Prediction of Flood in Jhelum River Using Hybrid SVM-PSO Approach

Flood prediction is an important aspect of disaster management, for areas prone to frequent flooding. This research endeavors to advance flood prediction capabilities in the Baramulla district of Jammu and Kashmir, an area characterized by its vulnerability to hydrological hazards. We are attempting to decipher the complicated interplay of factors that contribute to flood events in this geographically sensitive area through the extrapolation of historical data containing many hydro meteorological variables. To assess how the data-driven models can synergize to enhance the accuracy of flood forecasts on a monthly basis from 1975 to 2022, we have used an advanced computational capability from Support vector machines (SVMs), artificial neural networks (ANNs), and integration of PSO (SVM-PSO). To foster resilience and protect the population and infrastructure of Baramulla, from recurrent flooding threats, this pioneering approach, which brings together a series of inputs and leverages advanced AI algorithms to develop robust flood prediction models, underlines the importance of this research and has a significant impact in the field of hydrology and disaster risk reduction, by extending the scope of flood modeling. The model developed has shown exceptional performance through the use of a hybrid training algorithm (SVM-PSO) with a coefficient of correlation (R2) = 0.9739, Mean square error (MSE) = 4.9932, Nash–Sutcliffe Model Efficiency = 0.969. These results highlight the significant potential of development models for accurate flood forecasts.