Assessment of WRF microphysics and cumulus parameterizations in simulating heavy rainfall events over Badulu Oya catchment, Sri Lanka

Abstract

Extreme rainfall events leading to severe hydrological impacts warrant an accurate prediction of such events not only on time but also in magnitude. Sri Lanka is a South Asian country that is frequently affected by severe tropical storms. The primary aim of this study was to improve heavy rainfall events forecast during the North-east monsoon over the Badulu Oya catchment, Sri Lanka. This aim was accomplished by simulating precipitation for two extreme North-East monsoon rainfall events using the Weather Research and Forecasting (WRF-ARW) model. A detailed comparison was made between the 24-h spatial distribution of model rainfall and observations obtained from rainfall gauges. Verification was evaluated based on three deterministic approaches. Each rainfall event was simulated multiple times using 15 different parameterization scheme combinations including six microphysics and four cumulus schemes at a 3 km grid resolution. The filtered best model combinations were validated using observations from another two heavy North-East monsoon rainfall events. The key finding from these evaluations was that model configurations with WSM5, WSM6, Kessler and WDM6 microphysics, and KF, BMJ and MKF cumulus schemes displayed the overall best performances. Therefore, these combinations have a good potential for operational use in numerical weather prediction over the said catchment.

Weather forecast using WRFg at 2 km resolution over two provinces in northwestern Panama

“…In this regard, Sierra Lorenzo et al (2020) studied a set of fifteen different MPS (microphysics scheme) and CPS arrays (cumulus parameterization schemes) for Panama and using Pearson's correlation coefficient suggested the best MPS-CPS arrays for the country; these results are shown in Table 1. The choice of a "best model" will depend on the type of atmospheric event being analyzed, for example, for extreme events it is necessary to assess different configurations and determine which one has a better forecasting performance (Gimhan et al, 2022). (1994), Betts (1986), Betts et al (1986), Janjić (2000 2002 3…”