Geoscience Reference
In-Depth Information
converted to the temperature mode by inserting a small
paddle into the chamber through the slot and mixing. If
the temperature mode exists, the flow can be converted
to the salt mode by inserting the paddle near the salinity
source tube and suppressing the local turbulent mixing.
Soon, a layer of salty water forms along the bottom and
the salt mode forms.
The results are best summarized by plots of dimension-
less temperature and salinity versus scaled bath temper-
ature (Figure 13.4). For these runs, the value of salinity
difference, and the pumping rates of salt water for the box,
slot, and cavity experiments are each kept at one constant
value. Experimental runs with different values of pumping
rate were also studied in many of the papers, but results
are inferior due to poorer coverage over parameter space,
so they add little to the story. In all experiments, one com-
plete run must always be conducted over a long enough
time span to become convincingly steady. The necessary
time is at least an hour and sometimes much longer. In
regions of hysteresis, the two corresponding points with
the same driving parameters are determined by how that
particular run was initiated. One can start from an earlier
run or start the apparatus with either fresh or salty water
in the chamber. Transitions can be triggered by inserting
temporary blocks in some of the tubes or by stirring, as
mentioned above. Naturally, in points without hysteresis,
the same point is found no matter how the flow is started.
Scaling the laboratory results is quite simple. The bath
temperature,thetemperatureinthechamber,andthesalin-
ity of water in the chamber are plotted here using the
density difference between salt and fresh water:
T
∗
=
αT
∗
/βS
0
,
T
=
αT/βS
0
,and
S
=
S/S
0
,where
α
is the
coefficient of thermal expansion for water at 20
◦
C,
β
is
the density coefficient of salinity, and
S
0
is salinity of the
salt water pumped in at the source or, in the case of the
layered experiment, the salinity of deep water. Since warm
water has lower density than cold, the scaled temperature
is inversely proportional to density but the scaled salinity
is proportional to density. Thus, if scaled temperature and
salinityareplottedtogether,(Panelsaandc)thetwooppos-
ingeffectsoverlieeachotheranditisimmediatelyobserved
as to whether temperature or salinity affects density more
strongly. This is not true for Figure 13.4 panel b, where the
scaled salinity is subtracted from unity to make the figure
clear when compared to theory (shown by the curves).
Figure 13.4 shows a number of points. First den-
sity change from salinity always dominates over density
change from temperature in the Salt Mode and density
change from temperature always dominates over den-
sity change from salinity in the Temperature Mode, as
expected. Second, the range of hysteresis is greatest for the
box experiment and less so for the rest. A close relation
between the experiments and box model theory similar
to Stommel's are demonstrated in
Whitehead
(1996) and
Figure 13.2.
Side view of the slot experiment with the two dif-
ferent modes of flow at the same values of driving parameters.
The top panel shows the temperature mode and the bottom
panel shows the salt mode. BW rendition of a color figure
[
Whitehead et al.
, 2003].
Figure 13.3.
Layered experiment [
Whitehead
2009].
bath temperatures between those two temperatures. The
slot experiment is particularly appealing, because artifi-
cial mixing or suppression of mixing can trigger transition
back and forth. If the salt mode exists, the flow can be
