한반도 연안해역 해무 관측 및 예측 가이던스 구축에 관한 연구

Abstract

This study compared the characteristics of marine meteorologicalelements(T, SST, T-SST, Td, T-Td, WS, WD) by the coastal sea areas ofSouth Korea when the sea fog occurred using hourly data observed atAutomated Synoptic Observing System (ASOS) (7 sites), marinemeteorological buoys (4 sites), Automatic Weather System on lighthouse(lighthouse AWS) (2 sites) during 2013∼2017.It occurred generally the most in the southern part of the Yellow Sea,secondly Middle of the East Sea. It also occurred the most in summer out ofall seasons (spring∼winter), especially in July. The Southern part of the EastSea was excluded because there was very little time sea fog (14 hours). Inthe event of a sea fog, the spring temperature and SST showed the highestand lowest temperatures, and most sea fog events showed a value of positive(+) (excluding the Middle of the Yellow Sea), and there were mostlyadvection fog (excluding the Middle of the East Sea). Td, T-Td, RH werehigher than the seasonal average in all the coastal sea areas, becausesufficient water vapor was supplied in the atmosphere. In the summer, T andSST showed various changes and T-SST was similar or larger than theseasonal average. In the summer, air temperature and SST showed variouschanges, and the thawing car was similar to or larger than the seasonalaverage. Thus there has been an advection fog in all the coastal sea areas.Most of the dominant wind direction was south∼south west series in theevent of a sea fog, and the wind speed was weak in Middle and Southernpart of the Yellow Sea, western and eastern part of the South Sea, whereasstrong in the Middle of the East Sea and Southern part of the Jeju Island.This study predicted the possibility of the sea fog by applying statisticaltechniques such as Decision tree, Tree models, and Random Forest to theSouthern part of the Yellow Sea and Middle of the East Sea, where the seafog most frequently occurred. And it was verified using the 2019 yearobservation data and WRF (weather research and forecasting) model. As aresult, the prediction accuracy of the observation data (2019) was higher thanin Southern part of the Yellow sea in the case of Decision tree. In the caseof Tree models and Random Forest, it was generally high except for thesummer in the Southern part of the Yellow Sea. Verification using WRF.