Extreme Rainfall over West Africa: Current State and Projected Impacts of Climate Change

Due to the rainfall recovery in West Africa that commenced in the 1990s and in combination with population growth and urbanisation, extreme precipitation events, often accompanied by urban flash floods, have increasingly caused socio-economic distress in the region. Inhabited by one of the most vulnerable societies to extreme precipitation on earth, West Africa relies upon functioning early warning systems as well as sustainable adaptation measures in order to mitigate the flood-related risks. In light of climate change where the 1-day maximum rainfall are projected to increase in large parts of West Africa, these efforts are more pressing than ever. However, developing adaptation strategies generally require reliable and ideally long-term observational datasets which West Africa have become notoriously depleted from in recent decades. Consequentially, precipitation-based evaluations in all recent climate reports generally exhibit high uncertainties. These challenges have largely been tackled by the availability of satellite-based products which allow for a spatiotemporally near-seamless and high-resolution monitoring of precipitation and its extremes even in regions devoid of any in-situ measurements. Thus, exploiting the capabilities of the high-resolution Global Precipitation Measurement (GPM) dataset “Integrated Multi-satellite Retrievals for GPM (IMERG)” (dt=30 minutes, dx=0.1°) in its latest versions 6 and 7 in a systematic fashion, this study, which is part of the FURIFLOOD project (Current and future risks of urban and rural flooding in West Africa – An integrated analysis and eco-system-based solutions) funded by the Federal ministry of Education and Research (BMBF), aims towards a contribution to establish resiliency against climate extremes over West Africa. To approach this goal, this study evaluates the present-day and potential future statistics of daily precipitation extremes over West Africa (20°W-15°E, 0°-20°N) to extend the relatively thin body of current knowledge of West African precipitation extremes.

Utilizing an extreme value analysis approach, extreme precipitation values in IMERG for the period 2001-2022 are modelled with the generalized extreme value (GEV) distribution through yearly-based block maxima sampling of daily rainfall. In order to decrease the spatial uncertainty in a pure single-grid-point-based analysis, the approach is further extended by a regional frequency analysis (RFA) based on the so-called “Index-flood method”. With the assumption that neighbouring grid points exhibit similar rainfall characteristics, the sample of daily rainfall for a given grid point is increased following the “trade space for time” philosophy.

Results show that the spatial pattern of return values for a given return period are strongly correlated with the pattern of mean daily rainfall, which suggests that the magnitude of mean daily rainfall is widely driven by precipitation extremes. High return values, up to around 300 mm at a 50-year return period, are largely found over the coastal areas of West Africa, highlighting, among other things, the influence of the land-sea breeze convection on the formation of intense convection and orographic enhancement of rainfall along the Guinea Highlands. Thus, while extreme precipitation is prevalent along the highly urbanized coast, return values decrease with (a) distance from the coastline, and (b) towards the climatologically drier Sahelian region.

In a further step, the projection of future precipitation extremes is compiled using the statistically downscaled dataset of CMIP6 models “NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP-CMIP6)”. By determining the difference in the GEV parameters between future scenarios and the historical runs, an adjustment of the IMERG-based GEV parameters is accomplished to mimic a potential future state of the rainfall distribution (“Delta method”). First results with the most extreme scenario SSP5-8.5 projected onto the long-term period 2081-2100 suggest an increase of the return value magnitude by 50% and more, stressing again the need for reliable flood action plans in the future. These efforts will be expanded for additional scenarios and periods in line with the IPCC AR6 and also by the use of station data.