Application of a Hybrid Approach for Tide-Surge Modeling in the Persian Gulf

In this study a tide-surge model of the Persian Gulf area has been developed using an unstructured grid, Finite-Volume Coastal Ocean Model (FVCOM). The model used European Centre for Medium-Range Weather Forecasts Re-Analysis Interim wind field and atmospheric pressure forcings and the water level boundary condition derived from global ocean tide model TPXO 7.1. Postprocessing of the water level predictions in a coastal station was conducted by applying the back-propagation artificial neural network technique. This method was used to improve surge predictions of the hydrodynamic model in the region. The input factors of the network consist of all possible combinations of numerical model surge prediction, wind components and duration, and sea-level pressure anomaly. In order to select the best prediction model, all these parameters were tested and compared with observed surge heights. It was concluded that mean sea-level pressure anomaly is the main affecting input parameter to have the best surge prediction. The combination of all input parameters resulted in surge predictions with a skill factor of 94.82% vs. 90.86%, which is the FVCOM numerical model result. It was shown that excluding numerical model surge predictions will greatly reduce the accuracy of surge height predictions.