ارزیابی و بهینه‌سازی روش سالومونسون و همکاران در برآورد سطح پوشش برف حوزه‌های کرج و لتیان

The purpose of presenting this paper is to determine the snow covered area on Karaj and Latyan basins using MODIS Images, and evaluating Salomonson et al method which is applied in this study. The importance of snow cover_ Such as its impact on radiation budget, water balance and modeling_ has led to several researches. In this research, MODIS data for snow cover mapping and LISSIII_IRS image for accuracy assessment have been used. Up to now, various methods have been applied to compute pixels’ snow fraction. We have used Salomonson etal. method. The method has showed proper accuracy in global scale and doesn’t need priori knowledge of surface characteristics. Also, to increase accuracy, the coefficients of the Salomonson et al. model were modified using regional data. And the results were evaluated in a new region.Accuracy assessment results showed that Salomonson et al. method can calculate snow fraction of MODIS pixels with RMSE of 0.20 pixels. Furthermore, Kappa coefficients and overall accuracy of Salomonson et al. method were 0.84 and 92.12 respectively suggesting proper accuracy of the method. Local accuracy assessment showed that in Iran, river margins with low density tree cover and sparsely scattered orchards, this method has got more errors; therefore, it is important to exclude thisarea. Moreover, it is recommended to use proper masks which allow the narrow rivers to be removed. RMSE of the modified model was 0.258 while, RMSE of Salomonson et al. model was 0.266 at the same area. So, the results showed that modifying the coefficient could improve the result slightly. Keywords: NDSI, Salomonson method, Snow fraction, Subpixel.