IOT BASED SMART FLOOD DETECTION AND ALERT SYSTEM

Abstract

Floods are among the most frequent and destructive natural disasters, causing loss of life, damage to homes and crops, disruption of transport, and the spread of disease. Sudden floods from heavy rainfall, overflowing rivers, blocked urban drains, and the release of water from reservoirs often give people very little time to move to safety. In many areas, warnings depend on manual observation of water levels and announcements by local authorities, which may arrive late. This paper presents an IoT based Smart Flood Detection and Alert System that monitors water levels and rainfall continuously and warns residents and authorities automatically when flood risk increases. The proposed system consists of field nodes placed at rivers, canals, drains, low-lying roads, and near vulnerable settlements. Each node uses an ESP32 microcontroller connected to an ultrasonic water level sensor, a float switch for backup detection, a rain gauge, a water flow sensor, and a temperature and humidity sensor. The node measures the water level at short intervals, calculates how quickly the level is rising, and combines this with recent rainfall to assess the flood risk at that location. Because flood-prone areas often have poor internet coverage, the nodes communicate through LoRa radio to a gateway that may be several kilometres away, and a GSM module provides SMS alerts as a backup. The gateway forwards data through MQTT to a cloud platform. Local sirens and warning lights at each node are activated immediately when danger levels are reached, so that people nearby are warned even if the network is unavailable. Each node also reports its battery voltage and signal strength so that maintenance staff can find faulty units quickly. The cloud platform stores all readings, displays node locations and water levels on a map-based dashboard, and applies three warning levels, namely watch, warning, and danger, based on water level, rate of rise, and rainfall. Alerts are sent by SMS, mobile application notification, and email to registered residents, disaster management officials, and local bodies. Historical data helps authorities understand flood behaviour at each location and plan drainage improvements and evacuation routes. Officials can acknowledge alerts from the dashboard, which records who responded and when. Testing with a model river channel and a simulated rainfall setup showed that the system detected rising water reliably, classified warning levels correctly, and delivered alerts within seconds of a threshold being crossed. The use of rate of rise allowed warnings to be issued earlier than fixed level thresholds alone. The system is low in cost, solar powered, and suitable for deployment in villages and urban areas. Future work includes flood prediction using machine learning with weather forecast data and integration with government early warning services.