Modular tidal level short-term forecasting based on BP neural networks

Abstract

Accurate and real-time tidal level forecasting information is significant for ensuring safety of navigation and port operation. The conventional method for tidal level forecasting is the harmonic analysis method which only considers the effect of celestial bodies to tidal level. However, the cause of tidal level change is intricate which can be also influenced by environmental factors such as wind, rainfall and air pressure. Therefore the harmonic method alone can not adapt all parts well. In order to improve the precision of tidal level prediction, a modular prediction mechanism is proposed which contains the harmonious analysis module for predicting time-varying portion causing by celestial bodies and the BP neural network module for predicting the residual portion causing by other elements. To further determine whether the modular prediction mechanism model possess good effectiveness and efficiency for tidal level forecasting, tidal level data of Port Isabel have been chosen as the test sample, and the prediction results adapt well with the field data.