1. Introduction
Global sea level rose by 1.5–2 cm decade−1 over the past century (Wadhams and Munk 2004). About 0.5 ± 0.2 cm decade−1 of this trend was from steric effects, that is, changes in density that affect the volume and thus the height. Most of the steric change was from warming, although freshening played a role at high latitudes (Antonov et al. 2002).
Levitus et al. (2005) examined the contributions from both temperature and salinity to linear steric height trends using interpolated, gridded hydrographic data averaged into 5-yr running means over 1955–98. They found steric height increases in the North Pacific subpolar gyre and in the Nordic Seas, mostly from freshening. The North Atlantic subpolar gyre also freshened, but a compensating cooling meant that the net steric height change was small. Curry and Mauritzen (2005), using a different interpolation technique, found a large increase in North Atlantic Ocean freshwater content during the Great Salinity Anomaly (GSA) of the 1970s. Freshening continued at a slower pace until the 1990s, when the trend reversed and freshwater content began to diminish.
In the Arctic Ocean, Swift et al. (2005) examined annual hydrographic fields collected by Russian scientists over the latter half of the twentieth century and found a salinification over the upper few hundred meters of the Arctic Ocean during the mid-1970s. Also, tide gauge data were analyzed by Proshutinsky et al. (2004) to determine that sea level rose along the Russian arctic coast from the 1950s to the 1980s. They then used a numerical sea ice–ocean model to determine that 35% of this rise, or 6.4 mm yr−1, was from steric effects. The model also indicated that sea level was falling in the central Arctic Ocean.
In this study, we use hydrographic data over the period 1950–2000 to reexamine steric sea level change in the Northern Seas, which we define here as the North Atlantic Ocean north of 50°N, the North Pacific Ocean north of 40°N, and the Arctic Ocean. We consider both the thermosteric and halosteric components of steric change. The combination of several data sources allows us to consider the large-scale exchange of water properties between ocean basins, as well as possible effects on global ocean circulation. In section 2, we present a discussion of the data used in this study, followed in section 3 by a review of the mean field of steric height and its separate contributions from temperature and salinity. We present basin-averaged time series in section 4, including a detailed look at freshwater and heat transfers between basins. In section 5 we discuss how changes over the late 1960s to the early 1990s affected the large-scale steric height field and speculate about potential effects on the large-scale ocean circulation. We close with a discussion of our results.
2. Datasets
Our main dataset is the World Ocean Database 2001 (WOD01; Conkright et al. 2002), consisting in part of 90 905 paired temperature and salinity profiles during the years 1950–2000 over latitudes 50°–80°N in the Atlantic Ocean and 40°–60°N in the Pacific Ocean. We consider here only those profiles that extend to at least 1000-m depth. Baffin Bay is not included in our analysis, owing to a particularly severe summer bias in the small area enclosed by the 1000-m isobath (99.6% of data there are from July–October). The data density (Fig. 1) is best in the Nordic Seas and close to the North Pacific Ocean coastlines. Within the Nordic Seas, the North Atlantic subpolar gyre, and the North Pacific Ocean (see Fig. 1), profile pairs were discarded if their steric height (see section 3) exceeded three standard deviations from the mean in each 200-km bin. This eliminates less than 1% of the profile pairs.
In the central Arctic Ocean, WOD01 is too sparse for meaningful time series analysis, although it can be used to study trends over the Russian arctic shelves (Steele and Ermold 2004). Historical Russian temperature and salinity data are publicly available only as decadal averages from the 1950s through the 1980s (Timokhov and Tanis 1997, 1998; hereafter referred to by its common name, EWG for Environmental Working Group). However, gridded fields of annual average steric height from these same hydrographic data are available (Timokhov and Tanis 1997) on a roughly 182-km grid. These fields have been relatively underutilized by the scientific community to date. Unfortunately, steric heights were referenced to a relatively shallow 200-m depth. Using the decadal mean hydrography, we find that layer contains 71% of the Arctic Ocean mean 0–1000-m steric height. To better compare with the 0–1000-m steric height computed using WOD01 data, we have extended the Arctic Ocean mean steric height field discussed in section 3 to 1000 m by using the gridded Polar Science Center Hydrographic Climatology (PHC; Steele et al. 2001). Since PHC is a time-mean climatology, temporal variability in Arctic Ocean steric height in this study comes only from the upper 200 m.
3. Mean steric height in the Northern Seas
The Northern Seas play a vital role in two large-scale ocean circulation pathways: the Atlantic Ocean’s Meridional Overturning Circulation (MOC) and the Arctic Flow-Through (AFT). The first is well documented in the literature. The MOC represents the Atlantic branch of the “global ocean conveyor belt” (Broecker 1991), which at the surface brings relatively warm, salty water northward in the North Atlantic current. Along the way, its density increases as heat is lost to the atmosphere, until much of it sinks in the Nordic Seas and returns southward. The second circulation pathway is less well recognized. The AFT is the flow across the Arctic Ocean, usually assumed to be driven by high sea level in the North Pacific Ocean relative to the North Atlantic Ocean. While the volume flux is small [∼0.8 Sv (1 Sv ≡ 106 m3 s−1) mean flow through Bering Strait], the freshwater and chemical nutrient signals are significant (De Boer and Nof 2004; Jones et al. 1998; Wijffels et al. 1992; Woodgate et al. 2005). This pathway represents a shortcut, or “backdoor,” for the global ocean circulation between the North Pacific and North Atlantic Oceans. When this door closes (e.g., during an ice age), model simulations suggest that the MOC may be strongly affected (De Boer and Nof 2004; Goosse et al. 1997).
Figure 2 shows that the Nordic Seas (60°–80°N) contain a DH minimum, that is, a gravitational “potential well” drawing surface water from both the south (the North Atlantic Ocean), and from the north (the North Pacific Ocean via the Arctic Ocean). Of course the large-scale ocean current and sea surface height fields in the ocean arise from a complex combination of both thermohaline and wind forcings. The meridional pressure gradients represented in Fig. 2 are associated with meridional ocean transports (e.g., Griesel and Maqueda 2006) via geostrophy and basin geometry such that they are expressed as relatively narrow currents, for example, the North Atlantic Current from the south, and the East Greenland Current from the north.
Dynamic height is influenced by both temperature and salinity in much of our domain. Exceptions are the subtropical North Atlantic Ocean, where temperature dominates, and the Arctic Ocean and far northern Pacific and Atlantic Oceans, where salinity dominates. Figure 2 also shows that the North Pacific Ocean contains more freshwater than the Arctic Ocean. Surface salinity in the Arctic Ocean is quite fresh, but this freshwater layer extends only to ∼300 m, below which resides much saltier water of Atlantic Ocean origins. Another interesting feature in Fig. 2 is the steep change in all quantities across Bering Strait. Higher DH in the North Pacific Ocean is believed to drive year-round northward fluxes of volume, heat, and freshwater through Bering Strait, modulated by local winds (most recently, Woodgate et al. 2005).
Figure 3 shows the horizontal distribution of these quantities. A variety of gridded and ungridded data are used in this figure, described in detail in section 2. Only data from areas with depths of at least 1000 m are plotted. Figure 3 shows that the North Atlantic DH maximum in Fig. 2 actually contains a northwest–southeast gradient, along which the North Atlantic Current flows. A weaker east–west DH gradient is also evident in the Nordic Seas, driving flow into the Arctic Ocean. Note that strong slope currents along Norway and Greenland do not appear in this deepwater view.
The Eurasian Basin of the Arctic Ocean shares many properties with the inflowing waters from the Nordic Seas, that is, relatively salty and warm, with a low DH. On the other hand, the Canadian Basin of the Arctic Ocean contains a small DH maximum within the relatively fresh Beaufort Gyre. In the North Pacific Ocean, properties are more zonal than in the North Atlantic Ocean. Nonetheless, DH contours suggest a northwestward geostrophic flow south of Alaska, which contributes to northward flow through the Aleutian Islands and eventually through Bering Strait.
4. Regional time series
The following analysis focuses on change, and thus we compute steric anomalies relative to a long-term mean. For the WOD01 data, the anomaly is formed by subtracting at each profile position the interpolated mean value (Tref, Sref, P) from the gridded World Ocean Atlas 2001 (WOA01; Boyer et al. 2002; Stephens et al. 2002). For the EWG data in the Arctic Ocean, we formed a record-mean DH field from the individual years and used this, in combination with PHC between 200 and 1000 m, as the mean.
a. The Arctic Ocean
Figure 4 shows dynamic height changes over 0–200-m depth in the central Arctic Ocean during 1950–90, averaged over the colored grid boxes in Fig. 3a (i.e., where the ocean is at least 1000 m deep). The trends are nearly identical if the shallower gray boxes in Fig. 3a are also included. Dynamic height generally decreased during 1950–90, almost certainly because of increasing salinity since the thermal contribution to density is small at cold temperatures (Figs. 2 and 3). (A more definitive analysis could be performed using annual temperature and salinity fields, but as discussed in section 2, these are not yet publicly available.) A linear least squares regression line through the entire time series has a slope of −0.19 ± 0.07 cm decade−1 (mean ±1.96 × standard deviation of the slope, i.e., 95% confidence), implying a total DH decrease of 0.76 ± 0.28 cm over 40 yr. This central arctic DH decrease confirms the modeling results of Proshutinsky et al. (2004) and is opposite to the freshening and DH rise observed in recent decades on the Siberian shelves (Proshutinsky et al. 2004; Steele and Ermold 2004).
Arctic Ocean DH also demonstrates significant decadal variability. In fact, there seems to be an approximately 15-yr time scale during which DH slowly rises, then discharges over ∼5 yr. (Unfortunately, the time series is too short for this to be rigorously tested.) Since Arctic Ocean DH is controlled by freshwater (DH ≅ DHS; see Fig. 2), the 15-yr rises represent a freshwater “charging” of the arctic “capacitor,” similar to the concept of a Beaufort Gyre flywheel described by Proshutinsky et al. (2002). One charging period during 1958–72 includes the start of the GSA in the late 1960s (Dickson et al. 1988), when sea ice is thought to have streamed out of the Arctic Ocean and melted in the North Atlantic Ocean. This could be a response to changing winds associated with the North Atlantic Oscillation or its close relative, the Arctic Oscillation (AO; Thompson and Wallace 1998), although this connection is not robust for all time periods (Koberle and Gerdes 2003; Tremblay 2001). A weak DH drop at this time (marked “GSA-ice?” in Fig. 4) is consistent with the salinification expected from enhanced Arctic Ocean sea ice growth during the winter following enhanced GSA sea ice export. Model simulations by Rothrock and Zhang (2005) produce a single year of enhanced ice export in 1968, and a subsequent single year of enhanced Arctic Ocean ice growth in 1969. However, total sea ice volume does not completely recover to pre-GSA levels, owing mostly to slightly warmer surface air temperatures after the GSA.
A much larger DH drop of about 1 cm ≅ 40 cm of freshwater is also evident during the mid-1970s, and is marked “GSA-ocn” in Fig. 4. This is the time that Swift et al. (2005) found marked Arctic Ocean salinification, attributed by them to increased ice growth after the 1968 GSA. However, the model simulation cited previously (Rothrock and Zhang 2005) suggests that enhanced ice growth from the GSA ice export event happened within one year and did not last through the 1970s. Thus another explanation must be found. The Arctic Ocean freshwater layer (0–200-m depth) has a residence time of ∼10 yr (Serreze et al. 2006), with the bulk of it residing some distance away from Fram Strait in the Canadian Basin (Proshutinsky et al. 2002; Serreze et al. 2006). Thus, it seems probable that continued salinification through the 1970s was caused by a draining of liquid (ocean) freshwater from the Arctic Ocean, in response to winds associated with an increasing AO index. It is not clear whether the separation in Fig. 4 of GSA-ice and GSA-ocean by a slight DH increase during the early 1970s is real or an artifact of noisy data. The AO index does not monotonically increase over the late 1960s through the 1970s, which might delay liquid freshwater export relative to sea ice export, as winds shift annually. On the other hand, if the unusually large +1.3 cm annual DH anomaly in 1973 is eliminated, Arctic Ocean running-mean DH is nearly monotonic downward from the mid-1960s through the 1970s.
Thus we propose that the 1970s DH drop in Fig. 4 is a response to liquid (ocean) freshwater export from the Arctic Ocean (probably balanced to some extent below 300 m by increased salty Atlantic water inflow). Such a delayed response in ocean freshwater export has been noted in model simulations (Haak et al. 2003; Karcher et al. 2005). Our main idea here is that the AO-associated wind shift of the late 1960s forced an immediate Arctic Ocean sea ice export as well as a delayed and sustained liquid (ocean) freshwater export.
b. The North Atlantic Ocean
Figures 5 and 6 show DH, DHT, and DHS trends in the North Atlantic Ocean during 1950–2000, divided into the subpolar gyre (Fig. 5: between 50°N and the Greenland–Iceland–Scotland Ridge) and the Nordic Seas (Fig. 6: north of this ridge to 80°N). Anomalies were computed by subtracting interpolated monthly mean WOA01 values from WOD01 profile data, which were then averaged into 200-km bins to avoid spatial bias. The spatial coverage within each region of annual mean 200-km bins (gray dots in the figures) can be poor, which is why for most of our analysis, we focus on the 5-yr running-mean binned data (green dots in the figures). Collecting data over 5-yr bins generally provides 90% or greater areal coverage. Thus no spatial interpolation or smoothing was performed.
The linear DH trend over 1950–2000 in the North Atlantic subpolar gyre is not significantly different from zero, although the Nordic Seas DH trend is a positive 0.56 ± 0.23 cm decade−1. Further, there are obvious variations on decadal and longer time scales, especially in the temperature (DHT) and salinity (DHS) components. During the years 1965–90, both the Arctic Oscillation index and DHS rise in the subpolar and Nordic seas. This North Atlantic freshening was also found by Curry and Mauritzen (2005), although theirs is sustained (albeit at a slower rate) through the 1980s, while ours stops rising by 1980. The largest freshening occurs during the 1970s, starting in the subpolar gyre in 1968 and the Nordic Seas a few years later. This delay was proposed by Curry and Mauritzen (2005) to result from the GSA’s sea ice (and some resulting meltwater) flowing quickly down the East Greenland Current into the subpolar gyre, and only slowly diffusing eastward across the Nordic Seas.
The subpolar gyre freshened but also cooled during the GSA, possibly in response to cold Arctic water and sea ice (Fig. 5; see also Levitus et al. 2005). The cooling dominates the density change, creating a small net DH decrease of 1.40 ± 0.63 cm over 1965–90. Cooling is not so important to steric height in the Nordic Seas (Fig. 6), however, since density is controlled much more by salinity than by temperature in these cold waters. Thus DH rose in the Nordic Seas by 3.78 ± 0.78 cm over 1965–90 in response to GSA freshening.
Satellite altimeter observations since the early 1990s show rising sea surface heights in much of the North Atlantic subpolar gyre (Cabanes et al. 2001; Hakkinen and Rhines 2004; Levinson 2005), a result also seen in our DH data (Fig. 5). The altimeter data also indicate a weak 1990s sea surface height increase in the Nordic Seas (F. Mertz, CLS/DOS, 2006, personal communication), which is not seen in our DH time series (Fig. 6).
c. The northeast Pacific Ocean
Figure 7 shows dynamic height trends for the Northeast Pacific Ocean, defined as the region bounded by the North American coast to the east, 160°E to the west, Bering Strait to the north, and a line roughly between the southern tip of Kamchatka, Russia, and northern California (see following section). This is the region containing the inflows to the Arctic Ocean and shows spatially uniform trends in dynamic height that are opposite to nearby regions in the Pacific Ocean (see section 5a).
The northeast Pacific Ocean shows a positive DH trend over the entire record of 1.01 ± 0.21 cm decade−1, with contributions from both warming and freshening (see also Levitus et al. 2005). A major climate index for the North Pacific Ocean is the Pacific decadal oscillation (PDO), which is the first principle component in an empirical orthogonal function formed using sea surface temperature fields (Mantua et al. 1997). A major transition occurred in the mid-1970s from a negative to positive index, which brought warmer water to the North Pacific Ocean. Before 1960 the PDO index was weak and variable, while since 1990 the dominant climate forcing is best described by the second principle component (Bond et al. 2003), known as the Victoria Pattern (VP). This pattern is anticorrelated with warmer northeast Pacific Ocean temperatures, that is, a negative index implies a warming ocean. The net result of these climate forcings should be a DHT rise from the mid-1960s through the mid-1990s, which is what Fig. 7 shows. Freshening also occurs during this period, although not as strongly. The net result is a DH rise of 4.93 ± 0.80 cm over 1970–95. Note that the Victoria Pattern reversal in the late 1990s should result in a cooling, which would imply a DH fall if salinity changes are small. In fact, altimetric sea surface height does show a decrease of ∼1 cm in this region over 1993–2004 (Levinson 2005).
d. Domain-average changes
Table 1 shows a summary of regional steric height changes over two time periods, as described in sections 4a–c. Most height changes are consistent in sign and even magnitude for both time periods. The major exception is the North Atlantic subpolar gyre, which has a significant DH decrease over 1965–90 but not over 1950–2000.
Table 1 also shows the area-weighted total steric height changes, with and without including the northeast Pacific Ocean. These total height changes indicate that DH increased overall when including the northeast Pacific Ocean, mostly as a result of a DHS (i.e., freshwater) increase. This could be from increased net precipitation less evaporation at high latitudes in response to an accelerated hydrologic cycle (Curry et al. 2003), from increased net terrestrial and/or sea ice melting (Antonov et al. 2002; Wadhams and Munk 2004), or from net ocean freshwater transport into these high latitudes. While local DHT changes may be large, the overall effect when including the Northeast Pacific Ocean is small, consistent with its small influence on density at cold temperature.
On the other hand, total steric height change is insignificant when considering only the Arctic and North Atlantic Oceans. In this domain, net freshening (DHS increase) is essentially balanced by net cooling (DHT decrease). This is consistent with sea ice import into the North Atlantic Ocean, which cools and freshens as a result. Thus the Arctic/Atlantic domain experiences net cooling and net freshening of the ocean (but not necessarily of the total ice plus ocean system).
5. Trends over the late 1960s through the early 1990s
a. Gridded fields
Figure 8 shows how trends of dynamic height and its temperature/salinity components DHT and DHS vary spatially during the mid-1960s through early 1990s. This is the period when climate indices in all regions strongly varied (Figs. 5, 6 and 7). Although some noise is evident, spatial trends are remarkably coherent over the broad regions discussed in the previous sections: the central Arctic Ocean, the Nordic Seas, the North Atlantic subpolar gyre, and the northeast Pacific Ocean. The Arctic Ocean shows particularly smooth trends because the EWG data are only available as gridded, smoothed fields (section 3). In other regions, hydrographic data from WOD01 have been averaged into 200 km × 200 km bins (but not smoothed). Dynamic height trends from the Nordic Seas extend with the same sign into the Eurasian Basin of the Arctic Ocean. There is also some indication of northeast Pacific Ocean trends extending into the Chukchi Sea.
The drop during 1965–90 in Arctic Ocean DH ≅ DHS occurs mainly in the eastern Eurasian Basin and southern Makarov Basin, extending across the North Pole toward Greenland. This is the region that showed significant freshwater changes during the AO index rise in the early 1990s (e.g., Morison et al. 2000). Freshwater decreases as the Atlantic/Pacific front sweeps around cyclonically, bringing a more salty Atlantic influence to the central Arctic Ocean. Karcher et al. (2005) show in a model study how this might force freshwater out of the Arctic Ocean and into the North Atlantic Ocean.
In fact, Fig. 8 shows that the North Atlantic Ocean freshens during this time. Note that historical data coverage within the East Greenland Current, the Canadian Archipelago, and Baffin Bay is very poor. This means that we cannot explicitly follow the freshwater outflow, only its effects on the deep basins immediately adjacent. There is also a widespread cooling in the subpolar gyre, possibly caused in part by melting sea ice and mixing of cold melt water.
The North Pacific Ocean shows two distinct regions in Fig. 8: a fairly uniformly rising DH area in the northeast and a noisier falling DH area to the south and west. As shown in Fig. 7, the northeast Pacific Ocean DH increase arises from both freshening and warming over the 50-yr record (Levitus et al. 2005) but is more strongly forced by warming over the 1970s–90s. The similar spatial pattern of thermosteric and halosteric change over the North Pacific Ocean seen in Fig. 8 contradicts the finding of Overland et al. (1999), who found nearly orthogonal temperature and salinity differences in response to the mid-1970s PDO shift. However, our analysis does not extend quite as far south as theirs did.
b. Sector zonal means
Figure 9 shows sector zonal mean changes in DH, DHS, and DHT between the late 1960s and the early 1990s, using a format similar to Fig. 2. The dynamic height difference between the North Pole and Nordic Seas decreased by 32% (8 cm relative to the 25-cm mean), owing mostly to southward freshwater transfer. Similarly, the DH gradient between the Nordic Seas and the North Atlantic subpolar gyre declined by 18% (8 cm relative to the 45-cm mean), owing in part to cooling of the subpolar gyre. The net effect of these changes is a weakening, or “filling in” of the Nordic Seas DH minimum, indicated in Fig. 9 by shading. This leads to a decrease in the pressure gradients associated with flows from the south (the MOC) and from the north (Fram Strait and the AFT), possibly implying a slowing of these ocean transports during this period. Figures 4 –7 show that longer time series also imply a rise in Nordic Sea DH and a fall in Arctic Ocean DH, with little change in the North Atlantic subpolar gyre.
Is this evidence from the Northern Seas that the MOC is slowing, as recently proposed by Bryden et al. (2005)? Perhaps. However, no clear trend in MOC strength is evident in most coupled climate models until the first few decades of the twenty-first century (e.g., Schmittner et al. 2005). In fact, in situ data (Curry and McCartney 2001), altimetry (Hakkinen and Rhines 2004), and models (Böning et al. 2006) all indicate that the MOC increased in strength during roughly 1965–90, but decreased during the decade before and the decade after this period, similar to the AO index time series. Yet Fig. 9 suggests that steric height gradients, which help to drive the MOC, decreased over 1965–90 in the North Atlantic Ocean. The resolution to this contradiction might lie in a negative feedback process such as described in the modeling study by Jungclaus et al. (2005). The basic idea is that an increase in MOC strength eventually flushes sea ice and ocean freshwater out of the Arctic Ocean, which flow southward into the deepwater formation sites and slow the MOC. We find that the accumulation of arctic water builds over a time scale of about 25 yr, implying a total feedback period of 50 yr. Thus we propose that even as the MOC gained strength during 1965–90, cold and fresh arctic outflows were accumulating in the North Atlantic, eventually decreasing the steric height gradients and decelerating the MOC.
Of course, steric height is only part of the total sea surface height (SSH). Mass inputs (not included in this study) can also change SSH, although some studies have shown fairly good correspondence between large-scale patterns of SSH and DH trends (Cabanes et al. 2001; Proshutinsky et al. 2004), while others show a not unexpected decrease in the correlation of SSH and DH variability with increasing latitude (Fig. 3d; Guinehut et al. 2006)
Figure 9 shows rising Pacific Ocean DH trends, also sustained over the total 1950–2000 period (although recently reversing). This should accelerate, not decelerate, the pressure gradient forcing for the AFT. This acceleration might have occurred not via Fram Strait, but rather through the Canadian Arctic Archipelago and Baffin Bay to the Labrador Sea, which shows some DH increase in Fig. 8. Some support for this idea is provided by Zweng and Muenchow (2006), who analyzed Canadian data not yet incorporated into WOD01. They found a freshening in the southward-flowing Baffin Island current during the latter half of the twentieth century, although the severe summer bias in their data introduced a significant level of uncertainty. The data thus suggest that during 1965–95, the Fram Strait branch of the AFT may have slowed even as the Canadian Arctic branch accelerated.
c. Volume fluxes between the Arctic and North Atlantic Oceans
Can the changes observed in the North Atlantic Ocean during the Great Salinity Anomaly of the 1970s be completely explained by sea ice and liquid (ocean) fluxes from the Arctic Ocean? Freshening in the subpolar gyre amounted to ∼200 cm (Fig. 5), or ∼8400 km3 when multiplied by an area of 4.2 × 106 km2. For the Nordic Seas, Fig. 6 indicates a freshening of ∼160 cm, or ∼2400 km3 when multiplied by an area of 1.5 × 106 km3. Thus the North Atlantic Ocean freshened by ∼11 000 km3 over 10 yr. This is far more than can be explained by the mid-1970s Arctic Ocean freshwater loss of about 2000 km3 (Fig. 4; Arctic Ocean area = 4.7 × 106 km2). Further, sea ice export is not likely to explain the residual of ∼9000 km3, since this would require a positive annual export anomaly of about 900 km3 (i.e., 30%–50% of the mean) over 10 yr, a value far above that seen in model simulations (Karcher et al. 2005; Rothrock and Zhang 2005). Thus other factors must play a major role. A similar calculation for heat transports could also be performed, although it would be more complex, owing to the more complex relationship between thermosteric height and heat content [Eq. (4)]. A full accounting of freshwater and heat budgets is beyond the scope of the present study.
6. Conclusions
Our main conclusions are as follows:
Net steric height increase over our domain: Steric height over the Northern Seas increased during 1950–2000, mostly from freshening (Table 1). While local temperature changes were also important, the net effect over our domain was small.
Filling of the Nordic Seas DH minimum: Steric height plots centered on the Arctic Ocean (Figs. 2 and 3) clearly show two major ocean pathways: the AFT and the upper limb of the MOC. Our analysis indicates that the steric height minimum that in part drives both of these filled in over the North Atlantic and Arctic Oceans during 1965–90 (Fig. 9), possibly indicating a slowing of these circulations. A DH rise in the northeast Pacific Ocean during this time seemingly contradicts this result, possibly resolved by an accelerated flow through the Canadian sector of the Arctic and North Atlantic Oceans. More recent data indicate that some (but not all) areas show reversed DH trends relative to the previous 25 yr, highlighting the strong decadal-scale variability in these time series.
GSA-ice versus GSA-ocean: Anomalous Arctic Ocean sea ice export in the late 1960s (GSA-ice; Fig. 4) contributed to a freshening of the Nordic Seas (DH rise; Fig. 6), a freshening and cooling in the North Atlantic subpolar gyre (DH drop; Fig. 5), and a small salinification of the Arctic Ocean as sea ice growth rebounded in the following few years (DH drop). We also find a stronger Arctic Ocean salinification during the mid-1970s as liquid freshwater in the upper few hundred meters was exported southward (GSA-ocean; Fig. 4), which contributed to a sustained freshening of the North Atlantic Ocean. However, the freshening in our analysis levels off by the 1980s, about 10 yr earlier than in the study by Curry and Mauritzen (2005). The pattern of Arctic Ocean DH change over the period 1965–90 (Fig. 8) indicates a cyclonic shift in the front separating Pacific influence (high DH) from Atlantic influence (lower DH), similar to that observed during the 1990s. Our analysis of Arctic Ocean dynamic height fields supports the capacitor model of freshwater storage and release proposed by Proshutinsky et al. (2002), where we find a storage period of ∼15 yr and a release period of ∼5 yr.
Insufficient Arctic Ocean sources for observed GSA freshening: A freshwater volume analysis indicates that the 1970s North Atlantic Ocean freshening was much larger than can be explained by the sum of sea ice and liquid (ocean) freshwater export from the Arctic Ocean (section 5c). Thus other (as yet undetermined) sources must have played a major role.
Acknowledgments
We thank N. Bond and F. Mertz for data and graphics, H. Stern for statistics guidance, and R. Woodgate and two anonymous reviewers for comments on the manuscript. This work was sponsored by the Office of Polar Programs at the National Science Foundation.
REFERENCES
Antonov, J. I., S. Levitus, and T. P. Boyer, 2002: Steric sea level variations during 1957–1994: Importance of salinity. J. Geophys. Res., 107 .8013, doi:10.1029/2001JC000964.
Bond, N. A., J. E. Overland, M. Spillane, and P. Stabeno, 2003: Recent shifts in the state of the North Pacific. Geophys. Res. Lett., 30 .2183, doi:10.1029/2003GL018597.
Böning, C. W., M. Scheinert, J. Dengg, A. Biastoch, and A. Funk, 2006: Decadal variability of subpolar gyre transport and its reverberation in the North Atlantic overturning. Geophys. Res. Lett., 33 .L21S01, doi:10.1029/2006GL026906.
Boyer, T. P., C. Stephens, J. I. Antonov, M. E. Conkright, R. A. Locarnini, T. D. O’Brien, and H. E. Garcia, 2002: Salinity. Vol. 2, World Ocean Atlas 2001, NOAA Atlas NESDIS 50, 165 pp.
Broecker, W. S., 1991: The great ocean conveyor. Oceanography, 1 , 79–89.
Bryden, H. L., H. R. Longworth, and S. A. Cunningham, 2005: Slowing of the Atlantic meridional overturning circulation at 25°N. Nature, 438 , 655–657.
Cabanes, C., A. Cazenave, and C. Le Provost, 2001: Sea level rise during past 40 years determined from satellite and in situ observations. Science, 294 , 840–842.
Conkright, M. E., and Coauthors, 2002: Introduction. Vol. 1, World Ocean Database 2001, NOAA Atlas NESDIS 42, 160 pp.
Curry, R. G., and M. S. McCartney, 2001: Ocean gyre circulation changes associated with the North Atlantic oscillation. J. Phys. Oceanogr., 31 , 3374–3400.
Curry, R. G., and C. Mauritzen, 2005: Dilution of the northern North Atlantic Ocean in recent decades. Science, 308 , 1772–1774.
Curry, R. G., B. Dickson, and I. Yashayaev, 2003: A change in the freshwater balance of the Atlantic Ocean over the past four decades. Nature, 426 , 826–829.
De Boer, A. M., and D. Nof, 2004: The Bering Strait’s grip on the Northern Hemisphere climate. Deep-Sea Res., 51 , 1347–1366.
Dickson, R. R., J. Meincke, S. A. Malmberg, and A. J. Lee, 1988: The “Great Salinity Anomaly” in the northern North Atlantic 1968–1982. Progress in Oceanography, Vol. 20, Pergamon, 103–151.
Goosse, H., J. M. Campin, T. Fichefet, and E. Deleersnijder, 1997: Sensitivity of a global ice–ocean model to the Bering Strait throughflow. Climate Dyn., 13 , 349–358.
Griesel, A., and M. A. M. Maqueda, 2006: The relation of meridional pressure gradients to North Atlantic deep water volume transport in an ocean general circulation model. Climate Dyn., 26 , 781–799.
Guinehut, S., P-Y. Le Traon, and G. Larnicol, 2006: What can we learn from global altimetry/hydrography comparisons? Geophys. Res. Lett., 33 .L10604, doi:10.1029/2005GL025551.
Haak, H., J. Jungclaus, U. Mikolajewicz, and M. Latif, 2003: Formation and propagation of great salinity anomalies. Geophys. Res. Lett., 30 .1473, doi:10.1029/2003GL017065.
Hakkinen, S., and P. B. Rhines, 2004: Decline of subpolar North Atlantic circulation during the 1990s. Science, 304 , 555–559.
Jones, E. P., L. G. Anderson, and J. H. Swift, 1998: Distribution of Atlantic and Pacific waters in the upper Arctic Ocean: Implications for circulation. Geophys. Res. Lett., 25 , 765–768.
Jungclaus, J. H., H. Haak, M. Latif, and U. Mikolajewicz, 2005: Arctic–North Atlantic interactions and multidecadal variability of the meridional overturning circulation. J. Climate, 18 , 4013–4031.
Karcher, M., R. Gerdes, F. Kauker, C. Koberle, and I. Yashayaev, 2005: Arctic Ocean change heralds North Atlantic freshening. Geophys. Res. Lett., 32 .L21606, doi:10.1029/2005GL023861.
Koberle, C., and R. Gerdes, 2003: Mechanisms determining the variability of Arctic sea ice conditions and export. J. Climate, 16 , 2843–2858.
Landerer, F. W., J. H. Jungclaus, and J. Marotzke, 2007: Regional dynamic and steric sea level change in response to the IPCC-A1B scenario. J. Phys. Oceanogr, ., in press.
Levinson, D. H., 2005: State of the climate in 2004. Bull. Amer. Meteor. Soc., 86 , S1–S86.
Levitus, S., J. I. Antonov, T. P. Boyer, H. E. Garcia, and R. A. Locarnini, 2005: Linear trends of zonally averaged thermosteric, halosteric, and total steric sea level for individual ocean basins and the World Ocean, (1955–1959)–(1994–1998). Geophys. Res. Lett., 32 .L16601, doi:10.1029/2005GL023761.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 1069–1079.
Morison, J., K. Aagaard, and M. Steele, 2000: Recent environmental changes in the Arctic: A review. Arctic, 53 , 359–371.
Overland, J. E., S. Salo, and J. R. Adams, 1999: Salinity signature of the Pacific decadal oscillation. Geophys. Res. Lett., 26 , 1337–1340.
Proshutinsky, A., R. H. Bourke, and F. A. McLaughlin, 2002: The role of the Beaufort Gyre in Arctic climate variability: Seasonal to decadal climate scales. Geophys. Res. Lett., 29 .2100, doi:10.1029/2002GL015847.
Proshutinsky, A., I. M. Ashik, E. N. Dvorkin, S. Hakkinen, R. A. Krishfield, and W. R. Peltier, 2004: Secular sea level change in the Russian sector of the Arctic Ocean. J. Geophys. Res., 109 .C03042, doi:10.1029/2003JC002007.
Rothrock, D. A., and J. Zhang, 2005: Arctic Ocean sea ice volume: What explains its recent depletion? J. Geophys. Res., 110 .C01002, doi:10.1029/2004JC002282.
Schmittner, A., M. Latif, and B. Schneider, 2005: Model projections of the North Atlantic thermohaline circulation for the 21st century assessed by observations. Geophys. Res. Lett., 32 .L23710, doi:10.1029/2005GL024368.
Serreze, M. C., and Coauthors, 2006: The large-scale freshwater cycle of the Arctic. J. Geophys. Res., 111 .C11010, doi:10.1029/2005JC003424.
Steele, M., and W. Ermold, 2004: Salinity trends on the Siberian shelves. Geophys. Res. Lett., 31 .L24308, doi:10.1029/2004GL021302.
Steele, M., R. Morley, and W. Ermold, 2001: PHC: A global ocean hydrography with a high-quality Arctic Ocean. J. Climate, 14 , 2079–2087.
Stephens, C., J. I. Antonov, T. P. Boyer, M. E. Conkright, R. A. Locarnini, T. D. O’Brien, and H. E. Garcia, 2002: Temperature. Vol. 1, World Ocean Atlas 2001, NOAA Atlas NESDIS 49, 167 pp.
Swift, J. H., K. Aagaard, L. Timokhov, and E. G. Nikiforov, 2005: Long-term variability of Arctic Ocean waters: Evidence from a reanalysis of the EWG dataset. J. Geophys. Res., 110 .C03012, doi:10.1029/2004JC002312.
Thompson, D. W. J., and J. M. Wallace, 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25 , 1297–1300.
Timokhov, L., and F. Tanis, Eds. 1997: Environmental Working Group Joint U.S.–Russian Atlas of the Arctic Ocean—Winter Period. Environmental Research Institute of Michigan in association with the National Snow and Ice Data Center, CD-ROM.
Timokhov, L., and F. Tanis, 1998: Environmental Working Group Joint U.S.–Russian Atlas of the Arctic Ocean—Summer Period. Environmental Research Institute of Michigan in association with the National Snow and Ice Data Center, CD-ROM.
Tremblay, L. B., 2001: Can we consider the Arctic Oscillation independently from the Barents Oscillation? Geophys. Res. Lett., 28 , 4227–4230.
Wadhams, P., and W. Munk, 2004: Ocean freshening, sea level rising, sea ice melting. Geophys. Res. Lett., 31 .L11311, doi:10.1029/2004GL020039.
Wijffels, S. E., R. W. Schmitt, H. L. Bryden, and A. Stigebrandt, 1992: Transport of freshwater by the oceans. J. Phys. Oceanogr., 22 , 155–162.
Woodgate, R. A., K. Aagaard, and T. J. Weingartner, 2005: Monthly temperature, salinity, and transport variability of the Bering Strait through flow. Geophys. Res. Lett., 32 .L04601, doi:10.1029/2004GL021880.
Zweng, M. M., and A. Münchow, 2006: Warming and freshening of Baffin Bay, 1916–2003. J. Geophys. Res., 111 .C07016, doi:10.1029/2005JC003093.
Number of observations over our domain. For the North Atlantic and North Pacific Oceans, we show number of coincident temperature and salinity profiles from WOD01 (Conkright et al. 2002) in 200-km bins over 1950–2000. For the Arctic Ocean, the EWG annual 0–200-m depth DH fields do not provide data counts, so instead we show the 40-yr total number of 200-m depth observations for the four decadal-mean salinity fields also provided by EWG. This assumes that where a deep salinity observation was taken, shallower values were also obtained, and a coincident temperature profile as well. Geographic place names mentioned in the text are Bering Strait (BS), Chukchi Sea (CS), Canadian Arctic Archipelago (CAA), Baffin Bay (BB), Canada Basin (CB), Makarov Basin (MB), Eurasian Basin (EB), Fram Strait (FS), Greenland–Iceland–Scotland Ridge (GISR), Kamchatka (KC), Aleutian Islands (AI), and northern California (NC).
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Sector zonal means of DH (black), thermosteric dynamic height (DHT, red), and halosteric dynamic height (DHS, green) over 0–1000-m depth, using the PHC database (Steele et al. 2001). The Atlantic sector is defined as 90°W to 100°E north of 60°N, and 90°W to 0° between 20° and 60°N, while the Pacific sector is defined as 100°E to 90°W from 20° to 90°N, excluding the western Caribbean Sea (inset). Vertical lines show ±1 standard deviation from the zonal sector means. Bering Strait is ∼50 m deep. Arrows denote surface transport in the MOC and AFT.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Maps of integrated quantities over 0–1000-m depth. (a) DH, (b) halosteric dynamic height DHS, (c) thermosteric dynamic height DHT, and (d) mean buoyancy frequency 〈N〉. Gray squares in (a) are shallower than 1000-m depth. Gridded DH fields were used in the Arctic Ocean, while original hydrographic profile data were used elsewhere (see section 2). The PHC database was used in (d).
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Annual Arctic Ocean dynamic height anomalies over 0–200 m from the EWG atlas, relative to the 1950–89 mean. Areas shallower than 1000 m (gray boxes in Fig. 3) are not included. The years 1951–53 are absent in the atlas. The DH scale is exactly half that of subsequent time series (Figs. 5 –7). Shown are the annual anomalies (gray dots), the running 5-yr mean anomalies (green dots), 95% confidence range of these 5-yr anomalies (i.e., ±1.96 standard errors, vertical gray lines), and pentadal averages (red horizontal bars). A least squares trend (blue line) is plotted, with DH slope noted in the lower-left corner. The right-hand scale is freshwater content anomaly, using the conversion factor in Eq. (5). Freshwater charging and discharging periods (see text) are shown, while height decreases in the late 1960s and mid-1970s are denoted GSA-ice? and GSA-ocn.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Same as in Fig. 4, but for annual anomalies over 0–1000-m depth of (a) DH, (b) DHT, and (c) DHS in the North Atlantic Ocean subpolar gyre, using hydrographic profiles from WOD01 averaged into 200-km bins. Also shown is the 5-yr running mean January–March AO index (bottom panels, black curve, full scale ±2 normalized standard deviations). A least squares trend line is plotted for DH, with slope noted in the lower-left corner. Freshwater increase during the late 1960s through the early/mid-1970s is marked GSA.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Same as in Fig. 5, but for the Nordic Seas.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Same as in Fig. 5, but for the northeast Pacific Ocean (see Fig. 8 for definition). The PDO amplitude is shown for 1960–89, and the VP amplitude after that. A least squares trend line is computed for DH, with slope noted in the lower-left corner.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Gridded trends of (a) DH anomaly, (b) halosteric DH anomaly, and (c) thermosteric height anomaly, extending to 40°N in the Pacific Ocean and 50°N in the Atlantic Ocean. Atlantic and Pacific data are from WOD01 on a 200-km grid over 0–1000-m depth. Arctic data are from EWG on a ∼182-km grid over 200-m depth. Only DH is available annually from EWG. Trends are for 1970–95 (Pacific) and 1965–90 (Atlantic/Arctic). Grid squares are not plotted if there were no profiles within ±2 yr of the start and end years. White dots indicate 95% or greater significance. Black lines in (a) separate regions used in Figs. 4 –7 (starting from the bottom, the North Atlantic subpolar gyre, the Nordic Seas, the Arctic Ocean, and the northeast Pacific Ocean)
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Same as in Fig. 2, but for change over the late 1960s–early 1990s. Upper curves show zonal mean dynamic height (with a 5-degree latitude running mean smoother) in 1965 (black dots) and 1990 (black plusses) for the Atlantic/Arctic (left-hand scale), and 1970 (black dots) and 1995 (black plusses) for the Pacific (right-hand scale). Sectors are defined as in Fig. 2. Anomalies were computed from least squares trends over this period (see Figs. 4 –8) and then added to the PHC mean shown in Fig. 2. Lower curves show just the anomalies, including the breakdown of DH into its halosteric (DHS) and thermosteric (DHT) components. Shading indicates weakening of the Nordic Seas DH minimum, as described in the text.
Citation: Journal of Climate 20, 3; 10.1175/JCLI4022.1
Steric height changes (cm) over the Northern Seas for 1950–2000 and for 1965–90 (North Atlantic subpolar gyre, Nordic Seas, and Arctic Ocean), and 1970–95 (northeast Pacific Ocean). Arctic Ocean DHS changes are not available and thus are assumed to be identical to the observed DH changes (salinity dominates the density at cold temperature). Similarly, Arctic Ocean DHT changes are assumed to be zero, with half the uncertainty of the (slightly warmer) Nordic Seas. Also, Arctic Ocean trends from 1950 to 1990 are assumed to continue through 2000. Changes in the total North Pacific Ocean (north of 40°N) are of similar sign to those in the northeast Pacific Ocean shown here, but generally smaller owing to partly compensating trends to the southwest (Fig. 8). Uncertainty in each region is 1.96 × standard deviation of the slope (bold indicates 95% significance). Uncertainty in the combined regions (final two rows) is computed from the sum of individual regions’ slope variances and the variance of each region’s slopes relative to the total mean slope.
