Science Briefs

Satellites and Weather Models Study West African Storms

Weather forecasts made by computer-based models must be started with information from weather observations. The required observational datasets interpolate (fill in) the sometimes wide areas between weather observations, and they are made routinely several times a day on a relatively coarse grid (275 km spacing) over Earth's entire surface.

One computer model at NASA Goddard Institute for Space Studies and Columbia University's Center for Climate Systems Research produces weather analyses and forecasts over West Africa on a grid with only 50 km spacing over a limited area, confined to 70° of longitude by 55° of latitude. The relatively small dimensions of the grid elements in this Regional Model (RM) make it a very appropriate tool for studying weather phenomena with small spatial features, such as summer storms over West Africa.

Because the RM covers a limited area, its simulations require a constant stream of weather information at the geographic edges of the domain. In the experiments described below, these "lateral boundary data" were supplied from the same type of global analyses of interpolated observational weather data that were used to begin the RM simulations. In effect, the RM interprets coarser resolution global analyses by giving more detailed structure to the spatial distributions of weather variables, such as the circulation and rainfall patterns associated with storms.

We compared six June-September RM rainfall simulations to estimates of actual daily rainfall at each 1° square of latitude-longitude over a rainy area of West Africa. These estimates were based on meteorological satellite measurements available from the Tropical Rainfall Measurement Mission (TRMM).

Figure 1: Two time plots of TRMM rainfall. See following text

Figure 1: Side-by-side comparison of a) daily TRMM rainfall observations and b) rainfall in an RM simulation for the region of interest during early August 1998. The vertical coordinate is time, which here is the interval Aug. 1-15. The color keys indicate the rainfall rates. Click to see larger JPEG.

Since the TRMM satellite does not fly over and observe West Africa more than once per day, daily rainfall totals are based on a theoretical blend of TRMM data with readings from a second, geostationary satellite that is continuously poised over one equatorial location. Accordingly, TRMM daily rainfall rates are only approximations of what actually occurs.

Figure 1 shows daily RM rainfall forecasts for August 1998 compared to corresponding daily TRMM observations. The cross sections, which display daily values averaged over 5-15°N between 20°W-25°E, clearly show rainfall centers that move westward on consecutive days.

While the simulated amounts do not exactly match TRMM estimates, there is a remarkable similarity in the paths and timing of the rainfall maxima. For example, both panels indicate that heavy rainfall at 25°E on Aug. 5 spreads to 10°E longitude by Aug. 7 and reaches 15°W by Aug. 8. Similar results were obtained for five other June-September simulations.

Figure 2: Screenshot from animation

Figure 2: A screenshot of a frame from an animation showing one simulation for June-September 2003. Streamlines in the animation show the circulation and color indicates the daily precipitation rate. Click to see a QuickTime movie showing the 700 mb level circulation or the 925 mb level circulation (each about 15 MB). (Credit NASA/GISS, SCSU)

Interestingly, simulated rainfall rates do not usually correspond very well to TRMM estimates within an adjustment period during the first eight days of each simulation. However, a dramatic improvement occurs thereafter, and the comparison remains quite good throughout entire four-month summers. The implications of these results are that RM simulations, after a short adjustment to observational data, will afterwards create very realistic weather scenarios when high quality "lateral boundary data" are provided. Note that the model simulations were not supplied with any information from the satellite instrumentation, yet the RM patterns match the TRMM patterns quite well.

The day-to-day changes in meteorological fields over West Africa can be viewed in the accompanying animations of the RM simulation for June-September 2003, which feature streamlines of wind patterns at approximately 3 km altitude and near the Earth's surface, each superimposed on daily precipitation rates. Notice how some of the precipitation maxima move westward (to the left) and note also occasional wave-like crests in the 700 mb streamlines that invariably also advance westward. Counter-clockwise (cyclonic) circulations at the lower (925 mb) level move westward in tandem with the 700 mb waves. These features represent storms that move across West Africa on their way to the Atlantic Ocean.

These results strongly suggest that both TRMM estimates and RM simulations successfully depict actual westward moving summer storms over West Africa. TRMM daily observations show promise for monitoring rainfall over data-sparse areas, and the RM shows promise for making daily predictions of summer storm tracks over West Africa.


Druyan, L., and M. Fulakeza 2005. Mesoscale climate analysis over West Africa. CLIVAR Exchanges 10, 20, 34-35.

Druyan, L., M. Fulakeza, and P. Lonergan 2006. Mesoscale analyses of West African summer climate: Focus on wave disturbances. Climate Dyn. 27, 459-481, doi:10.1007/s00382-006-0141-9.

External Links


Thanks to Ruben Worrell, John DaPonte, Thomas Sadowski, Paul Thomas and Jian Lian Chan of Southern Connecticut State University for assistance creating the animation in Figure 2.


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