P3.5
Influence of local SSTAs and surface processes on the Sahel Rainfall

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 31 January 2006
Influence of local SSTAs and surface processes on the Sahel Rainfall
Exhibit Hall A2 (Georgia World Congress Center)
Marco Gaetani, IBIMET/CNR, Rome, Italy; and G. A. Dalu, V. Capecchi, and M. Baldi

Poster PDF (63.4 kB)

An intense and rapid cooling of the equatorial Atlantic waters of the Guinea gulf precedes the West Africa monsoon intensifying the sea-land temperature contrast. This temperature gradient exerts, from April to July a significant influence on the West African monsoon, since it intensifies the southwesterly flow which pushes the ITCZ inland. Using rainfall data from GPCP dataset and SST data from NOAA Optimum Interpolation SST dataset in the period 1979-2004, we correlate the rainfall in the Sahel, cumulated in July-September (JAS), with the SSTAs in the gulf of Guinea, averaged within a 3 month sliding window. We find that the correlation between the monsoonal rainfall and the previous fall SSTAs is very high. The rainfall in the year 0 and the October-December (OND) SSTA in the year -1 correlation is R=0.73.

The monsoonal dynamics in the Sahel is sensitive to the interannual fluctuations in the meridional gradient of the MSE, moist static energy in the PBL. Wet conditions are related to large gradients and dry conditions are related to weak gradients. In this scenario, the rainfall in the months preceding the rainy season plays an important role, because the boundary-layer MSE content is regulated by the soil wetness via re-evaporation. Applying the same method to the rainfall in the Guinea Coast region, we find a sizable correlation between the JAS Sahel rainfall in the year 0 and the OND Guinea Coast rainfall in the year -1 (R=0.39).

A multi-linear regression with cross validation is used to equate the JAS rainfall in the Sahel with the OND SSTA in the Gulf of Guinea and the OND rainfall in the Guinea region. It results a good correlation between the observed and the forecasted rainfall (R=0.65) and the hit rate is 0.55 for dry events and 0.89 for wet events.