Integrating Geospatial and Machine Learning Techniques for Landslide Mitigation and Risk Assessment in Mizoram, India

Joel TC Vanlalnunzira, Satya Prakash

Abstract: This study addresses landslide susceptibility in Mizoram, India, a region characterized by steep terrain and high rainfall. The objective is to improve risk prediction using an integrated geospatial and machine learning framework. Fifteen conditioning factors, including topographic, geological, hydrological, and land-use variables, were analysed using GIS and remote sensing data. Machine learning models such as Random Forest, Gradient Boosting Decision Tree, Extreme Gradient Boosting, and stacking ensembles were trained on historical landslide data using an 80:20 train-test split. Model performance was evaluated using accuracy, AUC, RMSE, and Kappa index. The XGB-based models achieved the highest predictive performance with AUC values above 0.90 and reduced error metrics. The resulting landslide susceptibility maps were integrated with population and infrastructure data to assess regional risk. The findings support improved land-use planning and disaster mitigation strategies, contributing to enhanced resilience in landslide-prone regions.

Keywords: Landslide susceptibility mapping, GIS-based modelling, XGBoost, machine learning, ROC analysis, Mizoram, Risk assessment

How to Cite?: Joel TC Vanlalnunzira, Satya Prakash, "Integrating Geospatial and Machine Learning Techniques for Landslide Mitigation and Risk Assessment in Mizoram, India", Volume 15 Issue 3, March 2026, International Journal of Science and Research (IJSR), Pages: 1849-1858, https://www.ijsr.net/getabstract.php?paperid=SR26330121913, DOI: https://dx.doi.org/10.21275/SR26330121913

Download Citation: APA | MLA | BibTeX | EndNote | RefMan