Science Briefs

Modeling Sahelian Rainfall

The summer monsoon season in Africa's Sahel region is punctuated by the periodic traversal of weather disturbances called "African waves" (AW). This nomenclature derives from the inverted U shape of the upper altitude circulation when AW are present. Since AW often trigger rainshowers, they are very relevant to the interannual cycles of drought that have plagued the Sahel.

Observational climate data from Sahel stations have been studied to advance our understanding of the structure of AW and their role in achieving seasonal precipitation totals. However, station data in this region are sparse, leaving large spatial gaps which limit the analysis. Computer-based simulation models can be used as a tool to interpolate climate analyses, i.e., to fill in the gaps. The Regional Climate Model (RCM) currently in use at GISS/CCSR is especially useful for studying AW because the spacing of its computational grid is only 50 km between adjacent elements. In the current application we limited the domain of the RCM to 70° of longitude by 50° of latitude. Because they cover limited areas, RCM simulations require a constant stream of climate information at the geographic edges of the domain. In the experiments described below, these "lateral boundary data" were supplied from global analyses of meteorological data based on a much coarser computational grid (275 km spacing). The data required to begin the RCM simulation also came from the same source. In effect, the RCM interprets coarser resolution global analyses by giving more detailed structure to the spatial distributions of climate variables, such as the circulation pattern associated with AW.

Fig 1
Fig. 1: Regional Climate Model predictions of upper air circulation (at 3 km) and 24-hr precipitation accumulations (left-hand panels) for Day 2 and 3 of a simulation initialized for Aug. 8, 1988 (12 GMT). Right-hand panels show satellite observed cloud imagery for 3 PM (at 0° longitude) on corresponding days. The spatial extent of satellite observed cloud cover is naturally greater than areas of active precipitation. Note the association of heavy precipitation with the cyclonic (counterclockwise) disturbances.

Using the RCM, a sequence of AW traversals during a 7-day period in August 1988 was studied with encouraging results. This period experienced three episodes of significant rainfall over the Central Sahel. Weather satellite imagery for the afternoons Aug. 9-15, 1988, confirmed that three major cloud clusters made their way, in succession, across West Africa from east to west. The first and the third moved in tandem with AW that are discernible on analyses of the circulation. In contrast, a northwest-southeast oriented cloud grouping that traversed the area between Aug. 12-14 was initiated in moist southeasterly circulation with no AW trigger. On most days, reasonable correspondence was noted between very bright cloud tops observed by satellite and co-located station precipitation measurements.

The figure shows a two-day sequence of RCM predicted upper air circulation (at about 3 km altitude) and predicted precipitation. A cyclonic (counterclockwise) circulation evident near the southwestern Nigeria resembles the pattern given by global observational analyses, but it provides more spatial detail. This feature is associated with several clusters of heavy rainfall, most of which are concentrated to the west of the axis of cyclonic circulation and imbedded within northeasterly winds. Satellite cloud imagery for the afternoon of Aug. 10 confirms the presence of a shield of bright clouds approximately over the location of the predicted precipitation. On Day 3 (Aug. 11) the circulation assumes the inverted U configuration signifying the organization of an AW, in this case located over Nigeria and Niger. While some rainfall lingers near the wave axis, the locus of rainfall within the northeasterlies has moved westward in response to the AW movement. The corresponding westward movement of satellite observed clouds confirms that the RCM prediction for Day 3 is indeed quite realistic. Of course, the area covered by clouds is broader than the area of active rainfall.

Our research indicates that the RCM can make credible predictions out to 5-7 days. Moreover, the quality of these predictions is not much compromised if the initial lateral boundary data are used for the entire period instead of updating them with new information from global analyses synchronized with the prediction. Summaries of monthly RCM simulations will be studied to learn more about the relationship between AW and seasonal rainfall. The RCM therefore shows promise for applications of weather prediction and climate research in the Sahel.


Druyan, L., M. Fulakeza, and W. Thiaw 2000. Regional model simulations of African wave disturbances. J. Geophys. Res. 105, 7231-7255.


Please address all inquiries about this research to Dr. Leonard Druyan.