천문우주과학과

일시:  2022년 11월 10일 목요일 오후 4시 자4514호(세미나실)
     

연사: 이혜숙 박사님(국립기상과학원)

제목: Generalized AI/ML model for Weather Forecast : Focused on How to Approach  

초록:  

KMA/NIMS has developed AlphaWeather in earnest in 2021 after 2-year pilot project AlphaWeather will be developed for 10 years. AlphaWeather is intelligent solution which AI/ML technique is applied and its goals are improvement, automation of structured procedure and tailor of meteorological data and information. For improvement, we’re focused on the precipitation forecasting. Firstly, we've studied the neural network (NN) emulator for the radiation parameterization and its application on KMA operational NWP models. We have studied the post-processing of multi-NWP models with recommendation module based on the bias pattern and developing precipitation forecast using the data of radar and raingauge. For automation or minimization of forecasters’ process, we have developed the intelligent solution which has the function of recognition on potential severe weather, analysis on NWPs bias pattern/trend, NLP based data/information retrieval agents and similar weather condition retrieval algorithm using 30 years of geostationary satellite image. This solution will be developed in the way to be incorporated into the existing forecast decision support system. existing systems. For tailoring the data and information, we’ve studied the data labeling. Here the first phase (2021-2024) of AlphaWeather project will be introduced briefly with recent progress. 

Topic Machine learning for emulation of model components, Forecast Process Automation