Study on the Soil Moisture Content Modelling and Data Assimilation Based on Remote Sensing and Land Surface Model

Abstract

Soil moisture content is a key parameter in water resources management. It is important to obtain reliable soil moisture content data that being consistent in space and time for water resources management. Land surface process model and data assimilation are effective means to obtain reliable spatial and temporal continuous soil moisture content. However, the lack of real observations and dynamically updated surface parameters often greatly affect the accuracy of stimulated results of the land surface model. Based on high-resolution remote sensing data, we continuously invert vegetation coverage which is the crucial parameter of land surface process model during the study period, dynamically update it, and consider the remote sensing inversion of soil moisture content as the observation input, combined with the data assimilation algorithm to carry out model simulation and data assimilation research. We choose China's Hetao Irrigation District as the research area, use Noah LSM model and Ensemble Kalman Filter (EnKF) algorithm to build the data assimilation system, and set the land use parcels of the irrigation district as the research unit. Based on the Chinese Gaofen-1 and environmental disaster reduction satellites’ remote sensing data, we retrieve the vegetation coverage fraction every ten days, and generate monthly vegetation coverage fraction data to update the Noah LSM model parameter. Meanwhile, we use the TVDI algorithm to invert the soil moisture content as the observation data in assimilation system and the simulation of the Noah LSM model. Model simulation and data assimilation experiments show that the remote sensing retrieval vegetation coverage fraction can improve the effectiveness of model simulation, which approaches to the true value, and also improves the data assimilation accuracy. The overall performance is better than that using the fixed state parameter.