1. Introduction
a. Physical oceanographic background
Until fairly recently, investigators of the Bering Strait (BS) flow focused on the long-strait pressure gradient without asking the question what forms that pressure gradient in the first place. Stigebrandt (1984) was the first to imaginatively argue that the flow through the BS [~1 Sv (1 Sv ≡ 106 m3 s−1)] is due to the atmospherically induced density difference between the Pacific and the Arctic (it evaporates over the Atlantic and precipitates over the Pacific). Accordingly, he treated the system as an estuary. His calculation was followed by more elaborate analytical models (Shaffer and Bendtsen 1994) and a series of local and global numerical experiments (e.g., Overland and Roach 1987; Hu et al. 2007, 2010). In line with these studies, the main understanding has been that local winds are responsible for the local pressure gradients and, hence, the flow. In contrast, using contours integrals in uncoupled ocean–atmosphere models, Nof (2000), Nof and Van Gorder (2003), and De Boer and Nof (2004a,b) argued that the mean sea level difference along the strait is setup by the global winds, particularly the strong Southern Winds (i.e., zonal winds along a latitudinal band connecting the tip of South America and Cape Town, which are also referred to as the Subantarctic Westerlies).
Although the BS is merely ~50 m deep and ~100 km broad, the 1 Sv that goes through it does not even come close to its hydraulic limit, which is one or two orders of magnitude larger (see e.g., Chaudry 1993 for details on hydraulic control). Similarly, as shown in De Boer and Nof (2004a), the form-drag exerted by the sill (in the BS) on the flow is not large enough to restrict the flow in any meaningful way. The mean flow state within the BS state appears to be independent of the local winds, but the local winds control the variability of the flow (Woodgate et al. 2006, and the references given therein). Of course, all of these estimates are based on measurements conducted over a period of merely 50 years. The paleoceanographic data that we will be working with here provides for the first time, an opportunity to examine these processes over a much longer period of ~12 000 calendar years before present (cal BP).
b. Models and paleoceanographic background
Here, we use a time-dependent version of the coupled analytical model of Sandal and Nof (2008b, hereafter SN). To do this, we invoke the slowly varying approach, where a steady model solution is taken to be valid when the forcing varies on a time scale much longer (~1000 yr) than the time scale of interest (~10 yr). The simplicity of the model, which incorporates only the fundamental physics needed to model the interaction of the wind-driven and the thermohaline circulation enables us to run the model with 100-yr time steps for a period of 12 000 years. As described below, we force the model with paleoceanographic proxies for the Southern Winds and the Northern Hemisphere freshwater flux, and evaluate the solution using an additional paleoceanographic record detailing the flow through the BS, not included in the model.
The locations of the cores from which these proxy records were extracted are shown in Fig. 1. The Palmer Deep site is located off the Antarctic Peninsula, not in Drake Passage proper. Analysis of the modern wind field indicates that it is highly correlated throughout the South Atlantic sector of the Southern Ocean. Most importantly, however, Nof et al. (2011) present a synthesis of the paleosouthern wind data suggesting that their approximately linear relationship holds on longer time scales as well. (Specifically, they examined the variations in the strength of the Agulhas retroflection during glacial periods.)
As we shall see, regression of our chlorite proxy against the model output provides a useful way of scaling the results into familiar units (because the paleo-proxy is based on standardized scores). The units of the comparison do not, however, affect the correlation between the paleoceanographic proxy and the model output because Pearson’s correlation is dimensionless.
Woodgate et al. (2010) note that the local winds act to slow the flow through BS by 0.4 to 0.2 Sv (out of a mean flow of ~0.8 Sv). Note, however, that Nof et al. [2011, autonomous underwater vehicle (AUV) glider and HF radar observations of circulation and stratification features in the Chukchi Sea, Ocean Sciences Meeting, 20–24 February 2012, Salt Lake City, Utah] concluded that the average magnitude of the through flow is closer to ~1.1 Sv. The local winds thus account for, at the most, 18%–36% of the mean flow. These values are similar to the paleo results that we shall later obtain and suggest that the uncertainty in our estimates is roughly 20%. Note that this comparison across time scales also argues in favor of a linear response of the flow through BS to the Southern Winds.
The SN model is based on the contour integrals first proposed by Nof (2000), and then later used by the uncoupled models of Nof and Van Gorder (2003) and De Boer and Nof (2004a,b). We show that the SN findings are consistent with a paleoceanographic proxy for the flow through the BS from 12 000 cal BP to present (Ortiz et al. 2009). We observe a period of enhanced flow into the Arctic from 6000 to 3000 cal BP, which occurred when the Southern Winds were weaker and the Northern Hemisphere freshwater fluxes were minimal following a lag time of several thousand years (dictated by the small vertical diffusivity of the ocean). The SN model provides a means of evaluating the global forcing and processes that drove the variations in flow through the BS during the Holocene. The shift in Southern Winds controls most of the flow through the BS. The difference between the SN model flow prediction and the paleo-BS flow record provides a measure of the contribution of local forcing to the flow through BS during the Holocene. This represents the first quantitative reconstruction of paleoflow through the BS, consistent with an underlying, theoretical physical oceanographic basis.
This paper is organized as follows: In section 2 we briefly review the SN model. In section 3 the model results are described, and in section 4 the outcome is summarized and discussed. Section 2 makes this paper self-contained. The reader who is anxious to see the final results without going through the detailed document is referred to the figures incorporated into section 5 where a remarkable correlation between the BS paleoflow variability and the Southern Winds variability is presented. The results are summarized in section 6.
2. The Sandal–Nof analytical BS model
Related aspects of island rule calculations can be found in (Pedlosky et al. 1997; Spall 2000; Pratt and Spall 2003). Because the integration contour does not cut through the deep-water formation region, it is permissible to include (the unknown) W in the island calculation. Also, note that the particularly chosen contour, as well as the distribution of vorticity, only affects how the net flow is partitioned between the Sverdrup interior and the western boundary current east of the island. Neither the vorticity nor the chosen contour affect the mean net flows (Q1 and Q2), which are the variables of interest here. It is precisely because one does not have to worry about the contour (or vorticity) that circulation theorems of the kind used here are so powerful.
In the heat Eq. (4), T (unknown) is the mean temperature in the convection box, whereas
The main difference between the coupled SN model and the uncoupled De Boer and Nof (2004a) model is the term on the right hand side of (4), which is the ocean–atmosphere coupling heat-flux term. It is fixed in De Boer and Nof (2004a) implying a fixed heat flux to the atmosphere regardless of the meridional overturning cell strength. By contrast, it is variable in SN implying that the ocean and atmosphere work in concert.
Equations (3) and (4) are nonlinear because they involve multiplications of the unknowns. Despite this nonlinearity, the flow regime is reversible as long as: (i) the AMOC does not completely collapse because of too large a freshwater input, and/or (ii)
3. The SN model output as a function of winds and freshwater flux
The SN analytical model suggests that the BS and the strong Southern Winds are intimately connected. Because of the stratification in the North Atlantic and the small vertical diffusivity (~0.1 cm2 s−1), the Atlantic is not large enough to allow compensating upwelling within its own limits. To sustain its surface flow, the AMOC must draw water from a much larger region outside the Atlantic itself (e.g., the Southern Ocean), providing a dynamical link to the Southern Winds (Nof et al. 2007). In the absence of North Atlantic Deep Water (NADW) formation (i.e., W = 0), the model reduces to the configuration shown on the top panel of Fig. 2, which is merely a reproduction of a figure shown earlier in Nof (2000) involving a simple flow around an island formed by the Americas. Because of north–south asymmetries, the modern Southern Winds (with no NADW) would transport about 4 Sv from the South Atlantic (Q1) northward through the Atlantic and Arctic basins and into the Pacific via the BS. As stated, the solution for this limiting case (upper panel of Fig. 2) is obtained simply by setting W = 0 in (1) and (2). Note the f is taken to be constant along AB but, as mentioned in Nof (2000), this introduces errors of less than 25%.
Figure 3 shows the resulting wind integral
To evaluate the parameter space of the model, we examined several model outputs in which we vary the freshwater flux (Ff) and the magnitude of the Southern Winds in a smooth hypothetical manner (Fig. 4). (Later on, in section 5, we will force the model with actual winds variability but this is not our focus right now.) The model implies that Q1 and Q2 are inversely correlated with the wind forcing in the south (Fig. 4, upper panels). As the winds increase, from zero to the strength of the modern winds, the temperature of the North Atlantic increases (Fig. 4 lower left panel) and the AMOC strength increases because of the enhanced cross equatorial flux of warm and salty surface water (Q1) toward the North Atlantic. Heat loss to the atmosphere of this warm and salty water (during winter convection) results in water denser than the colder, fresher water from the north (Q2).
Changing the freshwater flux (Ff) from zero to 0.03 Sv changes the magnitude of Q1 and Q2 but does not influence their dependence on the wind forcing (Fig. 4). In contrast, Ff does alter the relationship between the winds and North Atlantic temperature (Fig. 4, lower left panel) and between the AMOC and the Southern Winds (Fig. 4, lower right panel). Specifically, the gradient of the North Atlantic temperature dependency on the Southern Winds becomes steeper as freshwater fluxes increase (Fig. 4, lower left panel). For high freshwater flux, the AMOC increases with increasing wind strength, whereas for low freshwater flux, the AMOC weakens slightly with increasing wind strength (Fig. 4, lower right panel). This weakening occurs because stronger winds bring greater amounts of warm cross-equatorial water into the North Atlantic, decreasing surface density and making convection more difficult. Accordingly, decreasing the Southern Winds decreases the cross-equatorial flux, cooling the North Atlantic by increasing the relative proportion of Q2, the Arctic water entering the North Atlantic.
Note that the rate at which sea level rose is proportional to the derivative of the sea level curve, which we used as a linear measure of freshwater input to the global ocean. Because our model employs a 100-yr time step and integrates over the ocean basin, there is ample time for the freshwater flux to be well mixed. We assumed that 70% of the sea level rise was due to melting of Northern Hemisphere ice and scaled the input into the Atlantic relative to its surface area. The intercept for the relationship is a tunable parameter set to a minimal value consistent with modern freshwater flux as expressed in the model.
4. Paleoceanographic records
a. Wind variability during the Holocene
We focus on the last 12 000 cal BP because this is the most recent interval during which sea level was high enough for an open BS (Keigwin et al. 2006). Published paleoclimate data from two locations around the globe (Fig. 1) provide us with a means of testing the SN analytical model by comparing its results to past climate system variation. The magnetic susceptibility record from Palmer Deep (Domack et al. 2001) provides a useful measure of the strength of the Southern Winds during the past 12,000 cal BP. Located on the continental shelf along the Antarctic Peninsula, Palmer Deep is well situated for monitoring these winds.
The magnetic susceptibility signal recorded in sediment from Palmer Deep is correlated with enhanced delivery of ice-rafted debris (IRD) and the dilution effects of biogenic silica, with the latter factor related to wind strength. During cool, windy intervals the water column is well mixed, which forces phytoplankton down below the photic zone, limiting productivity and the flux of biogenic silica to the seafloor. By contrast, during warmer intervals with melting sea ice and reduced winds, the upper water column is strongly stratified, allowing phytoplankton to remain in the photic zone, increasing productivity and increasing the flux of biogenic silica to the seafloor (Leventer et al. 1996, 2002). The increase in biogenic silica dilutes the terrigenous component of the sediment, creating intervals of low magnetic susceptibility. Evidence of this relationship between magnetic susceptibility driven by terrestrial derived grains and stratification-controlled biogenic silica production can also be seen in the North Pacific over longer time scales (Haug et al. 1999).
Comparison of surface sediment magnetic susceptibility measurements across Drake’s Passage suggest that when the modern winds vary by a factor of two across the latitudinal range from 40°S to 60°S (Hellerman and Rosenstein 1983), magnetic susceptibility varies by two orders of magnitude in response to biogenic dilution (Pirrung et al. 2002). The relative changes in wind strength inferred from the Palmer Deep record (Fig. 5, upper panel) are also correlated with changes in pollen and sediment elemental ratios from South America. These are believed to arise from variations in the Southern Winds, supporting the above interpretation of the Palmer deep record (Gilli et al. 2005; Heusser et al. 2006; Villa-Martínez and Moreno 2007; Muratli et al. 2010).
b. Freshwater fluxes during the Holocene
Citing prior results from the uncoupled island model of De Boer and Nof (2004a) and Keigwin and Cook (2007) postulated that a freshening of the North Atlantic would tend to weaken the AMOC. However, they focused on cross-basin atmospheric water vapor transport from the Atlantic to the Pacific. While this creates the salinity difference between these oceans, it is not directly related to the changes discussed here. Here, we evaluated net global freshwater fluxes during the deglaciation by calculating the time rate of change of sea level from the deglaciation to present using the sea level curve of Milne and Mitrovica (2008). This integrates net changes in freshwater fluxes (Fig. 5, lower panel), the majority of which enter the ocean in the Northern Hemisphere.
c. Bering Strait flow variability during the Holocene
To monitor temporal changes in the flow through BS, we use a proxy based on the concentration of chlorite + muscovite inferred by visible derivative spectroscopy of a series of cores along the Alaskan Chukchi shelf (Ortiz et al. 2009). In the Western Arctic, chlorite mostly enters the Chukchi Sea from the Pacific, carried by currents or incorporated in sea ice flowing through the BS. Because chlorite is the dominant clay mineral in the NE Pacific and illite is the dominant clay mineral in the NE Chukchi Sea (Naidu and Mowatt 1983), increases in the concentration of chlorite on the continental shelf near Barrow Canyon are interpreted to indicate enhanced flow from the Pacific to the Arctic via the BS (Ortiz et al. 2009).
We selected core HLY0501-JPC6 raised from a location west of Barrow Canyon on the Alaska Chukchi Margin as the ideal core for this study because it contains the highest concentration of chlorite in a suite of eight cores (Ortiz et al. 2009). These cores were collected over a 1200-m range of water depths with an average vertical spacing of ~300 m between cores (Ortiz et al. 2009; Ortiz. 2011). The close proximity of these cores allows us to use them to generate vertical profiles documenting local variations in clay mineral distribution versus water depth (Ortiz. 2011). The relative contribution of chlorite in the sediment increases with depth into the Atlantic layer (200–900 m), reaching a maximum at 673 m, the depth of core JPC6, then decreases with depth for cores located within the cold salty bottom water (>900 m). The thickness of the postglacial sedimentary unit generally decreases with water depth on the Chukchi Margin. Exceptions are cores from a relatively steep part of the slope between 500–600 m. Here, sediment is either not accumulating rapidly or is lost because of a variety of sediment transport processes. These are as follows: winnowing by variable current strength (which is typically constrained by potential vorticity conservation), focused settling of particles from the Pacific layer above, and down slope transport by dense winter waters cascading via cross-shelf canyons. Weingartner et al. (1998, 2005) found that brine rejection produced water capable of penetrating a few hundred meters depth. Deeper downwelling flows are probably generated by the transformation of brine-rejected flows to sediment-laden, nepheloid flows down canyons (Eittreim et al. 1982). The distribution of these nepheloid flows to greater depth is very poorly known; they are likely concentrated along narrow depth intervals, such as the depth horizon sampled by JPC6.
5. Model–data comparison
For this comparison we made three different, time-dependent runs, which are referred to as runs 1, 2, and 3. We used the same numerical parameters values that were used by SN except that, to reflect present day conditions, S1 = 35.4 psu; U10 = 3.0 m s−1; Be = 0.4; and qs = 0.014 g kg−1 were changed to S1 = 35.2 psu; U10 = 5.0 m s−1; Be =0.65; and qs = 0.010 g kg−1, which are the same as those used in Nof et al. (2010, page 12, bottom of their section 4). These values produce a relatively high
In the first run, we forced the model with both the proxy-determined wind variability and the proxy-determined freshwater variability (Fig. 5). The results of this run are shown in Fig. 6. Note that the empty gap represents a single time step where the flow through the BS is negative so that our present model equations do not apply. We shall see shortly that our results display a dramatic correlation between the “predicted” (i.e., model outcome) volume flux through the BS and the paleoceanographic estimate of the flux over the last 12 000 cal BP. To verify that this strong correlation is indeed a result of the wind variability and not the freshwater variability, we made two more runs. For Run 2 we adopted a forcing where the wind was held constant, but the paleoceanographic freshwater flux varied (Fig. 7). We see from the results displayed in Fig. 8 that
The results shown in Fig. 6 could, in principle, be compared to the observed variability of
6. Discussion and summary
a. Role of the BS
Prior numerical modeling work has demonstrated that the AMOC is sensitive to the Southern Winds (Toggweiler and Samuels 1993, 1995) as well as to freshwater fluxes. The analytical model we use here is a convenient and transparent tool to study the impact of the BS on Holocene climate because it includes the most critical dynamics without adding unrelated and unnecessary complexity. This simplicity allows us to run the model in a time-dependent way that is not possible with more complex models. In the SN model, the flux through the BS into the Arctic (Q2) increases as Ff decreases. Enhanced flow through the BS into the Arctic Ocean exports fresher, Pacific water into the North Atlantic with the potential to weaken the AMOC by changing salinity in the North Atlantic deep-water formation region. During low stands, sea level drops of more than 50 m close the BS, isolating the Pacific and Arctic Oceans. (These cases are not discussed here but see Sandal and Nof 2008c and the references given therein.) As the winds increase, a larger freshwater flux is required to weaken or shut down the AMOC.
b. Time-dependent changes in the BS flow
The flow from the Pacific into the Arctic through the BS increased gradually from 10 000 to 8000 cal BP, and then dramatically at ~6000 cal BP, when the southern winds and freshwater fluxes were both minimal (Fig. 11). During that particular time, the flow from the Pacific to the Atlantic through the BS (Q2) increased. The Southern Winds remained weak until about 3000 cal BP when they abruptly increased and the flow through BS decreases.
As shown in Fig. 11, the timing of these changes is strikingly consistent between proxies and the “predictions” of the SN analytical model results. The sediment proxy was converted to Sv for comparison with the model output by least squares linear regression (y = 0.4965x + 2.8546), where x is the paleo data and y is the model output. The correlation of the two curves determined in this way is 0.69.
The model prediction of Q2 represents the mean flow of the BS. The difference in variability between the two records provides a measure of the impact of local processes on the flow through the BS. To quantify these differences, we calculated the correlation coefficient between the two cores, which as reported above is r = 0.69, indicating that 47% of the variance in the BS record
The results of the SN analytical model indicate that the flow through the BS after its initial flooding due to sea level rise is out of the Arctic as should be the case. As the freshwater anomalies from the melting of the Northern Hemisphere ice sheets are flushed out of the Atlantic, the strength of the AMOC grows until the flow through BS reverses, setting up the modern circulation with flow from the Pacific into the Arctic.
Comparison of the analytical model with paleoceanographic data provides an explanation for the climate response observed in the sedimentary record. On the basis of the SN analytical model and the paleoclimatic observations discussed here, we infer that the Southern Winds, which control not only the flow from the Southern Ocean to the North Atlantic, but also the flow from the Pacific to the Arctic, thus plays an important role in modulating the AMOC on millennial and glacial-interglacial time scales.
c. The relative importance of form-drag in the BS
It is a simple matter to determine how small the flow needs to be in order for the form- drag to be negligible. To do so, we use the Bernoulli to estimate the neglected pressure difference (
d. Bottom and side friction along the integration contour
Friction along the oceanic eastern boundaries and zonal coasts is neglected in our model. (By “friction” we include here bottom, shelf and topographic form-drag along the integration contour, save the form-drag on the BS sill, which was discussed above.) The coastal currents along Alaska and Canada northern coast are, however, relatively strong so they may involve relatively large frictional forces. It turns out that this is not the case because these flows are not a part of the interhemispheric exchange that we speak about. This can be easily seen by examining Fig. 3, which shows that 75% of the wind contribution to the steady stress integral comes from the very strong southern winds, not the winds over the region that are mentioned above. Hence, all the side and bottom frictional forces combined can, at the most, introduce an error of 25%, an error level acceptable in this kind of modeling.
Acknowledgments
We thank the Captain and Crew of the USCG Ice Breaker Healy. This study was supported by funds from the National Science Foundation’s Office of Polar Programs (ARC 0902835 and ARC 0453846) and the Kent State University Research Council. The manuscript was improved by comments from A. de Vernal, A. Mix, and S. Lee. Mike Spall, the JPO editor, provided very useful comments on the initially submitted version. L. Keigwin provided access to core HLY0205-JPC 16. All calculations related to the SN model were done by Stephen Van Gorder.
REFERENCES
Bard, E., 2002: Climate shock: Abrupt changes over millennial time scales. Phys. Today, 55, 32–38.
Chaudhry, M. H., 1993: Open Channel Flow. Prentice-Hall, 483 pp.
De Boer, A. M., and D. Nof, 2004a: The Bering Strait’s grip on the northern hemisphere climate. Deep-Sea Res. I, 51, 1347–1366.
De Boer, A. M., and D. Nof, 2004b: The exhaust valve of the North Atlantic. J. Climate, 17, 417–422.
Domack, E. W., A. Leventer, R. Dunbar, F. Taylor, S. Brachfeld, and C. Sjunneskog, 2001: Chronology of the Palmer Deep site, Antarctic Peninsula: A Holocene palaeoenvironmental reference for the circum-Antarctic. Holocene, 11, 1–9.
Eittreim, S., A. Grantz, and J. Greenberg, 1982: Active geological processes in Barrow Canyon, northeast Chukchi Sea. Mar. Geol., 50, 61–76.
Gilli, A., D. Ariztegui, F. S. Anselmetti, S. Flavio, J. A. McKenzie, V. Markgraf, I. Hajdas, and R. D. McCulloch 2005: Mid-Holocene strengthening of the Southern Westerlies in South America—Sedimentological evidences from Lago Cardiel, Argentina 49°S. Global Planet. Change, 49, 75–93.
Godfrey, J. S., 1989: A Sverdrup model of the depth-integrated flow for the world Ocean allowing for island circulation. Geophys. Astrophys. Fluid Dyn., 45, 89–112.
Hartmann, D. L., 1994: Global Physical Climatology. Academic Press, 411 pp.
Haug, G., D. Sigman, R. Tiedemann, T. Pedersen, and M. Sarnthein, 1999: Onset of permanent stratification in the subarctic Pacific Ocean. Nature, 401, 779–782.
Hellerman, S., and M. Rosenstein, 1983: Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr., 13, 1093–1104.
Heusser, L., C. Heusser, A. Mix, and J. McManus, 2006: Chilean and Southeast Pacific paleoclimate variations during the lat glacial cycle: directly correlated pollen and δ18O records from ODP Site 1234. Quat. Sci. Rev., 25, 3404–3415.
Hu, A., and G. A. Meehl, 2005: Bering Strait throughflow and the thermohaline circulation. Geophys. Res. Lett., 32, L24610, doi:10.1029/2005GL024424.
Hu, A., G. A. Meehl, and W. Han, 2007: Role of the Bering Strait in the thermohaline circulation and abrupt climate change. Geophys. Res. Lett., 34, L05704, doi:10.1029/2006GL028906.
Hu, A., and Coauthors, 2010: Influence of Bering Strait flow and North Atlantic circulation on glacial sea-level changes. Nat. Geosci., 1-4, doi:10.1038/NGE0729.
Keigwin, L. D., and M. S. Cook, 2007: A role for North Pacific salinity in stabilizing North Atlantic climate. Paleoceangraphy, 22, PA3102, doi:10.1029/2007PA001420.
Keigwin, L. D., J. P. Donnelly, M. S. Cook, N. W. Driscoll, and J. Brigham-Gretta, 2006: Rapid sea-level rise and Holocene climate in the Chukchi Sea. Geology, 34, 861–864.
Leventer, A., E. W. Domack, E. Ishman, S. Brachfeld, C. E. McClennen, and P. Manley, 1996: 200–300 year productivity cycles in the Antarctic Peninsula region: Understanding linkages among the Sun, atmosphere, oceans, sea ice and biota. Geol. Soc. Amer. Bull., 108, 1626–1644.
Leventer, A., E. W. Domack, A. Barkoukis, B. McAndrews, and J. Murray, 2002: Laminations from the Palmer Deep: A diatom-based interpretation. Paleoceanography, 17, 8002, doi:10.1029/2001PA000624.
Milne, G., and J. Mitrovica, 2008: Searching for eustacy in deglacial sea-level histories. Quat. Sci. Rev., 27, 2292–2303.
Muratli, J., Z. Chase, J. McManus, and A. Mix 2010: Ice-sheet control of continental erosion in central and southern Chile (36°–41°S) over the last 30,000 years. Quat. Sci. Rev., 23–24, 3230–3239.
Naidu, A. S., and T. C. Mowatt, 1983: Sources and dispersal patterns of clay minerals in surface sediments from the continental shelf areas off Alaska. Geol. Soc. Amer. Bull., 94, 841–854.
Nof, D., 2000: Does the wind control the import and export of the South Atlantic? J. Phys. Oceanogr., 30, 2650–2667.
Nof, D., and S. Van Gorder, 2003: Did an open Panama Isthmus correspond to an invasion of Pacific water into the Atlantic? J. Phys. Oceanogr., 33, 1324–1336.
Nof, D., S. Van Gorder, and A. De Boer, 2007: Does the Atlantic meridional overturning cell really have more than one stable steady state? Deep-Sea Res. I, 53, 2005–2021.
Nof, D., S. Van Gorder, and L. Yu, 2010: Thoughts on a variable meridional overturning cell and a variable heat-flux to the atmosphere. Geophys. Astrophys. Fluid Dyn., 105, 1–22.
Nof, D., V. Zharkov, J. D. Ortiz, W. Arruda, N. Paldor and E. Chassignet, 2011: The arrested Agulhas retroflection. J. Mar. Res., 69, 659–691.
Ortiz, J. D., 2011: Application of Visible/near Infrared derivative spectroscopy to Arctic paleoceanography. IOP Conf. Ser.: Earth Environ. Sci., 14, 012011 doi:10.1088/1755-1315/14/1/012011.
Ortiz, J. D., L. Polyak, J. M. Grebmeier, D. Darby, D. D. Eberl, S. Naidu, and D. Nof, 2009: Provenance of Holocene sediment on the Chukchi-Alaskan margin based on combined diffuse spectral reflectance and quantitative X-Ray Diffraction analysis. Global Planet. Change, 68, 73–84.
Overland, J., and A. T. Roach, 1987: Northward flow in the Bering and Chukchi Seas. J. Geophys. Res., 92, 7097–7105.
Pedlosky, J., L. J. Pratt, M. A. Spall, and K. R. Helfrich, 1997: Circulation around islands and ridges. J. Mar. Res., 55, 1199–1251.
Pirrung, M., C.-D. Hillenbrand, B. Diekmann, D. K. Fütterer, H. Grobe, and G. Kuhn, 2002: Magnetic susceptibility and ice rafted debris in surface sediments of the Atlantic sector of the Southern Ocean. Geo. Mar. Lett., 22, 170–180, doi:10.1007/soo367-002-0109-7.
Pratt, L. J., and M. A. Spall, 2003: A porous media theory for geostrophic flow through ridges and archipelagos. J. Phys. Oceanogr., 33, 2702–2718.
Sandal, C., and D. Nof, 2008a: A new analytical model for Heinrich events and climate instability. J. Phys. Oceanogr., 38, 451–466.
Sandal, C., and D. Nof, 2008b: The collapse of the Bering Strait ice dam and the abrupt temperature rise in the beginning of the Holocene. J. Phys. Oceanogr., 38, 1979–1991.
Sandal, C., and D. Nof, 2008c: Laboratory experiments on the paleo-jamming of the Bering Strait. Deep-Sea Res. I, 55, 1105–1117, doi:10.1016/j.dsr.2008.04.005.
Shaffer, G., and J. Bendtsen, 1994: Role of the Bering Strait in controlling North Atlantic ocean circulation and climate. Nature, 367, 354–357.
Spall, M. A., 2000: Buoyancy-forced circulations around islands and ridges. J. Mar. Res., 58, 957–982.
Toggweiler, J. R., and B. Samuels, 1995: Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res., 42, 477–500.
Toggweiler, J. R., and B. Samuels, 1993: New radiocarbon constraints on the upwelling of abyssal water to the ocean’s surface. The Global Carbon Cycle, M. Heiann, Ed., NATO ASI Series, Vol. 1, Springer-Verlag, 303–331.
Villa-Martínez, R., and P. Moreno, 2007: Pollen evidence for variations in the southern margin of the westerly winds in SW Patagonia over the last 12 600 years. Quat. Res., 68, 400–409.
Weingartner, T. J., D. J. Cavalieri, K. Aagaard, and Y. Sasaki, 1998: Circulation, dense water formation and outflow on the northeast Chukchi Sea shelf. J. Geophys. Res., 103, 7647–7662.
Weingartner, T. J., K. Aagaard, R. Woodgate, S. Danielson, Y. Sasaki, and D. Cavalieri, 2005: Circulation on the North Central Chukchi Sea Shelf. Deep-Sea Res. II, 52, 3150–3174.
Woodgate, R. A., K. Aagaard, and T. J. Weingartner, 2006: Interannual changes in the Bering Strait fluxes of volume, heat and freshwater between 1991 and 2004. Geophys. Res. Lett., 33, L15609, doi:10.1029/2006GL026931.
Woodgate, R. A., T. J. Weingartner, and R. Lindsay, 2010: The 2007 Bering Strait oceanic heat flux and anomalous Arctic sea-ice retreat. Geophys. Res. Lett., 37, L01602, doi:10.1029/2009GL041621.