Comparing typhoon intensity prediction with two different artificial intelligence models

Abstract

In typhoon-prone areas, in order to announce catastrophe warnings in advance accurate storm-intensity prediction is essential. In this study, forecasting typhoon intensity with their characteristics has been supported by two different artificial intelligence methods, a simple artificial neural network (ANN) and an adaptive neuro-fuzzy inference system (ANFIS) based on genetic algorithm (GA). The data used throughout this paper are from the reanalysis data set of the National Centers for Environmental Prediction and the best track database of the National Oceanic and Atmospheric Administration. The typhoons selected for this study happened during 1985–2008 over the South China Sea. By comparing the results from these two methods, their advantages and disadvantages were investigated. The results of the study confirm the superiority of the ANFIS-GA method, because its root mean square error was 3.78, which is significantly lower than the value of 6.11 for the ANN method in the best experiments. Considering the literature in typhoon intensity prediction by ANN and ANFIS models, the contribution of this paper is to investigate whether or not these methods are able to forecast typhoon intensity. The study represents that the ANFIS model can produce more applicable predictions for typhoon intensity than the ANN model.