In this study we objectively identified the optimal tropical SST pattern for forcing large North American drought responses (i.e., a map of the drought sensitivity to a unit SST change at each tropical location) through atmospheric GCM integrations in which the response of the Palmer Drought Severity Index (PDSI) was determined for each of 43 prescribed localized SST anomaly patches'' in a regular array over the tropical oceans. The robustness of this optimal forcing pattern was established through the consistency of results obtained using two different GCMs. We also demonstrated that the projection of an arbitrary tropical SST change field on this optimal pattern is an excellent predictor of the North American drought response to that SST field.
Clearly, how strongly the forthcoming changes in tropical SSTs project on this optimal drought sensitivity pattern will have important implications for North American drought. In this context it is noteworthy that this optimal pattern differs markedly in the Pacific Ocean from the dominant SST pattern associated with ENSO, and also shows a large sensitivity of the drought to Indian and Atlantic Ocean SSTs. Accurately predicting North American drought is thus not simply a matter of accurately determining how, for instance, ENSO will respond to global warming, but how the precise pattern of tropical SSTs will change. The poor performance of the IPCC/AR4 models in capturing the spatially varying pattern of tropical SST warming over the last 50 years is not reassuring in this regard. Hopefully, the AR5 models will prove better at representing the SST changes over the next 50 years.