Formation and transport of intermediate water masses in a model of the Pacific Ocean

Acta Oceanol. Sin, 2014, Vol. 33, No.5, P8- 16DOI: 10.1007/s13131-014-0480- .http:/ /www.hyxb.org.cnE-mail: hyxbe@263.netFormation and transport of intermediate water masses ina model of the Pacific OceanLI Yangchun', XU Yongful*1 State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute ofAtmospheric Physics, Chinese Academy of Sciences, Beijing 100029, ChinaReceived 11 December 2012; accepted 1 July 2013OThe Chinese Society of Oceanography and Springer-Verlag Berlin Heidelberg 2014AbstractA basin-wide ocean general circulation model of the Pacific Ocean was used to investigate how the interiorrestoration in the Okhotsk Sea and the isopycnal diffusion affect the circulation and intermediate watermasses. Four numerical experiments were conducted, including a run with the same isopycnal and thick-ness diffusivity of 1.0x103 m2/s, a run employing the interior restoration of temperature and salinity in theOkhotsk Sea with a time scale of 3 months, a run that is the same as the first run except for the enhancedisopycnal mixing, and a final run with the combination of the restoration in the Okhotsk Sea and large iso-relatively weak. An increase in isopycnal diffusivity can improve the simulation of both Antarctic and NorthPacific intermediate waters, mainly increasing the transport in the interior ocean, but inhibiting the outflowfrom the Okhotsk Sea. The interior restoration generates the reverse current from the observation in the Ok-hotsk Sea, whereas the simulation of the temperature and salinity is improved in the high latitude region ofthe Northern Hemisphere because of the reasonable source of the North Pacific Intermediate Water. A com-parison of vertical profiles of temperature and salinity along 50°N between the simulation and observationsdemonstrates that the vertical mixing in the source region of intermediate water masses is very important.Key words: intermediate water mass, Okhotsk Sea, isopycnal diffusivity, interior restorationCitation: Li Yangchun, Xu Yongfu. 2014. Formation and transport of intermediate water masses in a model of the Pacific Ocean.Acta Oceanologica Sinica, 33(5): 8-16, doi: 10.1007/s13131-014-0480-z1 Introductionresults of water mass distributions in the global world, the in-The formation and transport pathways of water masses have fluences of this mixing scheme on the water masses have beena great influence on the oceanic uptake of passive tracers such evaluated by several studies (Danabasoglu and McWilliams,ascartbon 'ioxide. In the Pacific Ocean, besides Antarctic Bot- 1995; England, 1995; Hirst and McDougall, 1996; McDougalltom Water (AABW), both Antarctic Intermediate Water (AAIW) et al, 1996; Duffy et al, 1997; England and Rahmstorf, 1999).and North Pacific Intermediate Water (NPIW) play an impor- The influences of the mixing scheme and its coefficients on thetant role in the uptake of anthropogenic gases. Many studies on transport of NPIW have also been examined, in which passivethe formation and transport of NPIW have been made (Talley tracers, including chlorofluorocarbons (CFCs) and radiocar-1993; Yamanaka et al, 1998a, b; You et al, 2000; You, 2003a, b). bon, were generally used.Those studies indicated that the Okhotsk Sea is important toYamanaka et al. (1998a) used an ocean circulation modelthe formation of NPIW, while in the ocean models, especially of the North Pacific north of 10°N to investigate the influencesin the relatively coarse models, the formation and transport ofof the mixing scheme and thermohaline restoration onNPIW are always underestimated because of the weak simulat- the formation of NPIW. Their results indicated that theed source of fresh water from the Okhotsk Sea and weak eddy simulation with a combination of the GM mixing schemetransport. In order to accurately simulate these water masses, and thermohaline restoration to climatological values in thesome artificial modifications, including unrealistic surface forc- Okhotsk and Bering Seas under perpetual wintertime forcinging, were employed in early studies (Yamanaka et al, 1998a, b;offered the best agreement with observations for NPIW, andXu et al, 1999).that larger isopycnal and thickness diffusivities produced betterAccording to the simulated distribution of pasive tracers distributions of water masses relative to observations. Thein the global ocean, Orr et al. (2001) pointed out that the eddy good results were also obtained in their simulations of CFCstransport, which is not explicitly included in coarse resolution(Yamanaka et al, 1998b), in which a large vertical diffusivity inmodels, plays an important role in thetransport of the water the Okhotsk and Bering Seas, which is one order of magnitudemass. Since Danabasoglu and Gent (1994) introduced the Gent- larger than that used in their ocean general circulation modelMcWilliams (GM) mixing scheme (Gent and McWilliams, 1990; (OGCM), was employed. Xu et al. (1999) have used a similarGent et al, 1995) of tracers into the Geophysical Fluid Dynam- basin-wide OGCM of the North Pacific north of 20°S to assessics Laboratory (GFDL) model and obtained some improved different tracer mi中国煤化Iptake ofCFCsFoundation item: The National Basic Research Program (973 program) of China under contracYHC N M H GNatural ScienceFoundation of China under contract Nos 41075091, 41105087 and 40730106. .*Corresponding author, E-mail: xyf@mail.iap.ac.cnLI Yangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No. 5, P 8-169in the North Pacific in terms of both annual mean forcing and resolution of 1.0*x1.0°. There are 26 unequal vertical levels withperpetual wintertime forcing. It was found that the results the maximum depth of 5 500 m, and the thickness of the upperfrom the model forced by perpetual wintertime climatologies 10 levels is 100 m.were much better than those from the model forced by the The primary features of the model include n-vertical coordi-annual mean climatologies. The results from the GM scheme nates, free surface primitive equations, and the Arakawa B-gridwere generally better than those from the horizontal mixing scheme. The GM mixing scheme and Richardson- dependentscheme compared with the observations. It is also knownmixing process (30°s to 30°N) are employed in the model. Thethat although the perpetual wintertime forcing can generate a 0GCM is forced at the suface by monthly mean climatologicalbetter distribution of CFCs in the North Pacific, the simulated boundary conditions of thermal fluxes, salinity and wind stress.surface distribution of tracers can hardly be compared with The Newton damping type for thermal forcing boundary con-the observations because there was no seasonal variation of dition is used. The datasets from Max Planck Institute -Oceansea surface temperature, which is extremely important for the Model Intercomparison Project (MPI-OMIP) are adopted fordetermination of solubility of interesting gases.wind stress and sea surface temperature (SST), including netThose works have indicated the importance of the (< short wave radiation, non solar fluxes and coupling cofficientsSea and large isopycnal dffusivity for NPIW and the crculation (Roeske, 2001). Surface flux of salinity is obtained by restoringof the North Pacific, and also shown that the wintertime model-predicted salinity in the first level to the observationsforcing is unreasonable. In order to understand the role of the with a timescale of 30 d.Pacific Ocean in the uptake of atmospheric CO2, a reasonable. Because of the prescribed domain of the model, the lateralsimulation of water masses with the seasonal cycle of sea boundary at both east and west also needs to be prescribed. Insurface properties is particularly important on relatively long order to obtain a reasonable Antaretic circulation, we treatedtime scales. Xu et al. (2006a) employed the same OGCM as Xuthe western and eastern boundary conditions as cyclic oneset al. (999 to further study the sensitivity of water masses and that are the same as those used in the global version of LICOM.passive tracers to tracer mixing schemes under the seasonallyTo do this, we atifcially set the eastern boundary (the first andvarying forcing in the North Pacific. Although the simulatedarea from the most southern point to 50°S is set as the waterresults could be improved in the model with large isopycnalgrids. Correspondingly, the Drake Passage is widened. In otherdiffusivity, it was still apparent that the simulated inventoryof CFC-11 in the westerm North Pacific was considerably words, under the condition of not afectin the interior regionunderestimated.of the Pacific Ocean, the model of the Pacific Ocean is modi-One of the characteristics of the Okhotsk Sea is that it is COV-fied to be zonally cyclic. The temperature and salinity near theered by seasonal sea ice from November to April and there is western boundary (sponge boundary) from 100 to 110°E arefreshwater input from the Amur River (Simizu and Ohshima,also resorted to the observations, and the resorting time scale2006), which can influence the salinity distribution of the Ok-smoothly increases from 30 d to infinity.hotsk Sea. In addition, the tidal currents in the Kuril StraitsPreliminary simulations have shown that the model couldare imprtant processes for the ventilation of NPIW because well reproduce main features of the Pacific Ocean, includingthey can cause strong vertical mixing (Nakamura et al, 2000;water mass and circulation, although the equatorial currentOhshima et al, 2002). All of these are difficult to be simulatedsystem could not be well resolved because of horizontal resolu-in the large-scale model. As a result, the interior restoration intion (Xu et al., 2006). It has also been confirmed that the resultsfrom the model with GM were generally better than those withthe Okhotsk Sea and other methods are still necessary for thelarge-scale model simulation. In this work, we have an attemptthe horizontal tracer mixing scheme. In these studies, the mod-to explore how the interior restoration in the Okhotsk Sea andel employed a relatively large area of restoration to the observa-isopycnal diffusivity affect simulations of NPIW and other cur-tions at both southern and northern boundaries. In this study,rent systems in the model of the Pacific Ocean under seasonallythe restoration to the observations at the northern boundaryvarying climatological forcing. The different roles of the Ok-is removed. At the southern boundary, the restoring method ishotsk Sea and large isopycnal diffusivity in the formation andstill used in the model. The restoring region is from 75° to 55°Swith the time scale of halfof 1 a in the Southern Ocean.transport of NPIW will be discussed.In this study, a series of four numerical experiments is de-The rest of this paper is arranged as follows. A simplesigned. In all four numerical experiments, the Laplace mixingdescription of the model and experiments is given in Sectionscheme of the momentum is used, with the horizontal and ver-2. The results and discussion about physical fields are given intical viscosities being 2.0x104 and 1.0x10-3 m2/s, respectively.Section 3. Finally, conclusions are given in Section 4.Besides the Pacanowski- Philander (P-P) scheme used in thetropical region of 30°S to 30°N, the GM tracer mixing scheme2 Model description and experimental designis used in the model. For the P-P scheme, background verticalThe model of the Pacific Ocean used in this work was con- viscosity and diffusivity are set to 1.0x10-4 and 1.0x10-5 m2/s,figured from a global ocean general circulation model called respectively, and maximum vertical viscosity and diffusivity ofLICOM (State Key Laboratory for Numerical Modelling for At- 5.0x10-3 m2/s are used. For the GM scheme, a thickness diffusimospheric Sciences and Geophysical Fluid Dynamics (LASG)，ity of 1.0x 103 m2/s is used in all four experiments.Institute of Atmospheric Physics (IAP) climate ocean model).The first simul中国煤化工loration in theThe techniques of LICOM have been detailed by Liu et al. North Pacific, withdiffusion coef-(2004). With the basic frame of LICOM, the calculated domain ficients being 1.0xYHCNMHGsingthesamefor our model of the Pacific Ocean is between 98°E and 690W values of tracer diffusivity as in Kun I, Run 2 employs the inte-(from the west to the east) and 75°S to 65°N with the horizontal rior restoration of temperature and salinity in the Okhotsk Sea10LIYangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No.5, P 8-16(50° to 60°N, 120° to 160°E) with a time scale of 3 months. The used in Run 1, simulated temperature is reduced somewhat inthird simulation (Run 3) is designed to investigate the effect of the region of transport of AAIW, while in the Bering Sea nearisopycnal diffusivity on the formation of circulation and water the northern boundary of the North Pacific, under the situationmasses, which is the same as Run 1 except that the isopycnal of absence of transport of Bering strait cold water, the cyclonicdiffusivity is doubled. The final simulation (Run 4) features the circulation of Alaskan Gyre is increased with the increase of iso-combination of restoration in the Okhotsk Sea and large isopyc- pycnal diffusion, which leads to the enhancement of northwardnal diffusivity, that is, Run 2+ Run 3. The Run 1 case is generally transport below the subsurface. Thu, the simulated tempera-taken as a reference in the following discussion.ture north of 50°N is higher in Run 3 than in Run 1. In the si-The temperature and salinity were initialized from Levi- multaneous use of both the interior restoration of temperaturetus94 (Levitus and Boyer, 1994; Levitus et al, 1994). All the ex- and salinity in the Okhotsk Sea and the enhanced isopycnal dif-periments were integrated under the seasonally varying clima- fusion, the simulated results in Run 4 show that compared totological forcing for 1 600 a from the static status. Equilibrium Run 1, the temperature is lower in the whole region of Pacificwas considered reached if the rates of change of the annually intermediate waters, whereas near 300 m depth of the northernaveraged domain-mean temperature and salinity at each level boundary, the temperature is still higher, but the area and mawere less than 0.005°C and 0.001 5 per century, respectively.nitude of higher temperature are obviously smaller than thosein Run 3. A comparison of four simulations demonstrates that3 Results and discussionboth enhanced isopycnal diffusion and restoration in the Ok-hotsk Sea favor the improvement of simulated results.3.1 Flowfields and water massesThe distribution of salinity with more characteristics of theFigure 1 shows the distributions of differences in zonally av- water mass gives an obvious variation during these four simula-eraged annual mean temperature between the simulation and tions (Fig. 2). Compared to the observations (Fig. 2a), the mainobservations, and among the simulations. The simulated results difference is that the simulated formation of intermediate wa-from Run 1 are relatively consistent with the observations from ters is weak in Run 1. The simulated salinities are higher thanWOA05 (Fig. 1a). The main difference between the simulation the observations in the intermediate waters. For example, theand observations is that the simulated results are warmer, par- maximum differences of 0.28 and 0.16 in the regions of NPIWticularly in the region of the Antarctic and North Intermediate and AAIW can be seen. Inclusion of the restoration of salinityWater Masses. A difference of over 2 °C occurs at a depth range of in the Okhotsk Sea in Run 2 can obviously reduce simulated sa-200 to 600 m at 50°N. Use of the restoration of temperature and linity in the North Pacific (Fig. 2b). The maximum reduction ofsalinity in the Okhotsk Sea can greatly improve the simulated over 0.22 is located near 50°N at the water depth of 400 to 600 m.results in the North Pacific. The temperature between 200 and Like temperature, a decrease in salinity can reach the equator600 m depth at 50°N can be reduced by about 1.6°C (Fig. 1lb). along the transport pathway of NPIW, and the deepest decreaseFurthermore, along the transport pathway of NPIW toward the can occur down to the region of more than 3000 m depth inequator, simulated temperature is lower in Run 2 than in Run 1. the Okhotsk Sea. This results in that the 34.3 isohaline of NPIWThat indicates that, just as the conclusion made by Yamanaka et can reach the deepest water of 800 m, while the 34.4 isohalineal. (1998a), the Okhotsk Sea is an important source of NPIW. For can reach as far south as 20°N (not shown). In the absence ofRun 3, in which isopycnal diffusivity is doubled from the value the interior restoration in the Okhotsk Sea, under the effect of6060*SPS， 20°S， 0°。°N 40°N 60°N=04.0.5200一20600-且.1000巨1000-下之。哥1500 i员15002 5002500-);4了500I3500504500.60S，40°s， 20°s， 0°， 20°N_ 40°N_ 60°N60°s 40°S， 20°S， 0°_ 20°N_ 40°N__ 60°N200量200--..2-0;2-2. -0-42 -o.2'=8:600日1000出三蔓1500受1500-3 500-4500 L中国煤化工MYHCNM HGFig.1. Difference of annual and zonal mean temperature (°C) between the simulations and observations, and among the simula-tions: Run 1-WOA05 (a), Run 2-Run 1 (b), Run 3-Run 1 (C), and Run4-Run1 (d).LI Yangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No. 5, P 8-1660°S 40°S， 20°S__ 0° _，20°N 40°N 60°N60*S 40°S， 20S， 0°_ ，20°N， 40°N 。60°N2002-09年-0.04600600-g 1000-f0.08-0.12-0.100.08--0.0615000、-0.0490.040.08e 1500-=0.2500-3500 只3500t_0.044 504 500609S， 40°S20°N__ 40°N， 60°N60°S，40S.。20°S，0°_20°N, 40°N、 60°N__0.-0.2“. -92 -0.2/=8:星00一且1000g 1000g15002-0.2员1500丁2 5002500.---.3 5003 500-Fig.2. Difference of annual and zonal mean salinity between the simulations and observations, and among the simulations: Run1-WOA05 (a), Run 2-Runl (b), Run 3-Run 1 (C), and Run 4 - Run 1 (d).enhanced diffusion in Run 3, the areas of simulated AAIW and 3b) because of changes in circulation (Fig. 4b). Because of theNPIW are obviously extended. Furthermore, as with tempera- enhanced isopycnal diffusion in Run 3, a variation in ACC canture, in the region north of 50°N, simulated salinity is increased, be seen at the boundary of 58s. It is clear that south of 58°Swith the maximum at about 400 m depth near the northern stream function is reduced by up to 1.5x106 m3/s, while northboundary. Simulated results in Run 4 are much closer to the of 58S it is increased by up to 2.5x106 m3/s. The variation ofobservation than those from other cases. Simulated salinity in circulation in both sides of the Antarctic front affects the con-Run 4 is lower than that in Run 1 in the whole Pacific (Fig. 2d). vergence band in the Southern Ocean, which further influencesThe decrease in salinity is enhanced by the combined effects of the Antarctic Intermediate Water. All of this is reflected well inthese factors. Meanwhile, increased salinity near the northern the distribution of simulated results of temperature and salinity.boundary in Run 3 due to the enhanced isopycnal diffusion is Compared to the Southern Ocean, there are no obvious changesgreatly weakened.in stream function in the other regions. Simulated results (notSimilar variations of temperature and salinity in these four shown) in Run 4 are the superposition of main variations in Runsimulations demonstrate that the Okhotsk Sea plays an impor- 2 and Run 3. In other words, the flow in the Northwest Pacifictant role in the formation of water masses in the North Pacific, is increased, and ACC undergoes a similar variation to that inwhile the isopycnal diffusion has an obvious effect on the en- Run 3.hancement of transport of temperature and salinity. Figure 3The annual mean circulation simulated in Run 1 (Fig. 4a)shows barotropic stream function (Eulerian velocity+eddy- shows that the inflow to the Okhotsk Sea from the Pacific Oceaninduced velocity) from Run 1, indicating two large subtropical is mainly through the northern part of Kuril Straits and the out-gyres with the maximum transport of more than 40x106 m3/s，flow to the Pacific Ocean occurs through the southern part ofwhich is consistent with the results by the coarse model (Hirst Kuril Straits. This cyclonic circulation is generally consistentand Godfrey, 1993) but smaller than the results by high-reso- with the previous knowledge (Simizu and Ohshima, 2006). Thelution models in the north subtropical gytes (Semtner and shortcoming in the simulation is that the anticyclonic currentChervin, 1992; Liu et al, 2004). The weak north subtropical gyre in the Kuril Basin (Ohshima et al, 2002) cannot be resolved be-is caused by the simulated weak Kuroshio Current, which is not cause of the coarse resolution adopted in our model. With thewell described in our model because of the relatively coarse interior restoration in the Okhotsk Sea, the circulation simu-resolution. The maximum Antarctic Circumpolar Current (ACC) lated in Run 2 is very different from that simulated in Runreaches 280x106 m3/s, which is much larger than the observa- The currents, especially the East Sakhalin Current, are reversedtions of 130x106 m3/s, but closer to that obtained by Hirst and from those in both Run 1 and observations with satellite-Godfrey (1993) with a global modular ocean model (MOM) sim- tracked drifters (Ohshima et al, 2002). This demonstrates thatulation. Large ACC is associated with the treatment of easternalthough the interior restoration can improve the simulation ofand western boundary conditions in the Southern Ocean of our temperature and salinity in the okhotsk Sea, the current fieldsmodel. An increase in the area of restoration in the Southern are unrealistic.Ocean can greatly reduce the strength of ACC. If the restoration中国煤化工extends from the southern boundary to 40°S, the maximum 3.2 Simulation ofACC can reduce to 160x10 m3/s (mot shown)..The interior resMYHC N M H Grects the circu-The interior restoration in the Okhotsk Sea in Run 2 leads to lation in the North Pacific, which also influences the distributionthe increase of up to 8x106 m3/s in the Northwest Pacific (Fig. of temperature and salinity in the North Pacific. Figure 5 shows12LIYangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No.5, P 8-16100° 120° 140° 160° E180°W160° 140° 120° 100 80°135°140°145°150°559160° E60°-8°-406°__2(20-4° -N29-0°2040°-4°60°262_46零280S==260760°⊥100°20°160° E 180°W40°120°100° 8030°40.500心0.505元6070°Fig.3. Annually averaged barotropic stream function (the Eulerian velocity+eddy-induced velocity, in units of 106 m3/s): Run 1 (a),Run 2- -Run 1 (b), and Run 3- -Run 1 (c).130° 135° 140° 145° 150° 1559 160°E 130° 135° 140° 145° 150° 155 160° E0°-60*-5°一5°-50°-0°L0.08 m/sFig.4. Annual mean circulation from 130 to 160*E at 150 m from Run 1 (a) and Run2 (b).the distribution of temperature and salinity along 50°N from these four simulations indicates that simulated insufficiencythe observations and simulations. The observations reveal a low of NPIW is associated with weak vertical mixing which may betemperature in the Okhotsk Sea. At the western boundary of the induced by the tidal currents in the Okhotsk Sea. Strong tidal50°N section, a minimum temperature of below 2°C is located currents occur in the Okhotsk Sea, and are responsible for thebetween 50 and 700 m. The low temperature in the Okhotsk strong vertical mixing in the Kuril Straits and thus for the waterSea induces an obvious east-west gradient above the 1000 m mass modification (Nakamura et al., 2000). Based on this rec-depth at 50°N. Although Run 1 can generate this characteristic ognition, the vertical mixing can be enhanced by the additionof low temperature in the subsurface under the seasonally vary- of tidal mixing process in our model, which may improve ouring forcing in the absence of ice, the area of low temperature simulation. The high temperature in the Okhostk Sea in Run 3is quite small, which demonstrates that the vertical mixing is indicates that enhancement of isopycnal diffusion cannot im-weak in the Okhotsk Sea in Run 1. Because of the interior res- prove the formation of NPIW.toration in the Okhotsk Sea, both Run 2 and Run 4 generate aIn the four sim中国煤化iribution ofsa-similar temperature gradient to the observations. Using large linity is quite simie observationsisopycnal diffusivity in Run 3, the model generates the spread- show that salinity[H.CNMHGslylowerthaning of Northeast Pacific warm water towards the west. The 6°C that in the other regons o1 ne Norun raclmc. unlike tempera-isotherm extends more 5° towards the west. A comparison from ture, its vertical distribution in the Okhotsk Sea monotonouslyLI Yangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No. 5, P 8-163140*E160°E 1800160°W. 140°W. 1200W 140°E160°E.180°1609W，1409W _ 120°W3.4200 .3:800 |_ 34.24000054.260034.2一 34.2.00-34.44.4800营1 0054.4受100034.6-1 2001 400-160016001 b1 800-L160°E180°160*W140°W.120*W40°E160W 140W. 120°WP7.33-200 "344.254.2.800-受10000034.6.200-1 600-d1 800I 800140°E80°160°W .1400W 1200W00 H34.2.00 tI0.14001600,1 800-Temperature/9CFig.5. Longitude-depth sections of annual mean temperature (in color) and salinity along 50°N: WOA05 (a), Run 1 (b), Run2 (c),Run 3 (d), and Run4 (e).increases with depth. Although Run 1 can generate this trend Northeast Pacific. The main difference for the four simulationsof distribution, its similar vertical mixing is not sufficient. The is restricted in the Okhotsk Sea. Compared with the observa-penetration of low salinity is too shallow. The 34.2 isohaline is tions, the simulated depth in Run 2 and Run 4 is more reason-about 500 m shallower than the observed one at 150°E. The low able than that in Run 1 and Run 3 in the Okhostk Sea, whereassalinity is partly induced by the transport of fresh water from too shallower in the North Pacific. This means that the restora-the Amur River (Simizu and Ohshima, 2006). Nevertheless, the tion is not good at the improvement of the physical fields out ofimprovement of the simulation of salinity in the Okhostk Sea the restoring region, and the density gradient between the Ok-needs a higher horizontal resolution which can produce a rea- hostk Sea and the Pacific Ocean is decreased in Run 2 and Runnable eddy transport and a reasonable runoff.4. The small difference between Run 1 and Run 3 indicates thatFigure 6 shows the calculated depth of the 26.8 σg isopycnal the large isopycnal diffusivity has a ltte effect on the density.Figure 7 shows the distributions of annual mean tempera-lowest isopycnal surface appears to be in the Okhotsk Sea, with ture on the 26.8 σq中国煤化工ations (Fig.7a)the depth of less than 100 m. The difference between the simu- indicate that the coEtsk Sea, mainlylation in Run 1 and the observations mainly occurs in the region moving eastwardsYHCNMHGth,thenwest-north of 30°N, where the depth of the 26.8 σg isopycnal surface wards along the direction ot depth gradients shown in Fig. 6a,is shallower in the simulation except the Bering Sea and the finally reaching the subtropical Northwest Pacific. The resultsLIYangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No.5, P 8-16100° 120° 140° 160 E180°W 160° 140° 120 100 80°100 120° 140° 160 E180°W 160° 140° 1200 100° 80°60°50°40-30°.Goo30*-2000--_-5C10-4C19-408品上1009 120° 140° 1600 E180°W 160° 140 120 100 80100° 120° 140° 160° E180°W 160° 140° 120° 100 80°60°e0° d50_ 1-40° :40°300° 90°0°一Fig.6. Depth (m) of the 26.8 o。isopyncal surface in the north Pacific Ocean calculated from WOA05 (a), Run 1-WOA05 (b), Run2-Run 1 (c), Run 3-Run 1 (d), and Run 4-Run 1 (e).100° 120° 140° 160 E180°W 160° 140° 120° 100 80°0° 120° 140° 160° EI80°W 160° 140° 120° 100° 80°°50.40400-800°0 120° 140° 160° E180°W 160° 140 120° 100 8000° 120° 140° 160° E180°W 160° 140° 120° 100 8(0°e60s0° .50°-10-0e-08-0°-0-o :g.B---0.4中国煤化工MYHCNMH GFig.7. Distributions of annual mean temperature (°C) on the 26.8 σ。isopyncal surface from WOA05 (a), Run 1-WOA05 (b), Run2-Runl (C), Run 3-Run 1 (d), and Run 4-Run 1 (e).LI Yangchun et al. Acta Oceanol. Sin, 2014, Vol.33, No. 5, P 8-165from Run 1 are quite similar to the observations. This means Our results are quite consistent with those from experimentthat the transport pathway of cold water from the Okhotsk Sea 4 of Yamanaka et al. (1998a). The 34.4 isohaline can reach theis basically consistent with that suggested by the observations. western boundary of subtropical North Pacific, but the trans-The main difference is that simulated temperature is higher port strength of the whole low-salinity water is weaker in ourthan the observation in the NPIW region. After adding the res- Run 2 than in the Yamanaka et al. experiment 4 (not shown).toration in the Okhotsk Sea, simulated temperature is greatly The difference in simulated salinity between Run 3 and Run 1decreased, which is mainly induced by the colder source in the is quite similar to the difference in simulated temperature. TheOkhostk Sea (Fig. 5). Only use of large isopycnal diffusivity in enhanced isopycnal diffusivity can increase the transport ofRun 3 can enhance the westward transport of cold water. How- lovw-salinity water along the isopycnal surface south of 40°N,ever, compared to Run 1, Run 3 also enhances the northward but do not avail the simulation north of 40°N This is also re-transport of warm water north of 40°N. The results from Run 4 flected in the comparison of experiments 3 and 4 in Yamanakaare actually a combination of the results from both Run2 and et al. (see Figs 9c and 11c in Yamanaka et al, 1998a). In our RunRun 3. In the subtropical region, the transport of cold water is 4 under the combined effect of the restoration in the okhotskfurther increased, while in the higher latitude region the trans-Sea and the enhanced isopycnal diffusion, the southward andport of cold water is quite similar to that in Run 2 (Fig. 7e).westward transports of fresh water are considerably increased,The transport of salinity is similar to that of temperature.although they is still weaker than the observations.The observations show that the low-salinity water with less than33.5 that flows out of the Okhotsk Sea into the Northwest Pacific 4 Summary and conclusionsmoves toward the east, and then continues to move southwardsIn the basin of the Pacific Ocean, in addition to the Antarc-and westwards. As an indicator of NPIW, the 34.3 isohaline can tic Bottom Water, both NPIW and AAIW play an important roleapproach to the western boundary (Fig. 8a). The distribution in the formation of the structure of temperature and salinity ascharacteristic of simulated salinity is basically consistent with well as the uptake and storage of passive tracers such as carbonthe observations, but the southward and westward transport dioxide. In this study we have examined the role of both the Ok-strengths are weaker with the higher salinity in the NPIW path- hostk Sea and isopycal difusion in determining the circulationway. To some extent our Run 2 is relatively similar to experiment and intermediate water masses in the ocean general circulation4 in Yamanaka et al. (1998a) except that they used the restora- model of the Pacific Ocean, particularly NPIW. Four simulationstion of temperature and salinity in the Bering Sea with the per- were conducted, which include the first simulation (Run 1) inpetual winter forcing, while our Run 2 is under the seasonally which there is no any restoration in the North Pacific, and bothvarying forcing and without the restoration in the Bering Sea. isopycnal and thickness diffusion coefficients are set to 1.0x 103100 120° 140° 160 E180°W 160° 140° 120° 1009 80°_00° 120° 140° 1600 E180°W 160° 140° 120° 100° 80°°a°一t50°4040030-302020°19100120° 140° 160 E180°W160 140° 120° 100 80100 1200 140° 160° E180°W 160° 140° 120° 100 8(°一60*S0-50°-40°-300.<. -0,1-- 0.05一0.05---10100 120° 140 160° E180°W 160° 140° 120° 100° 80°0°10°一;0-0--0.15----0.-中国煤化工YHCNMH GFig.8. Distributions of annual mean salinity on the 26.8 σ。isopyncal surface from WOA05 (a), Run 1-WOA05 (b), Run 2- Run 1 (c),Run 3 -Run 1 (d), and Run4 -Run 1 (e). .