De différents aspects de la variabilité de la pluviométrie en Afrique de l'Ouest et Centrale non sahélienne

La sécheresse observée depuis plus d'une vingtaine d'années dans les pays sahéliens se fait également ressentir plus au sud dans des régions d'Afrique aux climats plus humides. Cette baisse de la pluviométrie et la diminution des apports en eau de surface qu'elle entraîne y sont de nature à pénaliser les différents projets de développements liés à l'eau. Le programme ICCARE mené par l'ORSTOM a pour objet l'identification et les conséquences de cette variabilité climatique dans l'ensemble de la zone non sahélienne d'Afrique de l'Ouest et Centrale, en s'appuyant sur les données de deux cents postes pluviométriques et sur un ensemble de méthodes alliant représentations cartographiques et procédures statistiques de détection de ruptures dans les séries chronologiques, univariées et multivariées. La simple étude des séries chronologiques de hauteurs précipitées annuelles fait apparaître une nette et brutale fluctuation du régime pluviométrique dans toute la région considérée, à la fin des années 1960 et au début des années 1970. D'une manière générale, il apparaît que ce sont les zones à régime pluviométrique extrême qui ont subi les modifications les plus importantes : les plus arrosées (de la Guinée à la Côte d'Ivoire) et les plus arides (la bordure sahélienne au nord de la zone étudiée). Entre les deux, le phénomène est d'intensité plus nuancée. Les différentes procédures statistiques appliquées aux séries de hauteurs annuelles précipitées soulignent l'existence d'une rupture survenue à la fin des années 1960 ou au début des années 1970, et donc en phase avec ce qui a été observé et étudié au Sahel. D'autres variables permettant une caractérisation plus "qualitative" du phénomène ont également été étudiées. Elles apportent un complément d'information quant aux manifestations de cette variabilité pluviométrique et montrent que la variabilité climatique se traduit à différents niveaux (durée des saisons des pluies, quantités précipitées hors saisons des pluies, etc.). L'examen des séries chronologiques depuis l'origine des stations a permis de resituer l'événement observé dans une perspective historique faite d'alternances de périodes sèches et de périodes humides. Le phénomène observé à la fin des années 1960 et au début des années 1970 apparaît, cependant comme le plus significatif du point de vue statistique. Si les causes premières d'apparition du phénomène sont, à l'heure actuelle, encore insuffisamment expliquées, et ce même si certaines activités humaines y ont, sans aucun doute, contribué, cette baisse de la pluviométrie a, bien entendu, des conséquences importantes sur la disponibilité des ressources en eau dans ces régions. Si la carence pure et simple n'est pas à craindre dans ces régions où les quantités précipitées restent importantes dans l'absolu, les effets de cette variabilité climatique peuvent, malgré tout, se révéler désastreux, en ce sens qu'ils modifient les données d'un équilibre déjà souvent mis à mal par ailleurs (pression anthropique et déforestation par exemple).The drought observed for more than twenty years in the sahelian countries has also affected those located more to the South with more humid climates (SUTCLIFFE and KNOTT, 1987; NICHOLSON et al, 1988; MAHE and OLIVRY, 1991; OLIVRY et al, 1993 PATUREL et al, 1995). The decrease in rainfall and consequently that in runoff might penalise development projects linked with water supply. The ICCARE programme led by ORSTOM aimed at identifying and measuring the consequences of this climatic variability in the non-sahelian parts of the West and Central African region as a whole. The study was based on rainfall data from more than two hundred stations, break detection methods in the time series as well as cartographic tools were used. This study allowed to highlight the manifestations of the climatic variability observed for nearly 25 years in West and Central Africa. Whereas it had been thought for a long time that the rainfall deficit was restricted to the sahelian region, this study showed that the forest covered regions were also affected and generally speaking the so-called 'humid Africa'. The decrease in rainfall entails that in runoff and thus a change in water resources availability which is the cornerstone of a fairly great number of development projects. Hydrological regimes variability and possible modifications of rainfall-runoff relationship are to be the next stages of the ICCARE programme, partial results were already published (SERVAT et al, 1997). A simple study of the time series showed straightforward fluctuations of the rainfall patterns (figures 2 and 3), which happened during the late sixties or the early seventies over the whole region. The mapping of the time series analysis depicts a clear general trend towards a shift of the isohyets to the south-south-west from the 1950's to the 1980's. This shift reflects a sharp drop in the annual rainfall over the whole of the non-sahelian west and central Africa. Generally speaking it appears that the zones with extreme rainfall patterns underwent the most important modifications, namely : the wettest ones from Guinea to Ivory Coast and the driest ones, bordering the sahelian region to the North of the studied area. Elsewhere changes are less drastic (SERVAT et al, 1996). The different statistical procedures applied to the series of annual rainfall showed breaks taking place during the late 1960's or the early 1970's (figure 4), which was in keeping with what had been observed in the sahelian region. Rainfall deficits were in the order of 20% and they could reach values higher than 25% (table 1), in particular along the Atlantic Coast or in the North, which upholds the fact that 'humid Africa' was also severely affected by the rainfall variability. Other variables which allow a more qualitative characterisation of the phenomenon were also studied (table 2) (figures 5 and 6) ( PATUREL et al, 1997, SERVAT et al, 1997). They brought complementary information about the ways the rainfall variability expresses itself. The pattern of the rainy season was slightly different from what it had been before the 1970's, its length was generally shorter either because it started later than before or because it ended earlier. Likewise, the rainfall distribution was modified, which resulted in a more 'homogeneous' pattern for the zone with only one rainy season and in a sensible change in the ratio of the rainfall heights of the two rainy seasons. Some of the regions of the so-called 'wooded' savannah saw a modification of their climate with a shift from a 'guinean' climate toward a 'sudanese' one. In west Africa some regions saw also a decrease of the amount of precipitation occurring outside the rainy season, which led to a strengthening of the dry season and contributed, if need be, to the perception of the phenomenon by local populations. The decrease in the number of rainy days, where it was possible to study it, was in line with the rainfall deficit. A complementary statistical approach was carried out, it consists of a spatio-temporal study using a multidimensional exploratory analysis (KHODJA et al, 1998). This led to using a multivariate test for detecting a shift in the mean value. This approach confirmed the results obtained with univariate analyses whether it is for the time location of the break (late 1960's, early 1970's) or whether for the heterogeneous character of the phenomenon both from a spatial or temporal point of view. A major characteristic of this persisting rainfall deficit seems to be the existence of two axes of heterogeneity along the north-south and east-west directions (table 3) (figure 7). The survey of the rainfall time series from the origin allowed to place the drought in a historical perspective. So, it appears that, since the beginning of the century, the region underwent a succession of dry and wet periods, although it is difficult to speak of cycles. The phenomenon observed during the late 60's and the early 70's appears, however, as the most significant from a statistical point of view. Besides, the still lasting period of deficit has displayed a length and an intensity quite remarkable, in particular in the north and west sectors of the studied zone where the phenomenon presents an even more exceptional character (figures 8,9,10 and 11). Even though what brought about this diminution in rainfall remains, until now, unexplained, certain human activities undoubtedly contributed to the aggravation of the phenomenon. Although deforestation cannot be held entirely responsible for the drought, the fact remains that overlogging helped to increase the rainfall deficit in numerous regions along the Atlantic coast and the Guinean gulf. Of course, this rainfall deficit has important consequences on the availability of water resources in those regions. Agriculture, the filling of dams and therefore the hydroelectric production, to mention only a few domains, are strongly penalised by this decrease in resources. If a real shortage is not to be feared in those regions where the quantities of precipitation remain high in absolute terms, the effects of that variability can still prove to be disastrous, as they modify the elements of a balance that is already threatened by other factors (anthropic pressure and deforestation, for example). The ICCARE programme, which goes on with the study of the modifications of the river hydrological regimes, will give answers as to the effect of the rainfall deficit upon water resources availability