Development of a Web-Based Rainfall Forecasting System Using Long Short-Term Memory Networks

Nigeria has faced numerous challenges over time due to inadequate rainfall forecasting. This research aims to develop a predictive system for daily and monthly rainfall in Ondo State, using LSTM networks. The research uses a dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) comprising daily rainfall records spanning 31 years (1992–2022). The LSTM model was developed and trained within the TensorFlow framework. Model performance was evaluated through Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the Coefficient of Determination (R²). The daily prediction model achieved an RMSE of 32, an MAE of 35, and an R² of 0.68, whereas the monthly prediction model recorded an RMSE of 7.4, an MAE of 6.6, and an R² of 0.92. The daily and monthly rainfall prediction models are deployed as an interactive web application using Streamlit, enabling users to select date ranges and view rainfall forecasts, which bridges the gap between advanced modelling and practical application for non-technical stakeholders.