Comparing SSM/I Snow Depth Estimates to In-Situ and Interpolated Multi-Source Measurements

Abstract

Spaceborne remote sensing data from the Special Sensor Microwave Imager (SSM/I) have been used for several decades to estimate snow depth over large regions. The SSM/I snow depth accuracy is not well quantified in non-uniform terrain. In this study, SSM/I snow depth estimates for the Columbia River Basin and surroundings in the Western USA and Canada are compared with in-situ manual snow-course measurements and interpolated snow water equivalent from the National Operational Hydrologic Remote Sensing Center. Snow depth is estimated for 25-km pixels from SSM/I brightness temperatures with the widely used Chang algorithm, adjusted for canopy cover. Interactive Data Language and ESRI ArcGIS are used to generate maps and time-series graphs, and to analyze the agreement between SSM/I snow depth and the other data sources. Measures of agreement are cross-tabulated with quantitative landscape descriptors, including: mean pixel elevation, elevation standard deviation (a measure of terrain complexity), and evergreen canopy cover.