Evaluation of the Accuracy of Interactive Multisensor Snow and Ice Mapping System (IMS) 1 km Product Using Ground Snow Depth Data Across China

Abstract

This study evaluates the accuracy of the U.S. interactive multisensor snow and ice mapping system (IMS) 1 km product with recall, precision, false alarm rate, and overall accuracy by utilizing ground-observed snow depth data from meteorological stations in China over six snow seasons spanning from 2015 to 2020. Furthermore, it examines the correlation between recall and snow depth, as well as the correlation between overall accuracy and snow depth. The results reveal the following: The precision typically peaks during the middle of the snow season, predominantly exceeding 80% but rarely reaching 100%, with minimal spatial variation. The precision generally ranges from 75% to 80%, exhibiting slight spatial variations during this period. The overestimation rate remains below 5%, seldom exceeding 10%. The overall accuracy mostly exceeds 95%, showing no significant spatial variations. Across both China and its five regions, the recall demonstrates a significant logarithmic correlation with snow depth, while the overall accuracy exhibits an almost horizontal linear correlation with snow depth once it exceeds 3 cm.