Combining remote sensing and HYDRUS-RS model to simulate large-scale soil moisture dynamics

【Objective】 HYDRUS is a software widely used to simulate soil water movement, but it requires evapotranspiration which is difficult to obtain at large scales. In this paper, we investigated the feasibility of combining it with remote sensing to simulate spatiotemporal soil water dynamics at large scales. 【Method】 The experiment was conducted in a summer maize field in Daxing District, Beijing. The evapotranspiration estimated from the MODIS data and the surface energy balance system (SEBS) model was decoupled into soil-surface evaporation and plant transpiration, which were combined with the HYDRUS-1D to simulate spatiotemporal water dynamics in soil profile. Field-scale soil water dynamics was modelled by embedding the HYDRUS-RS model with the geographic information system (GIS) platform. The models were then applied to simulate spatiotemporal changes in soil water in the summer maize field during its growing season in 2018. 【Result】 The SEBS model was the best in modelling Rn, with the coefficient of determination (R2), bias and root mean square error (RMSE) being 0.81, 11 W/m2, and 54.8 W/m2, respectively. In the experimental area, the estimated daily ET during the growing season of the maize in 2018 varied between 3 and 6 mm/d. This differed slightly from that directly estimated from the soil moistures measured from pre-defined sampling points, but their temporal variation trends were the same. During growing season of the summer maize, soil moisture simulated by the proposed HYDRUS-RS model for the 40-60 cm soil layer was accurate, with the bias, RMSE, and R2 being 0.2%, 2.1% and 0.80, respectively. 【Conclusion】 Combining remote sensing with HYDRUS-1D provides a robust and accurate method for simulating large-scale soil water dynamics in agricultural lands. For the summer maize crop we studied, this integrated approach effectively predicts the changes in soil moisture, particularly in the 40-60 cm root zone.