Glacier retreat in Himachal from 1994 to 2021 using deep learning

Abstract

Traditional glacier identification using satellite data involves making basic band combinations, image processing, and calculations based on the spectral properties of glacier bodies. The results from computer processing of satellite data are sometimes affected by segmentation and thresholding techniques, due to which water bodies, rock surfaces, and other features are classified as glaciers. Deep learning (DL) based methods for image classification, object detection, and segmentation have proven to be very efficient and accurate in recent years. Deep learning-based simple semantic network U-Net has been utilized in various disciplines for image processing and efficient image segmentation. This paper emphasizes the application of this simple semantic segmentation network, i.e., U-net, for glacier identification using Indian Remote sensing (IRS) and Landsat satellite data with some improvement in segmentation and glacier identification. The glaciers in the Himachal Pradesh province of India, located in the Western Indian Himalayas are identified, segmented, and mapped using the U-net. The glaciers for the past three decades, i.e. from 1994 to 2021 mapped in the study area to assess the impact of climate change on glacial retreat. The suggested technique can automate the entire glacial mapping process, has a greater identification accuracy of 95 percent, and is less time-consuming. The glaciated area in Himachal Pradesh decreased at a rate of 67.84 km2 per annum from 4020.6 km2 in 1994–2198.5 km2 in 2021. It is observed that from 1994 to 2021, the glacier area decreased approximately by a percentage loss of 1.678 per annum, with decreasing decadal trend from 2.31% in 1994–2001 to 1.398 in 2011–2021.