Prediction of Convective Available Potential Energy and Equivalent Potential Temperature using a Coupled WRF and Deep Learning for Typhoon Identification Mamad Tamamadin (a,c), Changkye Lee (b), Seong-Hoon Kee (a,b), Jurng-Jae Yee (a,b)
a) Department of ICT Integrated Ocean Smart Cities Engineering, Dong-A University, Busan, Korea
b) University Core Research Center for Disaster-free and Safe Ocean City Construction, Dong-A University, Busan, Korea
c) Department of Meteorology, Institut Teknologi Bandung, Bandung, Indonesia
Abstract
To predict typhoon in the western North Pacific ocean, it is required to predict the determinants of typhoon activities. The formation of typhoon can be controlled by Convective Available Potential Energy (CAPE) and Equivalent Potential Temperature (EPT). To predict the variables, a mesoscale numerical model of Weather Research and Forecasting (WRF) can be used. However, the output of WRF needs to improve to obtain the more accurate CAPE and EPT prediction. This study uses a coupled WRF model and Deep Learning Multilayer Perceptron Regressor approach to increase CAPE and EPT prediction skill. Dataset scenario with WRF outputs as predictors and sounding data as predictors is developed and tested to obtain the most appropriate package of deep learning simulation. The study found that coupled models provide increased accuracy of CAPE and EPT using the dataset containing correlated predictors. This study also shows the difference in spatial distribution of CAPE and EPT between from WRF and its coupled model.
Keywords: accuracy improvement- CAPE- deep learning- EPT- typhoon- WRF
