The role of aerosols and greenhouse gases in Sahel drought and recovery

We exploit the multi-model ensemble produced by phase 5 of the Coupled Model Intercomparison Project (CMIP5) to synthesize current understanding of external forcing of Sahel rainfall change, past and future, through the lens of oceanic influence. The CMIP5 multi-model mean simulates the twentieth century evolution of Sahel rainfall, including the mid-century decline toward the driest years in the early 1980s and the partial recovery since. We exploit a physical argument linking anthropogenic emissions to the change in the temperature of the sub-tropical North Atlantic Ocean relative to the global tropical oceans to demonstrate indirect attribution of late twentieth century Sahel drought to the unique combination of aerosols and greenhouse gases that characterized the post-World War II period. The subsequent reduction in aerosol emissions around the North Atlantic that resulted from environmental legislation to curb acid rain, occurring as global tropical warming continued unabated, is consistent with the current partial recovery and with projections of future wetting. Singular Value Decomposition (SVD) applied to the above-mentioned sea surface temperature (SST) indices provides a succinct description of oceanic influence on Sahel rainfall and reveals the near-orthogonality in the influence of emissions between twentieth and twenty-first centuries: the independent effects of aerosols and greenhouse gases project on the difference of SST indices and explain past variation, while the dominance of greenhouse gases projects on their sum and explains future projection. This result challenges the assumption that because anthropogenic warming had a hand in past Sahel drought, continued warming will result in further drying. In fact, the twenty-first century dominance of greenhouse gases, unchallenged by aerosols, results in projections consistent with warming-induced strengthening of the monsoon, a response that has gained in coherence in CMIP5 compared to prior multi-model exercises.