Comparing the Satellite Image Transformation Techniques for Detecting and Monitoring the Continuous Snow Cover and Glacier in Cilo Mountain Chain Turkey

Abstract

In this paper, satellite image transformations (SIT) for detection and monitoring of the continuous snow covers and glaciers (SCG) were evaluated using Landsat 5 TM (30 m), Landsat 8 OLI (30 m) and RASAT (7.5 m) satellite images at a regional scale. The study was performed in two stages. Firstly, four potential SCG detection indices were produced from the 23rd of August 2013 Landsat 8 image as a Normalized Difference Snow Index (NDSI), a Normalized Difference Snow-Ice Index (NDSII), a Normalized Difference Principle Component Snow Index (NDPCSI) and Tasseled Cap Wetness transformation (WET) as indicators of the SCG. Continuous SCG amount was obtained classifying the August 2013 RASAT satellite image using an object based classification technique as ground truth data. Kappa accuracy co-efficiency of the NDSI, NDSII, NDPCSI and WET were calculated to be 0.74, 0.76, 0.4 and 0.77 respectively. In the second stage, WET SCG maps were produced from August 1984, 2000 and 2015 Landsat images. Changes in the two time periods (1984-2000 and 2000-2015) showed that total SCG loss was 247 ha from August 1984 to August 2015. Almost 47% of the SCG loss recorded in the region in 31 years. The highest loss was observed in the 1st period (198 ha). However, only 49 ha SCG loss was detected in the 2nd period although the 5 year mean temperature changes were found to be similar both in the 1st and 2nd period. Because the most sensitive SCG areas (regions that are located at a lower slope and thus receive more sunlight) had melted in the 1st period. Finally, physical variables were more significant than temperature in the 2nd period for the stability of SCG in the study area. © 2016 Elsevier Ltd. All rights reserved.