Flooding in Nigeria’s Niger Delta poses severe risks to lives and infrastructure, yet hydrometric data remain sparse. This study develops a predictive framework for water surface elevation (WSE) in the Bayelsa River by combining multi-mission radar altimetry with Artificial Neural Networks (ANN). Altimetric data from Jason-2, SARAL/AltiKa, and Sentinel-3 (for year 2008 to 2015) were preprocessed, corrected and merged into consistent WSE time series. Regression models revealed seasonal dependencies but achieved moderate accuracy (R² ≈ 0.41–0.89). By contrast, the ANN (39-20-1 feed-forward backpropagation) significantly improved prediction skill, attaining R² = 0.91, RMSE = 0.29 m, and MAE = 0.24 m against gauge observations. The ANN successfully captured double-peak seasonal hydrographs driven by bimodal rainfall, outperforming traditional statistical approaches. Findings confirm the potential of integrating radar altimetry and machine learning for flood forecasting in data-poor river basins. The framework offers a scalable, cost-effective solution for early warning systems and can be extended to other African basins and future SWOT mission datasets.
Ikharo et al. (Sun,) studied this question.