Glacial lakes outburst susceptibility and risk in the Eastern Himalayas using analytical hierarchy process and backpropagation neural network models
The Himalayan cryosphere is dynamic, and changing climate conditions threaten breach of glacial lakes. A number of glacial lake outburst floods (GLOFs) occurred in the Himalayas in the recent past, affecting people and infrastructures. Assessment of high-altitude glacial lakes is required to avoid a...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2025-12-01
|
| Series: | Geomatics, Natural Hazards & Risk |
| Subjects: | |
| Online Access: | https://www.tandfonline.com/doi/10.1080/19475705.2024.2449134 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The Himalayan cryosphere is dynamic, and changing climate conditions threaten breach of glacial lakes. A number of glacial lake outburst floods (GLOFs) occurred in the Himalayas in the recent past, affecting people and infrastructures. Assessment of high-altitude glacial lakes is required to avoid associated hazards and mitigate the impacts. In this study, we have made an inventory of naturally formed lakes within the Sikkim Himalayas, including Nepal, Bhutan, and China, and discussed the GLOF susceptibility. A total of 399 lakes have been identified, out of which 281 lakes have an areal coverage greater than 0.01 Km2. Monitoring temporal changes shows a higher rate of areal increment for the lakes close to the western boundary of Sikkim. Using an Analytical Hierarchy Process (AHP) based on fifteen parameters, a number of glacial lakes show medium to high GLOF susceptibility in the Himalayan and surrounding regions. Three backpropagation multilayer perceptron neural network (BPMLPNN) models with Bayesian Regularization (BR-), Levenberg-Marquardt (LM-), and Gradient Descent with Momentum and Adaptive Learning Rate (GDX-) optimizers are designed to have better prediction accuracies compared to the AHP target scores. The BR-BPMLPNN model is observed with maximum performance and close similitude with the results obtained from the LM-BPMLPNN model. |
|---|---|
| ISSN: | 1947-5705 1947-5713 |