Discrimination of Seasonal Snow Cover in Astore Basin, Western Himalaya using Fuzzy Membership Function of Object-Based Classification

Abstract

The temporal mapping of seasonal snow cover is generally being delineated through low resolution MODIS
data (250-500m resolution) due to daily frequency of image acquisition; however, it sometimes compromises the mapping
accuracies. In this study, the time-series of high resolution satellite imagery was used to evaluate the spatio-temporal
changes in the snow covered area of Astore basin during summer and winter seasons from 1990 to 2017. The Object
Based Image Analysis (OBIA) technique was applied on multi-spectral images of Landsat (TM and OLI sensors) of
respective years (1990, 2000, 2010 and 2017) in order to discriminate the snow covered area in both seasons. Although
OBIA is a strong technique that has been successfully applied in numerous research problems of remote sensing regarding
cryosphere, but due to hindrances (i.e. Clouds and haze), it is sometimes not highly efficient to detect the snow accurately,
therefore, Normalized Difference Snow Index (NDSI) has been calculated to distinguish snow covered area from snow
free areas. The range of 0.4-1.0 was used as a threshold value for fuzzy membership function in OBIA to delineate the
snow cover more precisely. The study suggested that the snow covered area is gradually increasing in winters during past
few decades in the basin; however, in summer season as compared to winters, no specific trend has been observed.