1. Introduction
The broad patterns of surface layer salinity (SLS) are set by the interaction of surface waters with the atmosphere (Wüst 1935). The descending branch of the Hadley cell drives anticyclonic flow over the ocean basins, causing evaporation to exceed precipitation, elevating SLS and creating six subtropical salinity maxima (hereinafter Smax). Five of the Smax lie primarily in the open ocean: the Pacific and Atlantic Smax are located around 25°N and 20°S; the south Indian Smax is around 30°S (Gordon et al. 2015). The north Indian (NI) Smax lies in the Arabian Sea and Gulf of Oman around 18°N. It is bounded by coasts to the north and west. For brevity, we will refer to each Smax by the first initial of its hemisphere and ocean basin, for example, NA Smax denotes the North Atlantic subtropical salinity maxima. The Smax are bounded by different isohaline values given in Table 1. We selected these values as those that define the maximum volume that is confined to the subtropical region of each basins’ corresponding hemisphere.
The isohaline contour bounding each Smax.
The annual cycle of evaporation minus precipitation (E − P) flux over the subtropical regions is asymmetric about the equator (Large and Yeager 2009). Over the Northern Hemisphere (NH) Smax, the semiannual harmonic dominates (peaking in local summer and winter), while over the Southern Hemisphere (SH) Smax, the annual harmonic dominates (peaking in local winter). In spite of this asymmetry in E − P forcing, the SLS of all six Smax reaches a maximum in local fall and minimum in local spring [open-ocean Smax (Gordon et al. 2015), NI Smax (Rao and Sivakumar 2003)]. On the eastward and poleward edges of each open-ocean Smax, the vertical gradient of salinity 
This study employs an eddy-resolving ocean general circulation model (OGCM) to compute salinity budgets for all six Smax to address two objectives: determine the importance of individual ocean processes in balancing climatological mean E − P forcing and to discover why, in spite of their different annual cycles of E − P, the annual cycle of SLS for all six Smax is dominated by the annual harmonic. Following Bryan and Bachman (2015), who studied NA Smax, these budgets are computed within a control volume that is bounded by an isohaline surface in order to eliminate the mean advection term from the budget. NA Smax is perhaps the most well studied Smax, as it was the site of the Salinity Processes in the Upper Ocean Regional Study (SPURS; Lindstrom et al. 2015) and is a region of high-coverage by the Ship of Opportunity Program (SOOP; Reverdin et al. 2007). Previous observation-based studies have emphasized the importance of different processes in the subtropical salinity balance: horizontal advection by Ekman and geostrophic flow (Yu 2011), vertical fluxes [Qu et al. (2011), who only examined NA Smax; Bingham et al. (2012)], and eddy fluxes [Gordon and Giulivi (2014), who only examined NA Smax].
While OGCMs have the ability to simulate regularly gridded fields of oceanographic variables, including diffusion terms that are difficult to measure in situ, they experience a particular challenge that is relevant to the salinity budget. Of the two components that determine the density and stability of seawater, potential temperature θ and salinity S, the latter has been more difficult for OGCMs to model with fidelity (Durack et al. 2013). Because modeled S distributions can be biased—even in this precise model configuration, as discussed by Bryan and Bachman (2015)—this study reinforces its key results with comparisons to Argo float data.
2. Background
The subsurface structure of each Smax creates gradients that constrain large-scale patterns of vertical mixing. The open-ocean Smax each have a subsurface salinity maximum known as subtropical underwater (STUW; O’Connor et al. 2002). The STUW are located westward and equatorward of the sea surface salinity maxima (shown schematically in Fig. 1). This displacement creates contrasting regions of 
A schematic of an open-ocean Smax. Solid lines represent isohaline contours with increasing contour thickness representing increased salinity. The dark (light) shaded region depicts that in the mixed layer eastward and poleward (westward and equatorward) of the sea surface salinity (SSS) max, ∂zS is stabilizing (destabilizing) because of the displacement of STUW with respect to the SSS maximum.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
In addition to these regions of contrasting 
The climatological annual cycle of surface heat flux (W m−2) over each Smax in CESM. The x axis is offset by six months for the SH Smax (top x-axis labels) with respect to the NH Smax (bottom x-axis labels) to account for the hemispheric phase shift of the seasons. The color legend is: NA Smax (light purple), SA Smax (light green), NI Smax (dark green), SI Smax (light blue), NP Smax (dark blue), and SP Smax (dark purple).
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
A notable manifestation of this air–sea interaction is the generation of various types of subtropical mode waters (MWs). In a review of subtropical MWs, Hanawa and Talley (2001) develop a classification system organizing them into types, two of which are relevant to this study: type I, the archetype of which is Eighteen Degree MW (Worthington 1959), which forms on the equatorward side of western boundary current extensions because of convection induced by intense winter cooling and is relatively uniform in potential density 
Type II MWs are of particular interest in regard to the subtropical cells (STCs; McCreary and Lu 1994), which are shallow overturning circulations that connect the subtropical and tropical oceans. STCs provide a mechanism by which the properties of subtropical oceans can propagate subsurface through the Central Waters of the thermocline (e.g., Schmitt 1981; Sprintall and Tomczak 1993) to the tropical oceans, where they can potentially influence the frequency or intensity of El Niño–Southern Oscillation events (e.g., Gu and Philander 1997; Cane et al. 1997). The formation region of type II MWs coincides with the source region of the descending branch of the STCs (e.g., Joyce et al. 1998; Johnson 2006), and these MWs have been identified within the vicinity of NA Smax (Käse et al. 1985), NP Smax (Hautala and Roemmich 1998), and SP Smax (Tsuchiya and Talley 1996).
Type II MW volumes vary significantly throughout the year. After decaying to minimum thickness in the summer, they build throughout the fall and reach peak thickness in the late winter as surface cooling weakens 
3. Numerical experiment and results
a. Model description
We used the ocean (POP2; Smith et al. 2010) and ice (CICE4; Bailey et al. 2009) components of the Community Earth System Model (CESM) in an eddy-resolving oceanic tripole grid (Murray 1996) with zonal resolution of 0.1°, meridional resolution of 0.1° cos(lat), and 62 vertical levels (10-m spacing above 160 m and gradually increasing spacing toward the sea floor).
b. Model experiment
We forced the ocean and ice components of CESM in accordance with the Coordinated Ocean–Ice Reference Experiments (CORE; Griffies et al. 2009) guidelines. The COREv2 dataset is composed of a blend of NCEP reanalysis (Kalnay et al. 1996) and a variety of satellite-based reconstructions of surface atmospheric state and flux fields (Large and Yeager 2009). COREv2 includes interannually varying forcing data, spanning the years 1948–2009 and a single “normal year,” which is an exactly repeating synthetic year derived from the interannually varying data. The interannual dataset incorporates observational and reanalysis data from a variety of sources with irregular start and stop times. In 1979, COREv2 begins incorporating data from the Microwave Sounding Unit (MSU; Spencer 1993), the Climate Prediction Center Merged Analysis of Precipitation (CMAP) product (Xie and Arkin 1996), and Global Precipitation Climatology Project (GPCP) data (Huffman et al. 1997). Coastal runoff is imposed using the Dai and Trenberth (2002) climatology. The model was initialized with temperature and salinity distributions from the World Ocean Circulation Experiment (WOCE) climatology (Gouretski and Koltermann 2004) and spun up for 15 model years with COREv2 normal-year forcing. This spinup was shared by the experiment described in Bryan and Bachman (2015), but the integrations branched after spinup: Bryan and Bachman continued by extending their integration for another 5 years under normal-year forcing, and the integration in this study continued for another 33 years under interannually varying forcing corresponding to the years 1977–2009. Since the transition from normal-year forcing to interannual forcing makes the forcing fields discontinuous, we started the interannual forcing in 1977 to mitigate the effects of these perturbations and restricted our analysis to the output from 1979 to 2009. Output was saved as 5-day averages and converted to monthly averages using a weighted mean.
c. Model analysis
d. Model results and discussion
1) Climatological description of the Smax
Consistent with observations of STUW, subsurface isohaline contours are displaced westward and equatorward of the surface outcrops (Fig. 3) of each open-ocean Smax (resembling Fig. 1). The subsurface isohaline contour is not displaced for NI Smax. The STUW of SA Smax is displaced more equatorward and less westward in comparison with the other Smax because it abuts the coast of Brazil. In the vicinity of the open-ocean Smax, the mean mixed layer depths (MLDs) range from around 50 m in the summer months to around 100–130 m in the winter months. The simulation’s annual SLS cycles, as indicated by outcrop extent (Fig. 4), are roughly in phase with MIMOC observations although the amplitude of the MIMOC annual cycles is larger for all Smax. The phase of these annual SLS cycles is also consistent with in situ observations [NI Smax (Rao and Sivakumar 2003), open-ocean Smax (Gordon et al. 2015)] with seasonal maxima (minima) in local fall (spring). The model tends to have a fresh bias (Fig. 5), with smaller outcrops for each Smax, with the exception of NP Smax. The open-ocean Smax outcrops tend to have a spatial bias, with the MIMOC outcrops extending farther eastward and poleward. This bias suggests the model may overestimate vertical diffusive fluxes in regions that have destabilizing 
The model climatological mean SLS. Black contours are the mean position of isohaline surface outcrops for each Smax. Gray contours are isohalines at 100 m depth. Each of the open-ocean Smax contours at 100 m depth is positioned westward and equatorward of the surface outcrop. The NI Smax surface and 100-m contours are at roughly the same position. White lines depict transects along which depth profiles are shown in Figs. 8 and 9.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
The annual cycle of outcrop extent anomaly (m2) from annual mean value for CESM (black) and MIMOC (light blue).
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
Model climatological mean SLS (psu, shaded) and comparison of climatological mean Smax outcrops of the integration (black contours) and MIMOC (light blue). Black contours are CESM; white contours are MIMOC. In all cases, with the exception of NP Smax, the MIMOC outcrops are larger, with the largest discrepancy in Smax, where MIMOC outcrop area is nearly double that of the model. The white areas along the eastward and poleward edges of each MIMOC outcrop cover the paths of the Argo floats over the 2-yr periods described in section 4.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
2) Climatological Smax salinity budgets
We consider the percentage of the climatological annual mean E − P that is balanced by each budget term (Table 2). Eddy advection balances the largest percent of E − P in NI Smax (−77%), SI Smax (−70%), NP Smax (−60%), and NA Smax (−55%). Vertical diffusion balances the largest percent of E − P in SP Smax (−60%) and SA Smax (−55%). The tendency term is small but of the same sign as E − P in all six Smax, indicating that the mean salinity of each Smax increased during the integration. While this is consistent with the amplification of SLS variance (Durack and Wijffels 2010) in response to a strengthening hydrological cycle (Held and Soden 2006), this result could also be a consequence of model drift (Durack et al. 2013).
Climatological annual mean magnitude of each budget term and the percent of E − P balanced by each of the remaining budget terms (Sv psu). Since the mean advection term should vanish from (7), it provides an estimate of the error in (8). We include its magnitude as an error estimate in the rightmost column.
Turning our attention to the annual cycles of each budget (Fig. 6), the SH Smax exhibit E − P minima in the late spring (November for SA and SI; December for SP) and maxima in the winter (June for SP; July for SA and SI). The NH Smax exhibit an annual cycle of E − P dominated by the semiannual harmonic with wintertime maxima in January and summertime maxima in June (NP) or July (NA and NI). In spite of this difference in the annual cycle of E − P, the annual cycle of net S flux out of the Smax into the ocean interior (which is the sum of the horizontal diffusion, vertical diffusion, and eddy advection terms) is roughly in phase for all of the Smax. In each of the open-ocean Smax, the magnitude of net flux is greatest in the winter while the net flux for NI Smax peaks in the late fall. The magnitude of the vertical diffusion term peaks in the winter for all of the open-ocean Smax. The magnitude of the eddy advection term nearly doubles in the winter for SI Smax and is slightly enhanced for NA and SA Smax. It does not have a noticeable seasonality in NP and SP Smax. In NI Smax, the magnitude of eddy advection reaches a minimum in April and maximum in October. This is likely related to the onset of the southwest monsoon, which develops during the summer and dissipates during the fall and is associated with freshening in the Arabian Sea (Rao and Sivakumar 2003).
The climatological annual cycle of the salinity budget (Sv psu) for each Smax. The net flux term is the sum of horizontal diffusion, eddy advection, and vertical diffusion.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
3) Vertical diffusion in the subtropics
Since the annual cycle of S flux from NI Smax into the ocean interior is governed by a distinct process, eddy advection, we continue by examining why the magnitude of the vertical diffusion term peaks in the late winter for the open-ocean Smax. However, since the vertical diffusion term peaks in winter in the open-ocean Smax and is approximately constant throughout the year in the NI Smax, we analyze NI Smax in parallel with the open-ocean Smax because this contrast demonstrates how the annual cycle of vertical diffusion is influenced by the annual cycle of SHF.
Figure 7 of Bingham et al. (2012) shows that in the regions eastward and poleward of the open-ocean Smax, a vertical flux term 
The simulation’s climatological mean vertical diffusive flux of salt though the base of the mixed layer (psu m s−1, shaded). Black and gray contours are isohaline contours as in Fig. 3. Light blue contours are regions where salt finger convection penetrates the base of the mixed layer. In the vicinity of the Smax, these regions of salt finger convection are on the eastward and poleward edges of Smax and correspond with regions of negative vertical diffusive salt fluxes through the mixed layer base (magenta-shaded regions).
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
Since the contrasting regions of stability are on diagonally opposite sides of the open-ocean Smax, we create depth transects through each Smax (Figs. 8, 9) along the paths in Fig. 3. Low Q water forms in the surface layer along each transect however, the waters beneath these low Q regions generate salt finger convection only on the eastward and poleward sides of the open-ocean Smax. When low Q water forms on the westward and equatorward side of the Smax, as visible in NP and SP, the waters underlying these low Q volumes do not generate salt finger convection because 
Transect plots through each NH Smax along the paths shown in Fig. 3. Climatological January planetary potential vorticity (m−1 s−1, shaded), isohaline bounding each Smax (white contours), and volumes where |Q| < 3.0 × 10−10 m−1 s−1 (magenta). Climatological mixed layer depth for January, March, and May is shown by mustard yellow contours ranging from dark–thick to light–thin. Climatological areas of salt finger convection for January, March, and May are shown by blue contours ranging from dark–thick to light–thin.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
As in Fig. 8, but for the SH with July, September, and November replacing January, March, and May.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
A simplified schematic of the seasonal deepening of the mixed layer. The solid lines depict isohaline contours. Dashed–dotted lines surrounding the white volumes depict low Q layers. Dotted lines surrounding dark gray volumes depict areas of salt finger convection.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
Though we have not computed a buoyancy budget in this study, it is important to note that the locations of these regions where the waters become susceptible to salt finger convection are significant because of the connection between type II mode waters and the descending branch of the STCs. This seasonal propagation of salt finger convection provides a diapycnal mixing pathway that transforms near-surface water into deeper, subsurface Central Water. Furthermore, since it mixes density upgradient, the newly transformed Central Waters can escape re-entrainment when the mixed layer deepens in subsequent winters.
4. Argo float selection and analysis
The mixed layer deepening mechanism described in section 3 is significant because it demonstrates how the annual cycle of the salinity budget, type II MW formation, diapycnal mixing induced by salt finger convection, and the descending branch of the STC may be connected. However, these results come with the caveat that the model has a low SLS bias in this region. We reinforce our results using Argo (Roemmich et al. 2009) profiles to demonstrate that this process can be observed in situ. This analysis follows Johnson (2006) for SP and Toyama and Suga (2010) for NP and shows that the process occurs on the eastward and poleward edges of all five open-ocean Smax.
a. Float selection and methodology
We select Argo floats along the eastward and poleward edges of each MIMOC Smax outcrop. To provide assurance that the observed processes are not merely single-year anomalies (since this study attempts to explain processes in the climatological annual cycle), we select Argo floats that remain in the region long enough to provide approximately 2 years of data (73 depth profiles). As a consequence of this requirement, the Argo data are reported with varying start and stop dates. Table 3 displays the ID numbers and duration of data analyzed from each float. The white areas in Fig. 5 depict the path each float traversed during this duration. We follow the analysis of section 3, using (2) to parameterize the salt finger convection diffusivity and (9) to detect the presence of low Q water masses using the same value as in section 3. There are many methods for determining the depth of the mixed layer from in situ data (de Boyer Montégut et al. 2004). We employ the criterion of Monterey and Levitus (1997), who use a threshold method specifically for the study of subtropical MWs: the base of the mixed layer is the pressure at which a profile’s 
The float IDs and data duration for the Argo data included in section 4. Each float provides 73 profiles (approximately 2 years) of data within the eastward and poleward region of each Smax.
b. Argo results
The Argo data (Fig. 11) from the open-ocean Smax depict a buildup of a low Q layer within mixed layer as it gradually deepens in response to surface cooling throughout the winter months. Salt finger convection diffusivity peaks beneath the low Q layer and above the base of the mixed layer as the mixed layer restratifies in the late spring. Salt finger convection diffusivity persists year-round but weakens throughout the summer months. The open-ocean Smax have relatively deep maximum mixed layer depths: NP and SP ~ 160 dbar, SA and NA ~ 225 dbar, and SI ~ 240 dbar. In the NI Smax profiles, the maximum mixed layer depth is much shallower (reaching ~90 dbar); the salt finger convection diffusivity is both weak and inconstant.
Seventy-three depth profiles from each of the Argo floats with data durations listed in Table 3. Depths where |Q| < 3.0 × 10−10 m−1 s−1 are depicted in magenta. Salt finger convection diffusivity as estimated by (2) (blue shading). Depth of the mixed layer depicted with a mustard yellow thick line.
Citation: Journal of Physical Oceanography 46, 10; 10.1175/JPO-D-15-0202.1
5. Summary and conclusions
This study aims to determine the relative importance of internal ocean processes for balancing climatological annual mean E − P fluxes and to explain why, in spite of their different annual cycles of E − P, the SLS of all six Smax reaches a maximum in local fall and minimum in local spring. Eddy advection plays a primary role in the climatological balance of E − P fluxes, from a high of balancing 77% of climatological annual mean E − P for NI Smax to a low of 41% for SP Smax. Vertical diffusion also plays a significant role in balancing E − P fluxes, from a high of 60% for SP Smax to a low of 28% for NI Smax. The climatological annual mean tendency term is of the same sign as E − P fluxes for all six Smax, indicating that they all increased in mean S during the integration. This is consistent with the observed pattern amplification of SLS (Durack and Wijffels 2010) but may be a consequence of model drift (Durack et al. 2013). The horizontal diffusion term is negligible in the budgets of all six Smax.
While the SLS of all six Smax reaches a maximum (minimum) in local fall (spring), the salinity budget of NI Smax is distinct: its annual cycle of net flux into the ocean interior is dominated by eddy advection. The annual cycle of SHF for NI Smax is governed by the seasonal evolution of the monsoon circulation, and thus it experiences a short period of surface cooling (typically from mid-November to early February) and relatively shallow mixed layers. The annual cycle of net flux in the open-ocean Smax budgets is governed by a combination of eddy advection and vertical diffusion for SI Smax and vertical diffusion for the remaining Smax.
The westward and equatorward displacement of STUW with respect to the surface outcrop of the open-ocean Smax creates contrasting regions of stability in their vicinity. On the equatorward and westward regions 
This finding, that low Q water is produced in response to wintertime surface cooling in regions of destabilizing 
Two other irregularities support this notion:
Type I mode waters are typically uniform in
Type I mode waters are typically associated with western boundary currents that experience intense wintertime SHF. Kallberg et al. (2005) estimate that the Kuroshio (adjacent to NP STMW) has a maximum wintertime SHF of −401 W m−2 and the Gulf Stream Current (adjacent to NA STMW) has a maximum wintertime SHF of −396 W m−2. In contrast, the Brazil Current (adjacent to SA Smax) has a maximum wintertime SHF of only −170 W m−2.
Our results come with two important caveats: the S fluxes due to salt finger convection are not differentiated from other contributions to vertical mixing in the diagnosed budgets, and the model has a low S bias in the regions eastward and poleward of the Smax. It should be possible to use the KPP mixing scheme to explicitly quantify the magnitude of S fluxes due to salt finger convection by exporting contributions to κ from each physical process in addition to the total vertical diffusion term. Additionally, St. Laurent and Schmitt (1999) developed (2) by considering mixing schemes in parameter space and found that in low Richardson number conditions, turbulent shear mixing best fit observations, while in high Richardson number and low 
Acknowledgments
Two awards supported this research: S.A.G. was supported by NNX12AF68G and F.O.B. was supported by NNX10AC16G. The model integration described in this paper was carried out on Yellowstone (Computational and Information Systems Laboratory 2012; ark:/85065/d7wd3xhc) provided by NCAR’s Computational and Information Systems Laboratory, sponsored by the National Science Foundation. Furthermore, we thank David Bailey for his assistance in keeping the model integration progressing. We found the comments from the two anonymous reviewers and the guidance of the editor to be invaluable.
REFERENCES
Bailey, D., M. Holland, E. Hunke, B. Lipscomb, B. Briegleb, C. Bitz, and J. Schramm, 2009: Community Ice CodE (CICE) user’s guide version 4.0. NCAR Doc., 22 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/cice/ice_usrdoc.pdf.]
Bingham, F. M., G. R. Foltz, and M. J. McPhaden, 2012: Characteristics of the seasonal cycle of surface layer salinity in the global ocean. Ocean Sci., 8, 915–929, doi:10.5194/os-8-915-2012.
Bryan, F. O., and S. Bachman, 2015: Isohaline salinity budget of the North Atlantic salinity maximum. J. Phys. Oceanogr., 45, 724–736, doi:10.1175/JPO-D-14-0172.1.
Cane, M. A., and Coauthors, 1997: Twentieth-century sea surface temperature trends. Science, 275, 957–960, doi:10.1126/science.275.5302.957.
Computational and Information Systems Laboratory, 2012: Yellowstone: IBM iDataPlex System (Climate Simulation Laboratory). National Center for Atmospheric Research, accessed 17 August 2016. [Available online at http://n2t.net/ark:/85065/d7wd3xhc.]
Dai, A., and K. E. Trenberth, 2002: Estimates of freshwater discharge from continents: Latitudinal and seasonal variations. J. Hydrometeor., 3, 660–687, doi:10.1175/1525-7541(2002)003<0660:EOFDFC>2.0.CO;2.
de Boyer Montégut, C., G. Madec, A. S. Fischer, A. Lazar, and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109, C12003, doi:10.1029/2004JC002378.
Durack, P. J., and S. E. Wijffels, 2010: Fifty-year trends in global ocean salinities and their relationship to broad-scale warming. J. Climate, 23, 4342–4362, doi:10.1175/2010JCLI3377.1.
Durack, P. J., S. E. Wijffels, and T. P. Boyer, 2013: Long-term salinity challenges and implications for the global water cycle. Ocean Circulation and Climate, G. Siedler et al., Eds., Academic Press, 727–758.
Gordon, A. L., and C. F. Giulivi, 2014: Ocean eddy freshwater flux convergence into the North Atlantic subtropics. J. Geophys. Res. Oceans, 119, 3327–3335, doi:10.1002/2013JC009596.
Gordon, A. L., C. F. Giulivi, J. Busecke, and F. M. Bingham, 2015: Differences among subtropical surface salinity patterns. Oceanography, 28, 32–39, doi:10.5670/oceanog.2015.02.
Gouretski, V., and K. Koltermann, 2004: WOCE Global Hydrographic Climatology. BSH Tech. Rep. 35, Bundesamtes für Seeschifffahrt und Hydrographie, 52 pp. [Available online at www.bsh.de/de/Produkte/Buecher/Berichte_/Bericht35/Bericht1.pdf.]
Griffies, S. M., and Coauthors, 2009: Coordinated Ocean-ice Reference Experiments (COREs). Ocean Modell., 26, 1–46, doi:10.1016/j.ocemod.2008.08.007.
Gu, D. F., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend on exchanges between the tropics and extratropics. Science, 275, 805–807, doi:10.1126/science.275.5301.805.
Hanawa, K., and L. D. Talley, 2001: Mode waters. Ocean Circulation and Climate, G. Siedler et al., Eds., Academic Press, 373–386.
Hautala, S. L., and D. H. Roemmich, 1998: Subtropical mode water in the Northeast Pacific Basin. J. Geophys. Res., 103, 13 055–13 066, doi:10.1029/98JC01015.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 5686–5699, doi:10.1175/JCLI3990.1.
Huffman, G. J., and Coauthors, 1997: The Global Precipitation Climatology Project (GPCP) Combined Precipitation Dataset. Bull. Amer. Meteor. Soc., 78, 5–20, doi:10.1175/1520-0477(1997)078<0005:TGPCPG>2.0.CO;2.
Johnson, G. C., 2006: Generation and initial evolution of a mode water θ–S anomaly. J. Phys. Oceanogr., 36, 739–751, doi:10.1175/JPO2895.1.
Joyce, T. M., J. R. Luyten, A. Kubryakov, F. B. Bahr, and J. S. Pallant, 1998: Meso- to large-scale structure of subducting water in the subtropical gyre of the eastern North Atlantic Ocean. J. Phys. Oceanogr., 28, 40–61, doi:10.1175/1520-0485(1998)028<0040:MTLSSO>2.0.CO;2.
Kallberg, P., P. Berrisford, B. Hoskins, A. Simmons, S. Uppala, S. Lamy-Thépaut, and R. Hine, 2005: ERA-40 Atlas. ECMWF Tech. Rep. 19, 191 pp.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Käse, R. H., W. Zenk, T. B. Sanford, and W. Hiller, 1985: Currents, fronts and eddy fluxes in the Canary Basin. Prog. Oceanogr., 14, 231–257, doi:10.1016/0079-6611(85)90013-8.
Kolodziejczyk, N., and F. Gaillard, 2013: Variability of the heat and salt budget in the subtropical southeastern Pacific mixed layer between 2004 and 2010: Spice injection mechanism. J. Phys. Oceanogr., 43, 1880–1898, doi:10.1175/JPO-D-13-04.1.
Kolodziejczyk, N., G. Reverdin, and A. Lazar, 2015: Interannual variability of the mixed layer winter convection and spice injection in the eastern subtropical North Atlantic. J. Phys. Oceanogr., 45, 504–525, doi:10.1175/JPO-D-14-0042.1.
Ladd, C., and L. Thompson, 2000: Formation mechanisms for North Pacific central and eastern subtropical mode waters. J. Phys. Oceanogr., 30, 868–887, doi:10.1175/1520-0485(2000)030<0868:FMFNPC>2.0.CO;2.
Large, W. G., and S. G. Yeager, 2009: The global climatology of an interannually varying air-sea flux data set. Climate Dyn., 33, 341–364, doi:10.1007/s00382-008-0441-3.
Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parametrization. Rev. Geophys., 32, 363–403, doi:10.1029/94RG01872.
Lindstrom, E., F. Bryan, and R. Schmitt, 2015: SPURS: Salinity Processes in the Upper-ocean Regional Study—The North Atlantic Experiment Introduction. Oceanography, 28, 14–19, doi:10.5670/oceanog.2015.01.
McCreary, J. P., and P. Lu, 1994: Interaction between the subtropical and equatorial ocean circulations: The subtropical cell. J. Phys. Oceanogr., 24, 466–497, doi:10.1175/1520-0485(1994)024<0466:IBTSAE>2.0.CO;2.
McDougall, T. J., D. R. Jackett, D. G. Wright, and R. Feistel, 2003: Accurate and computationally efficient algorithms for potential temperature and density of seawater. J. Atmos. Oceanic Technol., 20, 730–741, doi:10.1175/1520-0426(2003)20<730:AACEAF>2.0.CO;2.
Monterey, G., and S. Levitus, 1997: Seasonal Variability of Mixed Layer Depth for the World Ocean. NOAA Atlas NESDIS 14, 102 pp. [Available online at ftp://ftp.nodc.noaa.gov/pub/data.nodc/woa/PUBLICATIONS/Atlas14.pdf.]
Murray, R. J., 1996: Explicit generation of orthogonal grids for ocean models. J. Comput. Phys., 126, 251–273, doi:10.1006/jcph.1996.0136.
O’Connor, B. M., R. A. Fine, K. A. Maillet, and D. B. Olson, 2002: Formation rates of subtropical underwater in the Pacific Ocean. Deep-Sea Res. I, 49, 1571–1590, doi:10.1016/S0967-0637(02)00087-0.
Oka, E., and B. Qiu, 2012: Progress of North Pacific mode water research in the past decade. J. Oceanogr., 68, 5–20, doi:10.1007/s10872-011-0032-5.
Provost, C., C. Escoffier, K. Maamaatuaiahutapu, A. Kartavtseff, and V. Garcon, 1999: Subtropical mode waters in the South Atlantic Ocean. J. Geophys. Res., 104, 21 033–21 049, doi:10.1029/1999JC900049.
Qu, T. D., S. Gao, and I. Fukumori, 2011: What governs the North Atlantic salinity maximum in a global GCM? Geophys. Res. Lett., 38, L07602, doi:10.1029/2011GL046757.
Rao, R. R., and R. Sivakumar, 2003: Seasonal variability of sea surface salinity and salt budget of the mixed layer of the north Indian Ocean. J. Geophys. Res., 108, 3009, doi:10.1029/2001JC000907.
Reverdin, G., E. Kestenare, C. Frankignoul, and T. Delcroix, 2007: Surface salinity in the Atlantic Ocean (30°S–50°N). Prog. Oceanogr., 73, 311–340, doi:10.1016/j.pocean.2006.11.004.
Roemmich, D., and Coauthors, 2009: The Argo Program: Observing the global ocean with profiling floats. Oceanography, 22, 34–43, doi:10.5670/oceanog.2009.36.
Schmidtko, S., G. C. Johnson, and J. M. Lyman, 2013: MIMOC: A global monthly isopycnal upper-ocean climatology with mixed layers. J. Geophys. Res. Oceans, 118, 1658–1672, doi:10.1002/jgrc.20122.
Schmitt, R. W., 1981: Form of the temperature-salinity relationship in the central water: Evidence for double-diffusive mixing. J. Phys. Oceanogr., 11, 1015–1026, doi:10.1175/1520-0485(1981)011<1015:FOTTSR>2.0.CO;2.
Siedler, G., A. Kuhl, and W. Zenk, 1987: The Madiera Mode Water. J. Phys. Oceanogr., 17, 1561–1570, doi:10.1175/1520-0485(1987)017<1561:TMMW>2.0.CO;2.
Smith, R., and Coauthors, 2010: The Parallel Ocean Program (POP) reference manual. Rep. LAUR-10-01853, 141 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/pop2/doc/sci/POPRefManual.pdf.]
Snowden, D. P., and R. L. Molinari, 2003: Subtropical cells in the Atlantic ocean: An observational summary. Interhemispheric Water Exchange in the Atlantic Ocean, G. J. Goni and P. Malanotte-Rizzoli, Eds., Elsevier Oceanography Series, Vol. 68, Elsevier, 287–312, doi:10.1016/S0422-9894(03)80151-4.
Spencer, R. W., 1993: Global oceanic precipitation from the MSU during 1979–91 and comparisons to other climatologies. J. Climate, 6, 1301–1326, doi:10.1175/1520-0442(1993)006<1301:GOPFTM>2.0.CO;2.
Sprintall, J., and M. Tomczak, 1993: On the formation of central water and thermocline ventilation in the Southern Hemisphere. Deep-Sea Res. I, 40, 827–848, doi:10.1016/0967-0637(93)90074-D.
St. Laurent, L., and R. W. Schmitt, 1999: The contribution of salt fingers to vertical mixing in the North Atlantic Tracer Release Experiment. J. Phys. Oceanogr., 29, 1404–1424, doi:10.1175/1520-0485(1999)029<1404:TCOSFT>2.0.CO;2.
Stern, M. E., 1960: The salt-fountain and thermohaline convection. Tellus, 12A, 172–175, doi:10.1111/j.2153-3490.1960.tb01295.x.
Sugimoto, S., and K. Hanawa, 2007: Further evidence for non-reemergence of winter SST anomalies in the North Pacific eastern subtropical mode water area. J. Oceanogr., 63, 625–635, doi:10.1007/s10872-007-0055-0.
Toyama, K., and T. Suga, 2010: Vertical structure of North Pacific mode waters. Deep-Sea Res. II, 57, 1152–1160, doi:10.1016/j.dsr2.2009.12.004.
Tsuchiya, M., and L. D. Talley, 1996: Water-property distributions along an eastern Pacific hydrographic section at 135W. J. Mar. Res., 54, 541–564, doi:10.1357/0022240963213583.
Weller, R. A., A. S. Fischer, D. L. Rudnick, C. C. Eriksen, T. D. Dickey, J. Marra, C. Fox, and R. Leben, 2002: Moored observations of upper-ocean response to the monsoons in the Arabian Sea during 1994–1995. Deep-Sea Res. II, 49, 2195–2230, doi:10.1016/S0967-0645(02)00035-8.
Wong, A. P. S., and G. C. Johnson, 2003: South Pacific Eastern Subtropical Mode Water. J. Phys. Oceanogr., 33, 1493–1509, doi:10.1175/1520-0485(2003)033<1493:SPESMW>2.0.CO;2.
Worthington, L. V., 1959: The 18° water in the Sargasso Sea. Deep-Sea Res., 5, 297–305, doi:10.1016/0146-6313(58)90026-1.
Wüst, G., 1935: The Stratosphere of the Atlantic Ocean: Scientific Results of the German Atlantic Expedition of the Research Vessel “Meteor” 1925–1927. Amervid, 112 pp.
Xie, P. P., and P. A. Arkin, 1996: Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions. J. Climate, 9, 840–858, doi:10.1175/1520-0442(1996)009<0840:AOGMPU>2.0.CO;2.
Yeager, S. G., and W. G. Large, 2004: Late-winter generation of spiciness on subducted isopycnals. J. Phys. Oceanogr., 34, 1528–1547, doi:10.1175/1520-0485(2004)034<1528:LGOSOS>2.0.CO;2.
Yu, L. S., 2011: A global relationship between the ocean water cycle and near-surface salinity. J. Geophys. Res., 116, C10025, doi:10.1029/2010JC006937.
