Estimating actual maize yield with WOFOST in data-scarce small-scale cropping systems of South Africa: Data assimilation approach

Accurate farm-level yield estimation is crucial for informed agricultural decision-making and food security, particularly in data-scarce, small-scale rainfed systems such as those found in South Africa. This study explores the integration of remote sensing-derived leaf area index (LAI) into the WOFOST crop model using a recalibration-based data assimilation (DA) approach. Data from eight farms across two growing seasons (2020–2021) in the Eastern Cape were used. Key phenological parameters were calibrated, and a yield gap factor influencing daily gross assimilation was introduced and optimized alongside the specific leaf area using LAI observations from two farms in 2020. The optimized parameters were validated across additional farms and seasons. Results show that DA significantly improved yield predictions (RMSE = 472 kg.ha−1; NRMSE = 11%) compared to simulations without assimilation (RMSE = 4817 kg.ha−1; NRMSE = 112%). These findings highlight the method's potential to adapt crop models to data-limited contexts, quantify yield gaps, and support efficient resource management. The approach offers scalable benefits for decision-making in similarly constrained agricultural systems.