Evaluation of CORDEX Africa multi-model precipitation simulations over the Pra River Basin, Ghana

Abstract

Study regionPra River Basin (PRB) located in Ghana Study focusNow, hydrological and climate change impact studies are shifting towards reliance on openly accessible databases particularly for regions with limited observed datasets. There is therefore the need to evaluate the reliability of such datasets in order to reduce modelling uncertainties and boost confidence in modelling results. We present an evaluation of the performance of nine CORDEX RCA4 Regional Climate Model simulations in replicating the observed precipitation for a 31-year period (1975–2005). New hydrological insights for the regionOn the annual timescale, CanESM2, IPSL, CNRM-CM5 and HadGEM2-ES reproduced minimal annual mean biases (0.8–18.4 %) and thus selected for the seasonal and monthly timescale analysis. Generally, with the exception of spring (March, April and May), all the selected models were able to simulate quite well the seasonal climatology of the Pra River Basin (PRB) with noticeable distinctions in the reproducibility of the spatial patterns, variability and magnitude of the observed data. The multi-model ensemble means indicated strong correlation with observations (r > 0.75) but with weak spatial variability (σ < 0.25). It is recommended that for climate impact assessment and hydrologic modelling studies, multi-model ensembles of CanESM2, IPSL, CNRM-CM5 and HadGEM2-ES be used. However, on singular basis, the CanESM2 and HadGEM2-ES RCA4 simulation outputs present better representation of the climate of the basin.