Soil moisture content is an important factor of predicting natural disasters such as drought or flood in hydrology or agriculture area. In the past several decades, the brightness temperature products provided by passive microwave sensors have been used for retrieving soil moisture according to radiative transfer theory, and the received signals from land surface are dramatically affected by the land covers and surface roughness, so a modified method is developed to retrieve near surface soil moisture by combination vegetation attenuation with polarization index, and comparing with Variation Inflation Model simulated data, the result shows retrieval data has better accuracy and soil moisture distribution is obtained in catchment of Heihe River Basin of northwest China.