The temporal and spatial characteristics of surrogate tropical cyclones from a multi-millenial simulation

Output from a simulation with the CSIRO Mark 2 climatic model has been used to investigate the secular variability of tropical cyclone formation over the globe using Gray’s Seasonal Genesis Parameter. This simulation differs from previous surrogate studies in using a coupled atmospheric-oceanic model, instead of specified sea surface temperatures, as well as being of multi-millenial duration, compared with decadal length simulations used elsewhere. Mean climatological values for each season for a 5,000-year period indicate that the model replicated the broad patterns of spatial and temporal variability. Results are presented in some detail for three regions, the southwest and northwest Pacific Oceans and the low latitude North Atlantic Ocean. A marked range of temporal variabilities of surrogate tropical cyclone numbers was obtained in the simulation, possibly indicating that the present, observed increase in these numbers may not be outside that attributable to natural variability. The component terms of the Seasonal Genesis Parameter permit the contribution of individual climatic terms to the generation of tropical cyclones to be identified. This approach highlighted the important role of relative vorticity and relative humidity, in addition to the governing influence of vertical wind shear. The remote influence of ENSO, versus that of local sea surface temperature anomalies, on surrogate tropical cyclone numbers was examined and revealed different outcomes depending on the region under consideration. The global total of surrogate tropical cyclone numbers exhibited noticeable interannual variability. The simulation reproduced most of the observed correlations between tropical cyclones and relevant climatic variables, but many of the correlations were not stable within the 5,000-year time series used. This suggests that observed correlations based on, typically, 100-years or less of data may not be representative of possible future outcomes. With minor exceptions all climatological time series associated with the Seasonal Genesis Parameter were found to be Gaussian.