Predictability of Seasonal Sahel Rainfall Using GCMs and Lead-Time Improvements Through the Use of a Coupled Model

Abstract

The ability of several atmosphere-only and coupled ocean-atmosphere general circulation models (AGCMs and CGCMs, respectively) is explored for the prediction of seasonal July-September (JAS) Sahel rainfall. The AGCMs driven with observed sea surface temperature (SST) over the period 1968-2001 confirm the poor ability of such models to represent interannual Sahel rainfall variability. However, using a model output statistics (MOS) approach with the predicted low-level wind field over the tropical Atlantic and western part of West Africa yields good Sahel rainfall skill for all models. Skill is mostly captured in the leading empirical orthogonal function (EOF1), representing large-scale fluctuation in the regional circulation system over the tropical Atlantic. This finding has operational significance for the utility of AGCMs for short lead-time prediction based on persistence of June SST information; however, studies have shown that for longer lead-time forecasts, there is substantial loss of skill, relative to that achieved using the observed JAS SST.The potential of CGCMs is therefore explored for extending the lead time of Sahel rainfall predictions. Some of the models studied, when initialized using April information, show potential to at least match the levels of skill achievable from assuming persistence of April SST. One model [NCEP Climate Forecasting System (CFS)] was found to be particularly promising. Diagnosis of the hindcasts available for the CFS (from lead times up to six months for 1981-2008) suggests that, especially by applying the same MOS approach, skill is achieved through capturing interannual variations in Sahel rainfall (primarily related to El Niñ o-Southern Oscillation in the period of study), as well as the upward trend in Sahel rainfall that is observed over 1981-2008, which has been accompanied by a relative warming in the North Atlantic compared to the South Atlantic. At lead times up to six months (initialized forecasts in December), skill levels are maintained with the correlation between predicted and observed Sahel rainfall at approximately r 5 0.6. While such skill levels at these long lead times are notably higher than previously achieved, further experiments, such as over the same period and with comparable AGCMs, are required for definitive attribution of the advance to the use of a coupled ocean-atmosphere modeling approach. Nonetheless, the detrended skill achieved here by the January-March initializations (r 5 0.33) must require an approach that captures the evolution of the key ocean-atmosphere anomalies from boreal winter to boreal summer, and approaches that draw on persistence in ocean conditions have not previously been successful.Corresponding author address: Ousmane Ndiaye,

Variability and Predictability of West African Droughts: A Review on the Role of Sea Surface Temperature Anomalies

“…They found that the multimodel ensemble improved the spread-to-skill ratio and average skill score over the use of a single model by as much as 10% when measured as potential economic value. Philippon et al (2010) and Ndiaye et al (2011) found that skill can be enhanced by predicting modes of variability using model output statistics (MOS) instead of using direct model output and therefore such enhancements are used to produce forecasts.…”

Seasonal rainfall predictability over the Lake Kariba catchment area

“…MOS uses predictor values from the CGCM in both the development and forecast stages, subsequently reducing model errors. This notion of statistical post-processing has already been tested and successfully employed in studies of AGCM versus coupled model performance Ndiaye et al, 2011). In addition, the post-processing will also have as a result model forecast data directly applicable over an area of interest, such as the 0.5° x 0.5° grid of the CRU data across the Lake Kariba catchment (e.g., Landman and Goddard, 2002;Shongwe et al, 2006).…”

Seasonal Rainfall Prediction Skill over South Africa: One- versus Two-Tiered Forecasting Systems

“…Such empirical remapping of GCM fields to regional rainfall has already been successfully employed (e.g., Landman and Goddard 2002;Shongwe et al 2006). Model output statistics (MOS) equations are developed here because they can compensate for systematic deficiencies in the global models directly in the regression equations (Wilks 2006) and have already been successfully employed in a study of AGCM versus coupled model performance (Ndiaye et al 2011). Since MOS uses predictor values from the global models in both the development and forecast stages, these model errors are subsequently reduced.…”