Impact Of Land Cover On Snow Identification From MODIS Images

The snow is one of the most active natural elements on the Earth's surface,characteristics such as snow area,snow distribution,snow depth are the importantinput parameters of the global energy balance,the climate,the hydrological and theecological model. The high reflective properties of snow cover which decided to makeit a key factor in Earth's radiation balance, are an important part of regional and globalclimate change.China has many people use the EOS/MODIS data to monitor and analysis ofsnow in different region. But it also has many disadvantages, such as using the sameNDSI threshold to define snow area for different area, not fully consider the snow ofphysics and the atmosphere and surface coverage and other natural conditions. Due tothe effect of forest canopy shades, NDSI value usually very small. Athin snow cannotcompletely cover in cropland and grassland, using standard threshold value, it willunderestimate the snow cover. Through the rational utilization of the normalzedvegetation index, can greatly improve the regional forest area, Uniting NDSI andNDVI, it will be easilyto divide snow covered with land surface.In this paper, we took the Northeast of China as the research area, We Usedclimatic data as"ground trust"to validate the snow covered-area accuracy of theexisting snow product MOD10A1, MOD10A2 and MOD10C2, and explored theinfluence of the snow covered-area accuracy by the cloud cover, land-use type andsnow depth. The results showed that, under the different land cover type background,(excepting the cropland and grassland, because they have very similar identificationaccuracy.) snow identification accuracy were quite different; Along with snow depth'sincrease, the identification accuracy has the tendency which roughly increasesgradually; The thin snow (snow depth generally refers to 1-5cm) the missed detectionerror is big; The thin snow region recognition precision is low, affects one of MODISoverall snow recognition precision primary factors. The existing problems of snowproducts by meteorological stations to establish observation data in different types ofnortheast snow-covered land covers the area under the threshold, again throughsnow-covered NDSI index, vegetation index NDVI etc threshold, the snow bythreshold images generated and MOD10A1 snow product identification precision andaccuracyof classification is improved, and to enhance the accuracyof snow drawing.