OPTIMAL INTERPOLATION OF IN-SITU AND SATELLITE PASSIVE MICROWAVE DATA FOR GLOBAL SNOW DEPTH ESTIMATION

This paper presents an optimal interpolation approach to global snow depth estimation. The method uses 2-dimensional optimal interpolation to blend in-situ and satellite snow depth data weighted against their errors relative to a first guess and spatial correlations, the latter estimated from autocorrelation functions with respect to horizontal distance and elevation. The method is applied to ground-based and satellite passive microwave snow depth data within a 4-km resolution snow analysis system over the Northern Hemisphere. Assessment of the method is also presented using an extended ground-based snow depth dataset over US. Results show that the method is robust in estimating shallow and deep snow covers with mean bias and precision of 5 and 7 cm, respectively. Accuracy degrades over the mountains do to insufficient density of in-situ snow depth and inaccurate snow depth retrievals from satellite passive microwave data.