Informations et ressources scientifiques
sur le développement des zones arides et semi-arides


Université de Bourgogne (2008)




Auteur : ZOUNGRANA Joël

Etablissement de soutenance : Université de Bourgogne

Grade : Master II Recherche Discipline : Géobiosphère – Option : Climatologie /Environnement 2008

Intraseasonal variability of rainfall is studied at the scale of Burkina Faso, one of the Sahelian countries of West Africa. Sahel Drought is well documented (Charney¹, 1975 ; Folland², 1986 ; Dai³, 2004). Beside the negative impact of this long term interannual variability, intraseasonal variability of rainfall also affects the agricultural yields of the region (Sultan⁴, 2002). Sultan and Janicot⁵, (2000) found that the migration of the Inter Tropical Convergence Zone, accompanying the Northward movement of the monsoon in Sahel inlands, is not continuous. It is rather made of abrupt shifts between equilibrium positions. Following those results, Louvet⁶ et al. (2003) found that the monsoon onset in West Africa is characterized by a succession of intensive rainfall phases and stagnation or regression phases also called pauses. They established at the regional scale, a mean calendar for the pauses. Specifying these results at the local scale of Burkina Faso, would be a contribution to the understanding of this phenomenon, and would help for more effective local applications. We used daily observed rainfall data of Burkina Faso, covering 59 years from 1950 to 2008. Three indexes (North, Center-South, and South-West) were built using the objective method of hierarchical ascendant classification. The data were filtered by the low pass filter of Butterworth to eliminate frequencies less than 30 days, corresponding to daily and synoptic fluctuations. The pauses were detected year by year at each index with a simple criterion defining a pause, as continuous period of at least ten days, during which the differences of daily rainfall for consecutive days are less than zero. Four pauses were detected, but only those which are situated within the climatological period of the monsoon season in Burkina Faso were considered. So, three pauses were considered and classified as pause # 2, pause # 3 and pause # 4, to allow comparison with the regional calendar of Louvet⁶ et al. (2003). Then, a local calendar was established by making the arithmetical mean of the dates of each pause on the total years in which the pauses were detected. We found that the pauses of the three indexes are synchronal. The pause # 2 ends meanly between the 10th and the 17th of May, just at the beginning at the monsoon season. The pause # 3 and pause # 4 meanly end respectively between the 19th and 22nd of June, and between the 28th of July and the 4th of August. This local calendar is in accordance with the regional calendar established by Louvet⁶ et al. (2003). Then we studied, using statistical methods, the interannual variability of the calendar throughout our period of study. The aim was to seek for eventual correlations which could help in predicting the pauses and some useful parameters of the annual rainfall cycle, such as the rainfall seasonal amounts. We found significant linear positive correlations between the pauses, indicating that a change (advance/delay) in the date of a given pause should be followed by the same change for the following pause. Studying the impact of the Drought on the local calendar of the pauses, we found no significant change, in accordance with Louvet⁶ et al., (2003) at the regional scale and with Dieng⁷ et al. (2008) at Senegal scale. We compared the distribution of some useful rainfall cycle parameters (seasonal rainfall amount, date of end of the monsoon onset, maximum value of the difference of consecutive daily filtered rainfall) corresponding to advanced pauses years and delayed pauses years. We found two significant differences for the South-West index. Comparatively to the value for delayed pause # 3 years, for advanced pause # 3 years, the mean of seasonal rainfall amounts is more important (surplus of 90 millimeters), and the mean date of end of the monsoon onset is postponed 18 days. The local calendar that we established can be use to plan local activities subjected to seasonal rainfall variability, when the discovered correlations can help in predicting the pauses, the seasonal rainfall amounts and the duration of the monsoon onset.

Version intégrale

Page publiée le 3 décembre 2017