Research
Climate change in Africa
Changes in extreme weather under global warming
Technical details
The GHCN v2 rainfall stations for each region were clustered into homogeneous zones based on the 1921-1990 year-to-year variations in total precipitation. Monthly rainfall has been stratified into seasonal totals according to the main rainfall season for each homogeneous zone, taking into account biases of the models.
Coupled GCMs used
In this study, 12 models have been objectively selected on the basis of the realism with which they represent the observed 20th century seasonal cycle of African precipitation variation. For this we compared the monthly precipitation fields of all experiments available at the CMIP3 archive at PCMDI at the time of writing (2007) with the CRY TS 2.1 analysis in both spatial correlation and root mean square difference. The results are given in the table below.
Mean of monthly correlation and mean squared difference [mm/dy] of satellite estimates and 20th century GCM experiments with the CRU TS 2.1 analysis. We demand that the model is among the 6 best ones (denoted by blue correlation and difference values) in at least one region, and has enough data in the period 2050-2200 (failure on this criterium is denoted by a star)
| Obs | Africa | South | East | Northeast | West | |
|---|
| CRU TS 2.1 |
1.00/0.00 |
1.00/0.00 |
1.00/0.00 |
1.00/0.00 |
1.00/0.00 |
|
| GPCP v2.1 |
0.98/0.44 |
0.94/0.36 |
0.94/0.54 |
0.93/0.35 |
0.98/0.49 |
|
| CMAP |
0.98/0.50 |
0.96/0.30 |
0.92/0.65 |
0.92/0.37 |
0.97/0.57 |
|
| Model | Africa | South | East | Northeast | West | incl. |
|---|
| BCCR CM2.0 |
0.87/1.46 |
0.68/1.35 |
0.80/1.50 |
0.81/1.12 |
0.87/1.16 |
no* |
| CCCMA CGCM 3.1 T42 |
0.87/1.19 |
0.80/0.71 |
0.79/1.94 |
0.79/1.12 |
0.88/1.17 |
yes |
| CCCMA CGCM 3.2 T63 |
0.86/1.28 |
0.74/0.84 |
0.75/2.23 |
0.84/1.22 |
0.87/1.17 |
no* |
| CNRM CM3 |
0.82/1.57 |
0.78/1.08 |
0.81/1.43 |
0.79/1.23 |
0.77/1.57 |
yes |
| CSIRO Mk3.0 |
0.88/1.20 |
0.66/0.90 |
0.79/1.59 |
0.75/0.97 |
0.81/1.26 |
yes |
| GFDL CM2.0 |
0.91/1.27 |
0.84/0.79 |
0.79/1.76 |
0.82/1.00 |
0.92/1.11 |
yes |
| GFDL CM2.1 |
0.89/1.33 |
0.78/0.89 |
0.72/1.45 |
0.68/1.03 |
0.92/1.14 |
yes |
| GISS AOM |
0.78/1.47 |
0.66/1.12 |
0.53/1.46 |
0.59/1.60 |
0.87/1.25 |
no |
| GISS EH |
0.85/1.28 |
0.62/1.14 |
0.40/1.64 |
0.65/1.19 |
0.87/1.25 |
no |
| GISS ER |
0.85/1.31 |
0.67/1.29 |
0.18/2.93 |
0.71/1.18 |
0.88/1.27 |
no |
| IAP FGOALS 1.0g |
0.87/1.29 |
0.82/0.92 |
0.79/1.64 |
0.60/1.19 |
0.87/1.19 |
no |
| INM CM3.0 |
0.84/1.26 |
0.76/0.78 |
0.70/1.26 |
0.58/1.07 |
0.85/1.24 |
no |
| IPSL CM4 |
0.82/1.38 |
0.82/0.60 |
0.77/1.94 |
0.78/0.84 |
0.81/1.50 |
yes |
| MIROC 3.2 (hires) |
0.90/1.65 |
0.79/0.83 |
0.77/1.94 |
0.83/1.59 |
0.88/1.95 |
no* |
| MIROC 3.2 (medres) |
0.87/1.48 |
0.67/0.92 |
0.74/1.55 |
0.76/1.17 |
0.89/1.68 |
yes |
| MIUB ECHO-G |
0.87/1.17 |
0.65/1.02 |
0.75/1.18 |
0.61/1.56 |
0.89/1.07 |
yes |
| MPI ECHAM5 |
0.92/0.88 |
0.67/0.76 |
0.80/1.20 |
0.88/0.59 |
0.93/0.82 |
yes |
| MRI CGCM 2.3.2a |
0.83/1.40 |
0.79/0.98 |
0.68/2.62 |
0.81/1.78 |
0.91/1.09 |
yes |
| NCAR CCSM 3.0 |
0.80/1.58 |
0.73/0.99 |
0.72/1.66 |
0.54/1.79 |
0.67/1.59 |
no |
| NCAR PCM1 |
0.78/1.97 |
0.65/1.31 |
0.77/2.24 |
0.55/2.11 |
0.74/1.73 |
no |
| HadCM3 |
0.88/1.18 |
0.76/0.86 |
0.79/1.45 |
0.76/0.90 |
0.85/1.37 |
yes |
| HadGEM1 |
0.88/1.20 |
0.78/0.91 |
0.84/1.36 |
0.81/0.78 |
0.89/1.15 |
yes |
Not all models perform well everywhere in Africa. In northeast and West Africa, most models perform least well. But since the study covers almost the entire African Continent south of about 20° N, for consistency, the same set of models is used everywhere.
For these models, we investigated the likely changes in precipitation (mean and extremes) using the runs forced with the Special Report Emission Scenario (SRES) A1B scenario. In this scenario, the concentration of the major greenhouse gases reach its maximum around 2050, declining thereafter. After 2100, the forcing is constant at about 2× present-day CO2. For each of the selected IPCC models, we computed 10-year and 100-year return values by inverting a generalized Pareto distribution (GPD) fitted to the 20% driest or wettest seasons in the 20th century (control; 1901-2000) and future (2051-2200) climates. Likely changes in extreme precipitation were then calculated from the future to control climate ratio of these return values. The procedure has been repeated fitting 10-year block maxima to a generalized extreme value (GEV) distribution, the results were indistinguishable from the GPD fits. Changes in mean precipitation rates have been estimated analogously, but the ratios of the two (future and control climate) sub-sample means (averages) are used instead.
Global Coupled Climate Models used in this study. #20C3M (#SRESA1B) show the number of ensemble integrations by each model in the 20th century (future; 2051--2200) climate
Acknowledgments
We acknowledge the international modeling groups for providing their data for analysis, the Program for Climate Model Diagnosis and Intercomparison (
PCMDI) for collecting and archiving the model data, the JSC/CLIVAR Working Group on Coupled Modelling (WGCM) and their Coupled Model Intercomparison Project (CMIP) and Climate Simulation Panel for organizing the model data analysis activity, and the
IPCC WG1 TSU for technical support. The IPCC Data Archive at Lawrence Livermore National Laboratory is supported by the Office of Science, U.S. Department of Energy.
This research was partly funded by the Dutch Ministry of Foreign Affairs, Development Cooperation
