Virtually all metrics of Atlantic tropical cyclone activity show substantial increases over the past two decades. It is argued here that cooling near the tropical tropopause and the associated decrease in tropical cyclone outflow temperature contributed to the observed increase in tropical cyclone potential intensity over this period. Quantitative uncertainties in the magnitude of the cooling are important, but a broad range of observations supports some cooling. Downscalings of the output of atmospheric general circulation models (AGCMs) that are driven by observed sea surface temperatures and sea ice cover produce little if any increase in Atlantic tropical cyclone metrics over the past two decades, even though observed variability before roughly 1970 is well simulated by some of the models. Part of this shortcoming is traced to the failure of the AGCMs examined to reproduce the observed cooling of the lower stratosphere and tropical tropopause layer (TTL) over the past few decades. The authors caution against using sea surface temperature or proxies based on it to make projections of tropical cyclone activity as there can be significant contributions from other variables such as the outflow temperature. The proposed mechanisms of TTL cooling (e.g., ozone depletion and stratospheric circulation changes) are reviewed, and the need for improved representations of these processes in global models in order to improve projections of future tropical cyclone activity is emphasized.

© 2013 American Meteorological Society