Soil moisture affects many important hydrological and meteorological processes on various scales and it is important to know the global distribution of soil moisture. Microwave remote sensing is an indispensable method of obtaining this information. We used the first space-borne precipitation radar, on the Tropical Rainfall Measuring Mission satellite, for this purpose by examining backscattering not from rainfall but from the land surface under no precipitation conditions. The spatial pattern of the land surface backscattering coefficient (s°) is determined mainly by the incident angle and vegetation. The seasonal pattern of s° in general does not depend on different incident angles, except in the Sahel region, where there is a large impact of from the temporal change of vegetation. We propose a soil moisture estimation algorithm that considers a mosaic of different vegetation types in each scene. First, the vegetation fraction is determined by the s° observed at an incident angle of 3° and then the temporal change of s° for bare soil is calculated with observation at an angle of 12°. Because s° observed at 12° is not strongly affected by change of vegetation, the algorithm can simulate the seasonal pattern well even in the Sahel where vegetation changes drastically. This algorithm generally works well in regions without heavy vegetation, and we demonstrate this by validating it for estimating daily soil moisture in Oklahoma at a latitude of about 35°N.