Jenni Evans noted that Global Climate Model (GCM) reconstructions of 20th Century hurricane climatology haven't agreed well with observations. Since the GCM's don't reproduce hurricanes themselves (due to inadequate) resolution), what proxy for hurricanes is used in the GCMs to compare to observations? In what respect do they differ? (Note that Tom Knutson's recent work with 18 km resolution regional climate model reconstructions of hurricanes in the Atlantic claims pretty good agreement with observations--within 40%?)

We can reproduce the past 25 years or so of hurricane variability in the Atlantic using a regional downscaling approach and some colleagues are able to do pretty well with a global model using just observed sea surface temperatures.

As for the past century, a complication in comparing models and hurricane observations is that there is considerable uncertainty in the hurricane observations (and even in regional details of sea surface temperatures) going back into the presatellite era. This is a big topic in itself, and I will try to touch on parts of it in my talk.