Assessing maize growth and yield using a remote sensing–integrated crop model enhanced with lasso and ridge regression

The integration of remote sensing technology into crop models offers an innovative strategy that enhances the accuracy of crop performance predictions. This study aims to investigate the effectiveness of a machine learning (ML)-enhanced remote sensing–integrated crop model (RSCM) in simulating maize growth and yield across diverse environmental conditions. Field experiments were conducted from 2018 to 2021, involving multiple maize cultivars (Kwangpyeong, Ilmi, Daeryuk, Dream, and Kangil) across varied seeding dates. The ML-enhanced RSCM was employed to predict seasonal variations in leaf area index (LAI) and grain yield, integrating proximal sensing data and remote aerial observations. Ten ML regressors were tested or LAI simulation, with lasso and ridge showing the highest accuracy, with test performance scores of 0.914 and 0.924, respectively. Overall model performance was validated using statistical metrics, including percent bias (PB) and Nash–Sutcliffe model efficiency (ME), with PB and ME values ranging from 0.22 to 1.92 and 0.75 to 0.99, respectively. The results demonstrate strong alignment between simulated and observed values, confirming the robustness of the model across cultivars and conditions. The findings underscore the potential reliability of RSCM in precision agriculture, enabling accurate real-time predictions of maize growth and yield.