6533b7cffe1ef96bd1259856

RESEARCH PRODUCT

Sudan-Sahel rainfall onset: Definition of an objective index, types of years, and experimental hindcasts

Bernard FontaineSamuel Louvet

subject

Atmospheric ScienceSeries (stratigraphy)Index (economics)EcologyEquatorPaleontologySoil ScienceForestryContext (language use)Aquatic ScienceOceanographyFoyerMonsoonStandard deviationGeophysicsSpace and Planetary ScienceGeochemistry and PetrologyClimatologyEarth and Planetary Sciences (miscellaneous)Environmental sciencePrecipitationEarth-Surface ProcessesWater Science and Technology

description

Using rainfall estimates from the 5-day version of CPC Merged Analysis of Precipitation (CMAP) and Global Precipitation Climatology Project data along with multiscale spatial key descriptors of atmospheric dynamics from National Center for Environmental Prediction/Department of Energy reanalysis 2 (NCEP/DOE), we first define a West African monsoon onset index to determine its successive dates of occurrence over the period 1979-2004 (28-29 June in mean with a standard deviation of 8.5 days). Then we focus on the three main types of time evolutions of that index at the moment of the monsoon onset in terms of precipitation, pressure, temperature, and winds at different levels to detect the most robust associated signals and select on the mid-May, mid-June period different sets of potential predictors for onset dates. Basically, late (early) onsets are preceded by more (less) rainfall southward to the equator by the end of May, then by a clear decrease (increase) of the normal northward cross-equatorial gradient in mid-June. Finally, we present experimental cross-validated hindcasts of the dates of onset always based on four predictors using both reanalyzed atmospheric data and observed rainfall estimates. Accurate forecasting schemes are obtained, especially with CMAP rainfall (time tendencies and cross-equatorial gradients) over the Gulf of Guinea. In this context the hindcasted series are highly significant when compared to observations (75% of explained common variance with the sole rainfall predictors) and can reach 82% of variance when a zonal wind descriptor of regional scale is added. Basically, a late (early) onset is preceded by more (less) rainfall southward to the equator by the end of May, then by a clear decrease (increase) of the normal northward cross-equatorial gradient by mid-June.

https://doi.org/10.1029/2005jd007019