INTERNET OF THINGS SENSOR DATA AND META-ALGORITHMIC APPROACHES FOR ADVANCED CLIMATE CHANGE-RELATED NATURAL DISASTER PREDICTION

Abstract

Natural disasters are catastrophic events caused by natural phenomena such as earthquakes, floods, hurricanes, and wildfires, resulting in severe damage, destruction, and human suffering. Existing methods for predicting natural disasters often lack precision because of the complexity and variability of natural processes, limited data availability, and the unpredictable cascading effects of initial events such as earthquakes. In this work, a range of machine learning techniques was applied for predicting various natural disasters under climate change: Long Short-Term Memory (LSTM) networks for earthquakes, Support Vector Machines (SVM) for tsunamis, Convolutional Neural Networks (CNN) for cyclones, and Random Forest (RF) for extreme temperatures. These models were integrated into a comprehensive meta-algorithm, enhanced by the Internet of Things (IoT) for real-time data collection and analysis. Performance evaluation against traditional models, including the Ensemble Decision Tree model and the Logistic Discriminant model, showed that the meta-algorithm achieved 5 % greater accuracy, highlighting its effectiveness in natural disaster prediction.