Ph.D. Princeton University, 1982
Senior Research Scientist
NOAA/ESRL/Physical Sciences Division
Diagnosis, modeling, and predictability of large-scale weather and climate variations on time scales of days to millennia.
Critical Influence of the Pattern of Tropical Ocean
Warming on Regional Climate Trends Around
Climate models are now sufficiently advanced that they can reasonably simulate the globally averaged as well as some continental- scale aspects of recent climate change, and provide important guidance on future changes on these scales in response to anthropogenic changes in radiative forcing, as summarized in the 2007 Intergovernmental Panel on Climate Change (IPCC) report. This has led to increased interest and confidence in simulations and predictions of climate changes on even smaller sub-continental scales, that could be different and more severe than the globally averaged or continental-scale changes.
To assess the extent to which such confidence may be justified, we compared multi-model ensemble simulations of the last half-century with corresponding observations, focusing on the landmasses around the North Atlantic Ocean: North America, Greenland, Europe, and North Africa. We found that the patterns of the trends over these regions were generally not well captured by the state-of-the-art IPCC coupled atmosphere-ocean models with prescribed observed radiative forcing changes associated with anthropogenic greenhouse gases and other forcings. On the other hand, even uncoupled atmospheric models without the prescribed radiative forcing changes, but with prescribed observed sea-surface temperature (SST) changes only in the tropics were demonstrably more successful in this regard. The basic reason for the poor performance of the coupled models was thus their poor representation of the tropical SSTs. We showed that errors in representing both the observed SST climatology and the spatial pattern of the observed SST trends were important in this regard. The pattern error, in particular, had a large impact on the simulation of both the local and remote precipitation trends.
The sensitivity of the global mean climate to the pattern of tropical oceanic warming was already highlighted in some of our previous work. In this new study, we provided evidence of a similar large sensitivity also of regional climate changes, even in regions remote from the tropics. The fact that even with full atmosphere-ocean coupling, many current climate models with prescribed observed radiative forcing changes are not able to capture the pattern of the observed tropical oceanic warming suggests one of two things. Either the radiatively forced component of this warming pattern was sufficiently small in recent decades to be dwarfed by natural tropical SST variability, or the coupled models are misrepresenting some important tropical physics. Our study suggests that the discrepancy of the simulated trends with respect to observations is not just due to climate noise but also due to model errors. The existence of mean tropical SST biases in the coupled models, whose impact on remote trends is also significant, further supports our argument. Reducing such tropical SST errors is key to significantly improving regional climate predictions around the globe.