COMBINING SATELLITE MULTISPECTRAL IMAGE DATA AND A DIGITAL ELEVATION MODEL FOR MAPPING DEBRIS COVERED GLACIERS
Debris covered glaciers are widespread in mountains of Asia like Karakoram, Himalayas and Tien Shan which have spectral covered properties as surrounding terrain and we cannot spectrally distinguish from it.
Hence we developed a method to study the debris covered glaciers in which we study the combined advantages of multispectral remote sensing with DEM derived slope facet. The slope range (0-24) is not exceeded. As debris covered glacier has to be continuous with clean glacier we perform neighborhood analysis to exceed all remaining area not connected to glacier ice. In last step glacier fluctuation are used to separate flat and vegetation free glacier by means of multi temporal change detection analysis from sequential images.
In MULTI SOURCE METHOD the various steps include segmentation of a TM4/TM5 band ratio image into class glacier (black) and other (white) using threshold 2 TM4/TM5.
Then the hue component of Intensity Hue Saturation (HIS) color space transformation from TM bands 3, 4, 5 is used to map vegetation and vegetation free areas with threshold value 126. The vegetation map is then used to reduce misclassification of pixels during glacier mapping. Further the slope is calculated from DEM25. All slopes value < 24 are used for slope facet map. The class debris produced if pix = 127. That is when pix = 255 and (hue 345 > 126) and (slope < 24) then pix = 127. In order to reduce misclassification of pixel and noise a 3 by 3 median filter is applied. The result is depicts area with debris that are not connected to glaciers. They are marked with Inage Polygon Growing (IPG) module from digital image processing software. With IPG individual number (ID) are assigned to all areas that are connected to glacier or debris.
These numbers are written to text file for each pixel together with their respective grey value (0 or 255). A FORTRAN...