In contrast to tropical cyclone numbers, the theory of tropical cyclone intensities appears to be on a firmer foundation. The theory of tropical cyclone maximum potential intensity (MPI; Emanuel 1986, 1988) suggests that a tropical cyclone may be viewed as a Carnot cycle heat engine, with the warm reservoir being the sea surface temperature (or upper ocean heat content) and the cold reservoir being the upper tropospheric outflow temperature. The alternative, thermodynamic adjustment theory of Holland (1997) gives similar results. The application of earlier versions of these theories to the output of GCM simulations has suggested that increases in peak tropical cyclone intensities of 5-10% could occur some time after 2050 (Emanuel 1987; Henderson-Sellers et al. 1998; Walsh 2004).
Emanuel (2007) points out that the MPI predictions of Emanuel (1987) for the rate of change of intensity increase in the Atlantic since the 1970s, based upon the observed increase in SST, are considerably less than the observed changes in intensity in the Atlantic during that time. Emanuel (2007) has presented a new calculation based on the revised technique of Bister and Emanuel (2002). This version results in much better agreement with the observed intensity change in the Atlantic. Emanuel (2007) investigated the causes of the observed increase of tropical cyclone power dissipation index (PDI) over the period since 1950. He also created a diagnostic parameter that included the effects of changes in potential intensity, low-level vorticity and vertical wind shear. The results showed that the observed increase of PDI in the Atlantic since 1980 was consistent with changes in these three factors, including increases in low-level vorticity and potential intensity. The increases in potential intensity since 1980 were caused by increases in SST and decreases in upper troposphere temperature in the tropical Atlantic, thus increasing the thermodynamic efficiency of tropical cyclones. Note that the PDI is an integrated measure of cyclone characteristics and as such may not be the most sensitive variable for use in studies of detection and attribution.
Thus these results appear to indicate, with good skill, relationships between trends in large-scale variables and tropical cyclone PDI. Regarding attribution of these trends, it is clear that there is a relationship between the increases in Atlantic tropical SST and similar increases in global temperatures that have been well ascribed to global warming (Elsner 2006; Mann and Emanuel 2006; Trenberth and Shea 2006; Elsner 2007). In particular, Santer et al. (2006) showed this by using the standard formal attribution methodology in which a number of model simulations of 20th century climate were run with and without greenhouse gas forcings, so this is a conclusion with high confidence. It is also true that a number of studies have demonstrated a plausible statistical relationship between SST increases and intense tropical cyclone numbers in the Atlantic (Hoyos et al. 2006; Holland and
Webster 2007) and increases in SST are an expected consequence of global warming. Nevertheless, the decreases in tropical upper troposphere (100 hPa) temperature, which Emanuel (2007) found contributed to increased Atlantic tropical cyclone intensity, are not expected consequences of global warming. Meehl et al. (2007), among many others, show that tropical temperatures are expected to warm at this altitude in the 21st century, which is inconsistent with an anthropogenic cause for the observed cooling. It is also not clear why the low-level vorticity in the tropical Atlantic has been increasing. Nor is it clear what the relationship is between low-level vorticity and global warming. Thus the attribution of the increases in Atlantic tropical cyclone PDI to factors related to global warming is of less confidence as a result. For increased confidence, one would have to examine the simulation of Emanuel's (2007) diagnostic PDI in climate models forced with and without anthropogenic factors over the late 20th century. Additionally, the relationship between SST and PDI is much weaker in the northwest Pacific than in the Atlantic, where SSTs have also been increasing since 1980, but where the trend in PDI is not pronounced (Klotzbach 2006; Emanuel 2007). This is due to different trends in vertical wind shear and vorticity in this region. These different regional trends would also have to be seen in 20th-century climate simulations for confident attribution.
Was this article helpful?