Corresponding author: Y.-G. Park, Ocean Circulation and Climate Research Division, Korea Institute of Ocean Science and Technology, Ansan 426–744, South Korea. (firstname.lastname@example.org)
 Eleven climate models, one high-resolution and ten low-resolution, were analyzed to investigate the response of the northwestern Pacific under a global warming scenario. Application of scenario A1B of the Special Report on Emission Scenarios (SRES) weakens (intensifies) the southern (northern) part of the interior subtropical gyre both in high-resolution and low-resolution model. Such a dipole type change is mainly due to a basin-scale dynamic atmosphere-to-ocean process. Namely, under global warming the Hadley circulation is weakened and expanded poleward. The Ferrel circulation is also displaced poleward, leading to weakening of ascending (descending) air motion and a high (low) sea level pressure anomaly in the northwestern (southeastern extratropical) North Pacific. Finally, a negative wind stress curl anomaly developed along the zero wind stress curl line of the present-day climate to enhance the northern part of the gyre. The high-resolution model results show greater changes in the structure of the Kuroshio and Kuroshio Extension, with strong intensification of the Kuroshio Extension front and jet, while in the low-resolution models the changes are small. The Kuroshio between Taiwan and the southern coast of Japan is significantly intensified in the high-resolution model results, but is slightly weakened in the ensemble of the low-resolution models.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 In the North Pacific, the Kuroshio originating from east of Philippine coast is the western boundary current of subtropical gyre and, the Oyashio originating from the East Kamchatka Current is the western boundary current of western subarctic gyre. The two boundary currents play a critical role in meridional heat transport of the North Pacific. They meet off the eastern coast of Japan and flow eastward, forming the Kuroshio Extension (KE) and Oyashio Extension (OE). These extensions act to separate the warm subtropical and cold subpolar waters of the North Pacific, and thus generate two maxima of meridional temperature gradient, one at depths between 300 m and 400 m associated with the KE front, and the other in the surface layer associated with the OE front [Nonaka et al., 2006]. The KE consists of an eastward inertial jet near 35°N and is characterized by large-amplitude meanders and energetic pinched-off eddies [Qiu and Chen, 2005]. The OE flows eastward from 38°N nearshore to 40°–45°N offshore, and its offshore location is close to the line of zero wind curl [Mitsudera et al., 2004].
 On a multiyear time scale, the boundaries between the subtropical and subpolar gyres corresponding to the Kuroshio-Oyashio Extension (KOE) region are meridionally displaced, generating strong Sea Surface Temperature (SST) anomalies [Seager et al., 2001]. The SST anomaly in the KOE region changes the oceanic heat and moisture fluxes and thus has the potential to influence the mean state of the overlying atmospheric circulation, including the anchoring latitude of storm tracks in the North Pacific [Kwon et al., 2010]. Thus assessments of the changes in the Kuroshio and KOE are important.
 Using a high-resolution climate model in which the atmospheric CO2 concentration doubles in 100 years, Sakamoto et al. showed that flows at 100 m depth in the Kuroshio and KE region were intensified by spin-up of their recirculation gyres. This intensification was mostly due to changes in wind stress over the North Pacific. However, the atmospheric process responsible for the basin-scale surface wind changes was not examined.Lu et al.  provided a critical clue. Analyzing climate simulations included in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) under a global warming scenario, they found that the Hadley circulation weakens and expands poleward and the Ferrel circulation is displaced poleward, as proposed also by Yin  and Vecchi and Soden . This anomalous large-scale atmospheric circulation is presumed to affect the surface wind fields and thus the whole North Pacific gyre system.
 In most climate models, low- resolution ocean models are adopted. Considering the importance of the KOE in the climate of the North Pacific, the response of the KOE under a global warming in low-resolution models is also of much interest. Here we will elucidate the processes governing the aforementioned changes in the northwestern Pacific in conjunction with the atmospheric changes. To achieve this, we first assess changes in the Kuroshio and KOE regions under a global warming scenario using four climate models (one high-resolution, three low-resolution) selected from the IPCC AR4. Distinct changes of the high-resolution model results in the KOE region are also assessed in detail in comparison with those from the low-resolution models. We then investigate the changes in atmospheric variables.
 Eleven climate models (CCCMA-CGCM3.2(T47), CNRM-CM3, CSIRO-MK3.5, ECHAM5/MPI-OM, GFDL-CM2.1, GISS-ER, INM-CM3.0, MIROC3.2-Hires, MIROC3.2-Medres, MRI-CGCM2.3.2 and UKMO-HadCM3) are used in this study. The horizontal and vertical resolutions of the atmosphere and ocean components for these respective climate models are summarized inTable 1. Only MIROC3.2-Hires has an eddy-permitting ocean model; because of their lower resolutions the remaining models do not. It is well known that the Kuroshio in low-resolution models overshoots to the north, and the KE is thus located farther north than observed, [Thompson and Cheng, 2008]. However, our focus is not to reproduce accurately the Kuroshio and the KE, but rather to analyze their responses to a global warming scenario using available models.
Table 1. Horizontal and Vertical Resolutions for the Atmosphere and the Ocean in the Respective Climate Modelsa
Refer to Solomon et al. . Some variables are not available for analysis at least in part, and excluded from the analysis as listed. Here θocean is potential temperature of the ocean, and ω500hPa air vertical velocity at 500 hPa.
T106 (∼1.1° × 1.1°) L56
0.2° × 0.3° L47
T47 (∼2.8° × 2.8°) L31
1.9° × 1.9° L29
T63 (∼1.9° × 1.9°) L45
0.5°–2° × 2° L31
T63 (∼1.9° × 1.9°) L18
0.8° × 1.9° L31
T63 (∼1.9° × 1.9°) L31
1.5° × 1.5° L40
2.0° × 2.5° L24
0.3°–1.0° × 1.0° L50
4° × 5° L20
4° × 5° L13
4° × 5° L20
2° × 2.5° L33
T42 (∼2.8° × 2.8°) L20
0.5°–1.4° × 1.4° L43
T42 (∼2.8° × 2.8°) L20
0.5°–2.0° × 2.5° L23
2.5° × 3.75° L19
1.25° × 1.25° L20
 Gridded global monthly mean ocean current velocity, potential temperature, air vertical velocity at 500 hPa (represented as Lagrangian pressure tendency, ω500hPa), sea level pressure (SLP), and wind stress were retrieved from the Program for Climate Model Diagnosis and Intercomparison (PCMDI) website (http://www-pcmdi.llnl.gov). Note that some variables are not available for analysis at least in part. They are listed in Table 1and excluded from the analysis. We refer to these variables from the 20C3M scenario, in which greenhouse gases increase as observed through the 20th century, between 1901 and 2000 as “the present-day climate”. Scenario A1B of the Special Report on Emission Scenarios (SRES), in which atmospheric CO2reaches 720 ppm at the end of the 21st century, between 2051 and 2100, is referred to as “the future climate.” We averaged these analysis variables over the respective periods to eliminate high frequency internal variability. To compare results from the eddy-permitting model, MIROC3.2-Hires, with those from non eddy-permitting models, we took the ensemble mean of the ten low-resolution climate models (hereafter denoted as “EM LCM”).
3.1. Changes in the Kuroshio and the Kuroshio Extension
Figure 1shows stream functions of barotropic volume transport (BVT) and Sverdrup volume transport (SVT) for the present-day climate over the North Pacific north of 10°N, and for the future climate scenario. The BVT was directly calculated from the ocean velocity field:
where λ and ϕ are longitude and latitude, λe is longitude of the eastern boundary, v is meridional velocity, re is Earth's radius and D is bottom depth. The SVT was estimated from wind data based on the Sverdrup relation:
Where β is the meridional derivative of the Coriolis parameter, ρ0 is the mean density of the ocean, and is the wind stress curl.
 In the present-day climate, spatial patterns of BVT and SVT for the interior ocean are similar. The western boundary current is successfully reproduced in the BVT of both MIROC3.2-Hires and EM LCM. In MIROC3.2-Hires, this western boundary current is more intense than that in EM LCM, and the KE appears to meander. Such characteristics of pattern and amplitude in MIROC3.2-Hires are in agreement with other high-resolution ocean general circulation models (OGCMs) [e.g.,You, 2005]. Application of scenario A1B created distinct changes in the SVT. In both MIROC3.2-Hires and EM LCM, the southern (northern) part of the North Pacific subtropical gyre weakens (strengthens) and the southern part of the North Pacific subarctic gyre weakens. This dipole type volume transport change indicates that the entire North Pacific subtropical gyre is shifted poleward in the future climate. The BVT also shows a similar change, which means that the poleward shift of the North Pacific subtropical gyre is mainly attributed to changes in wind stress as inSakamoto et al. . The change in BVT, however, is smaller than that in SVT, and the northward shift of BVT is not substantial both in high- and low-resolutions.
 In EM LCM the anti-cyclonic recirculation gyre is observed only south of Japan, whereas in MIROC3.2-Hires it is divided into two: one south and another east of Japan. By applying scenario A1B, in MIROC3.2-Hires the BVT of the recirculation gyre east of Japan is increased by up to about 16 Sv (16%), indicating that the KE jet is intensified. The recirculation gyre south of Japan is shifted poleward, which leads to a partial intensification of the Kuroshio between Taiwan and the southern coast of Japan. Although in EM LCM the southern recirculation gyre is also shifted poleward, the Kuroshio between Taiwan and the southern coast of Japan appears to weaken slightly. This difference seems to be due to the intensity and size of the respective southern recirculation gyres. That is, the southern recirculation gyre of MIROC3.2-Hires is located in a relatively narrow area just south of Japan, whereas that of EM LCM extends farther northeastward toward the center of the North Pacific. Therefore, even for the similar change in the interior subtropical gyre the response of the Kuroshio, the western boundary current, can be resolution dependent.
 Changes in the KOE region are examined in Figure 2in terms of the zonally averaged zonal velocity averaged from the surface to 100 m depth, the upper ocean temperature and its meridional gradient between 145°E and 160°E. In the present-day climate, two zonal velocity peaks, which are indicative of the KE and OE jets, are distinctly observed in MIROC3.2-Hires along with the local maxima of meridional temperature gradient that represent the subsurface KE front at about 37°N and the surface OE front at about 40°N. It is in good agreement withNonaka et al. . However, in EM LCM these two fronts merge into one broad surface front, which is a typical feature of low-resolution climate models, as discussed by earlier studies [Hurlburt et al., 1996; Yamanaka et al., 1998].
 With the application of scenario A1B, in MIROC3.2-Hires the subsurface thermal front is significantly intensified, and its overlying zonal velocity peak, which is indicative of the KE jet, almost doubles in intensity from 0.26 m/s to 0.48 m/s. The OE front also intensifies while strengthening the jet from 0.09 m/s to 0.16 m/s. The intensification of the fronts results in clearer separation of the fronts under a global warming scenario. On the contrary, in EM LCM the scenario A1B does not have substantial influence on the KOE front. The thermal front intensifies slightly and shifts northward, but the change is not significant.
 To investigate if this intensification is a misinterpretation due to internal variability we display the time series of the KE jet from MIROC3.2-Hires for the 20th and 21st centuries inFigure 3. To find quantify the strength of the KE jet, we first zonal velocity averaged from the surface to 100 m depth and between 145°E and 160°E. We then defined the maximum in the meridional range between 30°N and 40°N as the strength of the KE jet. The KE jet reveals a slightly declining trend during the 20th century. Although there is strong short term variability, during the 21st century the long term inclining trend is significantly surpassing the declining trend during the 20th century, confirming that the intensification is not due to decadal or centennial scale internal variability. The standard deviations of the maximum zonal velocity for the 20th and 21st centuries are 0.07 m/s and 0.13 m/s, respectively. As discussed before, the KE jet is intensified by 0.2 m/s. This amount is well beyond the standard deviation and we could conduce that the intensification is not due to internal variability. For EM LCM similar to the KE jet we define the KOE jet between 30°N and 50°N. Note that we adopted a wider meridional range considering the meridional extent of the KOE jet in low resolution models. The KOE jet intensifies by about 0.01 m/s by the end of the 21st century, which is beyond one standard deviation, 0.004 m/s, for both the 20th and 21st centuries. However, the intensified amount is much smaller than that of MIROC3.2-Hires.
 One notable result from MIROC3.2-Hires is the enhanced internal variability of the KE jet during the 21st century. The intensified KE and OE fronts have a larger amount of potential energy available for baroclinic instability and the formation of transient eddies. The enhanced shear across the KE jet also could enhance eddy activities to generate the greater internal variability, but detail investigation on this subject is outside the scope of this paper.
3.2. Dynamic Atmosphere-to-Ocean Process
Figure 4 shows ω500hPa, SLP, and wind stress curl over the North Pacific outside the equatorial region for the present-day climate, and their changes under the future climate scenario. Positive (negative) values ofω500hPaindicate descending (ascending) air motion. In the present-day climate, ascending motion is dominant over the northwestern North Pacific off Japan, whereas descending motion is dominant over the southeastern North Pacific outside the tropics. One of the general features depicted in the future climate is the weakening of these vertical motions of air across the whole North Pacific (Figures 4a and 4d), which is interpreted as the weakening and poleward expansion of the Hadley circulation and the poleward displacement of the Ferrel circulation [Yin, 2005; Vecchi and Soden, 2007; Lu et al., 2007]. Over the northwestern Pacific including the Kuroshio and KOE region, the poleward shift of the subtropical subsidence region weakens the ascending motion. In the regions where these ascending motions are weakened, the SLP increases (Figures 4b and 4e).
 The increase of SLP over the northwestern Pacific, in turn, induces a negative wind stress curl anomaly along the zero wind stress curl line of the present-day climate via the geostrophic balance both in MIROC3.2-Hires and EM LCM, pushing the zero wind stress curl line poleward (Figures 4c and 4f). Consequently, the negative wind stress curl field driving the North Pacific subtropical gyre is shifted poleward to produce the dipole pattern in SVT (Figures 1c and 1d). Within the interior of the subtropical gyre the change in BVT is comparable to that in SVT except the poleward shift, and we can conclude that the atmospheric change induces ocean change over the interior of the northern Pacific subtropical gyre.
 Both MIROC3.2-Hires and EM LCM reveal that over the KOE region the upward vertical velocity at 500 hPa weakens, the SLP increases, and the negative wind stress curl anomaly occurs. This change generates an Ekman convergence anomaly to elevate the sea surface height. An anti-cyclonic current anomaly is formed in the KOE region to enhance the circulation gyre east of Japan in MIROC3.2-Hires, and the northern part of subtropical gyre in EM LCM. This dynamic atmosphere-to-ocean process indicates a link between changes in the large-scale atmosphere circulation and the KOE.
4. Summary and Conclusions
 Under a global warming scenario, analysis of eleven climate models (MIROC3.2-Hires with an eddy-permitting ocean model, and the ensemble mean of ten others with non-eddy-permitting ocean models) revealed that the southern part of the interior North Pacific subtropical gyre weakened while the northern part strengthened. This dipole pattern change appeared to be attributable to the basin-scale dynamic atmosphere-to-ocean processes. That is, weakening and poleward expansion of the Hadley circulation and poleward displacement of the Ferrel circulation induced a descending anomaly, and thus a high SLP, anomaly over the northwestern North Pacific. An ensuing negative wind stress curl anomaly developed along the zero wind stress curl line of the present-day climate to enhance the northern part of the gyre.
 The gyre-scale responses of the oceans–i.e., dipole type change of the interior subtropical gyre–to the anomalous large-scale atmospheric circulations are similar in both high- and low-resolution climate models, which suggests that the dynamic atmosphere-to-ocean process proposed above will remain valid even when higher-resolution climate models are used. On the other hand, detailed responses are resolution dependent. The Kuroshio between Taiwan and the southern coast of Japan was significantly intensified in MIROC3.2-Hires but was slightly weakened in the low-resolution climate models. Moreover, the KE front and jet were significantly intensified in MIROC3.2-Hires but were only slightly intensified in the ensemble of the low-resolution climate models. As shown inFigures 1 and 2, the present-day climate state of MIROC3.2-Hires, as well as its response to the global warming scenario, is more complex than those of the low-resolution climate models, probably because nonlinear effects are resolved better in the eddy-permitting climate model [Pedlosky, 1996]. Also, the aforementioned atmospheric forcing anomaly that is stronger over the KOE region in MIROC3.2-Hires than in the low-resolution climate models may play a significant role in the high-resolution model's response to the global warming scenario.
 As suggested by Frankignoul et al. , the meridional shifts of the KE and of the OE due to changes in wind stress are, in turn, capable of having a significant influence on the large-scale atmospheric circulation. Moreover, the KOE interfrontal zone is closely related to the formation of North Pacific Intermediate Water [Yasuda et al., 1996], and its modification is significantly influenced by the decadal variation of the KE jet [Qiu and Chen, 2011]. Thus further studies are necessary to assess the impact of the aforementioned changes, i.e., the intensification of the KE and OE fronts and jets and their variability, on the ocean interior and the overlying atmosphere. The differences between the eddy-permitting model and low-resolution models suggest that the responses could be resolution dependent. A systematic study investigating sensitivity to model resolution is also warranted.
 The authors thank to the Editor and two anonymous reviews for their critical and constructive comments and suggestions that helped us greatly. This research has been sponsored by CATER 2012–3052, PE98651 and PN65212.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.