查看文章

A few revisions were made in the MODIS snow cover algorithm with the objective to further minimize snow cover detection errors and increased data content relative to cloud masking for Collection 6.1 (C6. 1). The daily cloud-gap filled (CGF) product M* D10A1F that provides a “cloud free” daily snow cover map is included in C6. 1. The objective in C6. 1 is to minimize snow cover detection errors of omission and commission for the purpose of mapping snow cover extent (SCE) accurately on the global scale. To reach that objective the same “snow-conservative approach” from C6 was continued in C6. 1. The snow-conservative approach focuses on detection of snow wherever it might be present based on reflectance features, then screens for false snow detections. Detection of snow is pushed to algorithm limits eg, under low illumination conditions, high solar zenith angles (SZAs) and shadowed surfaces.

As compared to C6, minor revisions were made in the Level 2 snow detection algorithm. The thresholds in the low visible reflectance screen of MODIS bands 2 and 4 are lowered by 0.03 to reduce the occurrence of “no decision” results in low illumination conditions. Users should expect to see fewer “no decision” results in C6. 1 products. In general the lowering of the visible reflectance screen has been observed to result in no decision results changed to “snow” or “not snow”, with “no decision” results found for very low,< 0.07 reflectance situations. The MODIS cloud mask (M* D35_L2) flags for “probably cloudy” and “probably clear” are now included in the NDSI_Snow_Cover_Algorithm_Flags_QA dataset so they can be used to evaluate possible …