Snow cover in forests plays a crucial role in protecting the forest ecosystem, maintaining stability, and providing essential resources, particularly in snow-affected regions at mid- to high-latitudes. However, the presence of forests significantly impacts the accuracy of snow depth retrievals from passive microwave remote sensing. A new index, called normalized difference maximum stem volume (NDMSV), has been constructed by integrating the canopy height and tree cover to develop a novel algorithm for passive microwave snow depth retrieval. By considering both the vertical and horizontal canopy structures, NDMSV can depicts forest density in a more detailed manner than just fraction of forest cover. The validation and comparison of our work in forest perspective demonstrate that the accuracy of snow depth retrieval algorithm developed by us is higher than the algorithm which only consider forest cover fraction, especially in moderately dense or sparsely forested areas, against in situ snow depth data. In addition, our results exhibit high accuracy regardless of canopy height. Spatial-temporal comparison results indicate that our study exhibits the higher retrieval accuracy in the Northeast China and Eastern Siberian Mountains when validated and compared against in situ snow depth, as well as other algorithms and datasets such as ERA5, ERA5-Land and Globsnow. For different snow season, our results perform well during the months with more stable snowpack in the Northeast China, the Central Siberia Plateau, and Eastern Siberian Mountains. Moreover, the accuracy of our algorithm is significantly accurate not only in forested areas, but also in other land types, including farmland and grassland. In conclusion, NDMSV index can effectively capture the forest characteristics and helpful in enhancing snow depth retrieval accuracy.
 

Snow depth; Forest; Passive microwave remote sensing; CETB;