Modeling the uncertainty in responsiveness of climatic, genetic, soil and agronomic parameters in CERES-Sorghum model across locations in Kansas, USA

Kansas leads grain sorghum production in the USA. Crop models are useful tools which provide insight about the functioning of crops, agricultural systems, and their interactions. There is a temperature and precipitation gradient across Kansas. The CERES-Sorghum model in the DSSAT system (Decision Support System for Agro-transfer Technology) was applied to many locations within the state. We hypothesize that the degree of responsiveness to CERES-Sorghum parameters would vary due to these gradients. The objective of this study is to document the uncertainties in the responsiveness of the climatic, genetic, soil and agronomic parameters in CERES-Sorghum across many locations in Kansas using multiple response variables. The input parameter categories evaluated are: climatic (temperature, solar radiation, rainfall, and CO2); genetic (P1, P2O, P5, G2, G5); agronomic (planting date, planting depth, row spacing and plant population); and soil (drained upper limit, drained lower limit, pH, saturated water content, soil organic carbon, bulk density, runoff curve number and drainage rate). Uncertainty analysis was carried out for six output response variables (yield, biomass, anthesis days, maturity days, leaf area index and leaf number) Sensitivity analysis was carried out using the OAT (one at a time) method by perturbing one input at a time keeping rest of the input parameter constant. Both relative sensitivity (a mathematical approach) and a graphical method were used, Cumulative distribution functions were used for uncertainty analysis. Preliminary results showed that, responsiveness of input parameters varied with input parameters, response variable, and location.