| Title |
Deep Learning Based Prediction for Easterly Wind and Classification of Precipitation |
| Authors |
정광희(Kwanghee Jeong) ; 이성준(Sungjun Lee) ; 서기성(Kisung Seo) |
| DOI |
https://doi.org/10.5370/KIEE.2020.69.9.1387 |
| Keywords |
Easterly wind prediction; Precipitation classification; Deep learning; Long short-term memory; ERA5 |
| Abstract |
Understanding the characteristics of the easterly-related weather phenomena in the eastern coast in Korean Peninsula is very important to analyze abnormal atmospheric phenomena such as heavy rain, heavy snow, and hot-dry wind. As data science techniques have steadily improved, a data driven prediction models are becoming more powerful in the quantitative forecasting weather. In this paper, we apply the LSTM based deep learning method to predict the velocity of the easterly wind around the Korean peninsula. Bi-directional data shape of input data and cascaded LSTM structure are proposed. The modified LSTM based method for prediction of easterly wind is experimented in years form 2013 to 2017 for the Korean Peninsula and East Sea using ERA5 data. Experiments of precipitation classification for Gangwon and Gyeongsang area are executed in years form 2008 to 2017 |