Development of Tidal Prediction Functions Considering Ocean Meteorological Factors

Abstract

The accurate calculation of tide levels is important in the design, construction, and management of structures associated with the sea or water areas. However, conventional tidal predictions only consider astronomical factors. Therefore, it is difficult to predict based on storm surges, oceanic weather and other factors. Herein, we presented a predictive function to accurately determine tide levels by considering marine factors. For the eastern coastal area, past forecasts and survey data were collected from three tide stations at the Korea Hydrographic and Oceanographic Administration. In addition, data of maritime factors such as wind speed, wave height, and wave period were collected from domestic buoy stations at the Korea Meteorological Administration. Logistic regression analysis was conducted to predict the occurrence of abnormal tides, and a new tidal prediction function was proposed through multiple regression analyses. The Bayesian Information Criterion (BIC) method was used to select variables, and accuracy verification through k-division cross-validation and Normalized-Root Mean Square Error (N-RMSE) was performed. For the latter, the values ranged from 3.37% to 12.03%. If continuous meteorological data are accumulated and marine meteorological data are acquired near the tidal observation point, tidal levels can be predicted more accurately. Keywords: Abnormal Tidal, Multiple Linear Regression, Logistic Regression, Tidal Prediction