Evaluation Linear spectral Unmixing and NDSI Methods For Snow Cover Studying

摘要

In Iran, the monitoring of snow cover pattern and related parameters (e.g snow depth, snow water equivalent and snow melt)are traditionally based on the ground collection at the local scale. This process is both time consuming and expensive.On theother hand, due to the problems caused by inaccessible environmental conditions and irregularly-distributed ground stations,records are not always available for any areas or locations of interest,In this research, Modis satellite images after changing their PDS format to LIB, used for snow cover studing in Karaj andLatian Basins. These Basins that bounded between 51° 20' to 51° 36' longitude and 35° 52' to 36° 11' latitude supplyconsumption water for capital city of Iran.In this research two methods used for snow cover studying. NDSI used as a Pixel-base Method that determines snowosnow pixels and linear spectral unmixing algorithm used for Pixelbase method.Results shows that subpixel method in copmarison to NDSI. > 0.4 has better result but NDSI index, after atmosphericcorrection and changes in threshold according to snow characteristic, has best result. Inorder to atmospheric correction,Bulk correction method used. For subpixel method, Endmembers determined using IRS- P6 image and MNF algorithm usedfor decrease noises, thereafter, spectral curves determined for each endmembers and finally spectral unmixing performed.For evaluation the accuracy of both of the pixelbase and subpixel methods, IRS-P6 Images - 23.5 m spatial resolution -used.