SHIELD: Systematic Hazard Identification for Early Landslide Detection — Real-time Landslide and Flood Detection

This paper presents the development of a low-cost, highly accurate, and proactive landslide prediction and early warning system designed for high-risk regions such as the Western Ghats in Kerala, India. Existing monitoring frameworks primarily rely on reactive approaches that fail to provide sufficient warning time before catastrophic slope failures. To address this, the proposed system integrates Internet of Things (IoT) sensor nodes, edge computing, cloud-based Long Short-Term Memory (LSTM) neural networks, and physics-based Digital Twin simulations into a unified architecture. By combining real-time ground data, such as vibration, tilt, and soil moisture, with Factor of Safety (FoS) calculations, the framework delivers reliable early warnings up to 72 hours before potential events. The integration of multi-channel alerting mechanisms ensures timely dissemination to authorities and local communities. This research demonstrates that synergizing edge-level intelligence with cloud-based deep learning bridges the critical gap between theoretical disaster prediction and practical, life-saving implementation.