Spatial sorghum yield forecasting: Integrating remote sensing data and DSSAT simulation model in Belagavi

Abstract

Effective sorghum yield prediction plays a pivotal role in sustainable agricultural planning and food security. This research explores the integration of Sentinel-1A synthetic aperture radar (SAR) data and the Decision Support System for Agrotechnology Transfer (DSSAT) crop simulation model to map sorghum yield spatially in Belagavi district, Karnataka, India. The study processed SAR backscatter to delineate crop areas, achieving an overall classification accuracy of 85.2% with a kappa index of 0.70, demonstrating the utility of SAR for consistent monitoring under diverse weather conditions. Leaf Area Index (LAI) derived from backscatter was integrated with DSSAT outputs to estimate spatial yield, validated against Crop Cutting Experiment (CCE) data. The results showed an agreement of 85.4% between observed and predicted yields, confirming the robustness of this approach for precision agriculture. Future work will aim to refine model parameters and leverage advanced machine learning for enhanced adaptability to climate impacts.