An insight into POLDER database over Arctic through the angular information

• Angular information can be used in identifying snow cover effectively. • Microtopography affects the anisotropic scattering of snow. • Subpixel components introduce a difference in both forward and backward direction. Snow and ice are among the most important components of the cryosphere; therefore, accurate identification of snow and ice is key to improve the understanding of the influence of various studies on global warming, e.g., studies on global water cycling, climate change and the radiation budget balance. The spectral information at single-view, e.g., the normalized difference snow index (NDSI) has been widely used in the recognition of snow cover at both local and global scales. However, special information sources related to the angular pattern of snow reflectance have rarely been considered in such applications. In this study, we proposed an angular index named the snow anisotropic reflectance index (SARI) based on the RossThick–LisparseReciprocal–Snow (RTLSRS) bidirectional reflectance distribution function (BRDF) model to achieve snow identification. The POLarization and Directionality of the Earth’s Reflectances (POLDER) database is one of the most widely-used multiangular database because of the abundant measurements and the pixels marked as “snow” are served as the data foundation in this study. The purpose of this study is to explore the potential of multiangular information to differentiate these pixels over Arctic in various situations. First, we preprocess the POLDER snow database to select the pixels with a good BRDF sampling distribution. Then, the SARI is used to preliminarily classify these POLDER datasets as snow and nonsnow, together with the classification result derived by single-view reflectances as comparison in the red and NIR bands. Next, the classification result is validated by using the MODIS MOD10A2 product. Finally, further analysis of the BRDF variations, regarding the SARI, for the classification results is performed by using the ArcticDEM and the Arctic Vegetation Map, a more detailed snow BRDF database over Arctic is obtained. The main results are as follows: (1) The potential of multiangular information (SARI) in recognizing snow had been confirmed according to MOD10A2 product as indirect validation data, the overall accuracy can reach to 86.9%, higher than the conventional single-view method (81.7%). (2) Within–pixel variations in surface terrain and components have a significant influence on the variability in BRDF shapes in the forward and backward scattering directions; in turn, these variations can be easily captured by using several simple thresholds for subcategorization for this POLDER snow database in Arctic Circle. Finally, this study provides a detailed POLDER database over Arctic and explain BRDF variations in various situations, facilitates the future potential applications of the remote sensing community.