I recall a similar effect being observed in the experiments of Hagen [1], Couette [2], and Reynolds [3] on pipe friction.
Around the laminar/turbulent transition, they noticed that a pipe with the inlet held at a roughly constant head would discharge at an oscillating rate. Variations on the exact mechanism exist, but they tend to go as follows:
1. The initially laminar flow accelerates to a Reynolds number (Re, which is effectively a unitless measure of speed) slightly greater than the critical value Re_c.
2. Upstream disturbances are no longer damped out, and the onset of turbulence occurs near the entrance to the tube. This turbulence is then advected downstream.
3. As the turbulent plug grows, so does the hydraulic impedance. Eventually, the flow rate decreases to the point that Re goes below Re_c and no more turbulence is generated.
4. The flow rate is now constant and the turbulent plug is carried downstream. Once the plug exits the tube, the hydraulic impedance drops, causing the flow rate to rise again and the cycle repeats.
There are two import considerations to be made in order to determine applicability to our scenario.
The first regards the flow conditions in the cave. For a passage of hydraulic diameter 1m, the flow velocity corresponding to a transitional Reynolds number (somewhere in the low thousands) is a few centimetres per second, which seems reasonable.
The second regards the apparent complete flow reversal (as opposed to mere modulation), which is not explained by the mechanism above. This is the interesting bit from a digging perspective as it allows for myriad speculation to be made about the interior of the cave.
One configuration which may be consistent with the field observations is as follows. Consider a simple sink-passage-resurgence system displaying classic summertime stack effect behaviour. Now suppose that a constricted passage (a useful modelling assumption as it prevents it from influencing the main flow) connects from the dig site to the main passage. Suppose that the main passage is of appropriate dimensions to induce the above affects, and that during laminar flow the pressure at the intersection is lower than that at the dig (causing a slight in-draught).
As the turbulent plug grows, the flow resistance downstream of the intersection increases, so too increasing the pressure at the intersection until it rises above the dig site pressure and flow reverses, causing an out-draught.
This is certainly not the only possible explanation, but at least to me it seems at least plausible. What do people think?
1. Hagen, G., On the Influence of Temperature on the Movement of Water through Pipes (German), 1854.
2. Couette, M., Investigations on the Friction of Fluids (French), 1890.
3. Reynolds, O., An Experimental Investigation of the Circumstances Which Determine whether the Motion of Water Will Be Direct or Sinuous, and of the Law of Resistance in Parallel Channels, 1883.
