1854
JOURNAL OF CLIMATE
VOLUME 25
Changes in the Extratropical Storm Tracks in Response to Changes in SST in an AGCM LISE SELAND GRAFF AND J. H. LACASCE Meteorology and Oceanography Section, Department of Geosciences, University of Oslo, Oslo, Norway (Manuscript received 29 March 2011, in final form 12 August 2011) ABSTRACT A poleward shift in the extratropical storm tracks has been identified in observational and climate simulations. The authors examine the role of altered sea surface temperatures (SSTs) on the storm-track position and intensity in an atmospheric general circulation model (AGCM) using realistic lower boundary conditions. A set of experiments was conducted in which the SSTs where changed by 2 K in specified latitude bands. The primary profile was inspired by the observed trend in ocean temperatures, with the largest warming occurring at low latitudes. The response to several other heating patterns was also investigated, to examine the effect of imposed gradients and low- versus high-latitude heating. The focus is on the Northern Hemisphere (NH) winter, averaged over a 20-yr period. Results show that the storm tracks respond to changes in both the mean SST and SST gradients, consistent with previous studies employing aquaplanet (water only) boundary conditions. Increasing the mean SST strengthens the Hadley circulation and the subtropical jets, causing the storm tracks to intensify and shift poleward. Increasing the SST gradient at midlatitudes similarly causes an intensification and a poleward shift of the storm tracks. Increasing the gradient in the tropics, on the other hand, causes the Hadley cells to contract and the storm tracks to shift equatorward. Consistent shifts are seen in the mean zonal velocity, the atmospheric baroclinicity, the eddy heat and momentum fluxes, and the atmospheric meridional overturning circulation. The results support the idea that oceanic heating could be a contributing factor to the observed shift in the storm tracks.
1. Introduction Extratropical cyclones form preferentially in the storm tracks (e.g., Blackmon et al. 1977; Hoskins and Valdes 1990; Chang et al. 2002). These lie at midlatitudes and are intensified over the oceans. In the Northern Hemisphere (NH), where the oceans occupy only a fraction of the surface area, there are two distinct tracks: over the North Atlantic and the North Pacific. In the Southern Hemisphere (SH), where the oceans cover a greater area, a single storm track is found, and this is more zonal. The storm tracks are dynamically important in both hemispheres, at both short and longer time scales. The regions are associated with increased precipitation and winds and are subject to extreme weather events. The storm tracks are instrumental for meridional heat transport and thus the reduction of the equator-to-pole temperature gradient.
Corresponding author address: Lise Seland Graff, Meteorology and Oceanography Section, University of Oslo, P.O. Box 1022, Blindern, N-0315 Oslo, Norway. E-mail:
[email protected] DOI: 10.1175/JCLI-D-11-00174.1 Ó 2012 American Meteorological Society
There is growing evidence that the storm tracks are systematically changing in the warming climate, shifting poleward in both hemispheres. Observational studies suggest that the number of cyclones has decreased at midlatitudes in the NH during the last half of the twentieth century, whereas the number at high latitudes has increased (e.g., McCabe et al. 2001; Wang et al. 2006). The cyclone count between 408 and 608S has decreased during the same period (Simmonds and Keay 2000; Fyfe 2003), whereas it has slightly increased poleward of 608S (Fyfe 2003). The shift in the SH storm track is occurring simultaneously with a poleward migration of the westerly jet (e.g., Chen and Held 2007) and of the atmospheric baroclinicity (e.g., Fyfe 2003). Similar changes have been seen in models forced according to specific climate warming scenarios (e.g., Nakicenovic and Swart 2000), as documented by the Intergovernmental Panel of Climate Change (IPCC) (Meehl et al. 2007). In a study of an ensemble of 15 coupled general circulation models (GCMs), Yin (2005) found that the storm tracks intensified in the warmer climate and shifted poleward and upward along with the baroclinicity.
15 MARCH 2012
GRAFF AND LACASCE
Fischer-Bruns et al. (2005), Bengtsson et al. (2006), Ulbrich et al. (2008), and Wu et al. (2011) observed similar changes in simulations with the same and other warming scenarios. The storm tracks are clearly influenced by the ocean surface. Cyclones generated over open water preferentially do so close to oceanic frontal zones, in regions with strong sea surface temperatures (SST) gradients (Sinclair 1995). The track position and intensity are heavily influenced by low-level baroclinicity associated with such frontal zones (e.g., Nakamura and Shimpo 2004; Nakamura et al. 2008; Sampe et al. 2010). Sampe et al. (2010) found that, when removing such a frontal zone (based on southwestern Indian Ocean climatology) in an aquaplanet simulation, the storm tracks weaken and shift equatorward. Similarly, there is evidence that the ocean shifts the NH storm tracks poleward by transporting heat and thereby intensifying the SST gradients along the western boundaries of the ocean basins (Wilson et al. 2009). In addition, the storm tracks shift in response to other SST changes, such as those associated with El Nin˜o–Southern Oscillation (ENSO) (e.g., Chang and Fu 2002; Chang et al. 2002; Orlanski 2005). Therefore, it is reasonable to ask whether the poleward shift observed over the previous decades and that seen in climate simulations is also related to systematic changes in SST. Several studies have addressed such issues, using aquaplanet simulations. Caballero and Langen (2005) found an increase in poleward heat transport and a poleward shift in the storm tracks in response to an increase in the global-mean SST. Kodama and Iwasaki (2009) and Lu et al. (2010) found similar changes. Brayshaw et al. (2008) observed similar shifts but suggested they derived from changes in the SST gradient. In their simulations, the direction of shift was dictated not only by the sign of the change in SST gradient but also by where the gradient was altered. Increasing the gradient at midlatitudes produced a poleward shift in the tracks while increasing it in the tropics yielded an equatorward shift. However, the relative importance of the mean surface temperature and the gradients remains unclear. Caballero and Langen (2005), Lu et al. (2010), and Kodama and Iwasaki (2009) found little change in storm-track position when altering only the SST gradient. Observations suggest the oceans are indeed warming. The warming moreover is intensified at the ocean surface and at low latitudes (Levitus et al. 2005). SST has increased up to 18C over the past 50 yr in the tropics, with smaller changes at higher latitudes (bottom right panel of Fig. 5.3 in Bindoff et al. 2007; P. Durack et al. 2011, personal communication). Such changes would imply an increase in the temperature gradients at midlatitudes.
1855
The goal of the present study is to see how the storm tracks respond to such changes in SST in a realistic atmospheric model. The model, the atmospheric component of the National Center of Atmospheric Research (NCAR) Community Climate System Model (CCSM), has a more realistic lower boundary than in the aquaplanet simulations, including continents and orography. The latter significantly affect the circulation, particularly in the NH. We imposed various changes to SST and then observed the changes in the storm tracks. The primary focus is on warming at low latitudes, but we also consider scenarios where the warming or cooling is confined to high latitudes and to the tropics and where the warming is uniform with latitude. The results in many ways resemble those found in the aforementioned aquaplanet simulations, suggesting that the oceans exert a similar influence even when land is present. This is true in both hemispheres, even though the oceans occupy a different fraction of the total surface area. The results then support the notion that SST is central to storm-track dynamics (with implications not only for climate change but also for weather forecasting, particularly on seasonal time scales). The model and the SST modifications are described in section 2 and the analysis methods are described in section 3. The results are presented in section 4, a 20-yr control run first and then the experiments with altered SST. The results are discussed in section 5.
2. Model and data For the simulations, we used the NCAR Community Atmosphere Model version 3 (CAM3), the atmospheric component of the NCAR CCSM version 3 (Collins et al. 2004). CAM3 has been used previously to study the storm track (e.g., Hurrell et al. 2006; Alexander et al. 2006; Donohoe and Battisti 2009). In terms of bandpassfiltered eddy kinetic energy, the majority of structures and interseasonal changes are well simulated in CAM3 (Hurrell et al. 2006). We ran the model in its stand-alone version, with the Community Land Model (version 3.0) and a thermodynamic sea ice model. We used T42 resolution in the horizontal and 26 vertical hybrid levels. The SST was specified using a climatological dataset, with 12 monthly samples (i.e., the SST changes from month to month but not daily). For additional information on the model, see Collins et al. (2006) and Hurrell et al. (2006). We will focus on six experiments, one control and five sensitivity runs. All span the 20-yr period from 1 December 1980 to 28 February 2000. The control run employs standard surface forcing over the 20-yr period, including climatological SSTs. In the sensitivity runs, the SSTs were
1856
JOURNAL OF CLIMATE
VOLUME 25
FIG. 1. The SST modification (i.e., difference from the control run) in (a) the 2-K lowlat run, (b) the 2-K run, (c) 2-K highlat run, and (d) the 2-K tropics run. Contour interval (CI) is 0.5 K, as indicated in the gray-shade bar. Note that the SST change in the 22-K highlat run (not shown) is the same as in the 2-K highlat run, only negative.
increased or decreased by 2 K in different domains, as shown in Figs. 1 and 2. The initial surface temperature over land was left unaltered in all runs and SSTs were not changed where there is sea ice (e.g., in the Arctic and Antarctic). In the ‘‘2 K’’ run, SSTs were increased over the entire ocean. In the ‘‘2-K lowlat’’ and ‘‘2-K tropics’’ runs, the SSTs were increased equatorward of 458 and 158, respectively. In the ‘‘2-K highlat’’ and ‘‘22-K highlat’’ runs, the SSTs were increased and decreased by 2 K poleward of 458. When a domain boundary was located within the ocean basin, the SST increment was linearly relaxed to climatology over a latitude band of 88 along that boundary. Thus, in these cases, the SST gradient was altered as well. In the 22-K highlat and 2-K lowlat runs, the gradient was increased at midlatitudes around 458, whereas in the 2-K highlat run it was decreased. In the 2-K tropics run, the gradient was increased at 158. Unlike the aquaplanet simulations, the oceans cover only a fraction of the earth’s surface, to a degree that varies with latitude. As a result, warming the ocean uniformly, as in the 2-K run, also changes the zonally averaged SST gradients (Fig. 2). Moreover, the change is greater in the NH, where the land fraction is greater. Furthermore, because the temperatures were not altered
over ice, the gradients are affected where the sea meets ice (e.g., at Antarctica). We focus on the NH winter season [December– February (DJF)] for both hemispheres. An alternative is to use the SH winter season for that hemisphere and then to merge the figures (e.g., Sampe et al. 2010). However, the winter response in the SH is qualitatively similar to the summer response (section 4), so we will employ the DJF period for both. All data were interpolated to constant pressure surfaces, with a 50-hPa interval, from 850 to 50 hPa. For validation, we used the geopotential height field at the 500-hPa surface (Z500 hPa) from the National Center of Environmental Prediction (NCEP)–NCAR Global Reanalysis 1 (NCEP-1) [Kalnay et al. 1996; NCEP-1 data are provided by the National Oceanic and Atmospheric Administration (NOAA); see http://www. cdc.noaa.gov/]. The NCEP-1 data were taken from the same time period and season as the CAM3 data.
3. Analysis method Three methods of identifying the storm tracks are commonly used. One is ‘‘feature point identification’’ (e.g., Klein 1957; Reitan 1974; Serreze et al. 1997; McCabe
15 MARCH 2012
GRAFF AND LACASCE
FIG. 2. The initial zonally averaged DJF SST field from the control run (dashed line) and the corresponding difference field from the 2-K run (solid line). Vertical lines indicate the latitudes where SST modifications starts/stops in the other modified runs: namely, 158N/S (the 2-K tropics run) and 458N/S (the 2-K lowlat run and the 2-K highlat run). So, for example, the SST field from the 2-K tropics run is indicated by the part of the solid curve between the vertical lines 158S and 158N. Temperature is given in kelvins.
et al. 2001; Paciorek et al. 2002; Hoskins and Hodges 2002; Fyfe 2003; Benestad and Chen 2006), which identifies low pressure centers (commonly using surface pressure) in space and time. Then the storm tracks are evident as regions with high cyclone count densities. The second involves tracking cyclones, identified typically by surface pressure or vorticity (e.g., Hinman 1888; Klein 1957; Murray and Simmonds 1991; Sinclair 1994, 1995; Blender et al. 1997; Serreze et al. 1997; Simmonds and Keay 2000; Hoskins and Hodges 2002, 2005; Bengtsson et al. 2006; Wang et al. 2006). The storm tracks can then be identified as regions of high track density. An advantage with this method is that it allows studying changes in the storms during their life cycles. Using surface pressure, the approach captures the large-scale structure of the tracks but can also be affected by biased extrapolation over high topography or large-scale flows (e.g., the Icelandic low–fast jets) (e.g., Hoskins and Hodges 2002; Ulbrich et al. 2009). Using vorticity, the results preferentially capture the small scales, but the results also depend strongly on model resolution (e.g., Hoskins and Hodges 2002). The third method involves bandpass-filtered fields, where fluctuations with time scales between roughly 1 and 8 days are retained (e.g., Blackmon 1976; Blackmon et al. 1977; Hartmann 1974; Randel and Stanford 1985; Trenberth 1991; Branstator 1992; Chang and Fu 2002; Nakamura and Shimpo 2004; Yin 2005; Brayshaw et al. 2008; Ulbrich et al. 2008; Sampe et al. 2010; Wu et al. 2011). The storm tracks are evident as maxima in the
1857
variability of these bandpassed fields (e.g., Blackmon et al. 1977). The bandpass method is less selective than cyclone tracking, capturing both cyclones and anticyclones as well as other disturbances with the chosen time scales. However, the method is straightforward to apply, is easily reproducible, and can be carried out at any altitude, allowing for 3D reconstructions (e.g., Blackmon et al. 1977; Chang et al. 2002). The method has been used, among other things, to demonstrate the relation between the storm tracks and the variability of the large-scale flow (e.g., Branstator 1992, 1995). In the present study, we will use the bandpass method. In terms of bandpassed statistics, the storm tracks can be represented using a variety of fields including geopotential height Z, pressure, temperature, and velocity (e.g., Blackmon et al. 1977; Chang et al. 2002; Hoskins and Hodges 2002). Our focus will be on the standard deviation of the bandpassed Z field. Following Blackmon (1976) and Blackmon et al. (1977), we retain fluctuations with time periods between 2.5 and 6 days. We will also focus primarily on zonal averages, although horizontal (2D) fields are used to compare the control run with reanalysis fields and to identify zonal asymmetries. Then we use the zonally averaged fields to compare the control run with the various modified runs.
4. Results a. The control run We first consider the fields from the control run, comparing them to those from the NCEP-1 reanalysis and pointing out differences between the NH and SH. Similar fields have been described previously, so the presentation will be brief. However, the measures used in the subsequent sections are also introduced. Note that, in what follows, the fields are generally temporally averaged with the exception of the bandpassed Z fields for which we show the standard deviation, which is denoted with an SD subscript (ZSD).
1) HORIZONTAL FIELDS The bandpassed ZSD field on the 500-hPa surface Z500hPa,SD from the control run for the NH winter (DJF) is shown in Fig. 3a. The storm tracks appear as bands of increased variability in both hemispheres, situated between approximately 408 and 608. In the NH, the tracks are intensified over the Atlantic and Pacific and tilt to the northeast. The variability intensifies near the east coasts of North America and Asia, in the socalled entrance regions (e.g., Hoskins and Valdes 1990; Chang et al. 2002; Hoskins and Hodges 2002). In the SH, the variability increases east of the Andes, peaks in the
1858
JOURNAL OF CLIMATE
VOLUME 25
FIG. 3. The (a) DJF and (b) JJA bandpass ZSD standard deviation fields from the control run. (c) The DJF bandpass ZSD standard deviation field from the NCEP-1. (d) The mean bandpass y9T9 field from the control run. All fields are taken at the 500-hPa surface. CI is 5 m in (a)–(c) and 1 K m s21 in (d).
southern Indian Ocean, and then decreases over the South Pacific. The SH storm track is considerably more zonal than that in the NH (e.g., Trenberth 1991). The storm tracks from the NH summer [July–August (JJA)] are shown in Fig. 3b. The NH tracks are considerably weaker and shifted poleward compared to DJF. They are also more zonal, displaying less tilt. The SH storm track on the other hand has intensified and tilts somewhat to the southeast, particularly south of Australia. The result is a spiral structure (Trenberth 1991; Inatsu and Hoskins 2004; Hoskins and Hodges 2005). Nevertheless, the SH storm track has largely the same structure in both seasons; as such, we will focus exclusively on the DJF response in both hemispheres. The NH winter bandpassed Z500hPa,SD field from the NCEP-1 is shown in Fig. 3c. The structure closely resembles that in the CAM3 field, both in location and amplitude. However, the NH storm tracks in CAM3 are slightly stronger in the entrance regions and weaker in the exit regions. In addition, the simulated tracks are somewhat too zonal, as noted by Hurrell et al. (2006). In the SH, the two fields are very similar, except for small differences in amplitude. Thus, the CAM3 fields are realistic enough to warrant the sensitivity studies that follow.
The temporally averaged bandpassed eddy heat fluxes y9T9 from the 500-hPa surface are shown in Fig. 3d. The fluxes are calculated as follows: y9BP T9BP 5 y BP TBP 2 y BP T BP .
(1)
The subscript BP indicates bandpass filtered (dropped hereafter), and the bar indicates a temporal average; the remaining notation is standard. The heat transport is strongest in the storm tracks, particularly in the entrance regions (e.g., Blackmon et al. 1977; Hoskins and Valdes 1990; Chang et al. 2002). However, the fluxes are positive over much of the NH (and negative in the SH), indicating poleward transport of warm air and equatorward transport of cold air over the entire region. The largest maximum in the NH is in the Atlantic track, whereas the largest maximum in the SH is in the southern Indian Ocean.
2) ZONALLY AVERAGED FIELDS Hereafter, we focus on the zonally averaged fields, which are a convenient means for comparing the runs. However, it should be noted that such fields are perhaps more appropriate for the SH, where the (summer) storm track is more zonal. Because the fields are tilted in the NH, they are somewhat spread out in zonal average.
15 MARCH 2012
GRAFF AND LACASCE
1859
FIG. 4. Zonally averaged DJF fields from the control run. (a) The bandpass [ZSD] standard deviation field; CI is 5 m. (b) The [sB1 ] field; CI is 1 3 1021 day21. (c) The [u] field; CI is 5 m s21. (d) The [sB1,dT /dy ] field; CI is 1 3 1021 day21. (e) The [CM ] field; CI is 1 3 1010 kg s21. (f) The [sB1,N ] field; CI is 1 3 1021 day21.
Despite this, the changes are usually clear in both hemispheres. Zonal averages will be denoted by square brackets. The bandpassed [ZSD] field from the control run is shown in Fig. 4a. One maximum is seen in the extratropics of each hemisphere, between 408 and 608. The
NH maximum is wider, which is consistent with the aforementioned spreading. The temporally and zonally averaged zonal wind [u] is shown in Fig. 4c. Each hemisphere has a westerly jet, which lies equatorward and above the bandpassed [ZSD] maxima. As is well known, the jets represent a merger of
1860
JOURNAL OF CLIMATE
VOLUME 25
FIG. 5. Zonally averaged DJF fields from the control run. (a) The mean bandpass [y9T9] field, with CI of 1 K m s21. (b) The mean bandpass [u9y9] field, with CI of 4 m2 s22.
two structures, a baroclinic subtropical jet on the equatorward side and a more barotropic eddy-driven jet on the poleward side. It is also useful to consider the meridional overturning circulation streamfunction CM, which is defined as (e.g., Hartmann 1994) CM
ð 2pa cos(f) p 5 [y] dp. g 0
(2)
Here, a is the mean radius of the earth and f is the latitude. The [CM ] field (Fig. 4e) conveniently captures the secondary circulation over the range of latitudes. It is useful in the present context, because it shows how the Hadley and Ferrel cells shift with the other fields. As is typical during the NH winter, the NH Hadley cell is displaced southward across the equator and the Hadley cell in the SH is weaker. The eddy-driven Ferrel cells are evident in both hemispheres, with similar magnitudes. Thus, in the SH, the eddy-driven overturning accounts for a greater portion of the total. The storm tracks derive from the baroclinic instability of the westerly jets, and a frequently used measure of the latter is the Eady parameter of maximum baroclinic growth sB1 (Lindzen and Farrell 1980; Hoskins and Valdes 1990; Chang et al. 2002; Yin 2005). Here, sB1 is defined as sB1
g ›T 5 0:31 , N[T] ›y
(3)
where N is the Brunt–Va¨isa¨la¨ frequency. Thus, sB1 is proportional to the meridional temperature gradient and inversely proportional to the Brunt–Va¨isa¨la¨ frequency, so that increasing the temperature gradient increases baroclinicity whereas increasing static stability decreases it.
The value of [sB1 ] is plotted in Fig. 4b. In each hemisphere, there is a maximum near 400 hPa. There is also a secondary maximum higher up, between 100 and 50 hPa. The maxima occur in regions of strong vertical shear in u. They lie below the corresponding maxima in bandpassed [ZSD] and, in the NH, equatorward. Of the Brunt–Va¨isa¨la¨ frequency and the meridional temperature gradient, it is the latter that dominates the structure seen here. To show this, we decompose sB1 into two parts, one that varies only with the temperature gradient sB1,dT/dy and another that varies with the Brunt–Va¨isa¨la¨ frequency sB1,N (Yin 2005). They are defined as sB1,dT /dy sB1,N
›T g , 5 0:31 Nref [T]ref ›y g ›T 5 0:31 , N[T]ref ›y ref
(4)
where Nref, [T]ref, and j›T/›yjref are reference values derived from the spatially and temporally averaged control run fields. The terms [sB1,dT /dy ] and [sB1,N ] are shown in Figs. 4d,f. We see that [sB1,dT /dy ] captures the main features of [sB1 ], whereas [sB1,N ] is weaker and different in structure. Thus, [sB1 ] is dominated by the meridional temperature gradient, as in Yin (2005). The same result obtains in the sensitivity runs; as such, we only show [sB1 ] hereafter. As noted, the storm bands are regions of intensified eddy fluxes of heat and also momentum. The temporally and zonally averaged bandpassed eddy fluxes of heat and momentum [y9T9] and [u9y9] are shown in Figs. 5a,b. As expected, the fluxes are poleward and intensified in the storm tracks. The [y9T9] field displays largest values near the bottom boundary but also intensified at the jet level.
15 MARCH 2012
GRAFF AND LACASCE
In the [u9y9] field, one large maximum is evident in each hemisphere, slightly below the jet cores. Such fluxes yield a poleward intensification of the zonal flow, as (e.g., Holton 2004) ›[u] ›[u9y9] }2 . ›t ›y
(5)
The gradient of [u9y9] is positive on the equatorward side of the maximum and negative on the poleward side, forcing a poleward shift of the mean flow.
b. Sensitivity runs Hereafter, we consider the sensitivity runs. We examine the same fields as in the control run (bandpassed [ZSD ], [u], [CM ], [sB1 ], [y9T9], and [u9y9]). Differences from the control run are plotted rather than the full fields, to accentuate the response to the imposed changes in SST.
1) THE 2-K LOWLAT RUN SSTs were increased by 2 K equatorward of 458 in the 2-K lowlat run, as illustrated in Fig. 1a. Thus, both the low-latitude heating and the midlatitude SST gradients were increased in this case. The difference plots are shown in Fig. 6. The difference in the bandpassed [ZSD] field is shown in Fig. 6a. The storm tracks have intensified: the maxima are approximately 20% stronger, with the deviations at the lowest level larger by 10%–15%. In addition, the maxima have shifted poleward by roughly 78–88. In the NH, the track has also shifted slightly upward. The changes in [u] are shown in Fig. 6c. In the SH, both the subtropical and eddy-driven jets have intensified, illustrating the double-jet structure clearly. Note that the center of the eddy-driven jet is slightly equatorward of the bandpassed [ZSD] maximum. Similar changes are seen in the NH, although the separation between subtropical and eddy-driven jets is less clear. The impression is rather of a poleward shift and a strengthening of the barotropic flow. The change in [CM ] (Fig. 6e) is likewise clearest in the SH. The Ferrel cell has shifted poleward, in line with the bandpassed [ZSD] variability. The Hadley cell (which has a negative sense) has also shifted poleward, because the change is negative on the poleward side and positive on the equatorward side. In the NH, the Ferrel cell has also shifted poleward. The changes in the Hadley cell on the other hand are more skewed: the cell has intensified on the poleward side, near 308N, but has also weakened somewhat at height along its southern edge. It has also intensified near the upper boundary, at tropopause height.
1861
The change in [sB1 ] is shown in Fig. 6b. The strengthening of the two jets in the SH increases the shear and this is mirrored in [sB1 ], which displays a two-maximum structure. As noted, the latter is dominated by changes in the meridional temperature gradient and hence the shear. In the NH, where the two jets are merged, there is only a single maximum in the [sB1 ] difference. The eddy fluxes [y9T9] and [u9y9] (Figs. 6d,f) have intensified and shifted poleward, in both hemispheres. Interestingly, the [y9T9] difference has a nearly barotropic structure over much of the troposphere. The [u9y9] difference on the other hand is more localized at the jet level. As noted, we increased both the mean temperature and the temperature gradient at midlatitudes in the 2-K lowlat run. The subsequent runs are meant to distinguish these forcings.
2) THE 2-K RUN In the 2-K run, the surface ocean is warmed uniformly by 2 K, as shown in Fig. 1b. As noted though, such a change also affects the zonally averaged surface temperature gradients, because the oceans cover a varying fraction of the surface area. The difference fields are shown in Fig. 7. These resemble those in the previous run, but with weaker amplitudes. The bandpassed [ZSD] difference field suggest the storm tracks have intensified and shifted poleward, in both hemispheres, but the changes are less than before. Likewise, the [u] difference field shows an intensification and poleward shift in the jets, with the greatest changes occurring near the subtropical jets. The [sB1 ] difference field also resembles the previous one, and the changes in the vicinity of the subtropical jet are comparable. Likewise, the [CM ] difference field suggests the Hadley and Ferrel cells have shifted poleward and intensified, but to a lesser extent than in the previous run. The same comments apply to [y9T9] and [u9y9], whose maxima have shifted poleward. Thus, the mean SST also appears to be important for the strength and position of the storm tracks. Primarily, the increased diabatic heating at low latitudes strengthens the Hadley circulation, yielding stronger subtropical jets. Note that the changes in the [ZSD] field in the SH occur near 608S, where the ocean meets the Antarctic continent. This suggests that the changes in the SH may be due to an altered surface temperature gradient there. However, the NH bandpassed [ZSD] field changes comparably, and the storm track lies well south of the ocean–ice boundary. The changes in any case are weaker in this example than in the 2-K lowlat run.
3) THE 62-K HIGHLAT RUNS In the two 2-K highlat runs, the SSTs were increased and decreased by 2 K poleward of 458. The increase in
1862
JOURNAL OF CLIMATE
VOLUME 25
FIG. 6. Zonally averaged DJF difference plots from the 2-K lowlat run. The thick solid (stippled) contours are positive (negative) control run contours (CRCs) for reference. The thin filled contours are difference field contours. (a) The bandpass [ZSD] standard deviation difference field; CI is 2 m, and CRCs are 20, 30, 40, 50, 60, and 70 m. (b) The [sB1 ] difference field; CI is 2 3 1022 day21, and CRCs are 5, 6, 7, and 8 day21. (c) The [u] difference field; CI is 1 m s21, and CRCs are 10, 20, 30, and 40 m s21. (d) The [y9T9] field; CI is 2 3 1021 K m s21, and CRCs are 26, 24, 22, 2 and 4 K m s21. (e) The [CM ] difference field; CI is 2 3 109 kg s21, and CRCs are 24, 23, 22, 21, 1, 2, 3, and 4 3 1010 kg s21. (f) The [u9y9] field; CI is 1 m2 s22, and CRCs are 216, 212, 28, 24, 0, 4, 8, and 12 m2 s22.
15 MARCH 2012
GRAFF AND LACASCE
1863
FIG. 7. As in Fig. 6, but for the 2-K run.
the 2-K highlat run is shown in Fig. 1c. The SST difference in the 22-K highlat run is the same, with the opposite sign. Thus, the midlatitude SST gradient was weakened and strengthened in these runs, respectively, without changing the low-latitude temperatures. Consider the 22-K highlat run first (difference fields are shown in Fig. 8). In this, the temperature gradient at
458 has increased as much as in the 2-K lowlat run (Fig. 6). The bandpassed [ZSD] difference field indicates the storm band has intensified and shifted poleward. The changes are less than in the 2-K lowlat run but are quite comparable to those in the 2-K run. However, unlike in that run, the largest changes in [u] occur near the eddy-driven jets, with little change in the subtropical jets. The [sB1 ]
1864
JOURNAL OF CLIMATE
VOLUME 25
FIG. 8. As in Fig. 6, but for the 22-K highlat run.
difference field reflects this, with the largest changes occurring at midlatitudes. The Ferrel cells have also shifted poleward, as might be expected, but the Hadley cells have widened as well. The change in the midlatitude gradient has also produced an intensification of the eddy fluxes. In fact, the changes are greater than in the 2-K run and comparable to those seen in the 2-K lowlat run.
Now, consider the 2-K highlat run (Fig. 9). As noted, the midlatitude SST gradient is weakened here. The result is that the storm track weakens and shifts equatorward. Consistently, the eddy fluxes shift equatorward and the overturning cells have contracted. However, as in the 22-K highlat run, the subtropical jets are left largely unchanged. Thus, the two 2-K highlat runs are consistent with the midlatitude SST gradient influencing the position of the
15 MARCH 2012
GRAFF AND LACASCE
1865
FIG. 9. As in Fig. 6, but for the 2-K highlat run.
storm tracks. However, the subtropical jets are largely unaltered in these runs, unlike in the 2-K run. The changes in [ZSD] are comparable to those seen in the 2-K run but weaker than in the 2-K lowlat run. This suggests that the storm-track variability in that run stems from both the increase in tropical heating and the change in the midlatitude SST gradient.
4) THE 2-K TROPICS RUN The 2-K tropics run resembles the 2-K lowlat run in that both the low-latitude SSTs and the SST gradients have been increased. However, the heating is restricted to the tropics and the gradient is increased at 158 inside the NH Hadley cell and at the equatorward edge of the
1866
JOURNAL OF CLIMATE
VOLUME 25
FIG. 10. As in Fig. 6, but for the 2-K tropics run.
SH Hadley cell (Fig. 1). The resulting difference fields are shown in Fig. 10. There is a pronounced intensification in the subtropical jets, in both hemispheres, on the equatorward sides. Consistently, the Hadley cells have contracted and the baroclinicity reflected in [sB1 ] has increased. The bandpassed [ZSD] difference field also suggests an equatorward shift. However, interestingly, the magnitude of
the change is much less than in the 2-K lowlat run. Likewise, the eddy fluxes, though shifted equatorward, have changed less than in that run. Thus, the primary changes here occur in the subtropical jets, which have intensified and shifted equatorward. The storm tracks have also shifted equatorward. However, the changes are less than in the 2-K lowlat run, where the SST gradients were changed at midlatitudes.
15 MARCH 2012
GRAFF AND LACASCE
1867
FIG. 11. Horizontal projections of the bandpassed ZSD standard deviation field at the 300-hPa surface: (a) the 2-K lowlat plot, (b) the 2-K plot, (c) the 3-K highlat plot, and (d) the 2-K tropics plot. The thick solid lines are CRCs of 40 and 50 m, and the thin filled contours are the difference field contours with a CI of 3 m.
5) 2D DEVIATIONS In the preceding examples, the changes in the NH are frequently less clear than in the SH. This is because in the NH the storm tracks tilt and because the oceans cover less surface area. Here we return to 2D fields, to see aspects that may have been obscured by zonal averaging. Figure 11 shows the bandpassed ZSD difference fields from the 2-K lowlat, 2-K, 2-K highlat, and 2-K tropics runs. Note that these fields are taken at 300 hPa, closer to the jet level. The 2-K lowlat run is shown in Fig. 11a. The poleward shift in the SH is mostly zonal, but there is a marked increase in the Drake Passage (and this is seen in many of the other runs as well). The deviations in the NH are more tilted. The largest alteration in the North Atlantic storm track is actually in the exit region, over northern Europe and extending to central Asia. There are additional maxima in the entrance region of the North Pacific track and over northern Canada. The structure suggests
complex changes in the storm band, possibly reflecting the activity of mature cyclones. The 2-K run exhibits similar but weaker changes (Fig. 11b). The North Atlantic storm track has shifted to the northeast, the North Pacific track has shifted eastward, and the SH track has shifted poleward. However, there are also noticeable differences with the 2-K lowlat run. The changes seen previously over Asia are largely absent and those over the North Atlantic are more complex, indicating even a weakening in variability over the Nordic Seas. Note that significant changes occur near 608S, where the oceans meet the Antarctic ice sheet. These are conceivably the result of the forcing, because we have only warmed the oceans. However, we note that the changes in the 2-K lowlat run occur over the same latitudes, despite the SSTs being changed farther north in the 2-K run. In the 2-K highlat run (Fig. 11c), the storm tracks weaken on the poleward flanks in both hemispheres. The changes resemble those in the previous runs, being largely zonal in the SH and undulating in the NH. In
1868
JOURNAL OF CLIMATE
both NH storm tracks, the variability is altered primarily in the exit regions and most noticeably in the North Atlantic track. The 22-K highlat run response (not shown) is similar to the 2-K highlat run, but with opposite signs. The 2-K tropics run (Fig. 11d) presents an interesting contrast to the previous examples. The SH storm track has intensified on the equatorward flank, with large changes over the South Atlantic and southern Indian Oceans and south of Australia. In the NH, the field again has an undulating appearance, with a small decrease in variability along an arc extending from New England to England. In the Pacific, there is a decrease on the poleward side and increase on the equatorward side, most noticeably on the eastern side of the basin. Thus, the 2D plots largely confirm the previous conclusions. However, they also show that the changes in the NH are spatially variable, with the most pronounced changes occurring in the exit regions.
5. Summary and discussion We investigated changes in the storm tracks in both hemispheres induced by altering the SSTs in an atmospheric GCM (AGCM). We warmed/cooled the surface oceans by 2 K in various latitude bands and examined the changes when averaged over a 20-yr period. The storm tracks were shown in terms of the bandpass-filtered Z standard deviation field, retaining fluctuations with a time-scale range of 2.5–6 days. The results suggest the intensity and position of the storm tracks change in a consistent way in response to the forcing. Increasing the SST gradient at midlatitudes increases the baroclinicity and the eddy fluxes at midlatitudes, leading to an intensification of the storm tracks and a shift poleward. Increasing the gradient in the tropics also increases the variability but produces an equatorward shift in the storm band. Increasing the SSTs uniformly by 2 K strengthens Hadley circulation and the subtropical jets and produces a poleward shift in the storm track. Similar effects have been seen previously in simulations with aquaplanet boundary conditions. The dependence on SST gradient seen here is consistent with that described by Brayshaw et al. (2008), who found a poleward shift following an increase in the midlatitude gradient and an equatorward shift in response to an increased gradient in the tropics. Caballero and Langen (2005), Kodama and Iwasaki (2009), and Lu et al. (2010) found that increasing the mean temperature produced a poleward shift in the tracks, as in the 2-K run. Also, El Nin˜o, which is a warming of the ocean surface in the equatorial Pacific, causes the Pacific storm track to shift equatorward (e.g., Chang et al. 2002; Lu et al. 2008), as in the 2-K tropics run.
VOLUME 25
The present results suggest that changes both in the mean SST and in the gradients are important for the position and intensity of the tracks. Increasing the mean temperature affects the subtropical jets, and changing the midlatitude gradients alters the eddy-driven jets. Both changes affect the baroclinicity, as seen with the sB1 parameter. The latter is primarily affected by changes in the mean temperature gradient, rather than the stratification [as noted previously by Yin (2005)]. This would seem to imply that altering the surface temperature gradient will have a greater effect than simply increasing diabatic heating. However, the latter, which is greater in the runs where SSTs were increased in the tropics, also affects baroclinicity by altering the subtropical jets. It is perhaps unsurprising that our results are similar to the aquaplanet simulations in the SH, where the oceans cover a large fraction of the surface area. However, consistent changes are seen in the NH. Here, though, the greatest changes occur in the exit regions of the storm tracks. This is suggestive on a nonlinear interaction between the mature cyclones and the surface temperature gradient. The shifts in the storm tracks in the present runs were accompanied by an expansion or contraction of the Hadley cells. It is well known that the latter are currently expanding. This has been seen in observations (Fu et al. 2006; Hudson et al. 2006; Hu and Fu 2007; Seidel and Randel 2007; Seidel et al. 2008) and in models (Kushner et al. 2001; Previdi and Liepert 2007; Lu et al. 2007; Frierson et al. 2007; Johanson and Fu 2009). Indeed, it is difficult to separate the response in the tropics with that occurring on the flanks of Hadley cells and in the Ferrel cells. This suggests an important role for eddies in the Hadley circulation, as stressed by Schneider (2006). Recently, Polvani et al. (2011) suggested that similar effects to those seen here (e.g., a poleward shift of the midlatitude jet and a widening of the Hadley cell) could be induced in the SH as a result of stratospheric ozone depletion. In their simulations (also with CAM3), they fixed the SSTs to observed values. Given our results, it may then be difficult to distinguish which forcing is more important, particularly if changes in ozone also induce surface temperature changes (Sigmond et al. 2011). However, it should be noted that similar changes are found in the NH, where ozone-related forcing is weaker. As noted, the primary motivation of this study was testing the idea that the observed changes in the storm track could be triggered by oceanic warming. The pattern of warming found in observations is similar to that in the present 2-K lowlat run, in that the low latitudes experience the greatest warming and the temperature gradients at midlatitudes are increased. We found a poleward shift in the storm tracks, accompanied by an
15 MARCH 2012
GRAFF AND LACASCE
upward shift in the NH (although the latter was not as apparent in the SH). Moreover, the changes in both sB1 and the storm tracks in the 2-K lowlat run resemble those found by Yin (2005) (cf. our Fig. 6 with his Figs. 1, 2, left). Thus, the present results are consistent with the idea that oceanic warming is an important contributor to the observed changes in the storm tracks. It remains to be explained exactly why the storm tracks shift in the way they do. Increasing the SST gradients should intensify the heat fluxes, which are strongest near the surface. This in turn will strengthen the Eliassen– Palm fluxes acting on the mean circulation, particularly in relation to the eddy-driven jet. The latter could shift north or south in response. However, the present results and those of Brayshaw et al. (2008) indicate that the eddy fluxes themselves shift poleward. Therefore, there is evidently a feedback between the fluxes, the eddy-driven jet, and storm track. We are currently exploring how to capture this feedback in simplified models. Acknowledgments. The CAM3 simulations were performed on the Titan cluster at the University of Oslo. We would like to acknowledge three anonymous reviewers whose comments helped improve the paper. REFERENCES Alexander, M., and Coauthors, 2006: Extratropical atmosphere– ocean variability in CCSM3. J. Climate, 19, 2496–2525. Benestad, R. E., and D. Chen, 2006: The use of a calculus-based cyclone identification method for generating storm statistics. Tellus, 58A, 473–486. Bengtsson, L., K. I. Hodges, and E. Roeckner, 2006: Storm tracks and climate change. J. Climate, 19, 3518–3543. Bindoff, N. L., and Coauthors, 2007: Observations: Oceanic climate change and sea level. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 385–432. Blackmon, M. L., 1976: A climatological spectral study of the 500 mb geopotential height of the Northern Hemisphere. J. Atmos. Sci., 33, 1607–1623. ——, J. M. Wallace, N.-C. Lau, and S. L. Mullen, 1977: An observational study of the Northern Hemisphere wintertime circulation. J. Atmos. Sci., 34, 1040–1053. Blender, R., K. Fraedrich, and F. Lenkeit, 1997: Identification of cyclone-track regimes in the North Atlantic. Quart. J. Roy. Meteor. Soc., 123, 727–741. Branstator, G., 1992: The maintenance of low-frequency atmospheric anomalies. J. Atmos. Sci., 49, 1924–1946. ——, 1995: Organization of storm track anomalies by recurring low-frequency circulation anomalies. J. Atmos. Sci., 52, 207–226. Brayshaw, D. J., B. Hoskins, and M. Blackburn, 2008: The stormtrack response to idealized SST perturbations in an aquaplanet GCM. J. Atmos. Sci., 65, 2842–2860. Caballero, R., and P. L. Langen, 2005: The dynamic range of poleward energy transport in an atmospheric general circulation model. Geophys. Res. Lett., 32, L02705, doi:10.1029/2004GL021581.
1869
Chang, E. K. M., and Y. Fu, 2002: Interdecadal variations in Northern Hemisphere winter storm track intensity. J. Climate, 15, 642–658. ——, S. Lee, and K. L. Swanson, 2002: Storm track dynamics. J. Climate, 15, 2163–2183. Chen, G., and I. M. Held, 2007: Phase speed spectra and the recent poleward shift of Southern Hemisphere surface westerlies. Geophys. Res. Lett., 34, L21805, doi:10.1029/2007GL031200. Collins, W. D., and Coauthors, 2004: Description of the NCAR Community Atmosphere Model (CAM 3.0). NCAR Tech. Note NCAR/TN-4641STR, 226 pp. ——, and Coauthors, 2006: The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 2144–2161. Donohoe, A., and D. S. Battisti, 2009: Causes of reduced North Atlantic storm activity in a CAM3 simulation of the Last Glacial Maximum. J. Climate, 22, 4793–4808. Fischer-Bruns, I., H. von Storch, J. F. Gonza´lez-Rouco, and E. Zorita, 2005: Modelling the variability of midlatitude storm activity on decadal to century time scales. Climate Dyn., 25, 461–476, doi:10.1007/s00382-005-0036-1. Frierson, D. M. W., J. Lu, and G. Chen, 2007: Width of the Hadley cell in simple and comprehensive general circulation models. Geophys. Res. Lett., 34, L18804, doi:10.1029/2007GL031115. Fu, Q., C. M. Johanson, J. M. Wallace, and T. Reichler, 2006: Enhanced mid-latitude tropospheric warming in satellite measurements, Science, 312, 1179. Fyfe, J. C., 2003: Extratropical Southern Hemisphere cyclones: Harbingers of climate change? J. Climate, 16, 2802–2805. Hartmann, D. L., 1974: Time spectral analysis of mid-latitude disturbances. Mon. Wea. Rev., 102, 348–362. ——, 1994: Global Physical Climatology. International Geophysics Series, Vol. 56, Academic Press, 411 pp. Hinman, R., 1888: Eclectic Physical Geography. The Eclectic Geographies, American Book Company, 382 pp. Holton, J. R., 2004: An Introduction to Dynamic Meteorology. 4th ed. International Geophysics Series, Vol. 88, Academic Press, 535 pp. Hoskins, B. J., and P. J. Valdes, 1990: On the existence of stormtracks. J. Atmos. Sci., 47, 1854–1864. ——, and K. I. Hodges, 2002: New perspectives on the Northern Hemisphere winter storm tracks. J. Atmos. Sci., 59, 1041–1061. ——, and ——, 2005: A new perspective on Southern Hemisphere storm tracks. J. Climate, 18, 4108–4129. Hu, Y., and Q. Fu, 2007: Observed poleward expansion of the Hadley circulation since 1979. Atmos. Chem. Phys., 7, 5229–5236. Hudson, R. D., M. F. Andrade, M. B. Follette, and A. D. Frolov, 2006: The total ozone field separated into meteorological regimes– Part II: Northern Hemisphere mid-latitude total ozone trends. Atmos. Chem. Phys., 6, 5183–5191. Hurrell, J. W., J. J. Hack, A. S. Phillips, J. Caron, and J. Yin, 2006: The dynamical simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 2162–2183. Inatsu, M., and B. J. Hoskins, 2004: The zonal asymmetry of the Southern Hemisphere winter storm track. J. Climate, 17, 4882– 4892. Johanson, C. M., and Q. Fu, 2009: Hadley cell widening: Model simulations versus observations. J. Climate, 22, 2713–2725. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471. Klein, W. H., 1957: Principal tracks and mean frequencies of cyclones and anticyclones in the Northern Hemisphere. U.S. Weather Bureau Research Paper 40, 60 pp.
1870
JOURNAL OF CLIMATE
Kodama, C., and T. Iwasaki, 2009: Influence of the SST rise on baroclinic instability wave activity under an aquaplanet condition. J. Atmos. Sci., 66, 2272–2287. Kushner, P. J., I. M. Held, and T. L. Delworth, 2001: Southern Hemisphere atmospheric circulation response to global warming. J. Climate, 14, 2238–2249. Levitus, S., J. Antonov, and T. Boyer, 2005: Warming of the World Ocean, 1955–2003. Geophys. Res. Lett., 32, L02604, doi:10.1029/2004GL021592. Lindzen, R. S., and B. Farrell, 1980: A simple approximate result for the maximum growth rate of baroclinic instabilities. J. Atmos. Sci., 37, 1648–1654. Lu, J., G. A. Vecchi, and T. Reichler, 2007: Expansion of the Hadley cell under global warming. Geophys. Res. Lett., 34, L06805, doi:10.1029/2006GL028443. ——, G. Chen, and D. M. W. Frierson, 2008: Response of the zonal mean atmospheric circulation to El Nin˜o versus global warming. J. Climate, 21, 5835–5851. ——, ——, and ——, 2010: The position of the midlatitude storm track and eddy-driven westerlies in aquaplanet AGCMs. J. Atmos. Sci., 67, 3984–4000. McCabe, G. J., M. P. Clark, and M. C. Serreze, 2001: Trends in Northern Hemisphere surface cyclone frequency and intensity. J. Climate, 14, 2763–2768. Meehl, G. A., and Coauthors, 2007: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–845. Murray, R. J., and I. Simmonds, 1991: A numerical scheme for tracking cyclone centers from digital data. Part II: Application to January and July general circulation model simulations. Aust. Meteor. Mag., 39, 167–180. Nakamura, H., and A. Shimpo, 2004: Seasonal variations in the Southern Hemisphere storm tracks and jet streams as revealed in a reanalysis dataset. J. Climate, 17, 1828–1844. ——, T. Sampe, A. Goto, W. Ohfuchi, and S.-P. Xie, 2008: On the importance of mid-latitude oceanic frontal zones for the mean state and dominant variability in the tropospheric circulation. Geophys. Res. Lett., 35, L15709, doi:10.1029/2008GL034010. Nakicenovic, N., and R. Swart, Eds., 2000: Special Report on Emissions Scenarios. Cambridge University Press, 570 pp. Orlanski, I., 2005: A new look at the Pacific storm track variability: Sensitivity to tropical SSTs and to upstream seeding. J. Atmos. Sci., 62, 1367–1390. Paciorek, C. J., J. S. Risbey, V. Ventura, and R. D. Rosen, 2002: Multiple indices of Northern Hemisphere cyclone activity, winters 1949–99. J. Climate, 15, 1573–1590. Polvani, L. M., D. W. Waugh, G. J. P. Correa, and S.-W. Son, 2011: Stratospheric ozone depletion: The main driver of twentiethcentury atmospheric circulation changes in the Southern Hemisphere. J. Climate, 24, 795–812. Previdi, M., and B. G. Liepert, 2007: Annular modes and Hadley cell expansion under global warming. Geophys. Res. Lett., 34, L22701, doi:10.1029/2007GL031243. Randel, W. J., and J. L. Stanford, 1985: An observational study of medium-scale wave dynamics in the Southern Hemisphere
VOLUME 25
summer. Part I: Wave structure and energetics. J. Atmos. Sci., 42, 1172–1188. Reitan, C. H., 1974: Frequencies of cyclones and cyclogenesis for North America, 1951–1970. Mon. Wea. Rev., 102, 861– 868. Sampe, T., H. Nakamura, A. Goto, and W. Ohfuchi, 2010: Significance of a midlatitude SST frontal zone in the formation of a storm track and an eddy-driven westerly jet. J. Climate, 23, 1793–1814. Schneider, T., 2006: The general circulation of the atmosphere. Annu. Rev. Earth Planet. Sci., 34, 655–688. Seidel, D. J., and W. J. Randel, 2007: Recent widening of the tropical belt: Evidence from tropopause observations. Geophys. Res. Lett., 112, D20113, doi:10.1029/2007JD008861. ——, Q. Fu, W. J. Randel, and T. J. Reichler, 2008: Widening of the tropical belt in a changing climate. Nat. Geosci., 1, 21–24. Serreze, M. C., F. Carse, R. G. Barry, and J. C. Rogers, 1997: Icelandic low cyclone activity: Climatological features, linkages with the NAO, and relationships with recent changes in the Northern Hemisphere circulation. J. Climate, 10, 453– 464. Sigmond, M., M. C. Reader, J. C. Fyfe, and N. P. Gillett, 2011: Drivers of past and future Southern Ocean change: Stratospheric ozone versus greenhouse gas impacts. Geophys. Res. Lett., 38, L12601, doi:10.1029/2011GL047120. Simmonds, I., and K. Keay, 2000: Variability of Southern Hemisphere extratropical cyclone behavior, 1958–97. J. Climate, 13, 550–561. Sinclair, M. R., 1994: An objective cyclone climatology for the Southern Hemisphere. Mon. Wea. Rev., 122, 2239–2256. ——, 1995: A climatology of cyclogenesis for the Southern Hemisphere. Mon. Wea. Rev., 123, 1601–1619. Trenberth, K. E., 1991: Storm tracks in the Southern Hemisphere. J. Atmos. Sci., 48, 2159–2178. Ulbrich, U., J. G. Pinto, H. Kupfer, G. C. Leckebusch, T. Spangehl, and M. Reyers, 2008: Changing Northern Hemisphere storm tracks in an ensemble of IPCC climate change simulations. J. Climate, 21, 1669–1679. ——, G. C. Leckebusch, and J. G. Pinto, 2009: Extra-tropical cyclones in the present and future climate: A review. Theor. Appl. Climatol., 96, 117–131, doi:10.1007/s00704-008-0083-8. Wang, X. L., V. R. Swail, and F. W. Zwiers, 2006: Climatology and changes of extratropical cyclone activity: Comparison of ERA40 with NCEP–NCAR reanalysis for 1958–2001. J. Climate, 19, 3145–3166. Wilson, C., B. Sinha, and R. G. Williams, 2009: The effect of ocean dynamics and orography on atmospheric storm tracks. J. Climate, 22, 3689–3702. Wu, Y., M. Ting, R. Seager, H.-P. Huang, and M. Cane, 2011: Changes in storm tracks and energy transports in a warmer climate simulated by the GFDL CM2.1 model. Climate Dyn., 37, 53–72. Yin, J. H., 2005: A consistent poleward shift of the storm tracks in simulations of the 21st century climate. Geophys. Res. Lett., 32, L18701, doi:10.1029/2005GL023684.
