Modelling of sea water level during high tide using statistical method and neural network

Abstract

Recently, the rise of sea level has caused an increase in rising tides that affected about three million locations around the world. The tide rising phenomenon has been occurring in Peninsular Malaysia since the 20th century. The purpose of this study is to determine the most critical station and forecast three stations located on the West Coast of Peninsular Malaysia. The Box Plot analysis method has been used in this study; the results shown that Bagan Datuk Station is the most critical station. This is due to the maximum tide’s value of Bagan Datuk Station experienced the highest increment of 0.45 m, compared to Port Klang station and Permatang Sedepa Station with only 0.2 m increment in 10 years. However, these three stations are also experiencing rising tides. Thus, the focus of managing coastal structures should be given to all these three stations as well. In addition, for forecasting, the Artificial Neural Network (ANN) forecasting model provides better forecasting results compared to the Autoregressive Integrated Moving Average (ARIMA) model for long-term forecast. In this study, the artificial Neural Network (ANN) forecasting model obtained value of RMSE 0.05642 at Bagan Datuk Station compared to the RMSE value of 0.0928 obtained from the ARIMA model at the same station. Besides, MAE value of ANN method, 0.04387 compared to the MAE value of ARIMA which is worth 0.06391 at Bagan Datuk Station. This study can conclude that the Artificial Neural Network (ANN) forecasting model is better in high tide forecasting.