack of cold winter temperatures under
the climate change scenarios, and that climate change would bring about an increased demand for irrigation.
The CERES and Soygro models have been widely used in the US: in the Great Lakes (Ritchie et al.
1989), the Southeast (Peart et al 1989) and Great Plains (Rosenzweig 1990), to simulate change in crop yields
under different climate change scenarios. Rosenzweig (1990) characterized the direction, magnitude, and
uncertainty of potential climate change-induced alterations in wheat and corn yields. The study suggested that
adjusting the planting dates of winter wheat and corn would not significantly ameliorate the negative effects
of climate change and that change of cultivars only could be a possible adaptation strategy to compensate for
yield loss at some selected sites but not others. Her study also analyzed the direct effects of CO2 on crops (i.e.
increased photosynthesis and improved water use efficiency). Ritchie et al. (1989) suggested that climate
change induced maize yield reductions in the Great Lakes could be partly compensated by using new cultivars
with a longer growing cycle. Parry et al. 1988 reported possible yield decreases of rice in India, and some
increases in Japan under climate change conditions; Parry et al. (1988) reported that climate change may cause
rice yield to decrease in Thailand.
At the present time there is no published comprehensive study that evaluates the potential impacts
of climate change on large-scale rice production in China and that includes the beneficial direct effects of CO2
on rice production.
METHODS
Climate Data
Daily climate data (maximum and minimum temperatures, precipitation, and solar radiation) for a
period of 20-30 years (about 1957-1986) were taken from the Monthly Report of Chinese Meteorological
Record on the Surface (China Meteorological Bureau) and the Daily Solar Radiation Record of China (Beijing
Meteorological Bureau). The daily solar radiation at Xuzhou was not available and was calculated according
to Gao and Lu (1982).
Climate data for the calibration and validation experiments were collected from the local
meteorological stations in the following institutions: China National Rice Research Institute, the National
Academy of Agricultural Sciences, Guangdong Academy of Agricultural Sciences, Jiangxi Agricultural
University, Jiangsu Agricultural Bureau, Jiangsu Academy of Agricultural Sciences, Shaoguan Prefecture
CHINA-6
-------
Institute of Agricultural Sciences in Guangdon province, and Lixiahe Prefecture of Agricultural Sciences in
Jiangsu province.
Climate Change Scenarios
Mean monthly temperature differences, precipitation and solar radiation ratios (2xCO2/lxCO2 GCM-
generated climate) were combined with daily historic climate data to generate climate change scenarios for the
various sites. The GCMs used were: Goddard Institute for Space Studies (GISS) (Hansen et al. 1987);
Geophysical Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987); and United Kingdom
Meteorological Office (UKMO) (Wilson and Mitchell 1987).
GCMs have several advantages over other approaches for creating climate scenarios (Smith and Tirpak
1989): (1) the models estimate how global climate may change in response to increased CO2, and therefore,
regional outputs are internally consistent with global warming associated with doubled CO2; (2) the climate
variables estimated are in agreement with physical laws; and (3) the GCMs provide the information necessary
to run the crop simulation models.
GCMs, however, have some disadvantages: (1) the ocean models currently used in most GCMs are
relatively simple (most GCMs treat oceans as "swamps" or only simulate their upper layers); (2) they have low
spatial resolution and do not provide a good representation of major geographic features, such as plateaus,
mountains and basins, which have large impacts on local climate; and (3) GCMs include some simplified
assumptions about other factors, such as cloud cover, albedo, and the hydrology of the land surface.
Table 2 presents annual mean temperature and precipitation changes generated by the GISS, GFDL,
and UKMO scenarios for different locations in Southern China.
CERES-Rice Crop Model
The CERES-Rice model (Godwin et al. 1992; IBSNAT 1989; Alocija and Ritchie 1988) was used for
the study. The model simulates yield and other parameters of crop growth and development under different
management and climatic conditions.
The CERES-Rice model was chosen because (1) it simulates the effects of major factors, such as
climate and soil management, on rice growth, development, and yield; (2) the model also simulates the direct
effects of CO2 on crop photosynthesis and evapotranspiration; (3) the genetic coefficients that are used to
characterize the different cultivars can be calibrated based on the experimental data; and (4) the model has
been available and documented for several years and has been validated under a wide range of soil and climatic
conditions.
Management Variables and Soils
Crop data of the local rice cultivars (sowing date, soil type, sowing depth, transplanting density, row
spacing, maturity date, biomass, and grain yields) were taken from local field experiments. The following
institutions provided data: China National Rice Research Institute, the National Academy of Agricultural
Sciences, Guangdong Academy of Agricultural Sciences, Jiangxi Agricultural University, Jiangsu Agricultural
Bureau, Jiangsu Academy of Agricultural Sciences, Shaoguan Prefecture Institute of Agricultural Sciences in
Guangdon province, and Lixiahe Prefecture of Agricultural Sciences in Jiangsu province. Table 3 lists the main
rice production areas of Southern China, the sites selected for the study, the local rice cultivars used for
modeling, and the soil types at the various sites.
Soil data included soil type, albedo, organic matter, texture, structure, and bulk density. Representative
soil types and profiles were chosen according to the Soil Atlas of China (Institute of Soil Science 1986) and
CHINA-7
-------
Chinese Soils (Xiong Yi and Li Qingkui 1988).
Calibration of Genetic Coefficients
Eight genetic coefficients define a rice cultivar in the CERES-Rice model and characterize
quantitatively how that particular rice cultivar responds to environmental factors such as climate, management,
and soils. The genetic coefficients for the rice cultivars used in this study were determined by contrasting the
simulated results to local field experimental observations, and included growth duration, dry-matter
accumulation (if available), and yield. The management variables (planting date, plant density, and spacing,
eta) for each calibration site were taken from the field experiments and local climate data for the calibration
years was used. The specific genetic coefficients of a local cultivar were determined by trial-and-error so that
the simulated yield and maturation date were as close as possible to the observed values.
Simulation Scenarios
The three GCM climate change scenarios were simulated at each site, and each simulation was
completed under rainfed and irrigated conditions. Automatic irrigation was done to provide the crop with a
nonlimiting water situation, which is similar to practice. Since the climate change scenarios involve an
equivalent doubling of CO2, the physiological CO2 effects on crop yield and water use were also included in
simulations at each site (indicated as 555 ppm CO2 in the tables).
In addition, the study evaluated the possible adaptive strategies to rice management under future
climate. Scenarios were created by altering current agricultural practices in order to maximize yields under the
conditions of climate change.
Crop Index
A Crop Index has been used to evaluate different cropping systems in China (Gao et al. 1987).
Possible changes in the index were analyzed by using simulated rice growth output under climate change
conditions.
According to Gao et al (1987), the rice growing season in China is defined as the number of days
from the safe sowing date to the safe maturity date. It is also equal to the number of days from the safe sowing
date to the safe heading period, plus 40 and 30 days for the Japonica and Indica varieties, respectively. The
safe sowing date is defined as the time when the mean daily temperature is constantly above 10°C and the safe
heading period is defined as the time when the mean daily temperature is above 20°C, with no temperature
lower than 20°C for three consecutive days. Based on this index, it is easy to compute the lengths of the
growing season for rice under the current climate and for the three GCM climate change scenarios, and to
evaluate the changes in rice cropping systems in the future.
According to the China National Rice Research Institute (1988), a value of the accumulated
temperatures above 10°C for different rice-based cropping systems is:
(1) 2,000°C to 4,500°C for single-rice cropping;
(2) 4,500°C to 7,000°C for double-rice cropping (the northern limit for double-rice is 5,300°C);
and
(3) more than 7,000°C for triple-rice cropping.
These thermal values can also be used to estimate potential changes in the rice cropping systems under
climate change scenarios.
CHINA-8
-------
RESULTS AND DISCUSSION
Impacts on Rice Yields
Doubled CO2 climate change would directly influence rice yields through physiological processes and
indirectly through climate.
Climate Change Alone. The simulated rice yields decreased under the three scenarios at all locations.
Under the GISS scenario, rainfed yields decreased significantly (10%-78%) from baseline yield (Table 4).
Under the GFDL and the UKMO scenarios, yields decreased 6%-33% and 7%-34%, respectively. The yield
decreases were a result of the higher temperature under the climate change scenarios which shortened the
growing cycle of rice and also caused water stress in some regions. More severe yield decreases (-78% at
Chengdu) under the GISS scenario were due in part to significant growing season precipitation decreases. For
example, the July rainfall at Chengdu under the GISS scenario declined 90% in comparison with the current
rainfall, resulting in a growing season with total rainfall of less than 400 mm (Jin et al 1990). Although the
enhancement of respiration due to high temperatures may be another reason to decrease rice yields, it was not
taken into account in the CERES-Rice model.
The automatic irrigation simulations under the three climate change scenarios also resulted in
decreases in rice yields in comparison with the base data at all sites (Table 4): 15%-33% under the GISS
scenario; 14%-37% under the GFDL scenario; and 19%-33% under the UKMO scenario. These results
indicate that full irrigation did not completely offset the negative effects of increases in temperature on rice
yields because of the shorter rice growth duration, particularly of the grain-filling stage. Nevertheless, the
negative effects of climate change on rice yields were partially offset by full irrigation at the sites where the
rainfall during the rice growing season greatly decreases under the climate change scenarios. For example, at
Chengdu and Guangzhou under the GISS scenario, rainfed yields decreased 78% and 45%, respectively, and
the irrigated yields declined only 32% and 16%. These results suggest that an effective strategy for adapting
to climate change would be to improve the irrigation systems in such regions.
Table 4 also shows that the standard deviations of the percent yield changes-an indication of the
variability (Rosenzweig 1990)—of irrigated rice yields were much lower than the standard deviations of rainfed
rice yields.
Climate Change with Physiological CO2 Effects. Table 5 shows the physiological effects of CO2 on crop
yield and water use under the climate change scenarios. Under the GISS scenario, rainfed rice yields increased
at the northern sites (Xuzhou and Nanjing) and the eastern sites (Fuzhou and Nanchang). At these sites, the
scenario precipitation was not a limiting factor on rice production. In Wuhan, Changsha, and Shaoguan, the
direct effects of CO2 ameliorated, to a certain degree, the negative effect of increased temperature on rainfed
rice yield. However, in Chengdu and Guangzhou, because of large decreases in precipitation during the rice
growing season, the yields remained significantly lower than base case yields, even with the direct effects of
CO2. In these two southern sites, the GISS scenario yields were very low in comparison with the baseline, and
there were very small differences in yields simulated with and without CO2 physiological effects (Table 4).
Under the GFDL and UKMO scenarios, the direct effects of CO2 largely compensated for the climate change
impacts on rice yields in many but not all sites.
In the irrigated simulations, rice yields declined in the central-southern region (Nanchang, Changsha,
Wuhan, and Shaoguan) under the three climate change scenarios (Table 5). This was a result of the large
annual temperature increases projected by the GCMs in that region (4°C-7°C). In contrast, the direct effects
of CO2 resulted in an increase of the modeled irrigated yield at Xuzhou under the three climate change
scenarios. These results suggest that increased temperatures would benefit rice production in the northern
CHINA-9
-------
areas of the studied region.
As in the case of climate change alone, the yield stability (standard deviation) under the rainfed
conditions was less than that under irrigated conditions.
Impacts on Season Length
The increment of temperature under all climate change scenarios caused the rice growth duration to
decrease at all sites (Table 6). In the northern areas (Xuzhou, Nanjing, and Chengdu), the mean rice growth
duration decreased more than in the southern areas. Because the current temperatures in the northern sites
are relatively low, rice has a long growing period which is more sensitive to the temperature increase projected
by the GCM climate change scenarios. In contrast, the current temperatures at the southern sites already cause
the rice crop duration to be very short, and a further increase in temperature did not have a very large effect
on simulated growth duration. The physiological effects of CO2do not influence season length, as simulated
with the CERES-Rice model.
Impacts on Irrigation
An adequate water supply is one of the most important factors in rice production. The amount of
irrigation required for rice depends on many factors, such as growing season rainfall, temperature, solar
radiation, and evapotranspiration (Table 7), as well as crop and soil characteristics, depth of water table, and
topography. Among these factors, rainfall is the most important. Table 8 shows the percent change in
irrigation demand for rice under the climate change scenarios, including the direct effects of CO2. Simulations
were done with automatic irrigation (100% efficiency of application and availability of water).
Under the GISS scenario, the demand for irrigation rose at six of the nine sites. At Chengdu and
Guangzhou, the irrigation demand increased six and two times, respectively, over the present levels. This
increase was caused by large decreases in the summer rainfall and increases in temperature and solar radiation.
Under the GFDL scenario the irrigation demand decreased in most sites where precipitation during
the local rice growing season increases. Under the UKMO scenario irrigation demand results were mixed: they
increased in the sites where projected precipitation decreased, and increased at the rest of the sites.
Impacts on Cropping Systems
Under all the doubled CO2 scenarios, the temperature, as well as the >10°C accumulated temperature
index, increased and could result in an extension of the growing season for rice over large areas of Southern
China. However, the increased temperature shortened the simulated lifecycle of rice and had a negative impact
on yields. The combined effects of a prolonged growing season and shortened growth duration would shift the
northern limits of the various rice-based cropping systems towards higher latitudes. As a result, the Crop Index
in China would increase and the varieties and management systems would have to be adjusted to the new
conditions.
The current climatic classification for rice production in China would no longer be applicable under
climate change conditions. Table 10 lists the onset of the rice cropping season, length of the growing season,
and the >10°C accumulated temperatures during the rice growing season at selected sites. Under the GISS
scenario, the rice maturity dates in China advanced by an average of 19 days, the length of the rice growing
season was prolonged by an average of 45 days, and the 10°C accumulated temperature increased by an average
of 1,522°C. Figure 2 shows the northern limits for double- and triple-rice crops.
According to the GISS scenario, the thermal conditions at Beijing were more favorable for rice than
the current thermal conditions at Nanjing, and it would be suitable to grow rice after wheat. The thermal
CHINA-10
-------
conditions at Shenyang would be more favorable than those at Beijing, without any low temperature problems
to sustain the growth pattern of the three crops in two years. In the most northern area of China, Harbin, the
>10°C accumulated temperature would reach 3,696°C, with a rice-growing season of 159 days, and it would
be safe to grow single-rice; early Japonica with a late maturity date.
Similar results were obtained with the GFDL scenario since the temperature scenario increases are
comparable to those under the GISS scenario. The UKMO scenario resulted in an extreme situation: the
northern limit for double-rice would shift to the Shenyang region, and that for the triple-rice would move to
a line extended from Jinan to Zhengzhou. In the southern parts of China, triple-rice could be grown
throughout the year.
Currently, Indica rice varieties are grown in the south and Japonica varieties in the north. With
climate change, the patterns of rice varieties used throughout China would shift and the traditional practice
of "South Indica" and "North Japonica" would no longer hold. The high-temperature-tolerant varieties would
not only dominate in the southern parts of China, but would be likely to be more widely used in the north.
Based on the above analysis, a 2xCO2 climate change would result in a tremendous change in China's
thermal conditions. Therefore, effective strategies for adapting to climate change need to be developed, such
as improving crop breeding (especially the use of biotechnology to develop more heat- and drought-resistant
rice varieties); expanding the use of chemical fertilizers, pesticides, and herbicides; and improving irrigation
systems. These strategies might be quite costly, as they imply an increase in agriculture-related inputs.
Adaptation Strategies
This section presents and discusses some possible strategies for adaptation to climate change. The
GISS scenario was chosen because it had the most negative effects on crop yield. All adaptation simulations
included the physiological effects of CO2 was analyzed (Table 9).
Using crop model simulations, the following strategies for adaptation to climate change were
evaluated.
Change in rice cultivar. This strategy considers whether new cultivars would improve rice yield under
climate change conditions. An upland rice cultivar (UPLR15) was simulated at the sites where precipitation
declined dramatically under the scenario conditions (Chengdu and Guangzhou) (Table 9). With the new
cultivar, the yield at Chengdu increased in comparison with the yield of the originally used cultivar. Compared
with the baseline yield the decrease under climate change conditions was still significant (-54% with the new
cultivar and -72% with the original cultivar). At Guangzhou, however, introducing the same cultivar did not
improve the yield under the scenario conditions at all.
The cultivar IR43 was simulated at seven locations. At five sites (Nanchang, Fuzhou, Xuzhou,
Changsha, and Wuhan), the yields of the new cultivar were higher that those of the original cultivar under the
conditions of climate change. In the other two sites (Nanjing and Shaoguan), the change of cultivar did not
improve modeled yields.
Adjusted planting dates. Changes in the planting dates (by 10 day intervals) of the present cultivars may
increase yields under the climate change conditions. Changing the planting date caused rice yields to increase
at the northern sites (Xuzhou and Nanjing), but not at the southern sites.
Changes in both cultivars and planting dates. Rice yields increased significantly when a combination
and change in planting date was simulated at six of the seven locations tested (all except Nanjing). The yield
increases ranged from +3% to +43%, and the mean increase was +22% (the calculations assume equal area).
CHINA-11
-------
Improvement of the irrigation system. Yields simulated under rainfed and irrigated conditions were
compared at each site to determine if irrigation may mitigate the negative effects of high temperatures on
yields.
CHINA-12
-------
REFERENCES
Allen Jr., L.H., 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.
Alocilja, E.C., and J.T. Ritchie. 1988. Rice Simulation and its use in Multicriteria Optimization. IBSNAT
Research Report Series 01.
China National Rice Research Institute. 1988. A Classification for Rice Planting in China. Zhejing Scientific
and Technological Publishing House. Hangzhou.
Gao, L., L. Lin, and J. Zhiqing. 1987. A classification for rice production in China. Ag. and Forest Meteor.
39:55-65.
Gao, G., and Y. Lu. 1982. Radiation and Heat Balance on the ground surface in China. Scientific and
Technological Publishing House. Beijing.
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A user's guide to Ceres-rice. International
Fertilizer Development Center. Muscle Shoals, AL.
Hansen, J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
Review, Vol HI, No. 4: 609-662.
Institute of Soil Science, Academic Sinica. 1986. The Soil Atlas of China. Cartographic Publishing House.
Beijing.
IBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project (IBSNAT). 1989.
Decision Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy
and Soil Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii
96822.
IPCC. Intergovernmental Panel on Climate Change. 1990. First Assessment Report
Jin, Z., X. Zhen, J. Fang. D. Ge, and H. Chen. 1990. Potential effects of global climate change on winter wheat
production in China. Internal Report.
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. 194 pp.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
14. In: Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer,
Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
CHINA-13
-------
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Parry, M.L., T.R. Carter, and N.T. Konijn. 1988. The impact of climatic variations on agriculture. Vol 1
Assessments in cool temperate and cold regions. Vol 2 Assessments in semi-arid region. Kluwer,
Dordecht, Netherlands. 876 pp. and 764 pp.
Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, and L.H. Allen, Jr. 1988. Impact of climate Change on Crop
Yield in the Southeastern U.S.A. In J.B. Smith and D.A. Tirpak (eds). Report to Congress. The
Potential Effects of Global Climate Change on the United States. U.S. Environmental Protection
Agency. Washington, DC. EPA-230-05-89-050.
Ritchie, J.T., B.D. Baer, and T.Y. Chou. 1989. Effect of global climate change on agriculture: Great Lakes
region. In Smith, J.B. and D. Tirpak (eds.). The Potential Effects of Global Climate Change on the
United States. Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency.
Washington, DC. Appendix C-l, pp. 1-1 to 1-30.
Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study.
Professional Geographer 42(l):20-37.
Smit, B. 1989. Climatic warming and Canada's comparative position in agricultural production and trade. In
Climate Change Digest CCD 89-01. Environment Canada, pp. 1-9.
Smith, J.B., and D. Tirpak (eds.). 1989. The Potential Effects of Global Climate Change on the United States.
Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency. Washington, DC
423pp.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
Xiong Yi, and Li Qinqkui. 1987. Chinese Soils (Second Edition). Scientific Publishing House, Beijing.
CHINA-14
-------
Table 1.
Regional contribution to rice production in China (1980-86).
Regions *
Contribution to Total Rice
Production2
Contribution to Total Rice
Area
Yangzi River
South China
South West China
Yellow and Huir
69.9
15.7
7.6
6.0
68.1
17.7
7.8
6.0
1 Study sites includes in each region: Nanjing, Whuan and Changsha in Yangzi River
Region; Fuzhozi, Sshaoguan and Guangzou, in South China Region; Chengdou in South
West China; Xuzhou and Xiam in Yellow and Huir Region.
2 National Rice Production in China: 32,975,000 ha, 161,770,000 t, 4.92 t ha'1 (average
yield).
CHINA-15
-------
Table 2.
Annual temperature changes (°C) and precipitation differences (%) projected by the GISS,
GFDL and UKMO climate change scenarios for sites in southern China.
Temp. Diff. (°C)
Precip. Diff. (%)
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
GISS
4.1
4.1
3.5
6.0
6.0
4.7
3.5
6.0
3.5
GFDL
4.6
4.7
4.4
4.4
4.4
3.2
4.2
4.0
4.0
UKMO
6.5
6.5
5.7
6.5
5.7
6.7
5.7
3.3
3.4
GISS
43.7
45.2
17.8
-9.1
-7.3
-20.4
18.6
-7.5
79.9
GFDL
9.3
4.6
34.9
35.0
30.3
89.1
-11.4
6.8
9.3
UKMO
13.1
14.0
0.5
12.3
-0.6
34.0
1.6
26.2
31.8
CHINA-16
-------
Table 3.
Representative rice cultivars and soil types.
Region
Yellow & Huir
Yangzi R. Valley
South West
South China
Site
Xuzhou
Nanjing
Nanchan
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Latitude
34° 19' N
32°00' N
28°36' N
30°3T N
28°12'N
30°40' N
26° 05' N
24° 48' N
23° 08' N
Cultivar
Y.G.#2
S.U.#63
W.U.#64
Z.X.#26
Z.X.#26
G.C.#2
G.L.A.#4
S.U.G.#33
G.C.#2
Soil type
Fluoro-aquic
Yellow brown loam
Red earth
Yellow brown loam
Red earth
Yellow earth
Red earth
Lateric red earth
Lateric red earth
CHINA-17
-------
Table 4. Simulated yield changes (% of base yields) under GCM climate change scenarios without the
direct effects of CO2.
RAINFED
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
Y SD1
-20
-10
-23
-27
-27
-78
-21
-33
-45
-31
10
8
12
8
12
5
15
9
12
9
GFDL
Y
-33
-31
-5
-14
-6
-14
-19
-22
-27
-21
SD
10
8
14
9
13
2
6
9
4
7
UKMO
Y
-33
-23
-35
-28
-11
-28
-18
-8
-25
-21
SD
10
8
12
9
13
3
6
10
4
8
AUTOMATIC IRRIGATION
GISS
Y
-17
-15
-24
-22
-33
-32
-28
-25
-16
-24
SD
4
3
4
5
5
2
5
5
4
4
GFDL
Y
-20
-18
-19
-26
-37
-14
-25
-30
-21
-21
SD
4
3
4
5
5
2
5
5
4
4
UKMO
Y
-20
-19
-33
-26
-31
-20
-27
-26
-22
-24
SD
4
3
4
5
5
2
5
5
4
4
1 expressed as %
CHINA-18
-------
Table 5. Simulated yield changes (% of base yields) under GCM climate change scenarios with the
direct effects of CO2 included in the simulation.
RAINFED
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
Y SD1
4
20
2
-9
-2
-72
5
-14
-32
11
15
14
10
13
2
7
10
5
GFDL
Y
-13
-9
39
11
27
4
7
5
-6
SD
11
8
15
10
14
3
7
10
4
UKMO
Y
-12
3
-13
-6
21
-3
10
9
19
SD
11
8
13
10
14
4
7
10
4
AUTOMATIC IRRIGATION
GISS
Y
3
6
0
-3
-12
-6
-4
-3
3
SD
4
3
5
6
6
3
6
5
4
GFDL
Y
-2
2
4
-5
-15
3
0
-7
-2
SD
4
3
4
6
6
3
6
6
6
UKMO
Y
-2
1
-11
-5
-9
3
-3
-5
-5
SD
4
3
4
6
6
3
5
4
4
1 expressed as %
CHINA-19
-------
Table 6.
Season length changes (days??) with simulation of automatic irrigation.
330 ppm CO2
555 ppm CO2
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
-16
-22
0
-7
-8
-16
-12
-7
-13
-13
GFDL
-15
-20
-6
-14
-14
-15
-11
-17
-21
-16
UKMO
-17
-23
-6
-15
-13
-26
-15
-12
-16
-17
GISS
-15
-20
-1
-7
-8
-15
-11
-5
-11
-11
GFDL
-14
-19
-5
-13
-14
-14
-10
-15
-20
-15
UKMO
-16
-22
-5
-15
-13
-26
-14
-10
-15
-16
CHINA-20
-------
Table 7.
Simulated changes in evapotranspiration.
Percent ET Changes
RAINFED AUTOMATCIRRIG.
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
15
10
-1
0
3
-27
3
2
-13
-1
GFDL
0
-5
-10
-7
-2
-8
-6
-4
-8
-5
UKMO
8
5
-1
3
13
1
3
-7
-16
2
GISS
12
10
-5
12
8
25
0
21
7
10
GFDL
4
2
-18
-14
-17
-9
-7
-5
-5
-8
UKMO
14
12
-8
-3
-7
16
-4
-5
-6
1
CHINA-21
-------
Table 8.
Simulated changes in irrigation. The direct effects of CO2 were included in the simulation.
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
% Irrigation water
changes
GISS GFDL UKMO
-2
8
11
39
25
603
-16
69
194
4
14
-12
-41
-48
53
-9
-8
35
16
22
18
-6
-22
69
-22
-31
0
CHINA-22
-------
Table 9. Simulated yield changes for different adaptation strategies with the GISS climate change
scenario. The direct effects of CO2 were included in the climate change scenario simulation.
Strategy
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
ChengduH-
Fuzhou
Shaoguan
Guangzhou+
CC% CC+PD% CC+C% CC+PD+C%
3
6
-1
-3
-12
-72
-4
-3
-32
14
10
3
-1
-10
~
-2
5
-
-2
10
40
6
10
-54
29
-1
-47
0
11
43
21
25
-41
30
3
-36
+ Rainfed, the rest are automatic irrigation.
CC - Climate change effect (GISS scenario).
CC+PD - Effect of climate change plus change in the sowing
date.
CC+C - Eeffects of climate change plus change in the cultivar.
CC+PD+C - Combination effect of climate change, changes in
the sowing date and change
in cultivar.
CHINA-23
-------
Table 10. Crop Index indicators for baseline and GCM climate change scenarios. Initial possible date
for growing rice (Month/Day), number of possible days per year suitable for growing rice
(days), and index of more than 10 °C accumulated temperature (T).
BASE
Site
Harbin
Shenyan
Beijing
Jinan
Zhengzhou
Xuzhou
Nanjing
Chwngdu
Wuhan
Nanchang
Changsha
Fuzhou
Shaoguan
Guangzhou
Mean
(M/D)
5/14
4/24
4/10
4/8
4/5
4/9
4/5
3/22
3/29
3/30
3/26
3/9
3/13
2/26
4/1
Days
133
167
189
189
212
221
226
231
239
234
240
289
263
300
224
T
2848
3537
4216
4979
4822
4714
5076
5361
5433
5831
5731
6740
7011
7745
5289
GISS
(M/D)
4/24
4/14
3/30
3/24
3/20
3/22
3/14
2/27
3/7
3/10
3/9
2/15
2/10
2/4
3/13
Days
159
198
225
248
246
259
265
300
281
274
270
331
346
359
269
T
3696
4376
5215
6291
6185
6041
6565
7324
7634
7237
7905
8294
9465
9148
6812
GFDL
(M/D)
4/24
4/8
3/27
3/16
3/12
3/16
3/14
2/23
2/27
2/27
2/27
-
2/6
-
3/12
Days
191
204
233
261
240
258
254
281
276
300
278
365
326
365
274
T
3992
4772
5497
6469
6325
6274
6793
6761
7168
7722
7551
8839
8912
9558
6902
UKMO
(M/D)
4/9
2/24
3/9
3/3
3/8
3/8
2/25
2/9
2/26
2/13
2/18
-
2/25
2/3
2/27
Days
216
230
275
291
293
276
304
340
321
313
310
365
321
359
301
T
4815
5785
6353
7310
7197
7022
7598
8247
8038
8290
8103
9404
8470
9170
7557
CHINA-24
-------
CHINA
Figure 1. Map of China and location of sites selected for the simulation study.
-------
cd e£
e c
B
II
§ t"~-
X
S "»„
•o P
•° II
5*.
C 9
Si
« c
.2 «
H » a3
» t 3
u s fi
oo g U
"S ii ^
» ". rf
111
f a
r-< 00
I
E
-------
IMPLICATIONS OF CLIMATE CHANGE FOR JAPANESE AGRICULTURE:
EVALUATION BY SIMULATION OF RICE, WHEAT, AND MAIZE GROWTH
Hiroshi Seino
National Institute of Agro-Environmental Sciences
Japan
JAPAN-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Aims of the Study
Description of the Agricultural Regions and Systems
Food Trade and Vulnerabilities
METHODS
Baseline Climate Data and Climate Change Scenarios for the Region
Crop Models and Management Variables
Crop Model Validation
RESULTS AND DISCUSSION
Sensitivity Analysis
Crop Changes under GCM Climate Change Scenarios
Adaptation Strategies to Climate Change
IMPLICATIONS OF THE RESULTS
REFERENCES
JAPAN-2
-------
SUMMARY
This study used climate change scenarios derived from three General Circulation Models (GCMs) to
assess the possible impacts of climate change on rice, maize, and wheat production in Japan. Higher
temperature decreased simulated crop yields in many regions under the current management system. While
the direct beneficial effects of CO2 compensated for the yield decreases in central and northern Japan, they
did not compensate for the larger yield decreases in the southwestern part of the island, especially in Kyushu.
Early planting and irrigation are possible adaptation strategies to climate change. In most cases, simulated
yields increased under climate change conditions if an earlier planting date was adopted. However, in Kyushu,
because of high temperature stress, an earlier planting date did not improve simulated yields; the introduction
of new cultivars would be required. In Hokkaido, the major upland production area of Japan, climate change
increased simulated crop yields. However, these increases were dependent on the precipitation projections of
individual climate change scenarios and on the irrigation system practiced.
INTRODUCTION
Aims of the Study
The purpose of this study was to estimate changes in the yields of rice, wheat, and maize crops-in
both major production areas and vulnerable regions of Japan—due to climate change projected for increasing
levels of atmospheric greenhouse gases (IPCC 1990).
Description of the Agricultural Regions and Systems
The major crops cultivated in Japan are rice (2.11 million ha), wheat (0.40,million ha), potatoes (0.2
million ha), maize (0.33 million ha), and soybeans (0.16 million ha). The major rice production region is the
Tohoku district, located in the north area of the main island (Honshu). Two sites, Niigata and Miyagi, were
selected to represent this region (Figure 1). The production in Kyushu, iii southwestern Japan, is much lower
than in the northern area, but could be potentially affected by any increase in temperature as a result of global
warming. Therefore, a site, Miyazaki, was selected in this region for the simulation study. In Japan, most rice
fields are fully irrigated during the growing season. The soils in paddy fields are grey, lowland soils.
Major wheat production regions include Hokkaido (northern Japan), Tohoku (northern part of the
main island) and Kyushu (western Japan). In Hokkaido, wheat could be sown both in the spring and autumn.
In Tohoku and Kyushu, wheat is often planted in rice fields as a winter crop (i.e., sown in the autumn). The
sites selected to represent the main wheat-producing regions in this study are: Kitami (Hokkaido), Morioka
(Tohoku), and Fukuoka (Kyushu). While most wheat is rainfed, there are occasional, severe droughts that limit
production. The representative soils in these sites are grey lowland soils and andosols (volcanic ash soils).
Maize is primarily cultivated in Hokkaido (northern Japan), in central Japan, and in Kyushu. The sites
selected for this study are Obihiro (Hokkaido), Matsumoto (central Japan), and Miyakonojo (Kyushu). Most
of the maize is rainfed. The main soil type of these regions is andosols (volcanic ash soils).
Food Trade and Vulnerabilities
Japan is a net food importer (Appendix 1). Since Japan, depends on food imports, changes in world
crop yields and prices would directly affect domestic Japanese food prices.
JAPAN-3
-------
METHODS
Baseline Climate Data and Climate Change Scenarios for the Region
Climate data-daily maximum and minimum temperatures, precipitation, and solar radiation-from
1951 to 1988 were used as the baseline climate. Climate change scenarios were created based on the output
of three General Circulation Models (GCMs). GCMs provide the most advanced tool for assessing the
potential climatic consequences of increasing radiatively active trace gases. Climate models have been used to
simulate the equilibrium climate effects of an equivalent doubling of CO2. Some models have also been run
with gradually increasing trace gases to predict a set of transient responses. However, GCMs have not yet been
validated to project changes in climate variability, such as changes in the frequencies of drought and storms,
even though these could affect crop yields significantly. Therefore, the scenarios used in this study did not
include changes in variability.
Climate scenarios, which are sets of climatic perturbations, are commonly used to estimate the impacts
of potential climate changes on agricultural production. In this study, GCMs were used to derive climate
change scenarios for the crop models to estimate future changes in yield and other agronomically important
variables. The three climate change scenarios used were the Goddard Institute for Space Studies (GISS)
(Hansen el al. 1983); the Geophysical Fluid Dynamics Laboratory Model (GFDL) (Manabe and Wetherald
1987); and the United Kingdom Meteorological Office Model (UKMO) (Wilson and Mitchell 1987).
The characteristics of the climate change scenarios for Japan are summarized in Table 1. The GISS
model predicts a 2°C to 3.5°C rise in air temperature and a 1% to 5% increase in precipitation, except on the
southwestern islands. The precipitation increases are larger during the summer season (June-August). The
GFDL model predicts a 4°C rise in air temperature and a 10%-20% increase in precipitation, except on the
southwestern islands. The UKMO model predicts the highest temperature rise-3.5°C to 6°C. This model
predicts a decrease in precipitation in many regions, except during the summer season in eastern Hokkaido.
The solar radiation changes under the three GCM scenarios are less than 10%.
Crop Models and Management Variables
Crop models. The crop models used are the Crop Environment Resource Synthesis Models: CERES-
Rice (Ritchie et al. 1986), CERES-Maize (Ritchie et al. 1989), and CERES-Wheat (Ritchie and Otter 1985).
Soils and management. All soil data (type, initial soil water condition, etc.) were taken from soil survey
data (National Institute of Agricultural Science 1976). The organic carbon content, bulk density, and aluminum
saturation were estimated for each soil by the soil-data-retrieval program of the Decision Support System for
Agrotechnology Transfer (DSSAT). The planting date, row spacing, plant population, fertilizer, and cultivar
were taken from published data.
Irrigation. In Japan, most rice is well irrigated, and therefore, the option for automatic irrigation in
the crop model was chosen to provide the simulated crop with a nonlimiting water supply. Maize and wheat
are grown under rainfed conditions in all locations. Irrigation was tested as a possible adaptation to climate
change for these crops.
Crop Model Validation
The crop models were validated at nine locations by comparing experimental field data to simulated
crop yields and season lengths (Table 2). The simulated data correspond well with experimental observed data,
suggesting that the CERES models can be used for the simulation of rice, wheat, and maize growth in Japan.
JAPAN-4
-------
RESULTS AND DISCUSSION
Sensitivity Analysis
Arbitrary scenarios were created by applying changes in temperature and precipitation to the baseline
climate series. The physiological effects of 555 ppm of CO2 were considered in each of the scenarios. Tables
3, 4, and 5 show the changes in yield and season length for the three crops under the scenarios.
Rice. Because most rice fields are fully irrigated, the automatic irrigation option was selected in the
CERES model. Therefore, changes in precipitation did not affect rice. Rice yields decreased with a 2°C to 4°C
increase in temperature, and the length of the season from planting to maturity also decreased. With the
physiological effects of CO^ the simulated yield increased in each scenario. These increases compensated for
the negative effects of a 2°C increase in temperature and resulted in a net yield increase (6% to 10%) in
comparison with the baseline yields. However, a 4°C increase resulted in decreases in rice yields, even when
the physiological effects of 555 ppm of CO2 were taken into account.
Maize. In Japan, maize is grown under rainfed conditions. The effects of temperature increases on
maize production (Table 4) were quite different from those on rice production. Maize yields at Obihiro (in
Hokkaido) increased under a +2°C scenario, and even under a +4°C scenario there was a small increase. The
yield increases were intensified both by increases in precipitation and by the additional physiological effects
of CO2. These results suggest that current low temperatures limit the growth of the variety cultivated in
Obihiro, and that a moderate warming would have beneficial effects on maize production in that region.
However, associated precipitation (especially during the growing season) is an important factor in determining
the crop's final yield.
Wheat. Most wheat in Japan is cultivated under rainfed conditions. Wheat yields decreased under the
+2°C and +4°C scenarios at all locations. However, a precipitation increase combined with the higher
temperature scenarios produced increases in yields.
Crop Changes under GCM Climate Change Scenarios
The climate variables under each GCM are different, and therefore, the effects of climate change on
crop production depend strongly on GCM climate change scenario.
Rice. The rice yields at the three sites decreased under the 2xCO2 climate change scenarios with
climate change alone (Table 6). Under the GISS scenario, the physiological effects of CO2 compensated for
the negative effects of climate change alone at the three sites. However, under the GFDL scenario in Miyagi
and Niigata, and the UKMO scenario in Miyazaki, rice yields remained below the baseline yield, even with the
direct effects of CO2. At all locations, the length of the growing season was shortened by the temperature rise
associated with the climate change scenarios. The UKMO scenario produced the shortest season length of rice.
In fully irrigated paddy fields, there are very small differences among rice yields at each site. The direct
effects of CO2 increased simulated rice yields up to 20%. In rainfed upland fields, the direct effects of an
increased concentration of CO2 depended on the cultivar used and the site selected.
Maize. Table 7 shows changes in maize yields under the climate change scenarios considered. Maize
yields in Obihiro (Hokkaido) greatly increased under the GISS and GFDL scenario conditions, but
dramatically decreased under the UKMO scenario due to the lack of rain during the summer season. The
yields in Matsumoto (Kanto) decreased under climate change scenarios alone, but recovered if the direct
effects of CO2 were included in the simulation. In contrast, in Miyakonojo (Kyushu), maize was subject to high
temperature stress under all scenarios, and the yield decreases were not compensated when the the direct CO2
effects were included in the climate change simulation. Maize season length was shortened everywhere due to
the temperature increases associated with the GCM scenarios.
JAPAN-5
-------
t The results of the simulation of wheat under the GCM climate change scenarios (Table 8) were
similar to those of maize. The severe yield decreases under the UKMO scenario were caused by precipitation
decreases in the summer season in eastern Hokkaido. The yield decreases in Morioka and Fukuoka were not
offset by the direct CO2 effects.
Adaptation Strategies to Climate Change
In this study, early planting dates (15 and 30 days earlier) and irrigation were analyzed as possible
adaptation strategies to counteract the negative effects of climate change. The amounts of irrigation needed
to maintain soil moisture were calculated with the water balance submodel included in the crop models. In
the case of rice, we analyzed only the effect of early planting. We did not test irrigation, because most of the
rice production currently occurs without water restrictions, and it was assumed that water would be available
for irrigation in the future.
Table 9 shows that early planting was a very effective method of increasing rice yields under climate
change in northern Japan, but not effective in southern Japan (e.g., Miyazaki in Kyushu). Maize yields in
Obihiro (Hokkaido) increased with an early planting date and full irrigation, while the same practices did not
recover yield decreases in the Matsumoto region. In Miyakonojo (Kyushu), early planting improved yields;
however, irrigation did not have any effect since water is not limiting during the maize growing season at that
site (both in the current climate and in the climate change scenarios). We considered irrigation and early
planting date for spring wheat and late planting date for winter wheat. These strategies improved yields for
spring wheat in Kitami (Hokkaido), but they were not effective for winter wheat yields in Morioka and
Fukuoka.
IMPLICATIONS OF THE RESULTS
The results of this study suggest that climate change could affect regional agricultural production in
Japan. Increases in temperature decreased crop yields in many regions in the simulations with the present
management system. While the direct beneficial effects of CO2 compensated for yield decreases in central and
northern Japan, they did not compensate for the larger yield decreases in southwestern Japan, especially in
Kyushu. These results imply shifts in regional production patterns.
Early planting and irrigation are possible adaptation strategies to climate change in Japan. In most
cases, simulated rice, wheat, and maize yields increased under climate change conditions if an earlier planting
date was adopted. However, in Kyushu, because of high temperatures, an earlier planting did not improve
simulated yields, and the introduction of new cultivars better adapted to the climate change conditions would
be required. In Hokkaido, the major upland production area of Japan, climate change increased simulated crop
yields. However, the simulated yield increases are dependent on the scenario's precipitation level and on
regional irrigation systems. Potential future water availability under changed climate needs to be investigated
to improve agricultural projections.
JAPAN-6
-------
REFERENCES
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A user's guide to Ceres-rice. International
Fertilizer Development Center. Muscle Shoals, AL.
Godwin, D.C., U. Singh, R.T. Buresh, and S.K. DeDadda. 1990. Modeling of nitrogen dynamics in relation to
rice growth and yield. In 14th International Soil Science Congress. Kyoto, Japan.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R Ruedy, and L. Travis. 1983. Efficient three-
dimensional global models for climate studies. Model I and II. Monthly Weather Reviews., 3, 609-662.
Hansen, J., I. Fung, A. Lacis, D. Rind, S. Lebedeff, R. Ruedy, and G. Russell. 1988. Global climate changes
as forecasted by Goddard Institute for Space Studies three-dimensional model. /. Geophy. Res.,
93,9341-9364.
Horie, T. 1987. The effects on rice yields in Hokkaido. In The Impact of Climatic Variations on Agriculture. Vol.
1 Assessments in Cool Temperate and Cold Regions. M.L. Parry, T.R. Carter, and N.T. Konijn eds.809-
825.
IPCC. 1990. Intergovernmental Panel on Climate Change. 1990. First Assessment Report
Jager, J., and W.W. Kellog. 1983. Anomalies in temperature and rainfall during warm arctic seasons. Climate
Change, 5,39-60.
Jones, C.A., and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development.
College Station, TX: Texas A&M University Press.
Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
14. In: Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer,
Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.
Manabe, S., and R.T. Wetherald. 1987. Large-scale changes in soil wetness induced by an increase in carbon
dioxide. /. Atmos. ScL, 44,1211-1235.
Manabe, S., and R.T. Wetherald. 1986. Reduction in summer soil wetness induced by an increase in
atmospheric carbon dioxide. Science., 232,626-628.
Newman, J.E. 1980. Climate change impacts on the growing season of the North American corn belt. Biomet.,
7, Part 2, 128-142.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
model. In: Willis, W.O. ed. ARS Wheat Yield Project. Washington, DC: U.S. Department of
Agriculture, Agriculture Research Service, ARS-38,159-175.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
model. In W.O. Willis (ed.). ARS Wheat Yield Project. ARS-38. US Department of Agriculture,
JAPAN-7
-------
Agricultural Research Service. Washington, DC.
Ritchie J., U. Singh, D. Godwin, and L. Hunt. 1989. A user's guide to Ceres Maize- v2.10. International
Fertilizer Development Center. pp86.
Rosenzweig, C. 1985. Potential CCyinduced climate effects on North American wheat-producing regions.
Climate Change., 7,367-389.
Uchijima, T. 1987. The effects on altitudinal shift of rice yield and cultivable area in Northern Japan. In M.L.
Parry, T.R. Carter and N.T. Konijn eds. The Impact of Climatic Variations on Agriculture. Vol. 1
Assessments in Cool Temperate and Cold Regions. 797-808.
Uchijima, Z., and H. Seino. 1988. Probable effects of CCyinduced climatic change on agroclimatic resources
and net primary productivity in Japan. Bull Natl Inst. Agro-Environ. ScL, 4,67-88.
Wigley, T.M.L., P.D. Jones, and P.M. Kelley. 1980. Scenarios for a warm, high CO2 world. Nature, 283,17-21.
Willson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 climate sensitivity experiment with a global model
including a simple ocean. /. Geophy. Res., 92,13315-13343.
JAPAN-8
-------
Table 1. Average seasonal change in temperature and precipitation for the GCM climate change
scenarios.
SPRING SUMMER
(MAM) (JJA)
FALL WINTER ANNUAL
(SON) (DJF)
Temperature changes (°C)
Myyazaki, Fukuoka, Miyakonojo
GISS
GFDL
UKMO
Matsumoto, Morioka,
GISS
GFDL
UKMO
Obihiro, Kitami
GISS
GFDL
UKMO
4.0
4.6
4.1
Miyagi,
3.2
4.5
7.0
4.3
4.2
5.8
3.8
3.6
5.2
Niigata
2.9
3.7
4.8
3.1
3.5
7.0
3.3
4.1
4.3
2.8
4.4
4.9
3.1
5.4
6.9
3.0
4.6
3.7
2.3
4.0
5.3
3.5
4.2
5.0
3.6
4.2
4.3
2.8
4.2
5.5
3.5
4.3
6.2
Precipitation Changes (%)
Myyazaki, Fukuoka, Miyakonojo
GISS
GFDL
UKMO
Matsumoto, Morioka,
GISS
GFDL
UKMO
Obihiro, Kitami
GISS
GFDL
UKMO
15
7
-25
Miyagi,
7
19
2
12
37
-8
-6
18
35
Niigata
-4
14
7
-4
38
-18
1
5
15
22
24
-23
9
1
9
-5
9
-8
3
7
-13
2
1
7
1
10
4
7
16
-7
5
19
-3
JAPAN-9
-------
Table 2.
Simulated and observed yields and season lengths.
S. Length (d)
Crop
Rice
Rice
Rice
Maize
Maize
Maize
Wheat
Wheat
Wheat
Site
Miyagi
Niigata
Miyazaki
Obihiro
Matsumoto
Miyakonojo
Kitami
Morioka
Fukuoka
Data
(year)
75-87
80-88
80-88
76-86
68-82
70-82
67-83
70-86
72-87
DBS
145.8
134.7
118.0
136.3
125.6
114.3
103.3
288.1
196.8
SIM
143.1
136.8
119.8
134.0
127.3
113.5
102.8
285.6
191.6
Yield (t ha'1)
OBS
4.95
5.74
5.64
11.16*
16.64*
3.95
2.59
3.64
3.83
SIM
4.99
5.34
5.49
11.33*
16.67*
3.82
2.53
3.18
3.81
* Total biomass
JAPAN-10
-------
Table 3. Sensitivity of simulated rice yields and season length to temperature and CO2. Rice was
simulated under the automatic irrigation option of the CERES model.
330 ppm CO2
555 ppm CO2
Site
Miyagi
Niigata
Miyazaki
Temp.
Change
(°C)
+2
+4
+2
+4
+2
+4
Yield
Change (%)
-13
-22
-9
-16
-9
-16
S. Length
(d)
-15
-23
-12
-12
-7
-12
Yield
Change (%)
+6
-5
+10
+1
+9
+1
S. Length
(d)
-14
-22
-11
-17
-7
-12
Yield = percent changes from base; Season Length = differences from base.
JAPAN-11
-------
Table 4. Sensitivity of simulated maize yield, season length and evapotranspiration to temperature
(°C), precipitation (%), and CO2 changes.
Site
Obihiro
Matsumoto
Miyakonojo
T
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
PP
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
330 ppm
Yield
+43
+17
+34
+48
+8
+21
-8
+3
-12
-20
-27
-3
-31
-14
-18
+ 11
-20
-32
-20
-19
-32
-31
-1
+1
C02
SL
-4
-19
-4
-4
-19
-19
0
0
-13
-18
-15
-12
-19
-18
-4
+1
-7
-13
-7
-7
-13
-13
0
0
ET
+1
-10
-3
+3
-14
-7
-4
+2
-8
-10
-13
-4
-15
-6
-7
+4
-7
-12
-7
-6
-13
-11
-1
+1
555
Yield
+63
+35
+60
+66
+30
+37
+11
+14
+8
-5
-3
+15
-12
-1
+11
+36
-15
-27
-15
-15
-27
-27
+6
+6
ppm
SL
-4
-19
-4
-4
-19
-19
0
0
-12
-18
-13
-11
-19
-18
0
+3
-7
-13
-7
-7
-13
-13
0
0
CO2
ET
-14
-22
-17
-13
-24
-20
-18
-14
-17
-19
-21
-14
-22
-16
-14
-8
-19
-24
-20
-19
-24
-23
-14
-13
xicia — percent changes from base; SL = season length differences from base; ET = percent
evapotranspiration from base.
JAPAN-12
-------
Table 5.
Sensitivity of simulated maize yield, season length and evapotranspiration to temperature
(°C), precipitation (%) and CO2 changes.
330 ppm CO2
555 ppm CO2
Site
Kitami
(SW)
Morioka
(WW)
Fukuoka
(WW)
T
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
PP
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
Yield
(%)
-17
-30
-27
-12
-38
-25
-10
+4
-7
-15
-10
-7
-18
-15
-4
+1
-14
-27
-18
-11
-31
-24
-4
+2
SL
(d)
-8
-14
-8
-8
-14
-14
0
0
-4
-7
-4
-4
-7
-7
0
0
-3
-5
-3
-3
-5
-5
0
0
ET
(%)
-5
-9
-12
+1
-17
-4
-7
+5
+2
+4
0
+4
+1
+5
-2
+1
-2
-2
-5
0
-6
0
-3
+2
Yield
(%)
+10
-7
0
+14
-14
-1
+22
+31
+9
0
+8
+9
-1
0
+16
+18
+3
-13
-1
+5
-17
-10
+15
+22
SL
(d)
-8
-14
-8
-8
-14
-14
0
0
-4
-7
-4
-4
-7
-7
0
+3
-3
-5
-3
-3
-5
-5
0
0
ET
(%)
-9
-13
-16
-5
-19
-8
-11
-2
-5
-2
-7
-5
-5
-1
-10
-9
-8
-7
-11
-7
-11
-6
-9
-5
Yield = percent changes from base; SL = season length differences from base; ET = percent
evapotranspiration from base.
JAPAN-13
-------
Table 6. Changes in rice yield, season length and total season precipitation under GCM climate change
scenarios.
Site
Miyagi
Niigata
Miyazaki
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
330 ppm
Yield
-12
-27
-17
-7
-21
-11
-12
-17
-19
CO2
SL
(d)
-19
-22
-25
-15
-18
-21
-12
-12
-14
PP
-19
-6
-24
-16
0
-21
+5
-3
-5
555
Yield
+7
-11
+1
+12
-5
+6
+6
0
-2
ppm
SL
(d)
-18
-22
-25
-15
-18
-21
-12
-12
-14
CO2
PP
-19
-5
-24
-16
+1
-21
+5
-3
-5
Yield = percent changes from base; SL= season length differences from base; PP= recipitation percent from
base.
JAPAN-14
-------
Table 7. Change in maize yield, season length, evapotranspiration and total season precipitation under
GCM climate change scenarios.
330 ppm CO2
555 ppm CO2
Site
Obihiro
Matsumoto
Miyakonojo
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
Yield
+24
+25
-32
-19
-14
-15
-28
-34
-31
SL
(d)
-15
-16
-29
-17
-17
-20
-12
-13
-15
PP
-16
+9
-45
-19
+ 18
-17
+6
-4
-11
ET
-2
-9
-16
-7
-15
-7
-11
-16
-8
Yield
+51
+38
+5
+3
-5
-2
-23
-31
-27
SL
(d)
-14
-16
-29
-16
-17
-21
-12
-13
-15
PP
-16
+9
-45
-18
+18
-17
+6
-4
-11
ET
-15
-23
-22
-16
-26
-16
-23
-28
-20
Yield = percent changes from base; SL= season length differences from base; ET= percent cahnge of total
season evapotranspiration PP= recipitation percent from base.
JAPAN-15
-------
Table 8. Change in wheat yield, season length, evapotranspiration and total season precipitation under
GCM climate change scenarios.
•330 ppm CO2
555 ppm CO2
Site
Kitami
Morioka
Fukuoka
-.,...
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
Yield
-21
-17
-59
-16
-20
-31
-24
-33
-40
SL
(d)
-11
-12
-21
-7
-7
-12
-8
-5
-5
PP
-18
+8
-43
-1
+2
-23
-8
-6
-28
ET
-7
-5
-22
+5
+3
+6
-4
0
-2
Yield
+6
+8
-41
-1
-7
-18
-8
-20
-27
SL
(d)
-11
-12
-21
-7
-7
-12
-8
-5
-5
PP
-18
+8
-45
-1
+2
-23
-8
-6
-28
ET
-10
-11
-23
-2
-3
0
-9
-4
-7
1= percent changes from base; SL= season length differences from base; ET= percent cahnge of total
season evapotranspiration PP= recipitation percent from base.
JAPAN-16
-------
Table 9.
Evaluation of potential adaptation strategies: changed planting dates and levels of irrigation.
Miyagi
Niigata
Miyazaki
Obihiro
Matsumoto
Miyakonojo
Kitami
Morioka
Fukuoka
GCM Rainfed Planting
RICE
GFDL irrigated 15 d. early
irrigated 30 d. early
GFDL irrigated 15 d. early
irrigated 30 d. early
UKMO irrigated 15 d. early
irrigated 30 d. early
MAIZE
UKMO rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
GFDL rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
GFDL rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
WHEAT
UKMO rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
UKMO rainfed 15 d. late
rainfed 30 d. late
irrigated 15 d. late
irrigated 30 d. late
UKMO rainfed 15 d. late
rainfed 30 d. late
irrigated 15 d. late
irrigated 30 d. late
Base
Yield
4.30
5.44
5.56
5.81
4.87
2.64
1.60
2.69
2.69
2.69
2.69
2.94
2.94
2.94
2.94
Scenario Yield
with Adaptation
4.62
4.90
5.72
5.99
5.60
5.68
5.82
6.02
8.37
8.64
4.61
4.35
4.64
4.36
3.05
3.38
3.05
3.38
2.02
2.22
3.14
3.36
2.69
2.68
2.71
2.70
2.88
2.80
2.88
2.80
amoun
114.68
116.34
38.39
38.32
4.18
6.24
169.29
167.05
74.49
74.16
2.24
2.24
JAPAN-17
-------
Appendix 1. Food
Commodity
Cereals
Rice
Wheat
Barley
Naked Barley
Sweet corn
Sweet potatoes
Potatoes
Starches
Pulses
Vegetables
Fruits
Meat
Hen eggs
Milk&milk products
Fish&Shellfish
balance in Japan in
Domestic
Production
11,870
10,627
864
326
27
1
1,423
3,955
2,357
469
16,598
5,974
3,607
2,394
7,427
11,800
1987 (x 1000 t).
Imports
28,187
39
5,133
1,988
0
16,602
0
323
119
4,797
1,114
2,260
1,171
36
1,767
3,299
Exports
0
0
0
0
0
0
0
0
0
0
4
48
4
0
0
1,583
JAPAN-18
-------
JAPAN
Figure 1. Map of Japan and location of the sites selected for the study.
-------
-------
SECTION 7: AUSTRALIA
-------
-------
POSSIBLE EFFECTS OF GLOBAL CLIMATE CHANGE
ON WHEAT AND RICE PRODUCTION IN AUSTRALIA
Brian D. Baer
Department of Crop and Soil Science
Michigan State University, USA
Wayne S. Meyer
Division of Water Resources
Griffith Laboratory
CSIRO, Australia
David Erskine
Division of Water Resources
Griffith Laboratory
CSIRO, Australia
AUSTRALIA-1
-------
TABLE OF CONTENTS
SUMMARY
INTRODUCTION
Aims
Sites and Current Climate
METHODS
Climate Change Scenarios
Crop Models and Management Practices
Calibration and Validation of the Crop Models
Simulations
RESULTS
Crop Yield and Irrigation Demand
Sensitivity Analysis
Adaptation to Climate Change
DISCUSSION
Limitations of the Crop Models
Impact of Climate Change on Crops
REFERENCES
AUSTRALIA-2
-------
SUMMARY
This simulation study uses crop models and climate change scenarios generated from General
Circulation Models (GCMs) to determine the possible impacts of climate change on rice and wheat production
in Australia. The beneficial direct effects of CO2 on crop yield and water use are taken into account in the
simulations. In most sites, dryland wheat yields increase when the scenario projects a rainfall increase.
However, in the scenarios with the largest GCM temperature increases, yields generally decreased due to a
shortening of the crop growing season. Irrigated wheat yields also decreased due to the temperature increases.
Paddy rice yields decreased slightly under climate change conditions.
Simulation experiments showed that the most successful adaptation strategy to climate change was
changing the variety of rice to a more tropical one. Also, adjusting the sowing dates for dryland wheat to
obtain maximum water availability was helpful.
INTRODUCTION
Aims
The purpose of this study is to estimate the possible impacts of climate change on yield and irrigation
water demand of wheat and rice in different sites in Australia using simulation crop growth models and climate
change scenarios generated from GCMs.
Sites and Current Climate
Wheat is grown in the semi-arid regions of southern and eastern Australia, while rice is mainly
confined to inland southern New South Wales (NSW) (Figure 1). The main constraint on the extent of these
agricultural regions is water availability. In the southern part of the country, the rainfall pattern has a winter
maximum; in the middle regions of the eastern wheat belt, rainfall is evenly distributed throughout the year.
Winter rainfall is stored in the soil and is available to the wheat crop during the following winter until early
summer. In the north of Australia, the dominance of summer rainfall and the limited areas of suitable soils
restrict agriculture. Winter temperatures in the mainland are mild and do not present any constraint to crop
growth.
Five sites were chosen to represent the major wheat regions of Australia (Figure 1). Wheat is grown
under rainfed conditions in most areas; only in the Murrumbidgee Irrigation Area, represented by Griffith in
NSW, is wheat grown under irrigated conditions. Griffith is also central to the major rice-producing area of
southern New South Wales. The country's aridity and strong commitment of available water supplies to diverse
users make it unlikely that other grain-growing regions of the country will be able to acquire water for
irrigation in the future.
METHODS
Climate Change Scenarios
A series of climate data available for the period 1951-80 was obtained for each site (Wongan Hills
1966-80; Roseworthy 1957-80; Horsham 1957-59, 1961-80; Griffith 1951-80; Narrabi 1962, 1964-66, 1968-69,
1971-80). Daily maximum temperature, minimum temperature, and rainfall were available for all sites. Missing
AUSTRALIA-3
-------
values were estimated based on the values before and after that day. Daily solar irradiance data was available
for the Griffith site only. For the rest of the sites, we generated values based on maximum possible irradiation
modified with observed monthly average hours of sunlight (Richardson and Wright 1984).
Climate change scenarios for each site were generated from three GCMs: the Goddard Institute for
Space Studies Model (GISS) (Hansen et al. 1983), the Geophysical Fluid Dynamics Laboratory Model (GFDL)
(Manabe and Wetherald 1987), and the United Kingdom Meteorological Office Model (UKMO) (Wilson and
Mitchell 1987). The scenarios for each site were created based on the GCM output of climate variables applied
to the daily observed climate. This method uses the difference between lxCO2 and 2xCO2 monthly GCM
temperatures and the ratio between 2xCO2 and lxCO2 monthly GCM precipitation amounts. (lxCO2 refers
to current climate conditions, and 2xCO2 refers to the climate that would occur with a doubling of greenhouse
gases.) Table 1 shows the seasonal changes projected under the climate change scenarios. All GCMs project
temperature increases for all seasons. The GISS and GFDL models predict an average annual temperature
rise of around 4.5°C, and the UKMO model predicts temperature increases of approximately 5.7°C. The GISS
and UKMO models predict a slight increase in annual rainfall, while GFDL predicts a slight decrease. In most
sites and seasons, the three GCMs predict a slight increase in solar radiation.
Crop Models and Management Practices
The CERES-Rice (Godwin et al. 1992) and the CERES-Wheat (Godwin et al. 1989) models were used
for the crop simulations. In addition to the climate data, the crop models require a well-defined set of inputs
to simulate actual crop conditions. The inputs include soil parameters and management practices. The soil
parameters include data on texture, water-holding capacity, and nitrogen status of the soil. A representative
soil was determined and defined for each of the five sites. In general, the representative soils chosen were clay
loam soils, with the exception of a heavy clay soil for the rice growing area, a sandy loam soil for western
Australia, and a cracking clay soil for Narrabri. Information on soils in the wheat-growing areas was taken
from McGarity (1975). The crop management practices include sowing dates, plant population, irrigation,
nitrogen applications, and plant variety. These variables were determined according to current practices.
Wheat is grown in the winter at all sites to take advantage of the low evaporation and to maximize
available water. We simulated nonirrigated wheat with low inputs of nitrogen and relatively low plant
populations at all sites. In addition, we simulated irrigated wheat and paddy rice in Griffith. Table 2 shows the
conditions that were used for the crop simulations.
Calibration and Validation of the Crop Models
We determined the genetic coefficients that define a variety in the CERES models by comparing
model runs to actual field data and calibrating the coefficients by the method of trial and error. After defining
the genetic coefficients, the models were validated for use in the selected sites.
General data from a wide range of dryland wheat experiments in Australia were obtained from the
data base of Rimmington et al (1987) and were used to validate the wheat model for Roseworthy. For eastern
Australia, validation data were compiled from the data sets of MacKenzie et al. (1985) and supplemented with
more specific information from Angus et al. (1980) for an area similar to Narrabri. For Griffith, data for the
validation were taken from Mason and Fischer (1986); for Horsham, from Rimmington et al. (1987); and for
Wongan Hills, from Anderson and Smith (1990). Validation data for rice were obtained from the
comprehensive experiments of Muirhead et al (1989) and Humphreys et al. (1987). For irrigated wheat, local
data published by Meyer et al (1985) were used for validation.
Simulations
AUSTRALIA-4
-------
The simulations include the following sets of scenarios: (a) baseline climate (in all sites); (b) GCM
climate change scenarios alone (in all sites); (c) GCM climate change scenarios including the direct effects of
555 ppm CO2 in the crop growth simulation (in all sites) (Acock and Allen 1985); (d) sensitivity analysis,
where base daily temperature and precipitation were modified by fixed amounts (in Roseworthy and Griffith);
and (e) scenarios that incorporated possible changes in management conditions to analyze adaptive strategies
to climate changes. For dryland wheat at Horsham and Roseworthy we changed the current planting dates;
for irrigated rice, we replaced the current cold-resistent variety, Calrose, with a more tropical variety, IR-36.
RESULTS
Crop Yield and Irrigation Demand
Table 3 shows the yields (and standard deviations) and irrigation amounts simulated by the CERES
models for dryland wheat and irrigated wheat and rice for the baseline climate and the GCM climate change
scenarios. The direct effects of CO2, as well as the climate effects alone, are shown.
Crop Yield. Climate change scenarios had varied effects on wheat yields at each of the sites (Table 3).
While the UKMO model showed a decrease in yields in all cases but one, the other two models showed no
general trend. From the climate effects alone, the yield changes ranged from a 15% increase in Horsham under
the GFDL model to a 45% decrease in Wongan Hills under the GISS scenario. The direct effects of CO2
increased the yield under base climate from 25% to 39%. The beneficial direct CO2 effects on yield fully
compensate for the yield losses under the climate change scenarios in most cases.
Table 3 also shows the effects of climate change scenarios on irrigated wheat at Griffith. Simulated
yields decreased due to increased temperatures and a shortening of the growing season. The greatest yield
losses corresponded to the scenario that projects the largest increases in temperature. In Griffith, the
temperature increase was lowest under the GISS scenario and largest under the UKMO scenario. Simulated
wheat yields decreased 17% and 45% under the GISS and UKMO scenarios, respectively. With the direct
effects of CO2, the corresponding yield decreases were only 2% and 29%.
Flooded rice (Table 3) was much less susceptible to temperature increases than was wheat. Under
climate change alone, yields decreased from 9% (GISS) to 16% (UKMO). The direct CO2 effects compensated
for the rice yield decreases under the climate change scenarios alone.
Irrigation Demand. Two irrigated crops were simulated at Griffith. The irrigation needed for rice
increased or, in the case of GISS, did not significantly change (Table 3). However, the irrigation demand for
wheat showed an opposite trend under the climate scenarios, showing a decrease of more than 50% in the
UKMO scenario due to a shorter growing period. In all cases, irrigation demands were smaller with the
physiological CO2 effects because of increased stomatal resistance.
Sensitivity Analysis
Table 4 shows the results of the sensitivity analysis for dryland wheat in Roseworthy. Changes in yields
due to temperatures are relatively small in comparison with changes due to precipitation. There is an increase
in yields with a 2°C temperature increase, but a decrease from the baseline yields with a 4°C temperature
increase.
Adaptation to Climate Change
Table 5 shows the effect of changing planting dates on dryland wheat yields under climate change
scenarios at two sites. Planting dates of 15 and 30 days before and after May 30 (the most common planting
AUSTRALIA-5
-------
date) were chosen. Yields were higher with an earlier sowing date and they decreased with later planting dates.
Table 5 also shows the results of the flooded rice model runs at Griffith using variety IR-36. According to the
CERES model, this variety of rice produced higher average yields than Calrose (the current cultivar), even in
the current climate. However, the standard deviation of the current simulated yield increased significantly.
Under the climate change scenarios, yields increased substantially when using the variety IR-36, but without
an increase in standard deviation.
DISCUSSION
Limitations of the Crop Models
For this study the simulations lacked two components that might have affected the outcome. First,
the soil moisture was reset at the beginning of every season rather than being simulated for the entire year.
Since there is not enough water to supply the demands of the dryland wheat from rainfall during the growing
season, the water stored in the soil is critical to the development of the plants. The stored water comes from
rain that fell during earlier months when there were few or no plants growing. In this study we partially
compensated for this by starting the simulation two months before the sowing date. However, a full-year water
balance would have given more accurate predictions of water in the soil during the growing season. This
procedure might possibly underestimate the impacts of climate change on soil moisture.
Second, the rice model is not sensitive to cool temperatures. The rice model was mainly developed
with data from tropical areas where the temperatures are warm throughout the growing season. However, in
less tropical areas, such as southern Australia, there can be fairly cold nights during critical stages of the rice
growing cycle. These cold temperatures can induce sterility in the plants and reduce the yield substantially. The
CERES-Rice model does not account for this problem, and therefore, it probably overestimates yields in years
with cold nights. This effect should be tested in future climate change impact studies.
Impact of Climate Change on Crops
Dryland wheat yields in Australia increased when the scenario projected an increase in the amount
of rainfall (compared to the current rainfall). The sensitivity analysis at Roseworthy clearly shows the
importance of this variable. In most Australian regions, the water stored in the soil prior to sowing provides
a large percentage of the water needed by the crop, so the extra rainfall in the summer months contributes
directly to increases in yields. In contrast, at Wongan Hills, under the GFDL scenario yields decrease
substantially (about 30%) with the effects of climate change alone, due in part to projected decreases in
rainfall.
However, in the situations where the GCMs predict large temperature increases, the effect of rainfall
increases can be overcome by the effects of higher temperatures. In the cases of irrigated wheat and flooded
rice crops this high temperature effect is the only one to which the models are sensitive. Warmer temperatures
during the growing season decrease the length of time that the plant takes to reach maturity. This is
particularly critical during the grain-filling period, the duration of which is determined by accumulated
temperature. An increase in temperature stimulates the plants to mature more quickly, but with less grain
yield. Figure 2 shows the effect of temperature on the length of the growing season on simulated irrigated
wheat at Griffith.
An increase in atmospheric CO2 results in larger yields due to greater net photosynthesis and better
use of water by the plant due to increased stomatal resistance. In the plants that are not suffering from
drought stress these effects are particularly pronounced.
The rainfall changes associated with the climate change scenarios influenced the amount of irrigation
AUSTRALIA-6
-------
required for the crop. For irrigated wheat, the decrease in water demand is related to the shorter growing
season and smaller biomass accumulation. The plants that do poorly require less water, and therefore, less
irrigation. Rice, however, showed less of a decrease in yield and therefore, the water demand did not decrease
as much as in the case of wheat.
A change in management practices may compensate for the negative impact of climate change on
yields, as simulated in this study. In the case of dryland wheat, it is critical to grow the crop when there is
water available, and therefore, sowing dates may be adjusted to the most favorable moisture regime under the
climate change conditions. Changing the variety of rice may be a successful strategy for adaptation to climate
change.
Australia's farmers produce more than enough wheat and rice for internal consumption, so that none
of the predicted changes would threaten the food security of the country. However, farm products are major
export commodities, and yield variations could have large effects on farm export earnings, leading to significant
impacts on the national economy.
AUSTRALIA-?
-------
REFERENCES
Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
Department of Energy. Washington, D.C. pp. 53-97.
Anderson, W.K., and W.R. Smith. 1990. Field advantage of two semi-dwarf compared to two tall wheats
depends on sowing time. Aust. J. Agric. Res. 41, 811-826.
Angus, J.F., H.A. Nix, J.S. Russell, and J.E. Krruizinga. 1980. Water use, growth and yield of wheat in a
subtropical environment. Aust. J. Agric. Res. 31, 873-886.
Godwin, D., J. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES Wheat-V2.10. International
Fertilizer Development Center. Muscle Shoals. AL
Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A User's Guide to CERES-Rice. International
Fertilizer Development Center. Muscle Shoals. AL
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
Review, Vol III, No. 4: 609-662.
Humphreys, K, W.A. Muirheas, F.M. Melhuish, and R.J.G. White. 1987. Effects of time of urea application
on combine sown calrose rice in south east Australia. Aust. J. Agric. Res. 38, 101-112.
McGarity, J.W. 1975. Soils of the Australian wheat-growing areas. In: Australian Field Corps 1: Wheat and
Other Temperate Cereals. Eds. Lazenby, A. and Matheson, E.M. P 227-255. Angus and Robertson
Sydney.
MacKenzie, D.H., M.E. Robertson, M.F. Hutchinton, M.P. O'Connor, J.P. McMahon, and H.A. Nix. 1985.
Matched sets of crop, soil, and weather data from wheat lands in eastern Australia. Report CSIRO,
Division of Water Resources, Camberra.
Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Mason, I.E., and R.A. Fischer. 1986. Tillage practices and the growth and yield of wheat in southern New
South Wales: Lockhart, in a 450 mm rainfall region. Aust. J. Exp. Agric. 26, 457-468.
Meyer, W.S., F.X. Dunin, R.C.G. Smith, G.S.G. Shell, and N.S. White. 1987. Characterizing water use by
irrigated wheat at Griffith, New South Wales. Aust. J. Soil Res. 25, 499-515.
Muirhead, W.A., J. Blackwell, E. Humphreys, and R.J.G. White. 1989. The growth and nitrogen economy of
rice under sprinkler and flood irrigation in South East Australia. Irrig. ScL 10, 183-199.
Richardson, C.W., and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.
AUSTRALIA-8
-------
Rimmington, G.M., T.A. McMahon, and D.J. Connor. 1987. The Australian wheat field trial database (interim
report). Agric. Engin. Misc. Report. University of Melbourne.
Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
AUSTRALIA-9
-------
Table 1. Temperature differences and precipitation and solar radiation ratios between lxCO2 and
2xCO2 climate change scenarios at selected sites.
Temp. Diff. (°C)
Precip. Ratio
Solar Rad. Ratio
Site/Season
Wongan Hill
Summer
Autumn
Winter
Spring
Roscworthy
Summer
Autumn
Winter
Spring
Horsham
Summer
Autumn
Winter
Spring
Griffith
Summer
Autumn
Winter
Spring
Narrabri
Summer
Autumn
Winter
Spring
OISS
3.68
4.93
5.61
5.67
4.87
437
4.34
4.27
4.87
4.37
4.34
4.27
434
3.93
4.00
3.77
3.34
4.83
5.07
4.91
GFDL
4.63
4.32
3.77
4.84
5.27
4.53
4.09
4.00
4.86
4.46
4.46
4.29
5.27
4.53
4.09
4.00
2.36
3.82
4.12
4.41
UKMO
4.37
4.51
4.60
4.47
6.79
7.37
5.62
5.84
5.13
5.53
5.11
5.04
6.03
6.55
6.61
5.91
3.35
7.07
6.32
6.26
GISS
1.57
1.38
0.80
1.69
1.50
0.91
1.22
1.14
1.50
0.91
1.22
1.14
1.13
0.96
1.03
1.25
0.75
1.25
0.89
0.98
GFDL
1.01
0.96
0.77
0.50
0.71
1.01
1.06
0.98
0.92
0.87
1.12
0.79
0.71
1.01
1.06
0.98
0.75
0.78
0.95
1.08
UKMO
1.00
1.08
1.16
1.06
1.21
0.86
1.17
1.09
1.12
1.23
1.25
1.12
1.29
1.05
1.04
1.19
0.99
0.79
1.16
1.18
GISS
1.00
1.00
1.02
1.00
1.01
1.03
1.06
1.00
1.01
1.03
1.06
1.00
1.02
1.05
1.09
1.01
0.66
1.01
1.05
1.01
GFDL
1.01
1.02
1.07
1.07
1.02
0.97
1.04
1.07
1.05
1.02
0.95
1.09
1.02
0.97
1.04
1.07
0.66
1.09
1.05
1.05
UKMO
1.02
1.08
1.01
1.02
1.06 :
1.12
1.03
1.11
1.09
1.08
1.09 :
1.10
1.03
1.07
1.05 ;
1.04
0.67
1.10
1.04
1.04
AUSTRALIA-10
-------
Table 2.
Management variables used for the CERES models to simulate crop growth.
Site
Crop/Variety
Soil
Sowing Plant Population
Dates (plants m"2)
Wongan Hills
Roseworthy
Horsham
Griffith
Griffith
Griffith
Narrabri
Dryland wheat/Gamenya
Dryland wheat/Egret
Dryland wheat/Egret
Dryland wheat/Egret
Irrigated wheat/Egret
Flooded rice/Calrose
Dryland wheat/Egret
Deep sandy loam 1 May 150
Red brown earth 30 May 150
Red brown earth 30 May 150
Red brown earth 15 May 150
Red brown earth 15 May 180
Gray cracking 17 Oct 128
Vertisol 25 Jun 150
AUSTRALIA-11
-------
Table 3. Effects of climate change on simulated wheat and rice; (a) yield and (b) irrigation water.
(a) Yield (t ha'1)
Climate Scenario Alone
Climate Scenario with
Physiological CO2 Effects
Site
Wonhan Hills Yield
Roscworthy
Horsham
Griffith
Narrabri
Griffith
Griffith
(b) Irrigation
Griffith
Griffith
SD
Yield
SD
Yield
SD
Yield
SD
Yield
SD
Yield
SD
Yield
SD
water (mm)
mm
SD
mm
SD
BASE
3.24
1.22
2.48
1.17
2.64
1.08
2.73
1.23
3.55
1.96
6.65
0.55
9.16
0.42
240
83
693
84
GISS
2.19
1.10
2.57
1.24
2.73
1.08
2.71
1.48
1.96
1.43
5.46
0.53
8.30
0.77
177
63
689
105
GFDL
2.27
1.08
2.45
1.42
2.60
1.23
2.62
1.44
2.47
1.73
5.08
0.49
7.80
0.58
169
65
761
93
UKMO
Dryland Wheat
3.14
0.86
1.94
1.15
3.00
1.05
1.80
1.36
2.27
1.40
Irrigated Wheat
3.79
0.45
Flooded Rice
7.66
0.51
Irrigated Wheat
118
48
Flooded Rice
747
110
BASE
4.08
0.90
3.40
1.01
3.28
0.90
3.44
1.15
4.74
2.15
7.22
0.61
10.10
0.54
202
80
680
87
GISS
3.04
1.34
3.26
1.22
3.40
0.87
3.38
1.48
2.82
1.74
6.41
0.56
9.12
0.77
136
55
675
104
GFDL
3.18
1.12
3.11
1.40
3.32
1.14
3.27
1.39
3.43
2.10
5.99
0.53
8.60
0.82
133
50
749
94
UKMO
3.85
0.83
2.64
1.39
3.59
0.94
2.38
1.69
3.13
1.63
4.64
0.52
8.54
0.70
98
50
731
112
AUSTRALIA-12
-------
Table 4. Sensitivity analysis of the CERES-Wheat model to changes in temperature, precipitation and
CO2 levels. Dryland wheat at Roseworthy.
330 ppm CO2
555 ppm CO2
Changes in
Precip. Temp.
(%) (°C)
0% 0
2
4
20% 0
2
4
-20% 0
2
4
Yield
(t ha1)
2.55
2.82
2.51
3,29
3,57
3,05
1.59
1.86
1.66
SD
(t ha1)
1.24 .
1.32
1.49
1.26
1.11
1.27
0.75
1.03
1.12
Yield
(t ha-1)
3.45
3.69
3.18
4.00
4.19
3.75
2.69
2.85
2.51
SD
(t ha'1)
1.10
1.11
1.45
0.85
0.88
1.15
1.08
1.23
1.45
AUSTRALIA-13
-------
Table 5.
Adaptation strategies for climate change scenarios; effects of change in planting date on
simulated wheat yield at Horsham and Roseworthy (a); effects of change in cultivar on
simulated irrigated rice yield (b) and irrigation water (c) at Griffith.
(a) Dryland Wheat Yield (t ha'1)
Climate Scenario Alone
Climate Scenario with
Physiological CO2 Effects
Horsham
Roseworthy
(b) Irrigated
Griffith
30 Apr.
15 May
30 May'
15 June
30 June
30 Apr.
15 May
30 May*
15 June
30 June
Rice Yield (t ha"1) •
Calrose*
IR-36
BASE
3.94
3.19
2.64
2.29
1.80
3.72
3.15
2.48
1.91
1.37
9.16
9.30
GISS
2.24
2.77
2.73
2.66
1.77
1.60
2.40
2.57
2.53
1.79
8.30
10.17
GFDL
1.64
2.37
2.60
2.45
1.49
1.70
2.38
2.45
2.11
1.18
7.80
9.47
UKMO
2.09
2.92
3.00
2.75
1.72
0.80
1.39
1.94
2.09
1.32
7.66
9.63
GISS
3.11
3.53
3.40
3.20
2.37
2.22
3.19
3.26
3.22
2.39
9.12
12.05
GFDL
2.36
3.13
3.32
3.21
2.12
2.32
3.09
3.11
2.90
1.84
8.60
11.19
UKMO
3.04
3.78
3.59
3.28
2.29
1.16
2.03
2.64
2.85
1.97
8.54
11.40
(c) Water Used for Irrigation (mm)
Griffith
Calrose*
IR-36
693
825
689
728
761
783
747
769
675
707
749
760
731
741
AUSTRALIA-14
-------
NSW
Roseworthy
Figure 1. Australian study sites and major grain-growing regions.
-------
y
7
6
5
4
3
2
— — __r_
/
'V
X* /
T T/ /
T v^.//
• T J> x v
• /o V /
• / 9 V
"h^ /j^TV^W
"^ ,/ W^ T
•"> An xX *" V T V
m Ai D / «jfv
&E *"/ / v^ v
— • Bjr3^-i X „ T v
^f^vx * ' 9
*t Jfjfeb**'
'*5RA£ . O*
:x/ if% •
* xitl -#^A °
/V** * 6 D A
/ \ ^8,/r ^
/ * / A n
- /x * /^°o
/ -w%
X / v
/ o °o *
/ <>o
~xX
/
/* °
1 1 1 1
330 555
v T Base
D • GISS
A A GFDL
0 * UKMO
100 120 140 160 180 200
Season length (days)
Figure 2. Effects of season length on irrigated wheat yield at Griffith, NSW.
------- |