2.1 Selection of Simulation Locations The yield of two boro varieties BR3 and BR14 for the years 2008, 2030, 2050 and 2070 have been simulated for 12 districts of Bangladesh, which were selected from among the major rice growing areas in different regions of Bangladesh. Among them, Rajshahi, Bogra and Dinajpur were selected from northwestern region; Mymensingh and Tangail were selected from central region; Jessore and Satkhira from southwestern region; Barisal and Madaripur from southern region; Chandpur and Comilla from southeastern region; and Sylhet district from eastern regions. In addition to simulating yield for the selected years under the simulated climatic scenarios and the selected crop management conditions (described), potential yield (i.e., yield without any water and nitrogen stress) and vulnerability of the rice varieties under varying transplanting date was also assessed.
2.2 Crop Model The CERES-Rice model of the DSSAT modeling system is an advanced physiologically based rice crop growth simulation model and has been widely applied to understanding the relationship between rice and its environment. The model estimates yield of irrigated and non-irrigated rice, determine duration of growth stages, dry matter production and portioning, root system dynamics, effect of soil water and soil nitrogen contents on photosynthesis, carbon balance and water balance. Ritchie et al. (1987) and Hoogenboom et al. (2003) have provided a detailed description of the model.
2.3 Selection of Rice Variety The CERES-Rice model is variety-specific (e.g., BR3 boro) and is able to predict rice yield and rice plant response to various environmental conditions. In predicting crop growth and yield, the model takes into effect of weather, crop management, genetics, and soil water, C and N. The model uses a detailed set of crop specific genetic coefficients, which allows the model to respond to diverse weather and management conditions. Therefore, in order to get reliable results from model simulations, it is necessary to have the appropriate genetic coefficients for the selected cultivars. The two boro rice varieties BR3 and BR14 have been selected in the present study because genetic coefficients for these varieties are available in the DSSAT modeling system. Although these varieties are not widely used at present time, the effects of climate change and variability on these varieties provide insights into possible impact of climate change on rice yield in the future.
2.4 Soil and Crop Management Input The model requires a quite detailed set of input data on soil and hydrologic characteristics (i.e., pedological and hydrological data), and crop management. Input data related to soil characteristics include soil texture, number of layers in soil profile, soil layer depth, pH of soil in water for each depth, clay, silt and sand contents, organic matter, cation exchange capacity, etc. Required data on soil and hydrologic characteristics for the 12 selected locations (districts) were collected from Bangladesh Rice Research Institute (BRRI, Gazipur) and Soil Resources Development Institute (SRDI, Dhaka). As an example, the soil profile data used in the model for the North Eastern Barind Tract (i.e., AgroEcological Zone, AEZ-27) covering Dinajpur, Rangpur, Bogra, Gaibandha, and Joypurhat districts.
The crop management data (i.e., agronomic data) required by the model include planting date, planting density, row spacing, planting depth, irrigation amount and frequency, fertilizer application dates and amounts. The major crop management input data used in the model for all simulations in the present study are shown; these represent typical practices (BRRI, 2006) in Bangladesh. Using these inputs, the average (of 12 locations) yields of BR3 and BR14 for the year 2008, estimated by the model, were about 5500 kg ha-1 and 4050 kg ha-1, respectively; these values are close to the reported yields of these varieties (BRRI, 2006). These crop management inputs were subsequently used in all model simulations under the predicted weather scenarios for the years 2008, 2030, 2050 and 2070.
2.5 Weather data In this study, a regional climate model named Providing REgional Climate for Impacts Studies (PRECIS) was used to generate daily weather data needed for running the DSSAT model. The special report on emission scenarios (SRES) A2 of ECHAM4 has been used as PRECIS input. In this study PRECIS runs with 50-km horizontal resolution for the present climate (2008) using baseline lateral boundary conditions (LBCs). The model domain was selected 65–103°E and 6–35°N to cover Bangladesh and its surroundings. In the next step PRECIS run was completed for the year 2030, 2050 and 2070 using ECHAM 4 SRES A2 as the model input. The PRECIS outputs that were used in the DSSAT model include daily maximum temperature (Tmax), daily minimum temperature (Tmin), daily incoming solar radiation (Srad), daily precipitation. These parameters were extracted at 12 locations mentioned in subsection 2.1.