Robust direct effect of increasing atmospheric CO2 concentration on global tropical cyclone frequency - A multi-model inter-comparison

Rising concentration of carbon dioxide (CO2) is expected to affect tropical cyclone (TC) intensity, frequency, and genesis locations through an increase in global mean sea surface temperature (SST). This has been an area of intensive research for the past few decades with increasing use of high-resolution global climate models (GCMs) and various downscaling approaches. The assumption appears to be that the dominant effect of increasing CO2 on TCs is through an increase in tropical mean SST. However, recent modeling studies suggest that both spatial patterns of SST warming and higher atmospheric CO2 concentrations can significantly affect global and regional TC statistics independent of the global mean SST warming. In this presentation we focus on an examination of the direct or fast effect of CO2 on global TC frequency from the multiple models participating in the U.S. CLIVAR Hurricane Working Group (HWG). For each HWG participating model, a set of common experiments are requested. It includes one control experiment forced by monthly climatological SSTs with present-day GHG concentrations, and three perturbed-forcing experiments that are identical to the control, except 1) SSTs are uniformly increased by 2K (P2K), 2) atmospheric CO2 concentration is doubled (2xCO2), and 3) a combination of 1) and 2) (BOTH). We compare the response in global TC frequency from seven models that have provided sufficient data for this analysis. They include: two GFDL models (HIRAM and C180AM2), ECHAM5 simulations conducted in CMCC, NCEP GFS, USDOE/NSF CAM5H (CAM5.1 running at 25km resolution), NASA-GFSC GOES5, and COAPS simulations from FSU.

The seven models produce substantially different annual global TC frequencies and geographical distributions. Despite the large differences in simulated present-day TC frequency, all models produce a reduction in global TC frequency when both CO2 concentration is doubled and SSTs are uniformly increased by 2K. However, the responses to individual forcing changes (P2K or 2xCO2) show a larger discrepancy among the models. In particular, only two out of the seven models produce a significant reduction in global TC frequency under 2K SST warming while the rest exhibit insignificant changes. None of the models produces a significant increase in global TC frequency. In comparison, six out of the seven models produce a significant reduction in global TC frequency in response to 2xCO2 with GFS being the only one that generates an insignificant change. Hence, the response of global TC frequency to CO2 increases is more robust than its response to SST warming among these models. Among the various environmental variables, mid-tropospheric vertical pressure velocity appears to be the best in explaining global and regional changes in TC genesis frequency among the models.