Quantitative estimation of skin soil moisture with the Special Sensor Microwave/Imager

A physically based retrieval scheme is described to infer near surface moisture content of bare soil patches from time series of the Special Sensor Microwave/Imager (SSM/I) Polarisation Difference Temperature (PDT) at 19.35 GHz. The PDT is modelled as a function of soil moisture using Fresnel's relations for specular reflection at the air-soil boundary, together with a dielectric mixing model for the moisture-dependent soil dielectric constant. The problem of the specification of parameters describing atmospheric and soil roughness effects is avoided by considering temporal variations of the PDT. The retrieval method is applied to the FIFE site for a two-week period in summer, during which heavy rainfall occurred on a dry upper soil. Although differences in sensing depths made a rigorous validation impossible, remotely sensed soil moisture showed a high degree of consistency with ground measurements. An error analysis shows that the method is capable of providing an accuracy of approximately 20% of moisture content at field capacity. Finally, it is demonstrated that the method is relatively insensitive to soil textural characteristics when moisture content is scaled with field capacity.