435
2. 7'HE EYI'I~KIMt~N'lAL SITlIP
Only a w r y brief description of the experimental setup is giveti here: for
inore dctiiik the rcader is referred to Turner (1968) and Crappcr and Linden
( 1974). Hasically the apparatus consists of a tank containing two horizontal
grids which niay be oscillated vertically at varying frequencies and aniplitudes. The details of the turbulence so produced have been determined by
Thompson (1969) who made measurements of the rms velocity u and integral length scale I of the turbulence in a homogeneous fluid. Consequently all
the results are presented in terms of u and I directly, the values used being the
ones appropriate to the turbulence near the interface in each experiment.
Vertical profiles of temperature and salinity were taken by traversing
either a thermocouple or a conductivity probe through the interface. Each
probc had ;I vcrtical resolution of about 0.1 cm and was sampled approximately every 0.05 cm of its traverse. The output was digitized and stored on
magnetic tape and then analysed at leisure. The accuracy of the temperature
and salinity measurements were found to be 0.3 "/, and 0. I y,;; of full-scale,
respectively.
As was mentioned in t hc previous section this experimental configuration
is not completely steady since the density difference between the layers
decreases with time. However, the rate of change of Ap with time was very
slow and even when the entrainment was most vigorous a typical change of
0.001 gm!ml in Ap took about 60 min. Consequently, even measurements
taken over a period of 2 inin which were required for mean profiles can be
considered as representative of the quasi-steady structure of the interface.
3. T m EXPEHIMFNTAI. REsui.rs
3.1. 7'lJl~7pc~l.lllIrrc~ I l l 1 c r f a c - r
Some cxaiiiples of temperature depth profiles for varying Ri and Pe are
showa i n Fig. I . I t is seen that all the profiles show two well-mixed layers
scpilrilted by iI region of essentially constant temperature gradient. On occas i o n s the layer\ coiitiiin inversions (see Fig. Ib) presumably due to a mixing
event tahrng place. 11 is sccn thnt the interface becomes thinner as Pe
InCrL'iISc's.
In ordcr to quantify this last statement we define an "interface thickness"
by littirig a striiiglit line to the central SO ';,,of the data points in the interface
and extrapolating this line until it intersects the mean temperatures of the
upper and lower layers. The vertical distance between these two intersections is defined as the interface thickness /I.
2. 7'HE EYI'I~KIMt~N'lAL SITlIP
Only a w r y brief description of the experimental setup is giveti here: for
inore dctiiik the rcader is referred to Turner (1968) and Crappcr and Linden
( 1974). Hasically the apparatus consists of a tank containing two horizontal
grids which niay be oscillated vertically at varying frequencies and aniplitudes. The details of the turbulence so produced have been determined by
Thompson (1969) who made measurements of the rms velocity u and integral length scale I of the turbulence in a homogeneous fluid. Consequently all
the results are presented in terms of u and I directly, the values used being the
ones appropriate to the turbulence near the interface in each experiment.
Vertical profiles of temperature and salinity were taken by traversing
either a thermocouple or a conductivity probe through the interface. Each
probc had ;I vcrtical resolution of about 0.1 cm and was sampled approximately every 0.05 cm of its traverse. The output was digitized and stored on
magnetic tape and then analysed at leisure. The accuracy of the temperature
and salinity measurements were found to be 0.3 "/, and 0. I y,;; of full-scale,
respectively.
As was mentioned in t hc previous section this experimental configuration
is not completely steady since the density difference between the layers
decreases with time. However, the rate of change of Ap with time was very
slow and even when the entrainment was most vigorous a typical change of
0.001 gm!ml in Ap took about 60 min. Consequently, even measurements
taken over a period of 2 inin which were required for mean profiles can be
considered as representative of the quasi-steady structure of the interface.
3. T m EXPEHIMFNTAI. REsui.rs
3.1. 7'lJl~7pc~l.lllIrrc~ I l l 1 c r f a c - r
Some cxaiiiples of temperature depth profiles for varying Ri and Pe are
showa i n Fig. I . I t is seen that all the profiles show two well-mixed layers
scpilrilted by iI region of essentially constant temperature gradient. On occas i o n s the layer\ coiitiiin inversions (see Fig. Ib) presumably due to a mixing
event tahrng place. 11 is sccn thnt the interface becomes thinner as Pe
InCrL'iISc's.
In ordcr to quantify this last statement we define an "interface thickness"
by littirig a striiiglit line to the central SO ';,,of the data points in the interface
and extrapolating this line until it intersects the mean temperatures of the
upper and lower layers. The vertical distance between these two intersections is defined as the interface thickness /I.