Abstract: Retrieving land surface soil water content from remotely sensed passive microwave data has a good physical basis. Thus, it can provide dynamic monitoring of large-range soil moisture condition. The Advanced Microwave Scanning Radiometer for EOS (AMSR-E) data were used to derive soil moisture base value and daily variation for each image pixel, respectively, to build an inversion model for retrieving soil moisture information. A precipitation impact factor was proposed and incorporated into the modeling process to improve the accuracy of soil moisture retrieval. The IDL language was used to implement the proposed model as software modules of the System of Xinjiang Soil Moisture Inversion from Remotely Sensed Data. The in-situ measured soil moisture data by the WatchDog 2400 instrument and Loss-on-Drying method were used to derive empirical parameters for the regressive model that are suited to the conditions in Xinjiang, and to verify the proposed model output. The results show that, with reference to the data of in-situ measurements, our improved model can achieve better estimation of Xinxiang's soil moisture than the soil moisture products of US National Snow & Ice Data Center (NSIDC). The RMSE is improved from 8.4% to 4.25%. The software modules developed in this study can provide a tool for quick soil moisture monitoring in a large area such as Xinjiang.

Key words: AMSR-E, arid areas, ENVI/IDL, microwave, soil moisture