Remotely sensed snow-cover information has become a key tool to study temporal and spatial snow-cover patterns and to develop regional snow-cover climatologies. Aqua/Terra MODIS products provide about 20-year daily snow-cover data with 500 m spatial resolution. MODIS Collection 6 represents the most recent release of global snow-cover mapping algorithms and could further increase the high accuracy of previous collections: snow cover is now reported by its NDSI (Normalized Difference Snow Index) values, allowing more flexibility in using the datasets for specific regions than previous releases. We quantified the potential-added value of tuning the NDSI threshold as opposed to a global snow-detection algorithm by developing a 16-year snow-cover climatology for the Central Apennines (Italy) from daily observations in MODIS snow products Collection 5 (C5) and Collection 6 (C6). Seven ground-based stations were used as independent benchmark. Three versions of binary snow-cover maps were generated from the NDSI Snow Cover product (C6), using NDSI-threshold tests for snow detection. The most accurate snow-cover maps show an agreement with available ground data of 88% for Aqua and 89% for Terra MODIS, with an improvement compared to snow-cover maps obtained from C5 Snow Cover Area (SCA) products (yielding 86% for Aqua and 88% for Terra). NDSI thresholds in the range 0.10–0.40 provide an agreement higher than 83% but snow-cover duration, distribution, and spatial extent are sensible to the NDSI threshold: if compared to the optimal NDSI threshold for this region (0.20), the value of 0.40 reduces by 15% the snow-cover extent in all seasons due to increased underestimation. The lowest tested threshold (0.10) estimates at least 10% larger snow-cover fraction in winter and spring but increases commission errors. This high sensitivity to the NDSI threshold makes its choice an essential step for using MODIS C6 snow products in hydrologic or climatologic studies.
Comparing MODIS snow products Collection 5 with Collection 6 over Italian Central Apennines
De Michele C.;
2020-01-01
Abstract
Remotely sensed snow-cover information has become a key tool to study temporal and spatial snow-cover patterns and to develop regional snow-cover climatologies. Aqua/Terra MODIS products provide about 20-year daily snow-cover data with 500 m spatial resolution. MODIS Collection 6 represents the most recent release of global snow-cover mapping algorithms and could further increase the high accuracy of previous collections: snow cover is now reported by its NDSI (Normalized Difference Snow Index) values, allowing more flexibility in using the datasets for specific regions than previous releases. We quantified the potential-added value of tuning the NDSI threshold as opposed to a global snow-detection algorithm by developing a 16-year snow-cover climatology for the Central Apennines (Italy) from daily observations in MODIS snow products Collection 5 (C5) and Collection 6 (C6). Seven ground-based stations were used as independent benchmark. Three versions of binary snow-cover maps were generated from the NDSI Snow Cover product (C6), using NDSI-threshold tests for snow detection. The most accurate snow-cover maps show an agreement with available ground data of 88% for Aqua and 89% for Terra MODIS, with an improvement compared to snow-cover maps obtained from C5 Snow Cover Area (SCA) products (yielding 86% for Aqua and 88% for Terra). NDSI thresholds in the range 0.10–0.40 provide an agreement higher than 83% but snow-cover duration, distribution, and spatial extent are sensible to the NDSI threshold: if compared to the optimal NDSI threshold for this region (0.20), the value of 0.40 reduces by 15% the snow-cover extent in all seasons due to increased underestimation. The lowest tested threshold (0.10) estimates at least 10% larger snow-cover fraction in winter and spring but increases commission errors. This high sensitivity to the NDSI threshold makes its choice an essential step for using MODIS C6 snow products in hydrologic or climatologic studies.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
