Extraction of Zemu Glacier’s Boundary Using a Multi-Parametric Approach

Abstract

Though Zemu is the largest glacier in the Eastern Himalayas, it is one of the least monitored among the Himalayan glaciers. Thus, its sensitivity to global climate change should not be neglected. Glacier boundary delineation is a labor-intensive and time-consuming process. Therefore, the primary objective of this study is to create a semi-automatic processing chain that can recognize glacier boundaries using Synthetic Aperture Radar (SAR) data, Principal Component Analysis (PCA), and Connected components segmentation (CCS) techniques. SAR data processing provides weather-independent, high-resolution data that captures the surface characteristics of the glacier, including backscatter intensity and coherence, which are crucial for detecting glacier boundaries. PCA reduces data redundancy and enhances the spatial characteristics of the input data. CCS groups pixels with similar intensities into segments. Combining these methods results in a more accurate and reliable delineation of glacier boundaries. The parameters we have selected for the process are unique. Earlier researchers have used coherence and slope information. However, we have also considered the effect of radar backscattering intensity and curvature of the study area in addition to coherence and slope information. The use of CCS gives uniqueness to this study. A qualitative analysis of the study showed similarity with the GLIMS glacier outline, and the Intersection over Union (IoU) metric for segmentation accuracy was 0.67. Additionally, the two-pass Differential SAR Interferometry (DInSAR) technique was used for estimating the LOS velocity of the Zemu glacier. The significance is that LOS velocity measurements help track the movement of glaciers over time.