Work done by Eric Weeks, Yudong Tian, Jeff Urbach, Michael Ghil, and Harry Swinney.
Back in ~1996 we did experiments studying different flows we see in the rotating annulus where topography is present. We had two symmetrically placed ridges on the bottom of our experiment:
In the picture above, the ridges are lying on the bottom of the experiment, which is sloping upwards towards the center slightly. The sloping bottom models the change in the Coriolis force for different latitudes on a real planet (in other words, it helps us change from a spherical planet to a cylindrical experiment).
We find that two distinct types of flows occur. The time-averaged stream function for these two types of flows are shown below; these are streamfunctions taken from experimental data. For the two stream functions shown, the tank was rotating at 1.5 Hz and the pump was at a constant rate of 210 cm^3/s for the blocked flow (left) and 320 cm^3/s for the zonal flow (right).
The streamfunction is a way to represent a two-dimensional flow. Particles in the flow will follow the streamlines -- lines where the stream function is constant. Some of these lines are shown in the pictures below. Where the lines are close together, the flow is moving fast. These pictures are time-averaged streamfunctions so they show the general trends of the flow rather than the motions that a specific tracer particle would follow. The peaks of the mountains in both pictures are located at 6 o'clock and 12 o'clock. Left picture: blocked flow from experiment. Right picture: zonal flow from experiment.
For the blocked flow, the general flow is counter-clockwise, the same as the direction of rotation of the experiment. The four orange/red vortices rotate in the clockwise direction.
The picture above is taken from the book "Topics in Geophysical Fluid Dynamics: Atmospheric Dynamics, Dynamo Theory, and Climate Dynamics," by M. Ghil and S. Childress, Springer-Verlag (1987). It shows the monthly mean map of 500 mb geopotential heights for January 1963, and was provided by K. C. Mo. The contour intervals are 20 m. The lines in this indicate flow direction, similar to stream function contours.
The two regions marked H in the diagram are anticyclones, where the air moves in a clockwise direction (located at 4 o'clock and 7 o'clock). These correspond to the two dark red anticyclones in the blocking picture from the experiment (at about 4 o'clock and 10 o'clock). The general flow is counter-clockwise, which is the direction the planet rotates.