Study area: The study site was Mymensingh district situated in the north central side of Bangladesh. The study site belongs to the agro-ecological zone (AEZ) 9, which lies at 24.450N latitude and 90.250E longitude. The elevation of the site is 18 m above the sea level.
Data collection: Data used in the present study were collected from the Department of Irrigation and Water Management (IWM), Bangladesh Agricultural University (BAU), Mymensingh. The model used in this study requires three types of data such as i) weather data; ii) soil data and iii) crop growth and management data.
Weather data: Climate or weather data such as maximum and minimum temperatures, rainfall, and daily solar radiation were collected from the weather station situated at Bangladesh Agricultural University campus, Mymensingh. Weather data were collected of 30 (thirty) years spanning from 1985 to 2014.
Soil data: The texture of the experimental soil was silt loam underlain by sandy loam. The soil data such as soil profile information, nutrient status, chemical properties, and hydraulic properties of wheat experiment was collected from previous studies (Biswas 2012) conducted at BAU, Mymensingh. The organic matter content of the soil was low. The top soil was moderately acidic, but the sub-soil was neutral in reaction. The average field capacity and permanent wilting point of the soil were 39.94% and 18.70%, respectively. The average bulk density was 1.33 gm/cm3 for the 60 cm profile.
Generation of climate change scenarios: Total seven climate scenarios were generated: a ‘baseline’ scenario, representing current climatic conditions, and six future scenarios of climate change. The latter were produced using output from MAGIC/SCENGEN model (Wigley 2008) for three future time periods 2025–2054 centering 2040, 2055–2084 centering 2070, and 2085–2114 centering 2100. For each time period, two emission scenarios such as A2 (A future scenario affected by strongly changed climatic condition comprising maximum, minimum temperature, rainfall and solar radiation) and B2 (A future scenario affected by mildly changed climatic condition comprising maximum, minimum temperature, rainfall and solar radiation) were simulated according the Special Report on Emission Scenarios (SRES). Observed daily mean of climatic data (maximum and minimum temperatures, rainfall and solar radiation) from 1984 to 2014, also served as baseline, were used to calculate future climatic scenarios, which were constructed together with MAGICC/SCENGEN model output by using delta change approach.
Crop simulation model: The Decision Support System for Agrotechnology Transfer (DSSAT) (Jones et al. 2003) is a software package incorporating the effect of crop phenotype, soil, weather and crop management system on the basis of a database protocol and allows researchers to simulate experiments in computers in a moment that would take enormous time if conducted manually. DSSAT enables the user to predict the possible results from diverse managerial dimensions and strategies through separate independent programs functioning together. The inherent parts of the program include specific simulator models and databases for weather, soil, experimental condition and measurements and genotype information. The crop model simulates crop yield, growth and development based on some characteristics of the simulated crops: phenology, photoperiod, biomass accumulation as well as partitioning among its roots, stems and leaves, as defined by cultivar-specific genetic coefficients. The software enables researcher to calibrate inputs for each of the programs and compare simulation results with observation, building user’s confidence in models or estimate probable modification to achieve higher accuracy. Risk assessment associated with different crop production strategies through its multi-year simulation is another feature of DSSAT. Moreover, it allows changing weather variables without modifying original weather data outfitting for climate change impact studies through its built-in functions. Among the suits of the Cropping System Model (CSM) in DSSAT and CERES-Wheat modules were used to study the effect of climate change on wheat production.
Calibration and validation of the model: The calibration of each model was done such that the model parameters truly represented the characteristics and responses of the crops to soil and atmospheric conditions. The CERES-Wheat was calibrated using the observed data by changing one parameter at a time. The data from field experiments, conducted during 2007-2008 season, were used to calibrate CERES-Wheat models, each time, a model parameter was changed, and the model was run with the changed parameter value. The resulting model simulated LAI and yield of wheat were compared with the corresponding observed values. This process of simulation and comparison was repeated until satisfactory LAIs and yields were obtained, which were ensured by satisfactory values of the model performance indicators. The model calibration involved changing the values of the cultivar coefficients. The default and calibrated values of the cultivar coefficients for wheat is given in Table 3. After calibration, CERES-Wheat model was validated.
Model simulation with future climate data: The impacts of climate change on wheat production were deduced by comparing the results of model simulation under climate change projection to the baseline. The calibrated models were used to simulate growth and yield of wheat under changing future climatic conditions by selecting the generated future weather data as model weather inputs. Simulations were made to run for the baseline and three future periods of 2040, 2070 and 2100 of the projected climate change and for the two emission scenarios, B2 and A2. From the outputs of the models, yields, WUE, seasonal ET and LGS of wheat were estimated.