A GCM study of the impact of greenhouse gas increase on the frequency of occurrence of tropical cyclones

In order to make inferences on the possible future changes of tropical cyclogenesis frequency, we apply the diagnostic computation of the Yearly Genesis Parameter (YGP) proposed by Gray (1975) to the large-scale fields simulated by a GCM. The YGP is an empirical diagnostic of the frequency of Tropical Cyclones (TCs) based on six physical parameters computed from seasonal means of atmospheric and oceanic variables. In this paper, we apply the YGP diagnostic to the results of three climate simulations performed with the atmospheric General Circulation Model (GCM) of Meteo-France: ARPEGE-Climat. In a control simulation of the current climate, it is shown that the model has a realistic tropical climatology and that the computed YGP reproduces the geographical distribution of the tropical cyclogenesis frequency. The YGP is then applied to two simulations corresponding to two scenarios of doubled carbon dioxide concentration. The two experiments differ by the sea surface temperatures (SSTs) used as a lower boundary condition. In both simulations the YGP gives a large increase of total cyclogenesis frequency, but without extension of the area of possible cyclone genesis. The increase in YGP is due essentially to the contribution of the ocean thermal energy factor in the thermodynamical potential. The dynamical parameters, on the contrary, limit the cyclogenesis increase and are a major explanation of the difference between the two experiments. This is in agreement with the results of the previous similar study of Ryan et al. (1992) concerning the importance of large-scale atmospheric circulation modifications on tropical cyclone climatology. After discussing the observed relationships between ocean surface temperature and large-scale convection, and questioning the use of a fixed temperature threshold in the diagnosis of tropical cyclone frequency, we propose a modification to the YGP consisting in replacing the thermodynamical potential by a term proportional to the convective precipitation computed by the GCM. For the simulation of the present climate this modification affects only marginally the geographical distribution of tropical cyclone genesis, but for the doubled CO2 case, the modified YGP diagnoses a more limited increase in TC genesis in the Northern Hemisphere and a small reduction in the Southern Hemisphere, which seems in better agreement with other recent modelling studies with high resolution climate models (Bengtsson et al., 1996). We conclude that the modified YGP based on convective precipitation could serve as a useful diagnostic of tropical cyclone genesis, and should be tested in simulations with other GCMs.