Non-stationary analysis of spatial patterns of extreme rainfall events in West Africa

Heavy storm events frequently cause extensive damage, and often result in loss of life and property. The objective of this work is to build maps of Annual Maximum Daily Rainfall (AMDR) for various return periods for the Senegal River Basin. However, traditional stationary analyses are not suitable, since meaningful trends have been detected in historical hydrometeorological time series. Therefore, the GAMLSS (Generalized Additive Models for Location, Scale and Shape) tool is applied to fit the parameters of the probability density functions (pdfs). AMDR time series were estimated using observed daily rainfall grids and regional climate models (RCMs). The wide divergence in predicted trends from RCMs imposes the use of ensemble pdfs, which can be built using bootstrapping techniques. The plausible AMDR maps associated with various quantiles, interpolated from these ensemble pdfs, could be used by stakeholders to develop strategies of mitigation and adaptation to climate change impacts on floods events.