Enhancing root-zone soil moisture estimation using Richards' equation and dynamic surface soil moisture data
Xizhuoma Zha, Wenbin Zhu, Han Yan, Aifeng Lv
Abstract
Root-zone soil moisture (RZSM) is a critical variable for accurately modeling hydrological and ecological processes, but its monitoring is challenging due to the spatial and temporal variability at watershed scales. Richards' equation is a fundamental physical equation that accurately captures the dynamics of soil moisture transport in the root zone. However, due to its high sensitivity to input parameters, its application in large-scale spatial domains remains a significant challenge, particularly in regions with sparse meteorological data . This study addresses these challenges by proposing an innovative approach to estimating root-zone soil moisture by integrating dynamic surface soil moisture data into Richards' equation (SSMRE model). This approach encapsulates soil-atmosphere interactions using near-surface soil moisture, simplifying the computational framework and expanding the applicability of Richards' equation to broader spatial scales. Using the Lightning River Basin as a case study , simulations of different vegetation types and boundary conditions indicate that the correlation coefficient ( R ) for root zone soil moisture(50 cm) is generally greater than 0.7,SSMRE can accurately simulate root zone soil moisture under various lower boundary conditions and vegetation types. The HYDRUS-1D model, which is widely applied, typically uses atmospheric boundary conditions to simulate soil water movement under atmospheric influence. Comparative analysis of the HYDRUS-1D and SSMRE models against site-measured data reveals that for HYDRUS-1D, the correlation coefficients ( R ) across 5 cm,10 cm,20 cm,50 cm are 0.654, 0.621, 0.549 and 0.48, with root mean square errors ( RMSE ) of 0.03, 0.03, 0.03, and 0.04, respectively. The SSMRE model exhibits R values of 0.9, 0.85, 0.74, and 0.72, with RMSE values of 0.04, 0.02, 0.04, and 0.05. Demonstrating that our method provides improved accuracy in root-zone soil moisture simulations. The application of the Shuffled Complex Evolution-University of Arizona (SCE-UA) algorithm significantly enhances the model's accuracy. This research establishes a theoretical foundation for estimating multi-layer soil moisture over large spatial scales by integrating satellite-derived near-surface soil moisture data with Richards' equation.