Calibration and Validation of Ceres-Sorghum Module of Dssat Model for Summer Sorghum (Sorghum bicolor L.) under Different Sowing Windows
Dharmishtha S. Rathod *
Agricultural Meteorological Cell, Department of Agricultural Engineering, Navsari Agricultural University, Navsari, Gujarat, India.
Pravinsinh K. Parmar
College of Agriculture, Anand Agricultural University, Vaso, Gujarat, India.
Himani P. Vadodariya
Main Vegetable Research Station, Anand Agricultural University, Anand, Gujarat, India.
V. B. Virani
Agricultural Meteorological Cell, Department of Agricultural Engineering, Navsari Agricultural University, Navsari, Gujarat, India.
Sandip N. Saravadiya
Department of Agricultural Economics, Navsari Agricultural University, Navsari, Gujarat, India.
*Author to whom correspondence should be addressed.
Abstract
Sorghum (Sorghum bicolor L.) is an important staple crop for food-insecure populations across semi-arid and sub-humid tropics. The objective of this study was to calibrate and validate the CERES-Sorghum model (DSSAT v4.8.2) for improved summer sorghum varieties under different sowing dates in South Gujarat conditions and to assess its simulation accuracy for phenology and yield. .A comprehensive field experiment was conducted during the summer season of 2023 on the college farm of Navsari Agricultural University, Navsari, Gujarat (20°57′ N, 72°54′ E), to calibrate and validate the CERES-sorghum module of DSSAT v4.8.2 for three improved summer sorghum varieties (V1 = GJ-42, V2 = GJ-44 and V3= GNJ-1) under three different sowing dates (15th January, 31st January and 15th February) using a split-plot design with four replications. The CERES model was calibrated using the GLUE (Generalized Likelihood Uncertainty Estimation) algorithm to derive variety-specific genetic coefficients. Calibration results revealed differences between observed and simulated parameters consistently below 10% error. Independent field experimental data from 2017 and 2022 from Main Sorghum Research Station, Surat were used for rigorous validation. The model simulated 50% flowering and physiological maturity with percentage errors of 2.86% and -0.92%, respectively, in validation. For yield characters, grain yield and dry fodder yield expressed 1.70% and -7.19% error. Across both calibration and validation phases, root mean square error (RMSE) and normalized RMSE (nRMSE) metrics confirmed excellent model performance (PE < 10%, nRMSE < 10%). Regression analysis showed strong correspondence between observed and simulated values (R² > 0.92). The CERES-Sorghum model demonstrated excellent simulation capability for phenological stages, biomass and yield attributes, making it a robust tool for climate change impact assessment, optimal sowing date determination, irrigation scheduling and natural resource management applications in the South Gujarat heavy rainfall zone and similar agro-ecological regions.
Keywords: CERES-sorghum model, DSSAT v4.8.2, model calibration and validation, phenology and yield simulation, genetic coefficients, GLUE algorithm, summer sorghum