Application of Machine Learning and Deep Learning Algorithms for Landslide Susceptibility Assessment in Landslide Prone Himalayan Region

Landslides are an unpredictable natural disaster, but steps can be taken to reduce their impact. The Landslide Susceptibility Index plays a critical role in minimizing the risk of living in landslide-prone areas. Effective planning and management of these locations are essential. In recent years, statistical methods and, increasingly, machine learning-based approaches have gained popularity for landslide susceptibility modeling. This study employs various machine learning and deep learning algorithms, specifically Random Forest (RF), Artificial Neural Network (ANN), and Deep Learning Neural Network (DLNN), to estimate landslide susceptibility in Chamoli district, Uttarakhand, India—a region that witnessed over a thousand landslides in 2023. We carefully selected relevant metrics based on existing research and conducted a multicollinearity analysis on each parameter to ensure the model’s accuracy. We randomly split the data into training and validation sets in a 70/30 ratio. Among the models used, the DLNN outperformed others, superiorly predicting landslide susceptibility. These findings are valuable for local government efforts in disaster prevention and mitigation, particularly in the Chamoli District of Uttarakhand, where Geographical Information System (GIS)-based susceptibility mapping plays a critical role in identifying vulnerable areas. Overall, this model evaluation framework can be used as a guide to select the most suitable modelling strategy for assessing landslide susceptibility. This type of outcome is valuable to the decision-maker to implement a more optimal strategy for reducing the probability of landslides and its associated damages.