Optimized Cumulus Parameterization in Wrf Model for Simulation of Extreme Rainfall Over East Africa

Abstract

The use of numerical climate models for accurate simulation of weather and climate information is very crucial for optimum planning and management of weather related activities. The current study applied the Weather Research and Forecasting (WRF) model as the primary research tool in simulating the extreme rainfall characteristics over East Africa (EA) region. The study focused on the customization of the WRF model through the simulations and evaluation of Kain-Fritsch (KF), Kain-Fritsch with a moisture-advection based trigger function (KFT), Grell Dévényi (GRELL) and Betts Miller Janjic (BML) cumulus parameterization schemes (CPSs) to identify which scheme improves simulation of the various rainfall characteristics for the wettest and driest years over the EA region. For higher skill and accuracy in the model simulations and proper customization of WRF model over the region, three adjustable parameters from the cumulus schemes namely the sub-grid scale cloud–radiation coupling, adjustment time scale (ATS) and entrainment based on the lifting condensation level (LCL) were identified and tested. The main rainfall characteristics analyzed included the number of rainy days (NRD), intensity of rainy days (IRD) and the frequency of rainfall intensity (FRI) during the core rainfall seasons of March-April-May (MAM) and October-November-December (OND). The model evaluation was done using daily Climate Hazards Group Infrared Precipitation with Station data (CHIRPS) and ERA-Interim reanalysis. The evaluation/ error analysis metrics adopted in the study were root mean square error (RMSE), mean bias error, spatial correlation (SC), coefficient of variation (CV), Taylor diagram and box and whisker plots. The atmospheric dynamics calculated included the convergence and divergence analyses, vertically inter-grated moisture fluxes (VIMF) and vertical velocity. Generally, the cumulus schemes succeeded in reproducing the observed rainfall features associated with large scale systems like the Inter-tropical convergence zone (ITCZ). The GRELL scheme favored drier rainfall conditions over the eastern parts of the region. The KF and KFT schemes generated wetter rainfall conditions mainly confined to the western parts of the region. The analyses of observed NRD showed that the western and eastern parts of the region were characterized by more and fewer NRD respectively. These were fairly reproduced by the KF, KFT and GRELL CPSs. The BML scheme alternated the areas under wet or dry rainfall biases.

The Indian Ocean, Congo Basin and eastern parts of the region experienced persistent rainfall biases. More NRD over the western parts of the equatorial region were mostly related to light rainfall intensity (1-10mm/day). The biases associated with the heavy rainfall (>20mm/day) events were in most cases very low in the KF and GRELL cumulus schemes. The schemes simulated too many rainy days under the light rainfall category over the western and eastern parts of the region, the heavy rainfall events were quite well reproduced over the same locations by the KF scheme. The simulations of vertically inter-grated moisture flux (VIMF), vertical velocity circulations and upper level (200hPa) divergence/convergence showed a steady easterly moisture transport characterized by the development of anti-cyclonic flow over the northern part of Madagascar and along EA coasts were fairly simulated by the KF and GRELL schemes. The moisture convergence biases were found to be larger continentally as well as over parts of the nearby Indian Ocean, corroborating the observed larger rainfall biases noted over parts of East Africa region. In conclusion, the study has provided a unique way of isolating the deficiencies and limitations of four cumulus parameterization schemes used in the WRF model in simulating extreme rainfall. A robust analyses and evaluation of suite of model simulations resulted to the identification of the modified Kain-Fritsch scheme (KF-new) as the most suitable for application in WRF model in order to improve simulations of key rainfall characteristics of East Africa rainfall. By adopting a more robust user relevant metrics makes the findings more relevant contributing useful knowledge that can lead to improved numerical weather prediction (NWP) tools/systems over EA region. The tuning and testing of the specific adjustable parameters within the cumulus schemes led to the identification of the best cumulus scheme that can be used to improve the predictability of rainfall characteristics over the region. This contributes to the new knowledge and understanding of how choice and use of specific cumulus parameterization schemes in WRF model can influence the predictability and prediction of East Africa rainfall. The information from this study is an important contribution to the previous and ongoing research aimed at appropriately customizing numerical weather prediction models and climate modelling systems to skillfully/accurately predict and project climate and weather changes over the East Africa region.