Evaluation and improvement of snow cover detection from MODIS images

Information from snow cover as one of the major resources of the groundwater, drinking and agriculture water has a significant importance in water resources management. One of the important sources of data for snow cover detection are the MODIS satellite images, which snow products were produced using them routinely. This research aims to improve the snow cover detection algorithm of MODIS using topography correction and Land Surface Temperature thresholding. In this regard, Northwestern part of Iran was selected as a pilot area, and because of the lack of field snow area data, Landsat 8 images were chosen as ground truth. Five appropriate simultaneous MODIS and Landsat 8 images in 2014 and 2015 years were prepared and 5 samples with 10000 pixels in each MODIS image were selected and three algorithms were implemented. The first one with MODIS snow product method was achieved to the mean absolute relative error of 3.44 percent. In the second algorithm topographic correction was implemented which the error was 2.25 percent. The third algorithm investigated land surface temperature thresholding effect, which in the best value of LST with considering 278 K threshold yielded to mean absolute relative error of 2.58 percent. Hence, results indicate proficiency of the provided algorithms in the study area, comparing with the standard method.