Streamflow forecasting using machine learning for flood management and mitigation in the White Volta basin of Ghana

Abstract

Floods are a major threat to livelihoods and infrastructure in the White Volta basin of Ghana. Providing accurate streamflow information is essential for flood management and mitigation. This study, for the first time, used machine learning algorithms, specifically the Long Short-Term Memory (LSTM) and Random Forest (RF), trained on rainfall, temperature, soil moisture, and evapotranspiration data to predict streamflow at 1, 5, and 10-day intervals in the White Volta basin. The study further used these models (RF and LSTM) to forecast future streamflow using CMIP6 SSP5–8.5 scenario data. The model’s output was mainly evaluated using Mean Absolute Error, Mean Bias Error, and Kling-Gupta Efficiency. The result showed high variability in the streamflow, and both models performed well in capturing these variabilities. LSTM performed better in capturing peak flows, whereas RF provided stable long-term predictions for up to 10 days. The future predictions also showed high variability in streamflow, suggesting an increased risk of floods and droughts in the White Volta basin. Given that these models can capture the timings of streamflow (seasonal patterns and peaks), they are well-positioned to provide accurate and reliable forecasts to support effective flood risk management and mitigation in the basin. These models can be extended to similar basins, offering a replicable and sustainable framework for proactive flood early warnings.