JLE

Science et changements planétaires / Sécheresse

MENU

Analysis of the variability of the onset of the rainy season in the Sahel at hydrological and agronomical scales, based on EPSAT-Niger data Volume 16, issue 1, Mars 2005

Figures

See all figures

Authors
Laboratoire d’étude des transferts en hydrologie et environnement (LTHE), Institut de recherche pour le développement (IRD), 1023, rue de la Piscine, BP 53, 38041 Grenoble cedex 9, France

The onset of the rainy season in the Sahel is studied on a densely instrumented area (1990-2001), allowing an analysis of the spatial variability as well as of the interannual variability to be carried out. The area concerned is located in the region of Niamey (Niger) and covers about 16,000 km². Three criteria are used to define the onset of the monsoon: one is spatial (“climatological”) while the other two others are local (“hydrological” and “agronomical”). The climatological date is linked to the observation of the first organized mesoscale convective system. The hydrological criterion takes into account the first rainfall generating runoff, even if it is a local event. The agronomical criterion requires that no dry spell of more than 7 days should follow the onset. This study compares the onset dates derived from those three criteria over 12 consecutive years. On average, the “hydrological” onset occurs 6 weeks before the “agronomical” onset. The first organised convective system occurs in between. The agronomical onset takes place on average at a date close to the date of the monsoon jump (third decade of June). This is consistent with the fact that meteorological conditions at the monsoon onset lead to shorter dry spells. However, these average results conceal a great spatial variability in our study area. For both local criteria, in a given year, and depending on the station concerned, more than 3 weeks can elapse between the earliest and the latest onset. Similarly, there is an important interannual variability of the onset, whereas the end of the rainy season is much less variable. A correlation analysis between the date of the onset, the length of the season and the total seasonal rainfall shows that the total rainfall is poorly correlated to either the date of the onset or the length of the season. It follows that seasonal forecast techniques based solely on the date of the onset have little predictability potential. The implications of these results for millet crops are discussed as a conclusion.