CORDEX involves over 20 Regional Climate Modeling and Statistical Downscaling groups and involves 2 phases. The first is a downscaling evaluation phase, where for example, Regional Climate Models are run, for a common region, over the recent past (1989-2008) forced by ERA-interim reanalysis boundary data. These simulations are then used to evaluate the performance of the downscaling methods for this particular region. Similar runs and analysis are made for all the CORDEX regions.
The second phase uses output from a range of CMIP5 global climate projections for the period 1950-2100 as boundary data, to produce regionally downscaled climate projections for various regions at resolutions in the range 50km-10km.
Figure 1 shows the domain employed for all CORDEX integrations over Africa. In this article we present results from the first 5 Regional Climate Models (RCMs) to have completed the evaluation run over Africa-CORDEX. All RCMs were forced with the same lateral and surface boundary conditions, namely ERA-interim and were run for the period 1989-2008, all employing a horizontal resolution of 0.44°.
As of the time of writing this article, the number of RCMs that have now provided Africa-CORDEX results has increased to 9, with two more models expected soon. On figure 1 a number of regions are outlined, for which area averaged precipitation results will subsequently be presented.
Mean seasonal precipitation
In figure 2 and 3, we show the mean precipitation for seasons January-February-March (JFM) and July-August-September (JAS), as simulated by the individual RCMs, along with the 5 RCM ensemble average. Observations are derived from the Tropical Rainfall Measuring Mission (TRMM) 3B42 product (Kummerow etal. 1998) and from the gauge-based data of the University of Delaware (Legates and Wilmott 1990). The ERA-interim forecast/analyzed precipitation is also shown for comparison. All results are averaged over the period 1998-2008, which is the complete length of the TRMM data set.
In JFM all the RCMs capture the large scale structure of the observed precipitation, although it is noteworthy that ERA-interim is somewhat excessive over central Africa compared to TRMM and Wilmott. All the RCMs have a positive precipitation bias near their eastern boundaries over the Indian ocean. This maximum is also seen in ERA-interim, but to a much lesser extent in TRMM, suggesting the excess may be tied to erroneous surface or lateral boundary forcing. While there is some variability in the 5 RCMs, the ensemble mean precipitation aggrees with the TRMM observations extremely well.
By JAS the ITCZ has migrated to ~10° north, interacting with the west African monsoon flow. Again all the RCMs capture the main features, although all of the models simulate more precipitation than seen in TRMM over the Darfur mountains in Sudan (check). A maximum in this region is also seen in the Wilmott gauge-based data set. Whether TRMM underestimates in this region due to resolution or sampling problems is difficult to determine and will require a more detailed observational data set. While most of the RCMs capture the structure of precipitation and dry regions well in JAS, there is a general tendency to overestimate precipitation over equatorial land regions.
Annual cycle of precipitation for two regions in west Africa
Figures 4 and 5, illustrate the mean annual cycle of precipitation for two regions in west Africa, termed WA-N and WA-S on figure 1. We present these two regions as an example to see if the RCMs can capture a difference in the annual cycle of precipitation seen between the coastal strip of WA-S and the more interior-land climate of WA-N. In both the figures, the blue curve from TRMM should be seen as the observed precipitation.
The RCMs tend to capture the difference in the annual cycle of precipitation in these 2 regions, with a double-peaked annual cycle in the coastal region. The DMI model has excessive precipitation in both regions, while the SMHI-RCA model has a too rapid onset of monsoonal rains in the WA-N region, indicating too rapid northward movement of the west African monsoon. Overall the magnitude and timing of precipitation on these spatial scales appears well captured by most of the RCMs.
Kummerow C., W. Barnes, T. Kozu, J. Shiue and J. Simpson, 1998. The Tropical Rainfall Measuring Mission (TRMM) sensor Package. J. Atmos. Oceanic Technol., 15, 809-816.
Legates, D. R. and C. J. Willmott, 1990. Mean seasonal and spatial variability in gauge-corrected, global precipitation. International Journal of Climatology, 10, 111-127