Ensemble Physics of the Weather Research and Forecasting (WRF) Model for Predicting Heavy Rainfall in the Bandung area, West Java

Abstract

The complex topography of the Bandung region, with the presence of mountains and valleys, can affect air flow patterns and rainfall distribution. Accurate weather predictions and spatial precision are crucial for anticipating the impacts of heavy rainfall. This study aims to evaluate the capability of the WRF physics ensemble prediction system in forecasting heavy rainfall events in the Bandung region. The use of an ensemble prediction system is a viable approach to quantifying uncertainty in numerical weather prediction and provide more reliable information. The case study used is the heavy rainfall event that caused flooding on October 4, 2022, in the Pagarsih area. Global Forecasting System (GFS) data with a spatial resolution of 0.25 x 0.25 and a temporal resolution of three hours were used as input for downscaling in the WRF-ARW model. This study used 9 configuration schemes of the WRF-ARW model parameterization as ensemble members. The results of the study indicate that the WRF model (a combination of the Purdue Lin, Yonsei University Scheme, and Betts-Miller-Janjic Scheme) provided the most accurate heavy rainfall prediction, with an RMSE value of 2.13. The probability maps of rainfall products can effectively identify peak heavy rainfall between 1:00 PM- 4:00 PM. This is indicated by the large area with a greater than 90% probability of rainfall exceeding 10 mm. The ensemble mean product of rainfall predictions tends to underestimate heavy rainfall in the Pagarsih area. The ensemble mean product of surface air temperature can effectively identify the pattern of observational f luctuations with a low RMSE value (0.77), and the ensemble mean product of surface layer air humidity can identify the pattern of observational fluctuations with a relatively high RMSE value (13.28).