WRF Model-Based Regional Rainfall Prediction Over A River Basin With Tropical Hydro-Climatology

Abstract:

Accurate short and medium range predictions of rainfall is paramount to agriculture, urban and rural drinking water supply, industry, health, and transportation including the regional climate change studies. This study evaluates the efficacy of four cumulus parameterization schemes in the Weather Research and Forecasting (WRF) model to reproduce the spatiotemporal distribution of observed rainfalls over the Brahmani River Basin (BRB) in eastern India during two typical periods of two consecutive years characterized with high and low rainfall events. The purpose is to understand the occurrences of static and dynamic variability in model parameterization for these typical rainfall scenarios. The selected four parameterization schemes include Kain-Fritsch (KF), Betts-Miller-Janji’c (BMJ), Grell-Devenyi ensemble (GD), and Grell 3D ensemble (G3) forced with the Kessler microphysics scheme. The Yonsei University (YSU) PBL scheme is used as the Planetary Boundary Layer in the WRF. Note that, in the Numerical Weather Prediction (NWP) models such as the WRF, the amount of total precipitation accounts for both the convective and non-convective rainfalls. For this study, the WRF model setup involves single nested domain of 2 km resolution with a parent domain of 6 km resolution over the selected study area with the National Centers for Environmental Prediction (NCEP) datasets. The WRF-predicted rainfalls are compared with the observed rainfalls from the India Meteorological Department (IMD) using the performance indicators of root mean square error (RMSE) and sign test method (STM). The study results reveal that, among the four cumulus parameterization schemes selected, the Kain-Fritsch scheme performed the best followed by the Bettts-Miller-Janjic scheme. Conversely, the other two schemes have overestimated the rainfall amount which may be attributed to their inability to account for the altitude variations and different pressure distributions in estimating the convective precipitation. The presence of altitudinal variation shows an increased probability of prediction by 3.5% with an increase of 15% in altitude revealing the importance of altitude in the convection dynamics of rainfall in the WRF.