Variability of Thermohaline Circulation and Freshwater Storage in the Arctic Ocean
Principal Investigator
Andrey Proshutinsky
Woods Hole Oceanographic
Woods Hole, MA 02543
Phone: 508-289-2796
FAX: 508-457-2181

Project Description
First Year Project Results
Publications and Talks
Digital Data Access

This research has been supported by a grant from NOAA. NOAA’s project manager:

John A. Calder, Director Arctic Research Office
NOAA Oceanic and Atmospheric Research R/AR Silver Spring, MD 20910
First Year Project Results

Table of Contents
1. Overview
2. Conceptual Model
3. Hypothesis
During the first year of research we have reprocessed all available data collected in the archives of the Arctic and Antarctic Research Institute (AARI), St. Petersburg, Russia. In 1995-1997, the Arctic and Antarctic Research Institute (AARI), working with U.S. experts, have compiled an Arctic oceanographic database and developed climate scale hydrographic fields for the winter and summer periods. The available
Western data (Figure A) and Russian data (Figure B) on temperature (T) and salinity (S) were assembled into a combined Russian--American database for the Arctic Ocean which is now located at AARI. The hydrographic database contains a total of 354,287 stations of which 138,553 are for the winter season and 215,734 for the summer season. For the winter period analyses, March--May 1948--1993, 73,774 stations were used from the joint hydrographic database. Some winter and summer Russian data are classified but available for analysis by Russian scientists. A small portion of the western data remains proprietary. For more details go to

This is great new information but the problem is that the data for this atlas were averaged mechanically without taking into account natural processes and features of the climate variability in the Arctic. The data were averaged for decades, for periods 1950--1959, 1960--1969, 1970--1979 and 1980--1989. Analysis of the temperature and salinity fields obtained by this method shows that there are no significant changes in the Arctic’s circulation and freshwater content from decade to decade. See Figure C (pdf).

For our project, gridded water temperature and salinity fields from AARI archive were averaged for the years of cyclonic (1953-1957, 1964-1971, 1980-1983) and anticyclonic (1950-1952, 1958-1963, 1972-1979, and 1984-1988) circulation regimes (see Proshutinsky and Johnson, 1997). The method of data reconstruction for each year is based on EOF analysis and is decribed in the EWG atlas of the Arctic Ocean (see EWG, 1997). These fields averaged for the CCR and ACCR regimes are available in the grid 200x200 km. Each file for each depth (0, 10, 25, 50, 75, 100, 150, 200, 250, 300, 400, 500, 750, 1000, 1500, 2000, 2500, 3000, and 4000 meters) contains grid point number, grid point coordinates (latitude and longitude), water salinity, and water temperature. Last five digits in the file name mean a period of averaging, i.e. 50-52 means 1950-1952, 53-57 means 1953-1957, etc.) Click here to this data.

Salinity distribution for each circulation regime is shown in figures at depth 0, 10, 25, 50, 75, 100, 150, 200, 250, and 300m. Each figure shows mean salinity distribution during an anticyclonic circulation, a cyclonic circulation regime (based on the data analysis of the current project) and as decadal mean (based on EWG atlas data).

Salinity Distribution for each Circulation Regime
0 meters 10 meters 25 meters 50 meters 75 meters
100 meters 150 meters 200 meters 250 meters 300 meters

Differences in salinity between corresponding anticyclonic and cyclonic circulation regimes at depths 0, 10, 25, 50, and 75 meters are shown below:

Differences in Salinity between corresponding Anticyclonic and Cyclonic Circulation Regimes
0 meters 10 meters 25 meters 50 meters 75 meters

Vertically averaged salinity in the upper 50 meters for each year is shown below:

Vertically Averaged Salinity in the upper 50 meters
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Distribution of water salinity at the section from Point Barrow to Franz Josef Land is presented below:

Distribution of water salinity at the section from Point Barrow to Franz Josef Land
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

and in the following figures:

Figure D - section-1-1950-1961 (pdf)
Figure E - section-1-1962-1973 (pdf)
Figure F - section-1-1974-1985 (pdf)
Figure G - section-1-1986-1989 (pdf)

Maps of freshwater storage in the ocean corresponding to the anticyclonic and cyclonic climate regimes (based on gridded data) are shown in Figure H (pdf). Small crosses show locations of hydrographic stations taken into account. Note that only period 1972-1979 is very well covered by observations. During other years there are no enough data for a more or less good analyses of freshwater content in the Canadian Basin and the Beaufort Gyre.

A conceptual model (described below) of the freshwater accumulation and release during seasonal cycle and for a cyclonic and anticyclonic climate regimes was formulated as the first order approach. This is a new hypothesis along with supporting evidence that the BG plays a significant role in regulating the arctic climate variability. We propose and demonstrate that the BG accumulates a significant amount of fresh water during one climate regime (anticyclonic) and releases this water to the North Atlantic during another climate regime (cyclonic). This hypothesis can explain the origin of the salinity anomaly periodically found in the North Atlantic as well as its role in the decadal variability in the Arctic region.

A substantial release of the BG fresh water to the NA in response to changing climate conditions can be a source for a catastrophic salinity anomaly in the NA and consequently, a source for an abrupt global cooling. The above perspectives lead us to the conclusion that it is extremely important to understand the structure of the BG water properties, its currents, and their variability in space and time. Specially designed "Beaufort Gyre Exploration Program" ( was proposed in response to the NOAA’s 2001 Ocean Exploration Program announcement of opportunity in order to support project’s goals and hypotheses and to explore one of the most hostile and inaccessible areas of the globe. The proposal was evaluated by two reviewers and both recommended to fund the proposal but cost of the project was too high and NOAA declined this proposal. We have resubmitted the proposal to the NSF’s program "Arctic Freshwater Cycle: Land/Upper-Ocean Linkages. A contribution to the Study of Environmental Arctic Change (SEARCH)" in June 2002.

back to top

Conceptual Model
Our research shows that the origin of the salinity minimum in the BG can be inferred by a comparison of the seasonal change in wind and sea ice motion. Fig. 2 shows the wind and ice drift patterns seasonally averaged for the period 1979-1997. In winter (September-May), the wind Figure 2A (pdf) drives the ice and ocean anticyclonically Figure 2C (pdf) and the ocean accumulates potential energy through a deformation of the salinity field (Ekman convergence and subsequent downwelling, see Figure 1C (pdf). The strength of the horizontal salinity gradient and resultant geostrophic circulation depend on the intensity and duration of the anticyclonic winds. During the winter season the wind-driven and geostrophic currents coincide to set up a strong anticyclonic ice rotation Figure 2C (pdf).

In summer (June-August), the wind is weaker or it may even be cyclonic Figure 2B (pdf) but in the mean the ice still rotates anticyclonically Figure 2D (pdf). An obvious conclusion is that in summer the ocean geostrophic circulation prevails and drives the ice against the wind motion. The salinity anomaly and freshwater content (FC) in the BG Figure 1B (pdf) must decrease in summer, because without wind support, the ocean loses potential energy, i.e., Ekman pumping is reduced. During the following winter the ocean again accumulates potential energy. Hence, the climatic structure of the salinity and dynamic height distribution remain rather persistent (not shown) although exhibiting some seasonal and interannual variability. When viewed on a seasonal scale, the BG salinity anomaly stabilizes the circulation, remaining essentially anticyclonic throughout the year, thus permitting the BG geostrophic circulation cell to serve as a flywheel for the Arctic Ocean circulation.

Some modeling results confirming this mechanism are shown in Figure 3 (pdf). An idealized situation has been tested using a 3-D Blumberg and Mellor [1987] numerical model in a 2000x2000 km basin with 1500 m depth. The basin was initially horizontally uniform but vertically stratified, then it was forced for 9 months by symmetric anticyclonic winds followed by 3 months of cyclonic symmetric winds. The anticyclonic winds generate downwelling in the central basin and upwelling along the boundaries Figures 3A-B (pdf). The results after anticyclonic forcing only are similar to the winter Arctic conditions, and the salinity structure in Figure 3B (pdf) resembles that in Figure 1C (pdf). The addition of cyclonic winds leads to upwelling in the central basin and downwelling along the boundaries and to a reduction in the anomaly in the salinity field generated by anticyclonic winds. The distribution of salinity and currents after 3 months of cyclonic wind forcing are shown in Figures 3C-D (pdf). The circulation pattern in Figure 3C (pdf) is similar to the ice drift pattern in Figure 2D (pdf), i.e., it is still anticyclonic but is weaker than in winter. The salinity distribution in Figure 3D (pdf) resembles the summer salinity distribution in Figure 1B (pdf) when the cyclonic wind forcing leads to the release of FW from deep to upper layers. The seasonal variability of FW content in the central part of the basin is about 10% (not shown).

Figure 1. (pdf): (A) The salinity distribution at 25m. (B), (C) Salinity distribution along dashed line in summer and winter. (D) Dynamic heights relative to 200 db and direction of geostrophic currents.

Figure 2. (pdf): Winter (A) summer (B) sea level pressure (SLP, hPa) and geostrophic wind. (C), (D) Season sea ice drift

Figure 3. (pdf): Results of numerical experiments in the ideal basin. (A) Sea surface salinity (SSS) and surface currents. (B) Salinity section along dashed line. Both figures show results after 9 months of anticyclonic symmetric wind forcing. (C), (D) The same characteristics as in (A) and (B), respectively, but after an additional 3 months of symmetric cyclonic wind forcing.

Figure 4. (pdf): (A) The FC anomaly (solid blue line) from observations and SSHG (red dashed line). (B) The FC anomaly (solid blue line) from reconstruction and SSHG (red and yellow bars) as defined by {\it P&J}. The thick black line depicts the sea ice volume (km^3/year) anomalies from Hilmer and Lemke [2001]. Vertical axes show units of SSHG (x10^-6), sea ice volume anomalies (km^3/year), and the FC anomalies (km^3/year)

Figure 5. (pdf): Sea ice concentration (SIC) in the GIN Sea averaged for the two CCRs: 1980-1983 and 1989-1997 (A), and two ACCRs: 1984-1988 and 1998-2000 (B).

back to top

A hypothetical chain of relationships among atmosphere, ice and ocean in the Arctic at the decadal time scale has been proposed by Mysak and Venegas [1998], Proshutinsky et al., [1999] (hereinafter P99) and others but it is important to know what causes the variability. In order to explain the relationship between the wind-driven and geostrophic circulation and their influence on the accumulation and release of FW we examine the interplay between the atmosphere, ice and ocean in terms of the two circulation regimes identified by Proshutinsky and Johnson [1997] (hereinafter P&J) and P99.

During the anticyclonic circulation regime (ACCR), when high atmospheric pressure prevails in the Arctic, the Arctic Ocean accumulates FW through the increase of FW volume in the BG (Ekman convergence and subsequent downwelling, see
Figure 1C (pdf)) and through the increase of ice thickness and area due to enhanced ice growth (the Arctic is colder during an ACCR than a cyclonic circulation regime (CCR) as shown in P99. Ice is additionally accumulated in the BG during an ACCR due to convergence and ridging under anticyclonic wind forcing. River runoff is increased (trajectories of cyclones are shifted toward land) (P&J; Johnson et al., 1999) and more FW accumulates in the surface waters. When anticyclonic winds are prevalent, the flow of Arctic waters towards Fram Strait is reduced (P&J; Trembley and Mysak, 1998). Consequently, the ice and water flux from the Arctic Ocean to the Greenland Sea and the transport of Atlantic Water into the Arctic Ocean (as a compensation of outflow) are weaker than usual. Deep convection in the Greenland Sea is then enhanced because the vertical stratification is reduced (less FW in the surface waters). This decoupling of the Greenland, Iceland, and Norwegian Seas (GIN Sea) from the Arctic leads to their eventual warming.

Transition to a CCR
All of the above processes lead (with some time lag) to an increase in the gradient of dynamic height between the BG and the NA. The resultant geostrophic circulation increases as does the outflow of FW and ice from the Arctic. During warming of the GIN Sea, the Icelandic Low intensifies and moves to the north leading to an intensification of the transport of Atlantic waters into the Arctic Ocean. This increase in warm water flux to higher latitudes enhances the penetration of atmospheric cyclones into the Arctic, and ultimately decreases the atmospheric pressure in the Arctic. Warming of the Arctic establishes the CCR.

During the cyclonic circulation regime, when low atmospheric pressure prevails in the Arctic (see table characterizing different environmental features of CCR and ACCR in P99), the Arctic Ocean releases FW to the NA through the passages in the Canadian Archipelago and Fram Strait. Warming in the Arctic during the CCR increases ice melting and releases additional FW to the central basin. The accumulation and storage of FW in the BG is not favored by the CCR (even though the cyclonic regime leads to increased ice melt, the FW is not accumulated in the BG because of Ekman divergence and upwelling causing a decrease of freshwater volume in the BG), and hence more FW is available for transport to the NA. River runoff is lower during the CCR than during the ACCR but precipitation over the ocean is increased and hence more fresh water is available for immediate release to the NA from sea ice and surface waters during the CCR.

The stronger surface winds of the CCR in the Fram Strait area (P99) increase the transport of thick ice, and hence FW, to GIN Sea. At the peak of these processes, when all of them coincide, we observe low salinity anomalies in the GIN Sea.

Transition to ACCR
After several years of increased release of ice and FW to the GIN Sea, the surface layer becomes cooler and fresher, and the sea-ice extent increases in the Greenland Sea. Freshening associated with melting of the increased ice volume and increased flux of fresher surface waters leads to an increase in stratification and a decrease in the interaction between the deep ocean and the atmosphere; deep-water convection is consequently suppressed. After several years the dynamic height gradient between the BG and the NA (and consequently the geostrophic circulation) decreases, the Icelandic Low moves to the south and the interactions between the GIN Sea and the Arctic Ocean become weaker, reestablishing the anticyclonic circulation regime.

It is important to note that in this sequence of processes the accumulation and release of FW and ice plays a fundamental role in the interaction between the Arctic Basin and the GIN Sea.