Performance evaluation of Africa CORDEX regional climate models: case of Kpong irrigation scheme, Ghana

Abstract

This study examined four Regional Climate Models (RCMs) from Coordinated Regional Climate Downscaling Experiment-Africa which are widely used in Africa researches, to select the best performing model against gridded observational dataset over the Kpong Irrigation Scheme area with data from 1964 to 2005. Statistical tools such as correlation coefficient (r), RMSE, and standardized deviations (σ), and Mann–Kendall trend analysis were used to determine model performance. The capability of the models to reproduce measured yearly cycles and inter-annual variability for both precipitation and temperature were also assessed. The average precipitation and temperature biases for all the models are located in the ± 0.8 mm range with temporal correlations below 0.3. The Mann–Kendall trend analysis revealed the existence of predominant decreasing trends. On the other hand, all the models were able to reproduce the inter-annual variability, but were unable to capture the long and short rainfall range and deviations accurately. In general, the RCA4-CanESM2 of the four RCMs reproduces the precipitation and temperature climatology with reasonable skill and suggested as the most efficient for climate impact assessment researches over the study area. Overall, our results provide a systematic diagnosis of the strengths and weaknesses of the four models over a wide range of temporal scale.