Data Driven Spatio-Temporal Prediction of Landslide Susceptibility for The Himalayan Region

Abstract

Worldwide, landslides are the most frequently occurring disaster that are very destructive and unpredictable in nature. Tehri Garhwal in the Uttarakhand State of the Indian Himalayas is one such region where 850 landslide events were detected during 2005-2020. Many researchers have conducted landslide susceptibility mapping (LSM) studies for this region using different static landslide-causing factors. However, these studies lack consistency in selecting landslide causing factors for the susceptibility analysis and mapping. Also, studies considering dynamic factors in predicting future landslide susceptibility scenarios are inadequate. Hence in this study, initially, landslide causing factors were optimized for LSM, and then dynamic factors were utilized for future projection of LSM. The main objectives of this research include the development of scientific methodology for determining significant landslide causing factors for the Tehri region and validating them on two landslide prone sites of Himachal Pradesh with similar terrain conditions. Further, the LSM was prepared using the derived significant factors, and dynamic factors such as Land Use Land Cover (LULC) and climate variables were incorporated for future projection of the LSM. To achieve these objectives, first, the geospatial database in three temporal categories, 2005-2010, 2010-2015, and 2015-2020, was prepared for the historical landslide events. Second, the landslide-causing factors were optimized using multicollinearity analysis considering Pearson correlation and the Artificial Neural Network (ANN) model's sensitivity analysis. Third, the relevance of these significant factors in predicting landslide susceptibility was checked for the two test sites of the Himalayan region and utilized in LSM for 2010, 2015, and 2020. Fourth, the projected LULC map was generated for the year 2050 using the Artificial Neural Network Cellular Automata (ANN-CA) model. Fifth, CMIP6 climate projections maps were prepared using the Indian Institute of Tropical Meteorology Earth system model (IITM ESM) under four Shared Socioeconomic Pathway (SSP) scenarios. Finally, the projected maps were used as the driving parameter for the future prediction of LSM. The predicted maps were validated utilizing the Area under the Receiver Operating Characteristic (ROC) curve, and the Kappa coefficient verifies the reliability of the simulated future projected results. The results reveal that out of 21 parameters considered for the Tehri region, 11 were found to be significant for LSM and achieved the prediction accuracy of 0.93 Area Under Curve (AUC) value. Thus, this study recommends using the derived 11 landslide parameters and their hierarchy for carrying out LSM in the Himalayan region. Also, a high increase in the built-up area (5%) and agriculture land (4%) with a decrease in forest area (10%) in future LULC projections was observed. This LULC change and change in climate variable under four climate forcing scenarios of SSP 1-2.6, SSP 2-4.5, SSP 3-7.0 and SSP 5-8.5 has resulted in an increase of very high landslide susceptibility class by 2%, 4%, 7%, and 9% respectively. Future Prediction of LSM can help in the proper management and sustainable distribution of environmental resources. The target audiences can be land use policymakers who must decide which direction urbanization takes and which direction to restrict.