Employing Hybrid Support Vector Machine with Algorithm of Innovative Gunner for Streamflow Prediction

Prediction of streamflow (Qf) assists modelers in managing water resources in watersheds. It is important in water resource management, particularly for flood mitigation, reservoir operation, and drought warning. Water resource management is strongly reliant on hydrogeological prediction, and improvements in machine learning (ML) offer opportunities to improve predictive modeling capabilities. Artificial intelligence techniques cope with highly nonlinear relationships and complex hydrological processes, making them a superior option for Qf prediction. The support vector machine (SVM) model is trained using Algorithm of Innovative Gunner (AIG) in present study. The study reveals that the SVM-AIG approach performed exceptionally well across several metrics (Index of Agreement (IA) = 0.9858; Pearson’s correlation coefficient (R) = 0.994, and Mean Absolute Error (MAE) = 1.9514) when compared to SVM-only (IA = 0.9317; R = 0.9667, and MAE = 11.3679). The current study found that the SVM-AIG models were very good in predicting monthly Qf. This research shows that the SVR-AIG model is particularly useful for expressing real-world physical restrictions, and so has the potential to improve Qf prediction.