...
The evolution of the temperature observed at each sensor over time may be plotted.
Analysis of the Tank Experiment
What is perhaps the most fascinating finding from this plot is how the innermost sensor on the can is much warmer than either of the two sensors on the bottom. This stands to reason, however; as the warm water on the surface reaches the canister containing the ice, it only begins to cool. It only reaches its maximum coolness at the bottom of the cylinder, at which point it flows outward to reach the "second" and subsequently the "penultimate" sensor. As a result, the "second" sensor is the coolest, and the "penultimate" sensor appears second coolest. The outermost sensor barely cools at all, outside of the cooling due to radiation out of the tank.
Another notable pattern in the data is the periodic variation in the temperature observed at most sensors. This might be a result of an off-center alignment of the tank. This would create an imbalance and seiching of the water in the tank. As a result, each sensor would periodically move through warmer and colder water having the same period as the rotation, which this variation appears to have.
It is possible to numerically verify the shear created by the Hadley cell by applying the thermal "wind" equation for water which varies in temperature,
Where α is the thermal coefficient of expansion of water and Ω is the rotation rate of the tank. By taking the temperature difference between the two bottom sensors (approximately 2K over 8cm), we may obtain a value for ΔT, which may be substituted into the thermal wind equation to calculate the expected shear, 2.45 m/s per meter. As the depth of the water in the tank was 9 cm, the speed of the rotation of the dots observed on the top of the water would be expected to be approximately 0.22 m/s.
Several particles were tracked on the surface of the tank. These tracks can be used to plot the azimuthal velocity of the dots relative to the distance from the center of rotation (as averaged over a one second period around a given point in time).
What is surprising about this plot is the order of magnitude difference of the observed velocities (about 0.5–1 cm/s) and the expected velocity of 22 cm/s. Although we did not measure temperatures throughout the water column, the difference of the bottom temperatures from the side-temperatures suggests that the cool temperatures are only observable in a thin layer very close to the bottom of the water column. Assuming a perfect adherence to the thermal wind equation, the height of the cool water would then be only 4 mm.