The Gulf Stream and Kuroshio regions feature strong sea surface temperature (SST) gradients that influence cyclone development and the storm track. Previous studies showed that smoothing the SSTs in either the North Atlantic or North Pacific yields a reduction in cyclone activity, surface heat fluxes, and precipitation, as well as a southward shift of the storm track and the upper-level jet. To what extent these changes are attributable to changes in individual cyclone behaviour, however, remains unclear. Comparing simulations with realistic and smoothed SSTs in the atmospheric general circulation model AFES, we find that the intensification of individual cyclones in the Gulf Stream or Kuroshio region is only marginally affected by reducing the SST gradient. In contrast, we observe considerable changes in the climatological mean state as well as a reduced cyclone activity in the North Atlantic and North Pacific storm tracks that are shifted equatorward in both basins. The upper-level jet in the Atlantic also shifts equatorward, while the jet in the Pacific strengthens in its climatological position and extends further east. Surface heat fluxes, specific humidity, and precipitation also respond strongly to the smoothing of the SST, with a considerable decrease in their mean values on the warm side of the SST front. This decrease is more pronounced in the Gulf Stream than in the Kuroshio region, due to the larger decrease in SST along the Gulf Stream SST front. Considering the differences of the different variables occurring within/outside of a 750 km radius of any cyclone over their entire lifetime, we find that cyclones play only a secondary role in explaining the differences in the mean state between the smoothed and realistic SST experiments.
The Gulf Stream and Kuroshio regions with their strong sea surface temperature (SST) gradients are preferential locations for cyclogenesis
SST gradients influence individual cyclone intensification
Randomly selecting 24 individual cyclones that occurred in the Gulf Stream region,
In addition to low-level baroclinicity, upper-level forcing by the jet stream is known to contribute to cyclogenesis
In the light of this tight coupling between the jet and the storm track, it is not surprising that a smoothing of the SST can affect the upper-level flow. Indeed both the storm track
Focusing on mesoscale aspects of the atmospheric response to a smoothing of the SSTs,
While the spatial distribution of surface wind convergence into a narrow band has been linked to strong SST gradients
Extratropical cyclones strongly modulate the horizontal moisture transport
To shed light on these aforementioned issues, we assess the effect of a weak or strong SST gradient using an atmospheric general circulation model (AFES 3) based on simulations with realistic and smoothed SSTs in the Gulf Stream and Kuroshio regions. Our analysis of these simulations is twofold. Firstly, we follow the approach of
We use data from version 3 of the AGCM for the Earth Simulator (AFES) developed by the Earth Simulator Center of the Japan Agency for Marine-Earth Science and Technology
Using AFES 3,
We use SST, latent and sensible heat fluxes, large-scale and convective precipitation, temperature and specific humidity at 850 hPa, and wind at 925 and 300 hPa for our analysis. We also compare the model simulations with the same variables from the ERA-Interim reanalysis that was created using a four-dimensional variational data assimilation scheme and a spectral truncation of T255 and 60 levels in the vertical
We identify the position of SST fronts using an objective frontal detection scheme based on the “thermal” method, as described in detail by
To assess the potential impact of the SST smoothing on the upper levels, we detect the position of the
jet following the algorithm of
For the climatologies and composites we normalise the occurrence of both SST front lines and jet axis lines to account for the latitudinally varying area covered by grid cells. We achieve this by showing the average length of SST front line/jet axis line per unit area, hence the resulting unit of length per area. For details on the normalisation we refer to the jet climatology by
We employ the University of Melbourne cyclone detection and tracking algorithm
The cyclone density pattern is in good spatial agreement with previous studies (
We categorise the identified cyclone tracks with a maximum intensification in either the Gulf Stream region (30–50
We categorise the cyclones for the SMTHG and SMTHK experiments analogously to the CNTL experiment. However, as the SST gradient in the smoothed experiments is very homogeneous over a large region and too weak to qualify as a front, we instead use the front positions from the CNTL experiment for the classification. We use the same classification as in CNTL to be able to compare cyclones with geographically similar genesis locations and tracks across the experiments. For simplicity we still refer to C1–3 as the cold, warm, and crossing cases for the smooth (SMTH) experiments, even though, strictly speaking, no SST front is crossed.
In addition to the cyclone track classification, we present a decomposition of the winter climatology for selected variables, where we conditioned the two composites on either occurring within or outside an area with a radius around a cyclone centre throughout the life cycle of a cyclone. We performed this analysis for each ocean basin, irrespective of the direction of cyclone propagation and location of its maximum intensification. Consistent with the threshold on cyclone circumference in
Analysing the SST (Fig.
Climatological SST for DJF for
In the Kuroshio (black box in Fig.
Compared to CNTL, the SSTs in SMTHG are smoother and their gradient is more widely distributed (compare Fig.
The SST distribution in SMTHK (Fig.
The position of the North Atlantic jet coincides with the location of the SST front (Fig.
The changes in the SST field between CNTL and SMTHG/SMTHK introduce differences in both cyclone density and the climatological jet stream position. We observe an equatorward shift in the maximum cyclone density in both the North Atlantic and North Pacific, particularly in the central and eastern part of the basins (Fig.
While cyclone density (Fig.
To assess the effect of smoothing the SSTs on the evolution of individual cyclones, we now restrict our focus to cyclones with maximum intensification in the Gulf Stream or Kuroshio region (details in Sect.
In the Kuroshio region, the number of cyclones in C1 (86/81) and C3 (59/60) is more or less unchanged between CNTL/ SMTHK, whereas cyclones in C2 (24/14) decrease slightly in number. The small number of cyclones in C2, particularly in SMTHK, implies some uncertainty for the corresponding results. Note that in contrast to the Atlantic, there is no reduction in cyclones crossing the SST front, potentially because the SST gradient in the Pacific is already comparatively weak in CNTL (compare Fig.
Pressure tendency (hPa h
The more pronounced reduction of cyclones crossing the SST front in SMTHG compared to SMTHK highlights the significance of a strong SST gradient to anchor the position of the storm track, as discussed in previous studies
Among the three categories, Atlantic cyclones of C3 and C1 are deepening the most in CNTL, with a maximum 6 h intensification corresponding to approximately 28 and 25 hPa/d, respectively. Conversely, cyclones of C2 are characterised by a weaker intensification throughout their evolution (Fig.
These results support the findings of
In the Kuroshio region, cyclones of C3 are deepening the fastest (approx. 30 hPa/d), followed by cyclones of C1 (approx. 25 hPa/d; Fig.
SSTs (K) for the three categories relative to the time of maximum intensification for
In SMTHK, the cyclones in the three categories have similar pressure tendencies as in CNTL (Fig.
Considering the evolution of the SST underneath the cyclone core, cyclones of C1 propagate over comparatively low SSTs, because they remain on the cold side of the SST front in both regions (Fig.
In the smooth experiments, cyclones of C1 propagate over approximately 1 K higher SSTs in both regions and cyclones of C2 over 1–2 K lower SSTs. Further, cyclones of C3 in the smoothed experiments experience a less sharp decrease in SST across the SST front compared to CNTL (Fig.
Overall, the considerable reduction in the number of cyclones of C3 after the smoothing of the SST in the Atlantic highlights the anchoring effect of a strong Gulf Stream SST front on the storm track. On the other hand, the already weak SST gradient in the Kuroshio prior to the smoothing leads to minor SST differences between CNTL and SMTHK and to a similar number of cyclones that cross the SST front. The rather similar cyclone intensification between the experiments indicates that the SST gradient is not particularly important for the intensification of individual cyclones (consistent with
Our results thus suggest that the SST smoothing does not result in significant differences in the characteristics of individual cyclones, which is consistent with
In the CNTL climatology, we observe a maximum of latent and sensible surface heat fluxes on the warm side of both the Gulf Stream and the Kuroshio SST front (Fig.
The surface heat fluxes are similarly distributed in the ERA-Interim dataset, though latent heat fluxes in CNTL are considerably larger compared to ERA-Interim (compare Fig.
The SST smoothing affects the amount of surface heat fluxes in both regions, though to a different extent. In the Gulf Stream region, we observe considerably weaker surface heat fluxes (Fig.
To estimate the role of cyclones for the differences when smoothing the SST, we decompose the winter climatology considering the surface heat fluxes occurring within and outside of a radius of 750 km around the cyclones' centre over their entire lifetime (Fig.
The climatological differences between CNTL and SMTHG/SMTHK predominantly arise when we do not consider heat fluxes associated with cyclones (Fig.
Our results indicate that the smoothing of the SST front has only a minor effect when we consider cyclones, which is confirmed by a cyclone-relative composite analysis, where surface heat fluxes are only moderately reduced by the smoothing of the SST front (not shown). Our findings are also in line with
The precipitation distribution in the Atlantic in CNTL is characterised by a maximum of large-scale (
In the Pacific, we observe an analogous spatial distribution of the precipitation pattern, but the amplitude is somewhat larger than in the Gulf Stream region (compare Fig.
Analogous to the surface heat fluxes, the smoothing of the SST field affects precipitation in both regions. In the North Atlantic, the smoothing leads to a remarkable decrease in precipitation (Fig.
Among the two types of precipitation, convective precipitation is more sensitive to the SST smoothing. In SMTHG, the mean convective precipitation is reduced by half compared to CNTL, and the narrow convective precipitation band observed in CNTL largely disappears in SMTHG (not shown). This finding is in line with
Compared to the surface heat fluxes, the precipitation associated with cyclones is more influenced by the SST smoothing, in particular in SMTHG. There is a noticeable reduction in convective precipitation in the Gulf Stream region just south of the climatological position of the SST front (Fig.
Overall, cyclones account for a larger fraction of the precipitation differences than they did for the difference in surface heat fluxes when comparing CNTL and SMTHG. This result is also in line with
In the Kuroshio region we note a rather equal influence of the SST smoothing when we consider precipitation related or unrelated with cyclones. These differences mainly concern the large-scale precipitation and are more evident in the central North Pacific (Fig.
Higher values of specific humidity are observed to the south of the Gulf Stream and Kuroshio regions (Fig.
Analogous to the surface heat fluxes (Fig.
For specific humidity, cyclones account only for a small part of the climatological differences between CNTL and SMTHG (Fig.
Consistent with the results for the Atlantic, the North Pacific features larger differences in specific humidity, when the latter is not associated with cyclones propagating in the region (Fig.
Apart from the well-established Clausius–Clapeyron relationship between SSTs and moisture, several studies indicate the leading role of cyclones on the poleward transport of moisture
In CNTL, the strongest climatological winds reach 40 m s
In both basins, we observe decreasing (increasing) wind speeds to the north (south) of the climatological jet position with smoother SSTs (Fig.
In the North Atlantic, the displacement of the maximum wind speed at 300 hPa is overall present for both when cyclones are present and absent. However, Fig.
The higher contribution of cyclones to the observed differences in upper-level climatological wind speed in the Atlantic compared to the Pacific is consistent with previous studies indicating that the Pacific jet is externally (more thermally) driven, as opposed to the Atlantic jet, which is more eddy driven
We quantified and attributed differences in the atmospheric response when using realistic (CNTL) and smooth SSTs for the North Atlantic (SMTHG) and North Pacific (SMTHK), respectively, based on simulations with the AFES 3 model. The CNTL simulation compares well to ERA-Interim, except for considerably larger latent heat fluxes in CNTL, but these are most likely associated with the lower SST resolution in ERA-Interim prior to 2002. Overall, the AFES model successfully captured the distribution of pertinent variables in both ocean basins. Given the stronger SST gradient in the Atlantic, the effect of the smoothing on the SST front yields stronger SST differences between the CNTL and the respective smooth experiments (see Fig.
We first examined the impact of the smoothing of the SST on the intensification of individual cyclones. Considering only cyclones with a maximum intensification in the Gulf Stream or the Kuroshio region, we classified them into three categories based on their propagation relative to the SST front, where cyclones either always stay on the cold (C1) or warm (C2) side of the SST front or they cross the SST front from the warmer to the colder side (C3). Similar deepening rates for all these cyclone categories across all experiments reveal the rather minor role of the SST gradient on the intensification of individual cyclones. This result is valid for both ocean basins, though it is particularly relevant for the Gulf Stream region where the SST smoothing dramatically weakens the strong SST gradient.
Considering all cyclones propagating in either the North Atlantic or the North Pacific, irrespective of their direction of propagation, stage of evolution, and their location of maximum intensification, we found the cyclone density in the storm track to decrease when the SSTs are smoothed in the Kuroshio and even more so in the Gulf Stream region. We relate the different response of the cyclone densities for the two regions to the more pronounced reduction of the SST gradient in SMTHG for the Atlantic compared to SMTHK for the Pacific. Overall, we observe an equatorward shift in cyclone density in both regions, which is more pronounced over the central and eastern parts of the two ocean basins. Both cyclone density differences have a distinct SW–NE tilt, basically following the storm track. An analogous southward shift is observed in the upper-level jet for the North Atlantic, whereas for the North Pacific such a shift is absent and the difference between the experiments instead reveals a more meridionally focused and zonally extended jet with smoother SSTs.
We found a considerable decrease in both latent and sensible heat fluxes along the SST front when smoothing the SSTs, which was more pronounced across the SST front in the Gulf Stream region compared to the Kuroshio. Analogous to the surface heat fluxes, precipitation in the Gulf Stream region is strongly reduced when smoothing the SST front, which is particularly evident for convective precipitation on the warm side of the Gulf Stream SST front. However, both types of precipitation are only slightly affected by the SST smoothing in the Kuroshio region. Differences in specific humidity at 850 hPa feature a similar reduction after smoothing the SST. The weaker reduction of moisture and precipitation in the Kuroshio region is related to the smaller differences in the SST and SST front between CNTL and the smoothed fields in the Pacific compared to the Atlantic.
To clarify whether the differences between the CNTL and SMTH experiments stem directly from cyclones interacting with the SST and SST gradient, we considered selected variables within and outside an area with a radius of 750 km around cyclone centres propagating in either the North Atlantic or the North Pacific. We found that the surface heat fluxes that are associated with cyclones in both basins do not considerably contribute to the climatological differences between the CNTL and SMTH experiments. Differences in precipitation, however, were more closely associated with cyclones propagating in either the North Atlantic or the North Pacific.
For specific humidity, cyclones have only a minor contribution to the climatological differences between CNTL and SMTHG/SMTHK, with a more evident decrease in specific humidity in the Atlantic, arising from a considerable decrease in the SST gradient in the vicinity of the Gulf Stream. In contrast, both humidity and SST are not changed as significantly in the Pacific sector. Our results support that the underlying SST is the dominant factor determining the distribution of specific humidity, with cyclones playing a modulating role.
Similar to the surface heat fluxes and the specific humidity at 850 hPa, we found cyclones to only play a secondary role in explaining the upper-level (300 hPa) wind speed differences arising from the SST smoothing. Notwithstanding this secondary role, Atlantic cyclones contribute more to the climatological differences than Pacific cyclones, which is consistent with previous studies indicating that the Atlantic jet is more eddy driven than the Pacific jet.
Overall, our analysis highlights that SST fronts only have a minor impact on the characteristics and intensification of individual cyclones propagating in the Gulf Stream or Kuroshio region. Following the nomenclature of
The code to construct the figures in this study is available upon request.
ERA-Interim data are provided by European Centre for Medium-Range Weather Forecasts (ECMWF). Data availability is subject to the policies of ECMWF. The AFES data used in this study are publicly available at
The supplement related to this article is available online at:
LT carried out the bulk of the data analysis and writing. TS contributed to the detailed discussion about the methods and interpretation of the findings as well as to the writing process. CS contributed to both data analysis and writing.
The authors declare that they have no conflict of interest.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank ECMWF for providing the ERA-Interim data and Akira Kuwano-Yoshida and Akira Yamazaki, who provided the AFES 3 data. We also want to thank Irina Rudeva and two anonymous referees for their constructive feedback.
This research has been supported by the Research Council of Norway (grant no. 262220).
This paper was edited by Martin Singh and reviewed by Irina Rudeva and two anonymous referees.