Storm surge level prediction based on improved NARX neural network

Abstract

The northern Gulf of Mexico coast is affected by the North Atlantic hurricane season, which causes storm surge disasters every year and brings serious economic losses to the southern USA; therefore, it is necessary to make an accurate advance prediction of storm surge level. In this paper, a model with simple structure, fast computation speed, and accurate prediction results has been constructed based on nonlinear auto-regressive exogenous (NARX) neural network. Five types of data collected from observation stations are selected as the input factors of the model. To improve the model's computational efficiency, a neuron pruning strategy based on sensitivity analysis is introduced. By analyzing the output weights of the neurons in the hidden layer on the sensitivity of the model prediction output, the model structure can be adjusted accordingly. Moreover, a modular prediction method is introduced based on the tide harmonic analysis data so as to make the model prediction results more accurate. At last, a complete storm surge level prediction model, pruned modular (PM)-NARX, is constructed. In this paper, the model is trained by using historical data and used for storm surge level prediction along the northern Gulf of Mexico coast in 2020. The simulation test results show that the correlation between the predicted data and the observed data is stable above 0.99 at 12 h in advance and the model is able to produce the results within one minute. The prediction speed, accuracy, and stability are higher than those of conventional models. In addition, two sets of follow-up tests show that the prediction accuracy of the model can still maintain a high level. The above can prove that the pruned modular (PM)-NARX model can effectively provide early warning before the storm surge to avoid property damage and human casualties.