Publication

Abstract : This study focuses on enhancing disaster vulnerability analysis and plot price prediction in 16 villages within Thodupuzha Taluk, Idukki district of Kerala — an area increasingly prone to natural disasters such as landslides, floods, and storms. Using data from the NASA Giovanni Earth portal, we apply advanced machine learning models, including Bayesian-Optimized Random Forest, XGBoost, and Long Short-Term Memory (LSTM), to assess disaster risks and their effects on land values. Hyperparameter tuning is performed using the Tree-structured Parzen Estimator (TPE) for optimal model performance.Data preprocessing involves normalization, geographic calibration, and handling of missing values to ensure accuracy. A Disaster Susceptibility Index (DSI) is developed based on key environmental parameters such as precipitation, surface temperature, wind speed, deep soil temperature, and soil moisture. This index helps identify high-risk areas and provides a novel approach to understanding disaster potential.The DSI serves as a crucial link between environmental conditions and disaster events. Our findings reveal a clear inverse relationship between DSI and plot prices: higher DSI values signal greater risk and lower land values, while lower DSI values correlate with higher prices. This illustrates the economic impact of disaster vulnerability on real estate.The research contributes to better risk estimation and targeted disaster mitigation planning. By integrating environmental data and advanced analytics, the study offers a framework for sustainable urban development and land use planning. The localized focus on Thodupuzha Taluk enables detailed insights into the dual challenges of disaster vulnerability and its influence on property markets.