Soil moisture retrieval from multi-resource remotely sensed images over a wheat area

The aim of this study is to use the multi-resource remotely sensed images over land surfaces in a semi-arid region in northern China to infer the soil moisture content (SMC) in the presence of wheat.The multi-images include Advanced Synthetic Aperture Radar (ASAR),Landsat Thematic Mapper (TM) and Moderate-resolution Imaging Spectroradiometer (MODIS).The influence of vegetation on the satellite signal can be described by a water-cloud model.The vegetation water content (M_v),an important parameter in the water-cloud model,was synchronously measured with the corresponding satellite passes over the wheat area.The evaluation of the relationship between M_vand four satellite-derived vegetation indices (VIs) shows that the NDWI (normalized difference water index) from TM has the closest relationship with M_vas revealed by the correlation coefficient of 0.87.Based on the established M_v-NDWI relationship,combining the water-cloud model with the soil moisture retrieval over bare soil conditions,a semi-empirical model is developed for estimating SMC over a wheat area.A sensitivity analysis is also performed on the model parameters.In comparison with the insitu soil moisture observation,a satisfactory result is obtained in the soil moisture retrieval using the semi-empirical model as revealed by the RMSE (root mean squared error) of 3.83% and the correlation coefficient of 0.9.The spatial distribution of soil moisture deficits in the study area is also produced.