Validation of the Space-Time Variability of African Easterly Waves Simulated by the CNRM GCM.

International audienceThe ability of the Centre National de Recherches Météorologiques (CNRM) general circulation model (GCM) to properly simulate the space-time variability of the African easterly waves (AEWs) has been examined over the period 1982-88. The relative vorticity at 850 hPa in the analyses of the European Centre for Medium-Range Weather Forecasts analyses has been compared with the CNRM GCM simulations carried out in the framework of the Atmospheric Model Intercomparison Project. Detailed results of two statistical diagnoses used in this study have been presented for the selected year of 1985 in order to evaluate against the results of extensive studies available for this year. First, with the help of a space-time spectral analysis, it has been shown that the GCM produces only one main variance maximum for the waves located between the two analyzed variance maxima (consistent with the northern and southern components associated with AEWs). The waves in the GCM propagate more slowly than those from the analyses, mainly because of greater periods in the GCM. There is a qualitative agreement between the simulation and the analysis in terms of the seasonal evolution of the location and intensity of the variance maximum. Barotropic and baroclinic energy transfers seem to be quite correctly represented. Second, using a complex empirical orthogonal functions analysis, the existence of two main modes of variability of AEWs has been illustrated. A first hypothesis is proposed for the two modes' interpretation. The second mode seems to be well linked to the interannual rainfall variability over West Africa. A low-frequency modulation of the wave activity exists in the analyses. This modulation is only qualitatively duplicated in the GCM. Interannual variability of the variance shown by the analyses is quite well reproduced by the GCM, particularly the variance seesaw of the two main modes between dry and wet years. Nevertheless, the GCM does not seem able to correctly simulate the pattern of atypical years, such as 1987. Finally, by projecting the GCM dataset on the eigenvectors of the analyses, it has been shown that GCM time discrepancies could influence the quality of the space representation of the AEWs