Study on the spatiotemporal variation mechanisms of soil moisture in maize fields

Soil Moisture Content (SMC) is crucial for sustaining agricultural productivity, ecosystem health, and climate feedback processes. This study investigates the spatiotemporal variation of SMC during two key maize growth stages using a combined physically-based and data-driven approach, which synergizes Water Cloud vegetation correction, Dubois-Dobson dielectric retrieval. The developed SMC inversion method achieving high accuracy with determination coefficients (R²) of 0.75 in the maturity stage and 0.78 in the filling stage, yielding high-resolution SMC product. Machine learning methods, enhanced by Shapley Additive Explanations (SHAP), were employed to analyze the impacts of environmental factors on SMC based on the high-resolution SMC product. Land surface temperature (LST) was identified as the primary driver of spatial variation during maturity, while elevation dominated during the filling stage. The different levels of SMC in two stages were largely dictated by meteorological factors, but the role of maize was deemed inconsequential on SMC's temporal variation. Furthermore, the relationships between SMC and environmental factors were quantified. SMC exhibits a gradual decrease trend with rising LST, yet this trend escalates sharply when LST surpasses 15 °C. An optimal range of Normalized difference vegetation index (NDVI) value, between 0.2 and 0.5, was discovered to be most effective for preserving SMC. This research offers a comprehensive perspective on the drivers of SMC variation, which is pivotal for informed agricultural practices and the enhancement of climate models.