Browsing by Author "Othman, Faridah"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Enhancing Flood Simulation in Data-Sparse Niger Central Hydrological Area River Basin in Nigeria using Machine Learning-Based Data Fusion.
    (Springer Nature, 2026-03-12) Ganiyu, Habeeb Oladimeji; Ng, Cia Yik; Othman, Faridah; Wan Jaafar, Wan Zurina
    Flooding remains a persistent and devastating natural hazard in Nigeria. Accurate precipitation data are critical for hydrological simulation of floods. However, many river basins in Nigeria lack adequate ground-based meteorological data. This has created a need for alternative rainfall data sources, with satellite-based precipitation products (SPPs) offering promising solutions. However, SPP estimates often exhibit biases and uncertainties, particularly in complex tropical environments. Therefore, enhancing SPP accuracy through data fusion is essential for reliable hydrological modelling. This study enhances flood event simulation in a data-sparse Niger Central Hydrological Area River Basin in Nigeria by employing Extreme Gradient Boosting (XGB), Random Forest (RF), and Long Short-Term Memory (LSTM) models to fuse daily downscaled PERSIANN-CDR with observed rainfall data from 2013 to 2022. The RF-fused PERSIANN-CDR exhibited higher accuracy than the XGB and LSTM models. It improved the downscaled PERSIANN-CDR mean correlation coefficient (R) by 113.48% and reduced the mean MAE, RMSE, and Bias by 29.90%, 26.76%, and 75.71%, respectively. The reliability of the RF-fused PERSIANN-CDR was validated hydrologically using the HEC-HMS model. The simulated flow relative to the observed flow indicated improved NSE and R² values, alongside reduced RSR values during the calibration and validation periods, compared to the individual datasets. The robustness of the RF-fused data improvement was further validated using bootstrapped 95% confidence intervals and paired Wilcoxon tests (p<0.05). This study presents a practical framework that demonstrates the potential of data fusion to enhance satellite precipitation estimates for flood simulations, offering a viable approach for improving flood preparedness in data-sparse regions.