Designing Deep-Based Learning Flood Forecast Model With ConvLSTM Hybrid Algorithm

Efficient, robust, and accurate early flood warning is a pivotal decision support tool that can help save lives protect the infrastructure in natural disasters. This research builds hybrid deep learning (ConvLSTM) algorithm integrating predictive merits of Convolutional Neural Network (CNN) Long Short-Term Memory (LSTM) to design evaluate forecasting model forecast future occurrence events. Derived from precipitation dataset, work adopts Flood Index ( I F ), form mathematical representation, capture gradual depletion water resources over time, employed monitoring system determine duration, severity, intensity any situation. The newly designed utilizes statistically significant lagged , improved by antecedent real-time rainfall data next daily value. performance proposed ConvLSTM validated against 9 different datasets prone regions Fiji which faces flood-driven devastations almost annually. results illustrate superiority ConvLSTM-based benchmark methods, all were tested at 1-day, 3-day, 7-day, 14-day horizon. For instance, Root Mean Squared Error (RMSE) for study sites 0.101, 0.150, 0.211 0.279 four forecasted periods, respectively, using model. best model, RMSE values 0.105, 0.154, 0.213 0.282 same order horizons. In terms difference individual stations, Legate-McCabe Efficiency (LME) 0.939, 0.898, 0.832 0.726 horizons, respectively. demonstrated practical utility accurately its potential use disaster management risk mitigation current phase extreme weather