Downscaling hurricane climatologies from global models and re-analyses

We present a new technique for inferring tropical cyclone climatology from the output of global models and apply it to current and future climate states represented by re-analyses and simulated by a suite of global models developed in support of the most recent IPCC report. This new technique begins by seeding climate states with very large numbers of randomly distributed weak, fledgling vortices, which are then tracked using a beta-and-advection model driven by the explicit, large-scale flow. The CHIPS model is then integrated to determine the intensity evolution of each event. Vortices that fail to achieve maximum winds of at least 17 ms-1 are discarded. We show that this method is largely successful in reproducing the observed seasonal cycle and interannual variability of tropical cyclones in the present climate, and more modestly successful in simulating their spatial distribution. When applied to simulations of global climate with double the present concentration of carbon dioxide, this method predicts substantial changes and geographic shifts in tropical cyclone activity, but with much variation among the global climate models used. Basin-wide power dissipation and storm intensity generally increase with global warming, but the results vary from model to model and from basin to basin. Storm frequency decreases in the southern hemisphere and north Indian Ocean, increases in the western North Pacific, and is indeterminate elsewhere. We demonstrate that in these simulations, the change in tropical cyclone activity is greatly influenced by the increasing difference between the moist entropy of the boundary layer and that of the middle troposphere as the climate warms.