Oc 210 topic 2: circulation in the deep sea (Fall 2009)

OC 210 TOPIC 2: CIRCULATION IN THE DEEP SEA (Fall 2009)

Readings: Ocean Circulation Chap 6: sections 6.3-6.7.

PART 1

1. 
   Identify the major water masses that input water to the Deep Sea.
   
2. 
   Identify and quantify the presence of these deep water masses based on their seawater properties (T, S, nutrients, O₂).
   
3. 
   Describe the general circulation pathway in the Deep Sea.
   
4. 
   Use natural and anthropogenic chemical compounds (tracers) to estimate the rate of deep water circulation.
   
5. 
   Understand the impact of deep water circulation on climate and anthropogenic CO₂ uptake.
   

1. 
   How does the formation of deep waters affect the earth’s climate?
   

1. 
   The cooling of northward flowing warm surface water (initially via the Gulf Stream) in the far N. Atlantic has two effects (Fig. 1)
   
   1. 
      It increases the density of the water to the point where it can sink to great depth (~2000-3000m) and enter the circulation of the Deep Sea
      
   2. 
      It releases heat from the ocean to the atmosphere
      
      1. 
         This transfer of heat from the surface ocean to the atmosphere warms the air masses over the N. Atlantic
         
      2. 
         Air temperatures in lands surrounding the N America are warmer than land areas outside this region at similar latitudes because of this ocean-air heat transfer
         
   3. 
      Climatologists theorize that a slow down in this deep water formation process played a key role in the occurrence of Ice Ages during the last 1 million years.
      

2. 
   A reduction in the formation rate deep water in the N. Atlantic would have a cooling effect on air temperature over the N. Atlantic Ocean and surrounding land areas (Europe, Canada) (Fig. 2)
   
   1. 
      Less deep water formed means less heat would be released to the atmosphere and air temperatures would cool
      

3. 
   One concern among climatologists and oceanographers is about future changes in deep water formation rates and the subsequent impact on the earth’s heat budget.
   
   1. 
      One idea is that global warming will increase ice melt on the continents surrounding the N. Atlantic (e.g., Greenland, Canada, Scandinavia, etc.)
      
   2. 
      This increased ice melt will lower the salinity of the surface waters, decreasing their density and make it more difficult for surface waters to sink as they cool.
      
   3. 
      This reduction in deep water formation rate will, in turn, slow the rate at which surface currents (e.g., Gulf Stream) transport heat from the tropics to the polar latitudes.
      
   4. 
      This reduced poleward heat flow will likely make the tropics warmer and the polar latitudes colder. (Fig. 2)
      
      1. 
         This is the movie Day After Tomorrow scenario…..but a much less dramatic version.
         
   5. 
      Most ocean circulation models predict a future slowing of deep water formation in the N Atlantic as a result of a predicted surface salinity decrease due to increased ice melting in polar regions caused by the warming effects of the anticipated increase in atmospheric greenhouse gas concentrations. (Fig. 3)
      
   6. 
      Currently, it is difficult to determine the likelihood that this will happen and the magnitude and timing of the consequential temperature change (i.e., the accuracy of the model predictions is difficult to determine)
      

4. 
   Generally, oceanographers agree that the rate of deep water formation in the N Atlantic is a very important factor in maintaining the present distribution of temperature on earth and human induced perturbation of this formation rate could have significant climatic consequences.
   

B. characteristics of the Seawater in the Deep Sea

• 
  Cold temperatures, typically -1° to 3°C
  
• 
  Low salinity (than most surface waters), typically 34.5-35
  
• 
  High potential densities, σ_ = 27.7 to 28.0 (primarily due to cold temperatures)
  
• 
  High nutrient concentrations and low dissolved oxygen concentrations as a result of respiration
  
  • 
    Remember: no photosynthesis occurs in the Deep Sea because it is much too deep for sunlight to penetrate
    
  • 
    The main source of oxygen in the ocean is the dissolution of oxygen gas in air, which occurs at the surface of the ocean.
    

4. Current speeds are very slow, typically ~0.1 cm/sec

• 
  usually too slow to measure directly with current meters so circulation pathways are not as well known as for surface ocean
  
• 
  circulation pathways have largely been determined by theory and by using properties of seawater (like , S, oxygen, nutrients) to trace circulation pathways
  
• 
  however, in a few places (e.g., along the western edge of the deep N. Atlantic (where North Atlantic Deep Water is flowing southward) current speeds can be fast enough (>1 cm/s) to measure directly
  

5. The average water residence time in the Deep Sea is about 800 years

- thus, it takes ~800 years for the average water parcel in the Deep Sea (depths > 1500m) to return back to the polar regions of ocean’s surface where they originally sank.

- if we divide the path length of deep water flow by the average time it takes to

cover this path (~800 years), we find that the mean current velocity is ~0.1 cm/s (or 30km per yr or 1/3º per yr)
C. Geographic Regions where Surface Waters sink into the Deep Sea
1. Deep water is formed when surface water gets very dense (primarily cold and, to a lesser degree, salty) and sinks to depths >2000m. This process occurs in regions where the air temperatures are very cold and water column stability (Ds_q/DZ) is low (at polar latitudes).
2. Two primary sites of deep water formation are the far north Atlantic Ocean (off the coasts of Greenland, Labrador and Norway) and in the far south Atlantic Ocean off the coast of Antarctica primarily in the Weddell Sea. (Fig. 5).

-Very little deep water forms off the coast of Antarctica (Ross Sea) in the South Pacific Ocean. No deep water forms in the North Pacific.

- No deep water forms in the Indian Ocean.
3. In the far North Atlantic, deep water is produced at three locations (Fig. 5)

- Greenland (or Irminger) Sea, off the eastern coast of Greenland (~80% of deep water formation in the N Atlantic)

- Norwegian Sea, off the coast of northern Norway (some of this deep water may originate in the Arctic Ocean)

- Labrador Sea, between east coast of Labrador and western coast of Greenland

- During winter, these surface waters eventually cool sufficiently by the time they reach the far N. Atlantic (near Greenland, Labrador and Norway) to reach densities that allow surface water to sink into the Deep Sea

- The average rate of NADW formation is about 15 Sv (15x10⁶ m³/s)

1. 
   upwelling of deep cold water caused by winds
   
2. 
   cooling at the surface during winter because of extremely cold air temperatures
   
3. 
   increase in salinity due to sea ice formation in shallow ice shelves off the Antarctic coast.
   

• 
  as the surface water sinks, it entrains (mixes with) upwelling deep water in about a 50/50 mix of sinking surface water and upwelling deep water. (Fig. 6)
  
• 
  the estimated rate of AABW formation is ~20 Sv (~10Sv of sinking surface water and ~10Sv of entrained deep water)
  
• 
  AABW is dense enough to sink to the bottom of the Atlantic Ocean (denser and deeper than NADW)
  

9. The entrainment of deep water during the formation of AABW increases its nutrient concentrations and lowers its oxygen concentration (Why?)

-thus AABW has lower dissolved O₂ and higher nutrients (CO₂, PO₄, and NO₃) than NADW (which is a deep water with a surface water source, where typically nutrient concentrations are low and oxygen concentrations are high).
10. Deep water formation rates of both NADW and AABW are sporadic both in time and space

• 
  more deep water is formed during years with cold winter air temperatures and less is formed during years with warm winters
  
• 
  because deep water is formed in relatively small geographic areas (that are fairly inhospitable), it can be difficult to locate the exact formation region in any given year
  
• 
  it is difficult to directly measure deep water formation rates in any one year
  

• 
  our best estimates of deep water formation rates are ~15 SV for NADW and
  

11. There is no deep water formed in the Indian Ocean and very little in the Pacific Ocean

• 
  in the Indian Ocean, surface waters do not get cold enough to sink (the N. Indian ocean doesn’t extend past ~20ºN)
  
• 
  in the Pacific, surface waters have lower salinity than in the Atlantic which makes it them less dense and, thus, more difficult to sink
  
• 
  the N. Pacific does not extend geographically as far north as the N. Atlantic and thus surface waters cannot be cooled as effectively
  
• 
  in the S. Pacific off coast of Antarctica (Ross Sea), some deep water is formed but at a significantly slower rate than the rate of AABW formation in the Weddell Sea
  

-Intermediate Waters typically sink to depths of about 600 to 1500m, which is either in the thermocline or at the upper boundary of the Deep Sea.

• 
  AAIW is formed at around 50ºS - 60°S in all the ocean basins (Atlantic, Indian and Pacific) at the location of the Antarctic Polar Front at the northern edge of the Southern Ocean (Fig. 6).
  
• 
  Surface waters that form AAIW get dense (cool) enough to sink to 800-1000m
  
• 
  AAIW spreads northward penetrating to almost ~ 20°N in all ocean basins (Pacific, Atlantic, Indian oceans) (Fig. 7)
  

3. The outflow from the Mediterranean Sea (often called Med Water) is another Intermediate Water (Fig. 8)

• 
  Med Water (MW) sinks to about 1000m and spreads throughout the N. Atlantic only (not nearly as wide spread as AAIW)
  
• 
  MW has very high salinity (S~36.5), which increases its density, yet it is warm ( ~12ºC).
  
• 
  Thus the location of the tongue of high salinity at 1000m caused by MW input illustrates the clockwise circulation path of water parcels at this depth in the N. Atlantic. (Fig. 8)
  
• 
  MW is produced (formed) at a relatively slow rate (~2 Sv) and therefore has a much smaller distribution than AAIW
  

6. 
   In the N. Pacific, an intermediate water is formed (called North Pacific Intermediate Water or Pacific Subarctic Intermediate Water) in the northwest corner of the N. Pacific near the Kamchatka Peninsula and in the Sea of Okhotsk and sinks to a depth of about 500-700m. (Fig. 7)
   

-thus we can use potential temp and salinity to identify the presence of these various water masses in water parcels located in the Deep Sea.

• 
  the core (maximum presence) of each of these water mass sources are found at different depths: AABW at the bottom 4000-5000m, NADW at ~2000-3000m, AAIW at ~800-1000m and MW at ~1000-1200m
  
• 
  the depth at which these deep and intermediate waters are primarily found depends on the density at which these water masses reach when they sink
  

• 
  NADW has  = 3°C, S=34.95, s_q =27.84
  
• 
  AABW has  = -1°C, S = 34.65, s_q =27.87
  
• 
  AAIW has  = 3.5°C, S =34.2, s_q =27.2
  
• 
  MW has  = 12°C, S =36.5, s_q =27.7
  

4. The presence of water which originally comprised AABW, NADW and AAIW is found in water parcels throughout all three deep ocean basins (Atlantic, Indian and Pacific)

-water that originally sank in the regions of AABW, NADW and AAIW formation is transported throughout the Deep Sea via deep currents and mixing

-in contrast, the influence of Med Water is found only in the Atlantic Ocean because its formation rate is small and essentially diluted (by mixing) beyond recognition in the S. Atlantic and, likewise, the presence of N. Pacific Intermediate Water is found only in the N. Pacific because its formation rate is small

• 
  This situation would occur if there are no significant in situ sources or sinks of heat and salt in the Deep Sea and if there were no mixing with other (e.g., thermocline) waters.
  
• 
  A simple analogy is a bathtub with two faucets, one of which discharges blue water and the other yellow water so the water in the tub will be either yellow (close to the one faucet), blue (close to the other faucet) or green (a mixture of the two source waters)
  

-there is a S minimum at ~800m which results from AAIW input

-there is a S maximum at ~2700m which results from NADW input

-there is a S minimum at the bottom which results from AABW input

• to construct a T-S plot, the Pot Temp and Salinity values measured at each depth

in the depth profile are plotted in T-S space and then a line is drawn through these points

• 
  find where the  and S properties for the measurements at surface, 2000m and 4000m lie on T - S plot
  
• 
  the “S” shape in a T - S plot is caused by the inputs of deep and intermediate waters from AAIW, NADW and AABW
  
• 
  the input of water parcels with the distinct  and S properties of AAIW, NADW and AABW show up as bends (inflections) in the T-S plot
  

10. On a T-S plot for the Atlantic Ocean, the location of the AAIW, NADW, AABW and MW waters sources are distinct (separated from each other). (Fig. 10)

• 
  Recall that the mean  and S values for NADW is  =3°C and S=34.95, for AABW is  = -1°C and S=34.65 and for AAIW is  =3.5°C and S=34.2
  

AAIW

- Identify this density trend in Fig 10.
11.Let’s look at where the T-S values measured on a depth profile in the Atlantic Ocean lie relative to the T-S properties of AAIW, NADW and AABW. (Fig. 11)

– the numbers on the T-S curve represent depth where 6=600m, 40=4000m, etc.

-Below 2000m, the measured T-S values fall along on a straight line connecting the T-S properties of NADW and AABW

-this linear (straight line) portion of the T - S curve indicates that over this depth region the water is essentially a mixture of only two sources, that is, AABW and NADW (as will be discussed below)

-between 800m and 2000m, the T-S curve does not follow a straight line connecting the T-S properties of AAIW and NADW

-this means that there is another source of T and/or S in this depth interval, which is most likely input from Med Water

-one result of Med Water input is that water parcels with the T and S characteristics of pure AAIW (=3.5ºC and S=34.2) are not found at the site of this measured depth profile even though the measured T and S depth profiles are clearly influenced by input of AAIW water.

F. Determining the contribution that each water source (e.g., AAIW, NADW and AABW) makes toward the T and S measured on a water parcel in the Deep Sea
1.Let’s assume steady-state conditions for  and S in the Deep Sea

-that is, the measured  and S of a water parcel isn’t changing with time and that T and S are conservative (neither produced or consumed in-situ).

-the steady-state assumption is probably a very good one for the Deep Sea where the residence times are very long and changes are very slow.
2. We’ll start with the simplest situation, that is, when there are only two water sources contributing to the T-S properties of water parcels (Fig. 12).

-In this case, the T and S properties of the all the water parcels must fall on a straight line (in T-S space) that connects the T and S properties of the two water sources.

-an example of this situation is seen in Fig. 9 in the depth region between ~2700m and the bottom in the S. Atlantic (40ºS) where there is a linear section of the T vs S plot with the two water sources being NADW and AABW
Class Problem: Plot (on Fig. 12) the T and S properties represented by a 50-50 and 25-75 mix of the two source waters (I and II), where source water I has  = 10ºC and S = 36 and water II has  = 4ºC and S =34. Demonstrate that any mixture of these two water sources has a pot temp and salinity that falls along a straight line connecting the two water sources.
-A 50-50 mix of WM1 + WM2 would produce T = 5° and S = 34.5

-when you plot the T and S characteristics of the two end-member water masses and the mixture they all fall on a straight line

-A 75%WM1+25%WM2 mix would produce Tm = 6.5° and S = 34.75 which also falls on line

A 25%WM1+75%WM2 mix would produce Tm = 3.5° and S = 34.25 which also falls on line

-to calculate this proportion, one must know (assign) the  and S properties of the water sources in addition to measuring the  and S properties of the water parcel.

- for example, does the mixture of AABW, NADW and AAIW predicted by a circulation model or theory for a water parcel at a specific location in the Deep Sea agree with the mixture of these water sources calculated from measured  and S

- if not, maybe the theory or model estimates of formation rates of AABW, NADW and AAIW or mixing rates in the Deep Sea are incorrect

5. We can determine the fraction of each end-member water mass in any given sample by using either a graphical or mathematical approach. This works because the temperature and salinity most deep water parcels are not affected by in-situ processes, i.e. that is,  and S are conservative properties of seawater and are only affected by mixing

-this situation is the same as mixing being the only process affecting the color of water in a bath tub which has inputs of yellow and blue source waters, as described above

-a possible (but not too common) exception to the conservative characteristic of Potential Temperature can occur near mid-ocean ridges where input of geothermal heat and hot fluids can warm deep waters.

● on a T-S figure, plot the measured T and S of the two water sources (I and II, in this

example) and the measured T and S of the water parcel (R, in this example)

● on a T- S plot, the T and S properties of the water parcel must fall on a straight line connecting the T-S points of the two water sources (assuming there are only two water mass sources and that T and S are conservative)

• 
  on a T-S plot, measure (using a ruler) the length of the line between the observed T, S point measured on the water parcel [R] and the T, S point of the water source which is closest (source II in this case)
  
• 
  the ratio of this line length to the entire line length between water source I and II is the fractional contribution of I (F1), where F1 = b/(b+a), to the measured T and S of water parcel R
  
• 
  for example, if the measured  and S are 6.0°C and 34.7, respectively, then the water parcel contains 33% water source I and 67% water source II.
  

7. Finding the fraction of three water sources (e.g., AAIW, NADW and AABW) present in a water parcel using a graphical method. (Fig. 13)

• 
  When there are three water sources contributing to the T and S properties of water parcels, then the T-S of all water parcels must fall within the triangle defined by the T-S points for the three water sources.
  

• 
  Find the fraction of each source water contained by a water parcel with measured T and S
  

• 
  First, draw the triangle with corners determined by the T and S point for each of the three water sources
  
• 
  Second, draw lines from the measured T-S point of each of the three water sources (I, II and III in the figure) through the measured T-S point of the water parcel [R] until the side of the triangle is reached.
  
• 
  Third, measure the ratio of the length of the line between the measured T-S point and the point where the line intersects the opposite side of the triangle to the entire line length between the T-S point of the water source and the intersection with the triangle side. This ratio equals the fraction of the measured T-S property of the water parcel that was derived from that water source.
  

8. For example, in Fig. 13, the fraction of water source III present in water parcel “R” is the length of line segment “f ” divided by the length of line segment “f+e”

- e.g., similarly, the fraction of I = b/(b+a) and fraction of II = d/(c+d)

-the fractions of I, II and III are about 0.38, 0.44 and 0.18, respectively applying the graphical approach to Fig. 13

-NOTE: the sum of the three fractions should equal 1.0 (if not, you’ve incorrectly calculated the fractions)
9. For a real example, NADW is actually made up of three water sources formed separately in the Labrador, Norwegian and Greenland Seas. (Fig. 14)
10. NADW can be separated into a shallow (~2000m) and deep (~3500m) component with the

-the deeper component, called Lower NADW (seen as the circle inside the triangle in Fig. 14), is composed of water derived from all three source regions with highest contribution from Greenland Sea water

- In this figure, LSW represents Labrador Sea Water, GSW represent Greenland Sea Water and NSW represents Norwegian Sea Water.

- Use the centers of the circles to represent the T-S points of the sources waters and water parcel.

-You should get fraction estimates of around 0.20, 0.17 and 0.63 for LSW, NSW and GSW, respectively

13. A mathematical approach can be used to calculate the fraction of each water source in which you set up enough equations to calculate the fraction of each source water (2 equations for 2 sources, 3 equations for 3 sources, etc.). For three sources, the three equations describe the temperature, salinity and mass balances. By solving them simultaneously, each fractional contribution (F1, F2 and F3 representing the fraction of each water source) is determined.

equation 2: F1*T1 + F2*T2 + F3*T3 = Tm (Heat balance)

equation 3: F1*S1 + F2*S2 + F3*S3 = Sm (Salt balance)
-where F is fraction of the water source present in the water parcel. T1 and S1 are the Pot Temperature and Salinity water source #1 (same for water source #2) and Tm and Sm are the Pot Temp and salinity measured on the water parcel of interest.
-It is much easier to solve for F1, F2 and F3 using the graphical approach.
-Note: You only need to know the graphical approach, not the mathematical approach, to determining F1, F2 and F3 from T and S measurements.
F. How does mixing affect the depth profile of T and S?
1. Water parcels mix with each other as a result of turbulence in the ocean.

• 
  mixing occurs most easily (and thus fastest) along surfaces of equal density (isopycnals), (which are oriented sort of horizontally in most places) because there is no buoyancy effects to overcome by mixing
  
• 
  mixing occurs most slowly between or across isopycnals (sort of vertically) because buoyancy effects have to be overcome by mixing
  
• 
  mixing occurs at several spatial scales, that is, via large scale organized motions like currents, via large scale features like eddies, at smaller scales via turbulent (random) motions of water parcels and at the smallest scales via molecular motion.
  

• 
  the length scale being considered often determines which process dominates, generally turbulent mixing dominates on short length scales (<1m to 10km) whereas currents and eddies dominates over longer length scales (100-1000kms)
  

• 
  turbulence is an effective mixing process in both the Deep Sea and surface ocean
  
• 
  turbulence can mix both vertically and horizontally, however, the vertical mixing rate is much slower (~10⁶x slower) than the horizontal (along isopycnal) mixing rate because vertical mixing has to overcome stability (σ_/Z) in most regions of the ocean whereas along isopycnal mixing does not
  

2. We’ll use a hypothetical example to demonstrate how mixing affects the vertical shape of depth profiles of T and S and the shape of the T-S plot when three water sources are present (Fig. 15)

4. 
   Generally, the further away from the formation region of the source water (e.g., NADW) that one measures a T and S depth profile, the longer the waters parcels have had to mix and the smoother (less sharp) the T-S curvature. Also the fainter the original T and S properties of the source water appear.
   

-stability is maximized where the T-S line crosses isopycnals over the shortest depth interval (i.e., perpendicular to isopycnals) and yields the highest Δs_q/Δz

• 
  This is more easily seen in Fig 10 where the curvature of lines of constant density is more clear
  
• 
  if you make a mixture of two water masses with different T and S properties but which lie on the same isopycnal, then the T and S property of the resulting mixture follows the straight line connecting the two water mass sources. This mixture, however, is denser than the density of either source, that is, the density of the mixture lies on a denser isopycnal than the isopycnal of the two sources.
  
• 
  this occurs because the relationship between temperature and density is not linear, especially at cold temperatures (<2°C)
  
• 
  this phenomenon is called cabelling
  

• 
  Henry Stommel predicted this observation based on theory in the early 1950s (Fig. 17)
  
• 
  Stommel’s theory indicated that the deep equatorward flow should occur in intensified western boundary currents while the poleward return flow in the Deep Sea should occur via weaker currents in the middle and eastern portions of the deep ocean basins
  
• 
  In the late 1950s, Stommel’s postulated southward intensified flow of deep water along the western edge of the deep N. Atlantic was verified using neutrally buoyant floats
  

• 
  Because deep flows generally are very slow (except in the intensified western boundary currents), it has been difficult to verify whether the Stommel’s theorized flow pattern really exists
  

2. The reasons why deep western boundary currents are present are dynamical and related to the earth’s spin

• 
  Stommel theorized that there must be equatorward flow of deep water formed in the polar regions to balance the loss of deep water due to upwelling. This upwelling of deep water is needed to balance the input of deep waters (like NADW and AABW) to the Deep Sea and offset the downward diffusion of heat resulting from turbulent mixing.
  
• 
  Stommel hypothesized that the equatorward flow would be concentrated in intensified western boundary currents
  
• 
  Note: Our theoretical understanding of abyssal circulation is still evolving. As our observations of deep circulation pathways improve, our ability to verify the theory of deep circulation will improve.
  

-In contrast, most of the northward (equatorward) flow of the AABW gets entrained into the eastward flow of the ACC, but some passes underneath the ACC and flows northward at the bottom along the western boundary of the S. Atlantic

1. 
   Deep water circulation simulated by state-of-the-art General Circulation Models (GCM) verify the essential pathway of deep flow predicted by Stommel’s theorized circulation
   

6. How does the deep water produced in the Atlantic and carried to the Indian and Pacific oceans return to the surface in the N. and S. Atlantic to complete the round trip?

• 
  deep water warms during its trip from the AtlanticIndianPacific (via downward mixing of heat from the warm upper ocean)
  
• 
  this warming increases the buoyancy (decreases the density) of the deep water and causes it to rise. Eventually this density decrease is sufficient to allow the deep water to rise through the thermocline and, ultimately, become part of the surface ocean circulation.
  

- that is, there is a mass flow balance between deep water input (NADW+AABW) and loss (upwelling) in the deep sea at steady-state

-However, we don’t know well the pathway that returns deep water to the surface ocean

-oceanographers think a substantial amount of the upwelling from the Deep Sea occurs in the Southern Ocean (south of 50ºS) in the region of the Antarctic Circumpolar Current (see Fig. 6)

8. Ultimately the upwelled deep water must return to the site of deep water formation to complete the circulation circuit

● this general path of deep water flow gave rise to a Conveyor Belt analogy (Fig. 18)

-although the actual flow path is not this simple, the analogy is convenient

• 
  in the Pacific, the major surface return path is through the Indonesian archipelago
  
• 
  return flow of surface currents to the Atlantic, from the Indian and Pacific, occurs via the Agulhas Current around the tip of S. Africa
  
• 
  once in the Atlantic, ~15Sv gets incorporated into the Gulf Stream flow and is ultimately advected to the far N. Atlantic to cool, sink and start all over again as NADW
  

9. Some deep water returns to the Weddell Sea without ever contacting the surface.

- this is the deep water that is entrained during the sinking of AABW
10. Our best estimate is that ~25Sv of the upwelled deep water returns to the regions where deep water is formed in the far north and south Atlantic Ocean via surface circulation pathways

-about 15 Sv of surface water flow into the North Atlantic to form NADW and about 10 Sv of surface flow to Antarctica, where it sinks and entrains another ~10 Sv of deep water to form AABW.