Simulating widespread extreme rainfall events over the Drakensberg with WRF and MPAS models.

Abstract

Abstract This study investigates the characteristics of widespread extreme rainfall events (WERE) over the Drakensberg with the aid of observation, reanalysis, and simulation data during the period 1987–2016. WERE event over the DMR was defined as a rainfall event during which at least 40% of the grid points over the DMR experienced at least 95% percentile of daily rainfall at the respective grid points. The simulations were performed using the Weather Research and Forecasting (WRF) model and the Model for Prediction Across Scales (MPAS). The capacity of the WRF and MPASS models to represent past characteristics of extreme precipitation over the DMR was evaluated against five observation datasets and the forcing reanalysis data. Precipitation characteristics were represented with four precipitation indices. Both models (MPAS and WRF) simulate realistic rainfall characteristics over Southern Africa, especially over South Africa and DMR. For all the precipitation indices, the models capture the west-east precipitation gradient over South Africa and reproduce the local precipitation maxima over the DMR as well as along the south and southeast coasts of South Africa. Nevertheless, there are notable differences in the performance of the models. While MPAS performs better than WRF in some cases, WRF performs better than MPAS in other cases. All the observation datasets agree that WERE mostly occurs in three seasons over DMR and does not occur every year. However, there are substantial disagreements among the datasets regarding the climatology and annual frequency of WERE. Self-organizing map analysis of grid points where extreme rainfall occurred during WERE events shows that there are 5 major spatial patterns of strong rainfall areas during WERE events over the Drakensberg. The patterns are generally associated with frontal systems, tropical temperate troughs, and ridging highs. Patterns of strong rainfall areas during WERE events identified in this study could help in the management of extreme rainfall-related disasters around the Drakensberg.