Objective detection of tropical cyclones in climate models
Kevin J.E. Walsh, University of Melbourne, Parkville, Vic., Australia; and M. Fiorino, C. W. Landsea, and K. McInnes
Tropical cyclones are able to be generated by coarse-resolution atmospheric models, but their characteristics differ from those of observed tropical cyclones. Most importantly, their wind speeds tend to be lower, a direct result of limited resolution. This raises the question of what threshold wind speed should be used in a limited-resolution model, such as a climate model, for detecting the tropical cyclones generated by it. While many such studies have been performed, they have used numerous different criteria. In this work, it is argued that a resolution-appropriate threshold detection criterion can be established by examining the characteristics of observed tropical cyclones that have been degraded to the resolution of the model.
Observations of tropical cyclone wind profile characteristics are taken from the extended best track data set of Pennington et al. (2000) and are used to drive the Holland (1980) wind profile model. Only storms that are close in intensity to the observed threshold tropical cyclone wind speed of 17.5 ms-1 are analysed, as these are appropriate to establish a threshold criterion. The resulting 2-D wind field is then averaged at various resolutions for numerous storms, to obtain a resolution-dependent threshold criterion. HRD wind analyses are also similarly analysed for comparison.
The results show a roughly linear dependence of the detection threshold on resolution. At 300 km resolution, a storm that barely exceeds the observed tropical cyclone maximum wind speed will have a typical maximum wind speed of about 10 ms-1. It is suggested that this resolution-appropriate criterion be used for establishing the actual numbers of tropical cyclones generated in a limited-resolution model, and thus also for comparing results of simulations using models of different resolutions. The use of this criterion will enable determination of whether a model is simulating too few or too many tropical cyclones, which will aid in the diagnosis of the results and build confidence in them.
Extended Abstract (32K)
Session 1C, Tropical Cyclones and Climate I - Theory and Modeling
Monday, 24 April 2006, 8:00 AM-9:45 AM, Regency Grand Ballroom
Previous paper Next paper
Browse or search entire meeting
AMS Home Page