Geoscience Reference
In-Depth Information
(a)
(b)
Hysteresis
H
ystere
sis
Salt mode
Temperature mode
Salt mode
Temperature mode
1.5
0.75
.
.
.
.
.
o
o
.
1.0
o
.
.
.
0.60
.
.
.
o
o
.
.
.
.
.
o
o
.
.
.
.
.
.
0.5
.
.
.
0.45
o
o
o
o
o
o
.
0
0.30
.
.
o
-0.5
.
.
0.15
-1.0
.
.
0
0
-1.5
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0
0.15
0.30
0.45
0.
60
0.75
0.90 1.05
-T*
T
*
(c)
No hysteresis
Salt mode
Temperature mode
1
0.8
0.6
0.4
0.2
0
-0.2
0.55 0.6 0.65 0.7 0.7
5 0
.8 0.85 0.9 0.95
1
T
*
Figure 13.4.
Measurements of dimensionless temperature (circles) and salinity (rectangles) versus driving temperature for (a) the
box experiment, (b) the middle of the bottom layer of the layered experiment with the curves from the theory in
Whitehead
, [2000].
and (c) the middle of the cavity in the cavity experiment. The error bars in (b) are for the oscillation described in the next section
and Adapted from
Whitehead
[2009].
Whitehead
(1998), so it is not surprising to see sizeable hys-
teresis as the box model theory predicts. The theory for
the layered experiment is approximately confirmed by the
data in Figure 13.4b. The cavity experiment does not have
a box model counterpart and hysteresis is too small to be
resolved.
Figure 13.4 also indicates that density difference
between the chamber and the reservoir changes sign upon
an abrupt thermohaline transition from one mode to the
other. This is more clearly seen with plots of dimensionless
density from the box and cavity experiments (Figure 13.5).
The greatest uncertainly arises because it is impossible
to assign an exact value of temperature in the experi-
ments since it varies within a chamber. Also, temperature
at a point is time dependent, because active convection
takes place. Only in the box experiment was the tem-
perature in the chamber recorded over a long enough
time to make good averages, and since the convection
within the box is relatively even, the temperature is not
a very strong function of the location of the tempera-
ture probe. Therefore, scatter in Figures 13.4a and 13.5a
is relatively small and the data exhibit straight trends.
The cavity experiment temperature data are not as linear
(Figure 13.4c and 13.5b), because parcels of water of dif-
fering temperature and salinity are always present inside
the cavity. Although conceivably one could average over
long times to smooth out the time dependence, in prac-
tice, this was not possible with our existing equipment.
Also, for the cavity experiment in the temperature mode,
the average temperature is a function of elevation of the
probe above the bottom boundary layer. Therefore, the
scatter of the temperature and density data is larger than
optimal in Figures 13.4d and 13.5b.
13.3. OSCILLATIONS
Oscillations were first discovered in the slot experi-
ment. These did not prove to be reproducible in that
