ConvLSTM Hybrid Algorithm for Deep-Based Learning Flood Forecast Model Development

Abstract

In order to create and assess a flood forecasting model to predict the future occurrence of flood occurrences, this research develops a hybrid deep learning (ConvLSTM) algorithm incorporating the predictive advantages of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) Network. The performance of the proposed ConvLSTM model is evaluated against 9 distinct rainfall datasets. The outcomes show that the ConvLSTM-based flood model outperforms the benchmark techniques, all of which were evaluated at prediction horizons of one day, three days, seven days, and fourteen days. The results demonstrated practical utility of ConvLSTM in accurately forecasting