Next: Bravo Mooring
Up: Density Structure
Previous: Density Structure
A WOCE hydrographic cross-section in the Labrador Sea was most recently
occupied in June 1995 and Feb-Mar 1997. First, I will compare model
hydrographic data taken in Feb-Mar 1997 with model output. Secondly, I will
compare the temperature and salinity measurements taken in June 1995 with the
model output for June 1997 despite the two year time difference. The WOCE cross
sections took 2 weeks to complete, but I will compare it with the 10-day
average field during the most applicable time. The cross-sections give a
quasi-snapshot of the convective region. For convection to occur in the
Labrador Sea, heat loss to the atmosphere must erode the salinity-stabilized
upper layers of the ocean. Warm, salty Irminger Sea Water at depths of 200-700
meters lies directly below the surface-cooled surface water in the early
winter. The temperature profile appears unstable, but it is stabilized by the
salinity profile. The model shows a mid-depth local maximum in temperature and
salinity at 700 meters depth at this time of the year. As the winter
progresses, cold and dense surface water gradually erodes the mid-depth
temperature maximum. The convection scheme does not leave a density jump at the
base of the mixed layer, and therefore acts like nonpenetrative
convection. Most observations (Lilly el al., 1999) show this type of convection
in the Labrador Sea. As the mixed layer deepens, the mid-depth local maximum is
moved to greater depth until it is completely removed in the southwest side of
the basin.
Late Winter Hydrography
Late February and early
March is the time of greatest convection in the model run, and can be directly
compared to the WOCE hydrographic cross-section of Feb. 26-Mar. 7, 1997
(courtesy R. Pickart). It is instructive to compare the model's performance
separately in three depth levels: 0-500 meters, 500-1500 meters, and the deeper
waters (see Figs. 4-7). Due to a lack of mixed layer physics, the surface
temperature is too cold uniformly across the cross-section. The observed
cross-section shows a distinct asymmetry in the surface cooling which is not
captured by the model. On the other hand, the model's surface salinity has
similar values to the observed salinity distribution. The Irminger Current in
the northeast side of the transect is slightly too deep and large, but its
general T-S values are correct. As previously mentioned, the addition of mixed
layer physics may improve the upper 500 meters in a future run. At depths of
500-1500 meters, the model does accurately predict the strongest convection to
be in the southwest side of the transect. The newly-formed Labrador Sea Water
has the right water mass characteristics : 2.78 degrees Celsius and 34.84
salinity. The structure of the convection above 1500 meters is
reasonable. Below 1500 meters, the water mass characteristics are not greatly
affected by convection. The model is 0.2 degrees colder than the actual values
between 2300-2600 meters depth. This temperature discrepancy is due to the
initial conditions. The climatology of M. Visbeck does include Iceland-Scotland
Overflow Water (ISOW) as an incursion of relatively warm, saline water at
2300-2600 meters. However, the temperature inversion is not quite strong enough
in the initial conditions. Therefore, the model ISOW is colder and more dense
than the true ISOW. This will serve as a ``basement'' for any convection in the
model. However, Labrador Sea convection has never been observed to break
through this layer of ISOW. Deepest convection was observed in 1992 to a depth
of 2200 meters (Lab Sea Group, 1998). I think that the model ISOW will not
affect the depth of convection in any reasonable situation, and the initial
conditions do not need to be ammended at this depth. At depths below 2500
meters, water tends to warm throughout the year because of a slumping of
isopycnals from the deep boundary currents. This results from abnormally high
levels of friction necessary in GCM's. The deep boundary currents should not
play a role in transporting LSW, although the transport of Denmark Strait
Overflow Water will be affected, and simple resolution of the problem is not
available. The salinity structure has a homogeneous layer from 1500-2000 meters
deep at 34.85 salinity. The increased salinity of ISOW is also present. The
observed structure is very similar, except for a plume of fresh water in the
convective region which extends to 2000 meters. Plumes are typically of very
small scale, and are not likely to be captured by a relatively course
climatology. I believe that the initialized salinity structure at depth is
adequate. In conclusion, attention needs to be focussed on improving the model
above 500 meters depth, although the characteristics of the deeper ocean are
quite reasonable.
Early Summer Hydrography
Spring is a time of
rapid restratification in the upper layers of the Labrador Sea. The model
quickly restratifies and then settles into a summer state. June is sometime
after the convective season and most of the restratification, so the T-S
characteristics should not change drastically during the duration of the
transect. There is a thin upper layer of restratified fluid due to the seasonal
forcing in both model and observed fields (see Figs. 8-11). Below the surface
layer, there is a nearly homogeneous layer of fluid to a depth of 2000
meters. Finally, the deepest waters are restratified to the bottom depth of
3500 meters. Due to the previously documented lack of mixed layer physics, the
model's upper layer is too warm and shallow at the surface. Model surface
temperature of 6.5 degrees C exceeds observation by nearly 2 degrees. If the
model more strongly diffused heat downward in the mixed layer, both problems
would be fixed. It appears that the total heat flux from the atmosphere is
perhaps too strong in the model. Moving downward in the water column, a nearly
homogeneous layer of fluid sits between 500 and 2000 meters. The lateral
boundary of the homogeneous layer is nearly vertical, and the northeast side of
the transect is colder than observation. The coldness of the northwest side
should not be completely considered an artifact of the model's aversion to
restratification. Although the model preferentially cools the southwest side of
the transect in winter, the northwest side is also cooled considerably and its
effect remains in early summer. However, a strong horizontal density gradient
should eventually be smoothed. According to one hypothesis, geostrophic eddies
are very important in transferring boundary water properties into the interior
(Khatiwala and Visbeck, 2000). An ``eddy-induced'' velocity of
can
restratify most of the Labrador Sea within a few months. Because this model
does not resolve all eddies, it is probable that a lack of eddies does allow
the vertical boundary of LSW to remain until June. Another mechanism for
lateral exchange is simple lateral advection, but its effect relative to eddy
diffusion is uncertain here (Spall and Pickart, 2000). Not surprisingly, the
deepest bottom waters (2500-3500 meters) have almost exactly the same
properties in both model and observations. In the salinity fields, the mixed
layer has the right depth and correct value (Figs. ). The model salinity field
also captures the asymmetry of processes in the Labrador Sea. However, the
northwest side of the salinity cross-section also shows the model's difficulty
in restratification. In conclusion, the WOCE hydrographic sections prove that
the effect of convection in the model is reasonable, but questions the
representation of surface layer temperature. Also, the model should produce a
greater rate of mixing from the boundary into the interior.
Next: Bravo Mooring
Up: Density Structure
Previous: Density Structure
Jake Gebbie
2003-04-10