88th Annual Meeting (20-24 January 2008)

Monday, 21 January 2008
Statistical characterization of the dry spell risk during the west African monsoon from meteorological station data
Exhibit Hall B (Ernest N. Morial Convention Center)
Asher Siebert, International Research Institute for Climate and Society, Palisades, NY; and N. Ward, O. Ndiaye, and P. Kangah
Poster PDF (1.3 MB)
This analysis was motivated by a desire to understand the statistical properties of the observed rainfall pattern in several locations in Western Africa. Daily rainfall data are analyzed from three stations in Mali (Mopti, Kolokani and Segou for the period 1935-1997) and two stations in Senegal (Louga and St. Louis, for the periods 1960-2006, 1960-2000, respectively), as provided for this study by the meteorological services in Mali and Senegal. While these site locations are fairly specific, results from this study are intended to be more broadly replicable and to promote further inquiry. The primary variable emphasized by this study is the dry spell risk (the climatological probability of a dry spell of a given length starting on a given day), although other summary statistics are also included. Certain well established features of West African climate are noted in these summary results that have been found in numerous other studies of the climate of the West African Sahel: a high level of spatial coherence in seasonal rainfall anomaly, a more variable monsoon onset than termination, a south-north precipitation gradient, a trend towards drought from the 1950s to the 1980s and a slight rebound in the 1990s. However, the one feature of this study that is most unique is a “stutter” in the climatological onset of the monsoon, signified by a drop and then a subsequent rise in the dry spell risk at the outset of the season before the main “core” of the monsoon has developed. In some instances, there is a weaker “stutter” in the climatology of the termination phase of the monsoon, but this result is less consistent and robust. Further dynamical and statistical research will need to be conducted to understand the mechanisms, spatial and temporal extent of this phenomenon.

Supplementary URL: