Objective detection of tropical cyclones in high-resolution analyses

An automated procedure is used to evaluate objective criteria for the detection of observed tropical cyclones in high-resolution (1.125 degrees) European Centre for Medium-Range Weather Forecasts analyses. Best values of the objective criteria are determined for cyclones of tropical storm and hurricane strength. It is found that intense storms are more easily detected than weaker storms. For intense storms in regions of reasonable data coverage, the detection method has some skill, but detection generally is not reliable in regions remote from land. Moreover, the numbers of vortices detected in the analyses is sensitive to relatively small variations in the values of the objective criteria used. Aspects of the structure of analyzed hurricanes in regions of good data coverage are similar to those of observed hurricanes, although the radial variation of tangential wind speeds near the center of the simulated storms differs considerably from reality. A comparison is made between the representation in the analyses of two contrasting storms: Hurricanes Hugo and Andrew. The implications of the detection method for simulations of climate change are discussed.