Accurate flood prediction remains a critical challenge in hydrological modelling due to the nonlinear and dynamic interactions between climatic and catchment variables as Nigeria faces annual occurrence of devastating flood. This study presents a machine learning-based framework for runoff prediction using satellite derived and locally improved hydro-meteorological data from Ilorin, Nigeria, spanning the period 1990–2025 to enhance early warning signals and preparedness for inhabitants of Ilorin metropolis. The model incorporated important meteorological variables such as temperature, dew point , relative humidity, rainfall, and an engineered antecedent precipitation index (API_alpha) derived using an exponentially weighted moving average approach. Three models—Linear Regression, Random Forest, and Extreme Gradient Boosting (XGBoost)—were developed and evaluated using standard performance metrics, including Root Mean Squared Error (RMSE) and coefficient of determination (R²). Results indicate that Linear Regression performs poorly (R² = 0.026), highlighting its limitations in capturing nonlinear hydrological processes Random Forest achieved near-perfect performance (R² = 1.000), this result suggests potential overfitting and limited generalizability. In contrast, the XGBoost model demonstrated robust and reliable performance (R² = 0.854, RMSE = 0.000576), effectively capturing nonlinear relationships between predictors and runoff. The best performing model was used in projecting future run-offs for Ilorin using incorporating ARIMA (Autoregressive Integrated Moving Average) for rainfall and relative humidity projection. The findings emphasize the critical role of antecedent rainfall conditions in flood prediction and demonstrate the superiority of boosting-based models for hydrological applications. This study contributes to the growing body of knowledge on data-driven flood prediction and provides a scalable framework for improving early warning systems in data-scarce regions.
Keywords: Flood Prediction; Machine Learning; Runoff Modelling; Antecedent Precipitation Index; Hydrological Modelling; Climate Variability; Data-Driven Modelling;