Remote sensing based yield estimation of wheat crop at farm scale: A case study of Badsu village of Alwar district, Rajasthan

Accurate wheat yield estimation at the farm scale is crucial for food security, market strategies, trade planning, and storage decisions. However, predicting crop production using remote sensing at farm scale presents significant challenges. This research aimed to develop a field-scale wheat yield prediction model using multi-temporal vegetation indices derived from Sentinel-2 MSI imagery for the rabi seasons of 2018–19 and 2019–20 from Badsu village in Alwar district, Rajasthan. Vegetation indices derived from cloud-free Sentinel-2 images spanning the crop growth cycle were processed to generate multiple vegetation indices, grouped into greenness, chlorophyll content, and dryness indicators. Spearman’s rank correlation (ρ) assessed relationships between indices and wheat yield across various phenological stages and their combinations. Linear and multiple linear regression (MLR) models were developed using the most significant indices. Findings indicate that Wide Dynamic Range Vegetation Index (WDRVI), Normalized Green-Red Difference Index (NGRDI), and Normalized Difference Water Index-2 (NDWI2), representing greenness, chlorophyll, and water stress, respectively, exhibited strong correlations with yield, except during harvesting and crown root initiation. The best-performing model achieved an RMSE of 0.47 tons/ha and an R² of 0.74, demonstrating the effectiveness of remote sensing indices for precise wheat yield estimation at the field level in diverse agricultural Conditions.