The observed poleward trends in the annual-mean LMI are consistent with, and within the range of, the observed expansion of the tropics since about 1979 (refs 3, 4). Several climate-related features have been used to diagnose this expansion, which is thought to be due to increased concentrations of anthropogenic greenhouse gases. These features include ozone depletion in the stratosphere, which lies just above the lowest portion of the atmosphere (the troposphere); the height of the boundary between the stratosphere and the troposphere (the tropopause); and the width of the Hadley circulation, the main meridional overturning circulation in the troposphere, which is characterized by rising air and thunderstorms near the Equator and dry, sinking air at around 30° north and 30° south, where many of the world's deserts are found.
Kossin et al. suggest that two factors known to modulate tropical-cyclone development and intensity may have contributed to the observed poleward migration of annual-mean LMI: deep-layer vertical wind shear, that is, the absolute difference between wind speeds in the upper and lower troposphere; and potential intensity, a thermodynamically based theoretical upper limit of tropical-cyclone intensity that depends on local sea surface temperature and atmospheric temperature and humidity. Many storms never achieve their potential intensity because of competing influences, such as strong vertical wind shear and intrusions of dry air. However, in principle, increased potential intensity and decreased vertical wind shear should promote more-intense storms, all other factors being equal. That such trends moving away from the Equator have been observed over the past 30 years (see Fig. 2 of the paper2) therefore seems at least qualitatively consistent with the observed poleward migration of annual-mean LMI.
Despite the large and statistically significant global trends in the annual-mean latitude of LMI, substantial region-to-region and year-to-year variability is evident. For instance, the North Atlantic region, which has received considerable media attention owing to events such as hurricanes Katrina and Sandy, shows almost no poleward trend on the basis of historical 'best-track' data over the past 30 years. Moreover, when the authors used a state-of-the-art data set of tropical-cyclone intensity (ADT-HURSAT; ref. 5), an opposite, equatorward, trend is found for the North Atlantic (see Table 1 of the paper2). Such regional differences in trends are probably due to climate modes that extend in time beyond the period for which accurate satellite-based data are available.
This is one of the limitations of trend studies based on satellite-derived estimates of tropical-cyclone intensity. Although the post-1970s geostationary satellite era is considered to be the most accurate part of the historical tropical-cyclone record, the relatively short observation period hampers the detection of trends influenced by modes of climate variability whose periodicity spans decades or longer, such as the Pacific Decadal Oscillation6. Any such variability implies that regions in which the poleward migration of annual-mean LMI has been more pronounced over the past 30 years might experience less-pronounced trends in the coming decades, and vice versa. Even on a global scale, a trend of 1° of latitude per decade of tropical expansion (that is, a 10° shift per century, assuming a constant rate of expansion) cannot be sustained without implausible changes to fundamental physical constraints on the global atmospheric circulation, such as Earth's rotation rate.
On year-to-year timescales, variability in tropical-cyclone formation and track is dominated by the phase of the El Niño–Southern Oscillation (ENSO) — the episodic warming (El Niño) and cooling (La Niña) of the surface temperature of the tropical Pacific Ocean. El Niño often promotes an equatorward migration of tropical-cyclone activity, whereas during La Niña a poleward displacement is observed7, concomitant with changes in the width and intensity of the Hadley circulation8. It is therefore plausible that any trend in ENSO could project onto trends in tropical-cyclone activity. Kossin et al. attempt to remove this contribution by accounting for the effect of ENSO on the linear trend of annual-mean LMI latitude and then examining the residual data. The poleward migration remains pronounced and statistically significant, suggesting that ENSO plays only a minor part in the long-term hemispheric and global trends.
Kossin and colleagues' findings provide insight into the response of global tropical-cyclone activity to a changing climate. However, several questions remain unanswered. For instance, will future changes in wind patterns cause storms to move towards or away from coastlines9? What are the key mechanisms driving the observed tropical expansion, and how do these tie in with factors known to modulate tropical-cyclone intensity? Such questions remain the subject of future research.
- Nature Geosci. 3, 157–163 (2010). et al.
- Nature 509, 349–352 (2014). , &
- Ch. 2, 226–229 (Cambridge Univ. Press, 2013). et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.)
- Nature Geosci. 1, 21–24 (2008). , , &
- J. Clim. 26, 9960–9976 (2013). , &
- J. Clim. 25, 3282–3290 (2012). , , &
- Clim. Dyn. 39, 897–917 (2012). , &
- J. Clim. 26, 3357–3376 (2013). , , , &
- Proc. Natl Acad. Sci. USA 110, 15211–15215 (2013). , &
- Bull. Am. Meteor. Soc. 91, 363–376 (2010). , , , &