Assessment of heavy rainfall forecasts over the southern Vietnam by using WRF-ARW with different physical parameterization schemes

Abstract

This research aims to investigate heavy rainfall forecasting over southern Vietnam (SVN) by using the Weather Research and Forecasting with the Advanced Research dynamical solver (WRF-ARW) model. The selection of 34 heavy rainfall events from 2012 to 2016 over SVN during the active phase of the Southeastern Asian summer monsoon was studied. Sixty different WRF-ARW configurations are determined from three cumulus schemes (Kain–Fritsch with three different types of trigger mechanisms, Grell–Devenyi and Betts– Miller–Janjic), three planetary boundary schemes (Yonsei University, Mellor–Yamada–Janjic and Bougeault-Lacarrère) and four cloud microphysics schemes (Lin, Thompson and two WRF microphysics schemes with five and six moments). Typical skill scores are calculated based on the surface observation data and spatial verifications are carried out using the fraction skill score (FSS) and grid rainfall observations from the satellite-based GSMaP product. It is clear that the most sensitive to heavy rainfall forecasts were cumulus scheme selections, especially at thresholds below 50 mm/24 h. The configurations using the Kain–Fritsch scheme have wet bias issues. However, overall, in terms of a high probability of detection (POD) and high threat scores (CSI), the most skillful configuration was the use of the Kain–Fritsch scheme with the third trigger mechanism in combination with local diffusion planetary boundary schemes.