Prediction of Flood Discharge Using Hybrid PSO-SVM Algorithm in Barak River Basin.

Area of Science:

• Hydrology and Water Resource Management
• Artificial Intelligence in Environmental Science
• Climate Change Adaptation

Background:

• Accurate flood forecasting is essential for integrated water resource management.
• Traditional hydrological models face challenges due to the complex, time-varying nature of flood prediction parameters.
• Artificial Intelligence (AI) offers promising advancements in hydrological modeling and prediction.

Purpose of the Study:

• To investigate the efficacy of Support Vector Machine (SVM), Back Propagation Neural Network (BPNN), and a hybrid Particle Swarm Optimization-SVM (PSO-SVM) model for monthly flood forecasting.
• To assess the impact of meteorological parameters on forecasting accuracy.
• To compare the performance of the selected AI models using statistical metrics.

Main Methods:

• Utilized monthly river flow discharge data from 1969-2018 for BP ghat and Fulertal gauging sites on the Barak River.
• Employed SVM, BPNN, and PSO-SVM models, with PSO optimizing SVM parameters.
• Assessed various input combinations of Precipitation (Pt), Temperature (Tt), Solar Radiation (Sr), Humidity (Ht), and Evapotranspiration loss (El).

Main Results:

• The inclusion of five meteorological parameters significantly improved the forecasting accuracy of the hybrid PSO-SVM model.
• The PSO-SVM model demonstrated superior performance (RMSE: 0.04962, NSE: 0.99334) compared to standalone SVM and BPNN models.
• The PSO optimization algorithm proved to be computationally efficient, easy to implement, and theoretically simple.

Conclusions:

• The hybrid PSO-SVM model offers a highly reliable and accurate alternative for monthly flood discharge forecasting.
• AI-driven optimization techniques enhance the predictive capabilities of hydrological models.
• The study highlights the potential of advanced AI methods for improving water resource management strategies.