Major “storm tracks”, where migratory cyclones and anticyclones recurrently develop, are observed around midlatitude oceanic frontal zones with strong meridional gradient of sea-surface temperature (SST). A set of atmospheric general circulation model experiments is performed with zonally uniform SST prescribed at the model lower boundary. The latitudinal SST profile for each hemisphere is characterized by a single front. The frontal latitude is varied systematically from one experiment to another, while the intensity of the frontal gradient is kept unchanged. Though idealized, the experiments reveal a climatological tendency for a low-level storm track to be organized along or slightly poleward of the SST front if located in the subtropics or midlatitudes. As a surface manifestation of an eddy-driven polar-front jet (PFJ), surface westerly axis tends to form on the poleward flank of the front. This anchoring effect of the SST front is also hinted at upper levels, but the climatological positions of the storm track and PFJ are less sensitive to the frontal latitude. For the SST front at subpolar latitude, the joint primary axes of the upper-level storm track and PFJ form in midlatitudes away from the SST front. Their positions correspond to their counterpart simulated with a particular SST profile from which frontal gradient has been removed, suggesting that the anchoring effect of a subpolar SST front on the storm track and PFJ is overshadowed by atmospheric internal dynamics, namely, the self-maintenance mechanism of a midlatitude storm track and PFJ through their interactions.
 A midlatitude oceanic frontal zone is a confluent region of warm and cool ocean currents, characterized by strong meridional gradient in both sea-surface temperature (SST) and surface air temperature (SAT).Nakamura et al. pointed out that core regions of the major “storm tracks”, zonally elongated regions where migratory cyclones and anticyclones recurrently develop, are observed along or just downstream of the major oceanic frontal zones. They hypothesized that this collocation arises from the anchoring of a surface baroclinic zone through effective restoration of cross-frontal SAT gradient via differential heat supply from the ocean (“oceanic baroclinic adjustment” or “oceanic restoration of surface baroclinicity”), which has been verified by numerical studies byNakamura et al. , Nonaka et al. , Taguchi et al. , Sampe et al. , and Hotta and Nakamura .
 In addition to moisture supply from warm ocean currents [Hoskins and Valdes, 1990; Minobe et al., 2008], the anchoring effect can be important for maintaining a storm track and the associated polar-front jet (PFJ), which is driven by westerly momentum transport from a subtropical jet (STJ) by transient eddies [Lee and Kim, 2003; Nakamura et al., 2004]. In the presence of poleward heat flux due to baroclinic growth of those eddies, PFJ accompanies strong surface westerlies. They in turn drive ocean currents [Trenberth et al., 1990], whose heat transport acts to confine tight SST gradients into narrow frontal zones [Nakamura et al., 2004]. Therefore, the anchoring effect on both storm track and PFJ can be of certain importance in understanding the extratropical atmospheric general circulation interacting with the underlying ocean. In fact, through their experiments with an atmospheric general circulation model (AGCM), Nakamura et al.  and Sampe et al.  showed that the anchoring effect of SST on a storm track would be reduced substantially without its frontal gradient.
 In the present study, a set of AGCM experiments with zonally uniform SST prescribed as the model boundary condition is conducted to examine the sensitivity of the climatological-mean latitudes of the primary storm track and PFJ to the latitude of frontal SST gradient.Brayshaw et al.  and Chen et al. have revealed notable sensitivity of the latitudinal positions and intensities of a storm track and attendant PFJ to prescribed SST gradient anomalies. Each of the SST profiles they used is, however, characterized by an unrealistically broad zone of moderate gradient without realistic frontal gradient. Furthermore, modifications they added to the SST profile are likely to alter SST gradient not only in midlatitudes but also in the Tropics. The latter alternation can directly modifies the Hadley cell and associated STJ that acts as an upper-tropospheric Rossby waveguide, which makes it difficult to isolate the influence of midlatitude SST gradient. In the present study, in contrast, the extratropical SST profile is characterized by a single frontal zone with realistic intensity. We perform sensitivity experiments by varying the frontal latitude from one experiment to another with Tropical and subtropical SST kept unchanged, to isolate the influence of the extratropical frontal SST gradient on the troposphere.
2. Experimental Design
 The AGCM we used is called AFES (AGCM for Earth Simulator) [Ohfuchi et al., 2004], with 56 vertical levels. Its horizontal resolution (T79; equivalent to ∼150 km grid interval) is sufficient for resolving the effect of an oceanic frontal zone on large-scale atmospheric circulation. The lower boundary of the AGCM is set as the fully global ocean with six different latitudinal profiles of zonally uniform SST (Figure 1a). With this idealized “aqua-planet” setting without any landmass, we can eliminate planetary-scale atmospheric stationary waves forced by land-sea thermal contrasts and topography as observed in the Northern Hemisphere (NH). As in the work bySampe et al. , one of these SST profiles was taken from the OISST data for the South Indian Ocean [60∼80°E], where the warm Agulhas Return Current is confluent with the cool Antarctic Circumpolar Current to maintain frontal SST gradient. The NOAA Optimum Interpolation (OI) SST V2 is available at http://www.cdc.noaa.gov/cdc/data.noaa.oisst.v2.html. For any of its profiles, SST poleward of 70° is set to −1.79°C and linearly interpolated equatorward to the poleward flank of the SST front, to realize ice-free condition and eliminate strong thermal contrast across the sea-ice margin. Examining its climatic impact is, however, beyond the scope of this study. The profile for austral winter (Jun.–Aug.) was assigned to the model Southern Hemisphere (SH) and the corresponding summertime profile (Dec.–Feb.) to the model NH. With this SST profile characterized by the frontal gradient at 45° latitude in both hemispheres, the AGCM was integrated for 60 months under insolation fixed to its solstice condition after six-month spin-up in order for obtaining robust statistics. In our aqua-plant setting without any landmass, seasonality of the tropospheric circulation arises mainly from the equatorial asymmetry in the Hadley cell and associated STJ, which is controlled mainly by the prescribed SST profile in the Tropics and subtropics through determining the position of the rising branch of the Hadley cell. In this particular profile, SST peaks in the NH Tropics, whereas subtropical SST is apparently lower in the SH. The model SH (NH) can thus be regarded as the “winter (summer) hemisphere”. It should be noted that our idealized experiment is not designed for reproducing the observed circulation in the SH. The mean states of storm tracks and westerlies simulated with this SST profile are almost identical to those for the CTL experiment ofSampe et al. , which are nevertheless similar to their observational counterpart for the Indo-Australian sector.
 Five 60-month integrations were then repeated with modified SST profiles in which the frontal latitude had been shifted artificially from 45° equatorward to 30° or poleward to 55° with 5° intervals. For the modifications, not only the intensity of the frontal gradient but also the SST profile equatorward of 25° was kept unchanged (Figure 1b), so as to fix the thermal forcing on the model Hadley cells [cf. Brayshaw et al., 2008; Chen et al., 2010]. Our experiments are thus designed to give some insight into the climatological sensitivity of the atmospheric general circulation to the latitude of an extratropical SST front. Owing to our aqua-planet setting, the simulated climatological-mean state exhibits a high degree of zonal symmetry. In the following we therefore present zonally averaged statistics obtained from our experiments.
3. Latitudes of Surface Baroclinic Zone and Midlatitude Low-Level Storm Track
 The mean-flow baroclinicity can be measured by the Eady growth rateσ [Eady, 1949], which is proportional to meridional temperature gradient and inversely to static stability [Hoskins and Valdes, 1990]. Near-surfaceσ, which is of critical importance for baroclinic development of synoptic-scale eddies, has been evaluated from the climatological zonal-mean temperature at the 850 and 1000 hPa levels.Figure 2ashows meridional profiles of the near-surfaceσ in the “winter” hemisphere for the six experiments. Each of the profiles exhibits a distinct peak (dots) collocated with the SST front, reflecting the anchoring of a surface baroclinic zone through effective oceanic restoration of SAT gradient across the SST front [Nakamura et al., 2008].
 Activity of baroclinically developing transient eddies can be measured by their poleward heat transport [v′T′], where the bracket denotes zonal averaging and the primes denote fluctuations that have been extracted through high-pass filtering with a half-power cutoff period of eight days. It is evident in the climatological-mean profile of 850 hPa [v′T′] for each of the experiments (dots in Figure 2b), a well-defined single low-level storm track forms in the “winter hemisphere”. (The storm track latitude exhibits no significant trend during the analysis period.) As summarized inFigure 2c, the climatological storm track in the “winter hemisphere” shows a strong tendency to form in the immediate vicinity of the surface baroclinic zone anchored by the SST front whose latitude is 40° or higher. Meanwhile, the mean storm track forms slightly poleward of the SST front whose latitude is 35° or lower. In fact, the surface storm track in these experiments, defined as the latitudinal maximum of 1000-hPa [v′T′] is situated at SST front (not shown). At higher levels, however, the storm track axis shifts toward a mid-tropospheric baroclinic zone associated with the midlatitude PFJ, which may be a factor that deviates the eddy activity from its linear relationship with Eady growth rate.
 In the model “summer hemisphere” (Figure 2d), the low-level storm track forms systematically poleward of the surface baroclinic zone if the SST front is located at 45° or equatorward. In contrast, the mean storm track forms equatorward of the front if located at 55°. The simulated “summertime” storm track exhibits a clear tendency to stay in midlatitudes, although it nevertheless shows a noticeable tendency to follow the latitudinal shift of the SST front.
4. Axial Latitudes of the Surface Westerlies and Their Eddy Forcing
 As a surface manifestation of a PFJ, midlatitude surface westerlies are maintained mainly through the downward transport of westerly momentum via poleward eddy heat transport, while the upper-tropospheric PFJ is maintained by the meridional eddy momentum transport mainly from STJ [Lee and Kim, 2003; Nakamura et al., 2004]. In the transformed Eulerian mean (TEM) framework, eddy forcing of the surface westerlies can be approximated as the near-surface divergence of the Eliassen-Palm (E-P) flux [Andrews and McIntyre, 1976]. The meridional and vertical components of the E-P flux are proportional to the meridional eddy transport of westerly momentum [u′v′] and heat [v′T′], respectively. Figures 3a and 3bshow the peak latitude of the climatological-mean eddy forcing for the “winter” and “summer” hemispheres, respectively, as approximated by 925-hPa E-P flux divergence associated with sub-weekly disturbances [Edmon et al., 1980].
 In the “winter” hemisphere (Figure 3a), the peak latitude of the eddy acceleration follows the latitudinal shift of the SST front and low-level storm track (Figure 2c), while being displaced systematically poleward from the front. The displacement is consistent with a slight poleward tilt of the maximum [v′T′] with height (not shown). In a manner consistent with the eddy acceleration, the axis of the climatological-mean 925-hPa westerlies [U] is systematically poleward of the front (Figure 3a). The near-surface [U] axis tends to be farther poleward of the maximum eddy acceleration if the SST front is located at 40° or a lower latitude. For the SST front at subpolar latitude (50° or 55°), in contrast, the climatological-mean near-surface westerlies exhibit double jet structure with another axis around 40°. Driven by eddies away from the SST front, the presence of the secondary branch of the near-surface [U] implies the importance of atmospheric internal dynamics that is not directly related to the thermal influence of the frontal SST gradient.
 In the “summer” hemisphere (Figure 3b), the surface westerly axis is situated systematically poleward of the SST front as long as it is at 45° or lower latitude. This poleward displacement is consistent with the positions of the storm track and associated eddy forcing. Unlike in the “winter” hemisphere, no double jet structure emerges for the SST situated at subpolar latitude (50° or 55°). In this case, the surface westerly axis again forms in midlatitudes away from the SST front. This result suggests a large contribution from internal dynamics via eddy-mean flow interaction to the maintenance of the surface westerlies. The particular contribution tends to be comparable to or even dominant over the anchoring effect on the surface westerly jet to the SST front when located at subpolar latitude, in agreement withBrayshaw et al. . Interestingly, the midlatitude westerly axis almost coincides with the single westerly axis simulated in the NF experiment by Nakamura et al.  and Sampe et al. , where the frontal gradient is artificially removed from their SST profile (their “NF” profile). This implies that the thermal forcing of a subpolar SST front on the lower-tropospheric circulation tends to be less effective than that of a midlatitude or subtropical SST front, leading to the dominance of the internal dynamics postulated byRobinson .
5. Latitudes of Upper-Level Storm Track and PFJ
Figure 3cshows the climatological-mean axis of an upper-level storm track in the “winter hemisphere” simulated in each of the experiments, defined as the peak latitude of 300 hPa meridional wind variance [v′v′]. Though somewhat less obvious than near the surface, the storm-track axis still exhibits a tendency to follow the latitudinal displacement of the SST front as long as it is located in the subtropics or midlatitudes. For the SST front at subpolar latitude (50° or 55°), in contrast, the storm track forms near 45°, away from the front.Figure 3calso shows a tendency for the upper-level secondary jet to coincide with the midlatitude storm track, reflecting its characteristics of an eddy-driven PFJ. In fact, the convergence of eddy momentum flux [u′v′] tends to maximize just a few degrees poleward of the secondary jet. Meanwhile, the strongest divergence of [u′v′] occurs on the poleward flank of a STJ, indicating systematic poleward transport of westerly momentum by eddies from the STJ toward the midlatitude PFJ located poleward of the SST front. As observed in the SH [Nakamura and Shimpo, 2004], a secondary storm-track forms along the STJ without any notable baroclinic eddy growth (Figure 3c), featuring the role of the STJ as a waveguide.
 In the model “summer” hemisphere (Figure 3d), the upper-level sensitivity of the latitudes of the midlatitude storm track and PFJ to the latitudinal position of the SST front is qualitatively the same as its wintertime counterpart, but this particular sensitivity is diminished if the SST front is located at subpolar latitude (50° or 55°). Located just a few degrees equatorward of the latitudes of storm track and maximum eddy acceleration, the summertime midlatitude jet is essentially a PFJ.
6. Summary and Discussion
 Though performed under the idealized setting, our AGCM experiments have revealed certain sensitivity of the climatological-mean positions of a midlatitude storm track and an eddy-driven PFJ to the latitudinal position of an SST front. The sensitivity, which is unambiguous for the SST front situated in the subtropics or midlatitudes, is manifested as the tendency for the positions of the storm track and PFJ to follow the latitudinal displacement of the SST front. This tendency arises probably from the anchoring effect of the SST front on the storm track and associated PFJ by maintaining the surface baroclinic zone along the SST front, via effective “oceanic restoration of surface baroclinicity” [Nakamura et al., 2008] and by supplying moisture to individual storms and thereby energizing them. The particular sensitivity tends to be stronger in the lower troposphere than in the upper troposphere, while it is more obvious in the “winter” hemisphere with an intensified STJ than in the “summer” hemisphere. If the SST front is situated at subpolar latitude (50° or 55°), however, both the PFJ and upper-level storm track are situated in midlatitudes away from the front. Rather, their positions correspond to those realized without frontal SST gradient [Nakamura et al., 2008; Sampe et al., 2010]. In other words, the anchoring effect of the SST front, if located at subpolar latitude, tends to be overshadowed by atmospheric internal dynamics, especially in the upper troposphere. We argue that the internal dynamics involve self-maintenance mechanisms of a midlatitude storm track and eddy-driven jet via eddy-mean-flow interactions [Robinson, 2006], can be operative regardless of the thermal forcing from the surface.
 It should be pointed out that our results must be regarded as an upper bound of potential impacts of an extratropical SST front on the atmosphere, as it is prescribed for the model in a zonally symmetric manner. Nevertheless, our preliminary results provide some insight into the fundamental nature of the extratropical atmospheric general circulation, especially the mean state of a midlatitude storm track and PFJ, while invoking some fundamental questions. Specifically, what mechanisms cause the “winter-summer” difference in the sensitivity of their axial positions to the location of the SST front, and why are their upper-level positions less sensitive to the frontal location than their low-level positions? We hypothesize that these hemispherically and vertically contrasting sensitivities may arise from the corresponding contrasts in the relative importance between the thermal forcing with an extratropical SST front and the internal dynamics that acts to situate the storm track and PFJ around 40° latitude regardless of the surface influence. Verification of our hypothesis and investigation of variability around the mean state both remain for our future work.
 We thank the two anonymous referees for their sound criticism and constructive comments and suggestions on the earlier version of this paper, which have led to its substantial improvement. We used the Earth Simulator in support of JAMSTEC. We thank the AFES/CFES working team of JAMTEC, Y. Kosaka, T. Sampe and A. Goto for their advices for our experiments. This study is supported in part by the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT) through the Grant-in-Aid for Scientific Research 22340135 and that on Innovative Areas 2205 and by the Japanese Ministry of Environment through the Global Environment Research Fund (S-5).
 The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.