Abstract: Global Positioning System Interferometric Reflectance (GPS-IR) is a new remote sensing technique, which can be used to estimate soil moisture content near the surface. From the view of multi-satellite fusion, an estimating method of surface soil water content based on multi-satellite fusion is proposed. Firstly, the direct and reflected signals of multiple satellites are separated by low order polynomial fitting, and then the sinusoidal fitting model of the reflected signals is established to obtain the relative delay phase. Finally, a multiple linear regression model is established based on the relative delay phase of multiple satellites. Based on the monitoring data provided by the plate boundary Observatory (PBO), the inversion effects of different modeling sequence lengths are compared and analyzed to determine the best modeling length. The experimental results show that multiple linear regression model can effectively integrate multiple satellites, and it is feasible to estimate soil moisture.

Key words: GPS-IR, soil moisture, multi-satellites fusion, multivariable linear regression model, the optimal input variables set, accuracy evaluation