Landslide Prediction Using Deep-Learning Frameworks and Wireless Sensor Networks

Landslides are destructive natural disasters caused by various factors, such as climate change, heavy rainfalls, rising temperatures, soil erosion, earthquakes, and hurricanes, and predicting them is crucial for safeguarding lives and properties. This paper explores landslide prediction using wireless sensor networks (WSN), remote sensing data, and deep learning techniques. By creating a sensor network with multiple nodes, data can be collected and analyzed to detect landslides. Additionally, deep learning-based image segmentation models are applied to WSN data, while satellite imagery combined with the U-Net deep learning model effectively maps and detects landslides. The advantage of using deep learning is its ability to cover a vast area through satellite images, while WSN data offer robustness and customizable data collection frequency. The combination of both methods yields accurate predictions of landslides, enabling early detection and warning systems. Furthermore, the study yielded significant outcomes, showcasing proficiency in node management, sensor data acquisition, and efficient database integration.