Estimating maize root zone soil moisture by assimilating high spatiotemporal resolution optical and radar remote sensing into the WOFOST-HYDRUS coupled model

Abstract

Root zone soil moisture (RZSM) has important applications in agricultural water resource management, drought monitoring and warning. Previous studies primarily assimilated one type of satellite data or data with coarse resolution; therefore, RZSM accuracy estimation by their methods was limited. This study builds a framework to couple crop and hydrology models through daily dynamic parameter transfers to estimate the RZSM per centimeter throughout the crop growing season. In the coupled framework, surface soil moisture (SSM) and leaf area index (LAI) were derived from high spatial resolution Sentinel-1 and Sentinel-2 data with average root mean square error (RMSE) of 0.054 cm3/cm3 and 0.40 m2/m2 respectively. The high-frequency (∼7 d) estimated SSM and LAI were assimilated into the coupled model using the ensemble Kalman filter (EnKF) method, and the results of the single-sensor (SSM or LAI) and dual-sensor (SSM and LAI) assimilations were compared with in-situ observations. Four RZSM estimation results were compared with the field observations; the RZSMWOHY (WOHY: World Food Studies (WOFOST) and HYDRUS-1D coupled model), RZSMWOHY_LAI (WOHY assimilate LAI), RZSMWOHY_SM (WOHY assimilate SSM), and the RZSMWOHY_SM_LAI (WOHY assimilate SSM and LAI). RZSMWOHY_SM_LAI had the highest accuracy among the three assimilation strategies and WOHY, with the lowest unbiased RMSEs (ubRMSE) of 0.050 cm3/cm3, 0.050 cm3/cm3, and 0.060 cm3/cm3 and high correlations of 0.63, 0.50, and 0.57 for 5, 10, and 60 cm soil depths, respectively. The results highlight that the framework can accurately capture vegetation growth processes and soil moisture dynamics in the root zone, providing data and methodological support for efficient water resource utilization and precision agriculture.