94
Land-Ocean Systems in the Siberian Arctic: Dynamics and History
one) the dependence of buoyancy flux on the density gradient is quasi-linear and local
perturbations of density profile tends to be smoothen by turbulent diffusion. In this case layers
does not form.
This effect was demonstrated in different laboratory and numerical experiments. It was shown
that in both cases of shear (Barenblatt et a!., 1993; Krylov, 1993; Kan and Tarnai, 1994) and
shear-free (Ruddick et a!., 1989; Park et a!., 1994) turbulent flows the formation of step-like
density profile structure may occur. In order to explain the mentioned above observations made
in the Kara sea the results of laboratory experiments with shear turbulent flows were used by
Golovin et a!. (1996). It was obtained that in order to satisfy the laboratory criteria of the
pycnocline splitting 0.02 < Riu < 2 (Krylov, 1993) (here Riu=(g'·h)lU 2 -the Richardson
number, g' - the reduced gravity, based on the density difference across the pycnocline, h - the
thickness of the pycnocline, U - the horizontal velocity difference across it), U must be
approximately twice of the ice drift velocity. In other words there must be enough strong
current in the water layer below the pycnocline of opposite direction to the direction of ice drift.
Although it is quite possible (there was no direct velocity measurements), the suggestion was
made that the critical Richardson number Rierit for step-like structure formation may be a
monotonously growing function of the Reynolds number for turbulence (the higher is the
Reynolds number, the higher is the value of critical Richardson number), at least for low values
of the Reynolds number (Nishida and Yoshida, 1994). In real ocean conditions the Reynolds
number for turbulence is larger then in the most of laboratory experiments, so if the mentioned
above suggestion is true, the critical Richardson number may also be larger. In these
circumstances the shear across the pycnocline, required for the step-like structure formation
may be smaller.
In order to check the assumption of the Ricrit dependence on Re and to study more about
conformities of the step-like structure formation in turbulent stratified fluid, we provided a
series of shear-free experiments described below.
Experemental set-up
The scheme of preliminary experimental set-up is shown on the Figure 1. Here (1) is the tank
(25* 16*30 cm 3 ), made from 2.0 cm thick organic glass sheet, and filled by linearly stratified
salt (NaCl) water solution. The turbulent mixing is produced by the system of horizontally
oscillating grids (2). Approximately similar method of mixing was used first by Ruddick et a!.
(1989). In most of experimental runs there were six grids situated on the same oscillating rod
(3) at the distance of about 3.5 cm from each other and from opposite small side walls of the
tank. Each grid consists of six cylindrical vertical glass rods 0.7 cm in diameter with the
distance of about 2.8 cm from each other. The oscillations are produced by electric motor with
eccentric drive (4). The period of oscillations is fixed: T = 2 s, the amplitude is changed from
one run to another. The larger is the amplitude of oscillations the higher is the intensity of
turbulence. In order to make visible the density inhomogeneities in the turbulent stratified fluid
the simple shadowgraph device (5) is used. The shadowgraph picture is monitored by video
camera and photo camera.
Observation and results
Twenty five experimental runs were provided with six grids on the rod in order to observe
different regimes of turbulent mixing in initially linearly stratified fluid. The initial salinity
gradient was changed from 1.0 to 12.5 unit/cm and the amplitude of the oscillations - from 0.5
Land-Ocean Systems in the Siberian Arctic: Dynamics and History
one) the dependence of buoyancy flux on the density gradient is quasi-linear and local
perturbations of density profile tends to be smoothen by turbulent diffusion. In this case layers
does not form.
This effect was demonstrated in different laboratory and numerical experiments. It was shown
that in both cases of shear (Barenblatt et a!., 1993; Krylov, 1993; Kan and Tarnai, 1994) and
shear-free (Ruddick et a!., 1989; Park et a!., 1994) turbulent flows the formation of step-like
density profile structure may occur. In order to explain the mentioned above observations made
in the Kara sea the results of laboratory experiments with shear turbulent flows were used by
Golovin et a!. (1996). It was obtained that in order to satisfy the laboratory criteria of the
pycnocline splitting 0.02 < Riu < 2 (Krylov, 1993) (here Riu=(g'·h)lU 2 -the Richardson
number, g' - the reduced gravity, based on the density difference across the pycnocline, h - the
thickness of the pycnocline, U - the horizontal velocity difference across it), U must be
approximately twice of the ice drift velocity. In other words there must be enough strong
current in the water layer below the pycnocline of opposite direction to the direction of ice drift.
Although it is quite possible (there was no direct velocity measurements), the suggestion was
made that the critical Richardson number Rierit for step-like structure formation may be a
monotonously growing function of the Reynolds number for turbulence (the higher is the
Reynolds number, the higher is the value of critical Richardson number), at least for low values
of the Reynolds number (Nishida and Yoshida, 1994). In real ocean conditions the Reynolds
number for turbulence is larger then in the most of laboratory experiments, so if the mentioned
above suggestion is true, the critical Richardson number may also be larger. In these
circumstances the shear across the pycnocline, required for the step-like structure formation
may be smaller.
In order to check the assumption of the Ricrit dependence on Re and to study more about
conformities of the step-like structure formation in turbulent stratified fluid, we provided a
series of shear-free experiments described below.
Experemental set-up
The scheme of preliminary experimental set-up is shown on the Figure 1. Here (1) is the tank
(25* 16*30 cm 3 ), made from 2.0 cm thick organic glass sheet, and filled by linearly stratified
salt (NaCl) water solution. The turbulent mixing is produced by the system of horizontally
oscillating grids (2). Approximately similar method of mixing was used first by Ruddick et a!.
(1989). In most of experimental runs there were six grids situated on the same oscillating rod
(3) at the distance of about 3.5 cm from each other and from opposite small side walls of the
tank. Each grid consists of six cylindrical vertical glass rods 0.7 cm in diameter with the
distance of about 2.8 cm from each other. The oscillations are produced by electric motor with
eccentric drive (4). The period of oscillations is fixed: T = 2 s, the amplitude is changed from
one run to another. The larger is the amplitude of oscillations the higher is the intensity of
turbulence. In order to make visible the density inhomogeneities in the turbulent stratified fluid
the simple shadowgraph device (5) is used. The shadowgraph picture is monitored by video
camera and photo camera.
Observation and results
Twenty five experimental runs were provided with six grids on the rod in order to observe
different regimes of turbulent mixing in initially linearly stratified fluid. The initial salinity
gradient was changed from 1.0 to 12.5 unit/cm and the amplitude of the oscillations - from 0.5