Simulation of Extreme Rainfall and Flood Events in West Africa

Abstract

It is well known that extreme rainfall events that usually devastate the socio-economic activities in West Africa are inadequately simulated by many conventional uniform-grid global climate models, but little is known about how well they are simulated by the emerging variable-resolution global climate models. The present study examines the performance and sensitivity of the Model for Prediction Across Scales-Atmosphere (MPAS-A or simply ‘MPAS’) in simulating extreme rainfall characteristics over West Africa. Eight indices were used to characterise extreme events. Firstly, the uniform grid version of MPAS (60km resolution) was applied to simulate global climate for the period 1981–2010, and the capability of the model was quantified to capture the characteristics of extreme rainfall events over that period. Secondly, a series of simulations were performed with the variable-grid version of the model to study the sensitivity of the simulated extreme rainfall events to local enhancements in model resolution (i.e., 15km, 10km, and 3km) over West Africa, using two cases of extreme rainfall over the Oti River basin. The results show that MPAS gives a realistic simulation of the spatial distribution of most of the eight extreme rainfall indices with a high pattern correlation coefficient (r>0.8). However, the model overestimates the magnitude of some indices (e.g., the annual number of wet days and the maximum number of consecutive wet days) over the Guinea highlands and along the Guinea coast and Cameroon Mountain and underestimates others over the entire region. The local refinement of model resolution improves its performance in simulating extreme rainfall events over the river basin. The results of the study have applications in improving and implementing MPAS for extreme rainfall predictions over West Africa. Floods are among the most destructive natural disasters with associated adverse impacts on society and the environment, and the present study also assesses the capability of HEC-RAS model version 6.1 in simulating flood events in Mango along the Oti River in Togo. Actual flood events in October 2020 were initially simulated as a model verification, and hypothetic modeling scenarios were simulated to explore the effects of fluvial and combined fluvial and pluvial floods over selected areas A and B. The results indicate that HEC-RAS gives a realistic simulation of the flood extent, which agrees with the local topography. However, while the model underestimates the expanse in some parts of the simulation area, it overestimates it in others, especially in areas A and B. In addition, the combined fluvial and pluvial floods aggravate the impacts of the events over the two areas, as the flood extent and depths are more significant than the ones produced by a single type of flooding. The results of the study have applications in improving and implementing HEC-RAS for flood events predictions over West Africa.