IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis

3.8.3 Evidence for Changes in Tropical Storms

The TAR noted that evidence for changes in tropical cyclones (both in number and in intensity) across the various ocean basins is often hampered by classification changes. In addition, considerable inter-decadal variability reduces significance of any long-term trends. Careful interpretation of observational records is therefore required. Traditional measures of tropical cyclones, hurricanes and typhoons have varied in different regions of the globe, and typically have required thresholds of estimated wind speed to be crossed for the system to be called a tropical storm, named storm, cyclone, hurricane or typhoon, or major hurricane or super typhoon. Many other measures or terms exist, such as ‘named storm days’, ‘hurricane days’, ‘intense hurricanes’, ‘net tropical cyclone activity’, and so on.

The ACE index (see Box 3.5), is essentially a wind energy index, defined as the sum of the squares of the estimated six-hour maximum sustained wind speed (knots) for all named systems while they are at least tropical storm strength. Since this index represents a continuous spectrum of both system duration and intensity, it does not suffer as much from the discontinuities inherent in more widely used measures of activity such as the number of tropical storms, hurricanes or major hurricanes. However, the ACE values reported here are not adjusted for known inhomogeneities in the record (discussed below). The ACE index is also used to define above-, near-, and below-normal hurricane seasons (based on the 1981 to 2000 period). The index has the same meaning in every region. Figure 3.40 shows the ACE index for six regions (adapted from Levinson, 2005, and updated through early 2006). Prior to about 1970, there was no satellite imagery to help estimate the intensity and size of tropical storms, so the estimates of ACE are less reliable, and values are not given prior to about the mid- or late 1970s in the Indian Ocean, South Pacific or Australian regions. Values are given for the Atlantic and two North Pacific regions after 1948, although reliability improves over time, and trends contain unquantified uncertainties.


Figure 3.40. Seasonal values of the ACE index for the North Indian, South Indian, West North Pacific, East North Pacific, North Atlantic and combined Australian-South Pacific regions. The vertical scale in the West North Pacific is twice as large as that of other basins. The SH values are those for the season from July the year before to June of the year plotted. The timeline runs from 1948 or 1970 through 2005 in the NH and through June 2006 in the SH. The ACE index accounts for the combined strength and duration of tropical storms and hurricanes during a given season by computing the sum of squares of the six-hour maximum sustained surface winds in knots while the storm is above tropical storm intensity. Adapted and updated from Levinson (2005).

The Potential Intensity (PI) of tropical cyclones (Emanuel, 2003) can be computed from observational data based primarily on vertical profiles of temperature and humidity (see Box 3.5) and on SSTs. In analysing CAPE (see Box 3.5) from selected radiosonde stations throughout the tropics for the period 1958 to 1997, Gettelman et al. (2002) found mostly positive trends. DeMott and Randall (2004) found more mixed results, although their data may have been contaminated by spurious adjustments (Durre et al., 2002). Further, Free et al. (2004a) found that trends in PI were small and statistically insignificant at a scattering of stations in the tropics. As all of these studies were probably contaminated by problems with tropical radiosondes (Sherwood et al., 2005; Randel and Wu, 2006; see Section 3.4.1 and Appendix 3.B.5), definitive results are not available.

The PDI index of the total power dissipation for the North Atlantic and western North Pacific (Emanuel, 2005a; see also Box 3.5) showed substantial upward trends beginning in the mid-1970s. Because the index depends on wind speed cubed, it is very sensitive to data quality, and the initial Emanuel (2005a) report has been revised to show the PDI increasing by about 75% (vs. about 100%) since the 1970s (Emanuel, 2005b). The increase comes about because of longer storm lifetimes and greater storm intensity, and the index is strongly correlated with tropical SST. These relationships have been reinforced by Webster et al. (2005, 2006) who found a large increase in numbers and proportion of hurricanes reaching categories 4 and 5 globally since 1970 even as the total number of cyclones and cyclone days decreased slightly in most basins. The largest increase was in the North Pacific, Indian and Southwest Pacific Oceans.

These studies have been challenged by several scientists (e.g., Landsea, 2005; Chan, 2006) who have questioned the quality of the data and the start date of the 1970s. In addition, different centres may assign different intensities to the same storm. The historical record typically records the central pressure and the maximum winds, but these turn out not to be physically consistent in older records, mainly prior to about the early 1970s. However, attempts at mutual adjustments result in increases in some years and decreases in others, with little effect on overall trends. In particular, in the satellite era after about 1970, the trends found by Emanuel (2005a) and Webster et al. (2005) appear to be robust in strong association with higher SSTs (Emanuel, 2005b). There is no doubt that active periods have occurred in the more distant past, notably in the North Atlantic (see below), but the PDI was evidently not as high in the earlier years (Emanuel, 2005a).

There is a clear El Niño connection in most regions, and strong negative correlations between regions in the Pacific and Atlantic, so that the total tropical storm activity is more nearly constant than ACE values in any one basin. During an El Niño event, the incidence of hurricanes typically decreases in the Atlantic (Gray, 1984; Bove et al., 1998) and far western Pacific and Australian regions, while it increases in the central North and South Pacific and especially in the western North Pacific typhoon region (Gray, 1984; Lander, 1994; Kuleshov and de Hoedt, 2003; Chan and Liu, 2004), emphasizing the change in locations for tropical storms to preferentially form and track with ENSO. Formation and tracks of tropical storms favour either the Australian or South Pacific region depending on the phase of ENSO (Basher and Zheng, 1995; Kuleshov and de Hoedt, 2003), and these two regions have been combined.

The ACE values have been summed over all regions to produce a global value, as given in Klotzbach (2006), beginning in 1986. The highest ACE year through 2005 is 1997, when a major El Niño event began and surface temperatures were subsequently the highest on record (see Section 3.2), and this is followed by 1992, a moderate El Niño year. Such years contain low values in the Atlantic, but much higher values in the Pacific, and they highlight the critical role of SSTs in the distribution and formation of hurricanes. Next in ranking are 1994 and 2004, while 2005 is close to the 1981 to 2000 mean. The PDI also peaks in the late 1990s about the time of the 1997–1998 El Niño for the combined Atlantic and West Pacific regions, although 2004 is almost as high. Webster et al. (2005) found that numbers of intense (category 4 and 5) hurricanes after 1990 are much greater than from 1970 to 1989. Klotzbach (2006) considers ACE values only from 1986 and his record is not long enough to provide reliable trends, given the substantial variability.

Box 3.5: Tropical Cyclones and Changes in Climate

In the summer tropics, outgoing longwave radiative cooling from the surface to space is not effective in the high water vapour, optically thick environment of the tropical oceans. Links to higher latitudes are weakest in the summer tropics, and transports of energy by the atmosphere, such as occur in winter, are also not an effective cooling mechanism, while monsoonal circulations between land and ocean redistribute energy in areas where they are active. However, tropical storms cool the ocean surface through mixing with cooler deeper ocean layers and through evaporation. When the latent heat is realised in precipitation in the storms, the energy is transported high into the troposphere where it can radiate to space, with the system acting somewhat like a Carnot cycle (Emanuel, 2003). Hence, tropical cyclones appear to play a key role in alleviating the heat from the summer Sun over the oceans.

As the climate changes and SSTs continue to increase (see Section, the environment in which tropical storms form is changed. Higher SSTs are generally accompanied by increased water vapour in the lower troposphere (see Section and Figure 3.20), thus the moist static energy that fuels convection and thunderstorms is also increased. Hurricanes and typhoons currently form from pre-existing disturbances only where SSTs exceed about 26°C and, as SSTs have increased, it thereby potentially expands the areas over which such storms can form. However, many other environmental factors also influence the generation and tracks of disturbances, and wind shear in the atmosphere greatly influences whether or not these disturbances can develop into tropical storms. The El Niño-Southern Oscillation and variations in monsoons as well as other factors also affect where storms form and track (e.g., Gray, 1984). Whether the large-scale thermodynamic environment and atmospheric static stability (often measured by Convective Available Potential Energy, CAPE) becomes more favourable for tropical storms depends on how changes in atmospheric circulation, especially subsidence, affect the static stability of the atmosphere, and how the wind shear changes. The potential intensity, defined as the maximum wind speed achievable in a given thermodynamic environment (e.g., Emanuel, 2003), similarly depends critically on SSTs and atmospheric structure. The tropospheric lapse rate is maintained mostly by convective transports of heat upwards, in thunderstorms and thunderstorm complexes, including mesoscale disturbances, various waves and tropical storms, while radiative processes serve to cool the troposphere. Increases in greenhouse gases decrease radiative cooling aloft, thus potentially stabilising the atmosphere. In models, the parametrization of sub-grid scale convection plays a critical role in determining whether this stabilisation is realised and whether CAPE is released or not. All of these factors, in addition to SSTs, determine whether convective complexes become organised as rotating storms and form a vortex.

While attention has often been focussed simply on the frequency or number of storms, the intensity, size and duration likely matter more. NOAA’s Accumulated Cyclone Energy (ACE) index (Levinson and Waple, 2004) approximates the collective intensity and duration of tropical storms and hurricanes during a given season and is proportional to maximum surface sustained winds squared. The power dissipation of a storm is proportional to the wind speed cubed (Emanuel, 2005a), as the main dissipation is from surface friction and wind stress effects, and is measured by a Power Dissipation Index (PDI). Consequently, the effects of these storms are highly nonlinear and one big storm may have much greater impacts on the environment and climate system than several smaller storms.

From an observational perspective then, key issues are the tropical storm formation regions, the frequency, intensity, duration and tracks of tropical storms, and associated precipitation. For landfalling storms, the damage from winds and flooding, as well as storm surges, are especially of concern, but often depend more on human factors, including whether people place themselves in harm’s way, their vulnerability and their resilience through such things as building codes.