One difficulty with applying theoretical concepts to predict tropical cyclone formation for this purpose is that there is no widely accepted quantitative theory of tropical cyclone formation (Emanuel 1986; Rotunno and Emanuel 1987; Bister and Emanuel 1997; Simpson et al. 1997; Ritchie and Holland 1997; Ritchie et al. 2003; Montgomery and Enagonio 1998; Reasor et al. 2005; Tory et al. 2006). In the absence of such a theory, tropical cyclone genesis parameters have been developed that statistically relate large-scale atmospheric and oceanic fields to formation of tropical cyclones. The earlier work of Gray (1975) and the more recent parameters of Royer et al. (1998), DeMaria et al (2001) and Emanuel and Nolan (2004) all show an ability to diagnose tropical cyclone formation when forced by large-scale fields, but since they are diagnostic parameters, none of them necessarily constitute a predictive theory of formation that would be valid in a changed climate. In particular, Gray's parameter is unrealistically sensitive to changes in SST (Ryan et al. 1992), which severely limits its application to climate change studies.
One way to build confidence that these parameters may be useful in a changed climate would be to compare their predictions with the number of tropical cyclones directly simulated by a climate model in current and enhanced greenhouse climates, applying the large-scale fields generated by the models to the genesis parameters. This approach was employed by McDonald et al. (2005), who found reasonable agreement between the predictions of the Royer et al. (1998) Convective Yearly Genesis Parameter (CYGP) and the model simulation of tropical cyclone formation. Chauvin et al. (2006) reached a similar conclusion, while Camargo et al. (2007) showed mixed results. Royer et al. (2008; this volume) apply the CYGP to the enhanced greenhouse predictions of fifteen general circulation models, finding a wide variation of responses of the CYGP, due to the considerable differences in the models' SST predictions in a warmer world. These conclusions are also subject to the criteria used to identify tropical cyclones in the output of climate models (Walsh et al. 2007); if different selection criteria are used, different numbers of tropical cyclones would be detected. One potential use of these cyclone genesis parameters in a detection and attribution study would be to apply them to a suite of forced and unforced model simulations to determine whether there are any systematic differences in the genesis potential between the two and compare the differences to observed trends.
Was this article helpful?