FSU Physics Science Tallahassee

Linking conceptual models with laboratory experiments

A recent project is leading my adviser and I to design and build a large, acrylic tank which will be able to hold several hundred litres of seawater. Since a tank like this will take both time and money to build, it is logical to develop multiple experiments to run in it. These experiments will also help inform the design process and ensure that the tank is as useful as possible. I would like to take a few minutes and share with you an experiment I’ve been thinking about for a couple days and about how it developed.


In our lab Mike and I focus on understanding how much organic carbon manages to escape from the surface ocean and reach deeper depths (e.g. 100-1000 meters), since once down there the carbon–no matter what biological processes go on–will stay down there for decades or longer. This carbon is termed “export production” and it is a small fraction of the total carbon cycling within the surface ocean. Biological systems are nothing if not efficient. To study this export, the oceanographic community has developed a number of tools and techniques over the years, yet we will only discuss one of them today.

If you imagine that this export production is sinking through the water column, you may be tempted to conceptualize export as the “raining down of organic carbon”. So in order to measure this fraction of carbon you can setup a “rain gauge” in the ocean to collect all the sinking particles. You can then determine how much carbon is in the trap and divide that by the effective area of the opening and the time the trap was out to get the mass of carbon sinking per unit area per unit time (e.g. mg C m^{-2} d^{-1}).


Since conceptualized models are ALWAYS over simplifications, our mental picture here is missing a few key considerations. First, a rain gauge works since we assume the individual rain drops are both randomly placed (in space and time) and that they fall much quicker than the medium they travel through. While these conditions are generally true, there are important cases where it is invalid (such as in gale force winds). Both of these assumptions are generally accepted in the water column as well, but the room between being a valid assumption and an edge case is much narrower and more grey. Large aggregates can brake the randomness assumption while currents and the high viscosity of the water can violate the relative speed requirement.

Second, unlike a stationary rain gauge, the sediment trap’s I work with are floating at a particular depth in the ocean (say 100 metres). The floating, drifting nature of these traps can be a benefit since we don’t need to worry so much about the horizontal velocity of the water (as mentioned above) since the trap is moving horizontally at the same rate. Vertical velocities are different and therein lies an issue. Water masses in the surface can become more dense (e.g. cooling, evaporation) and sink. But this sinking will actually move particles past a sediment trap without allowing us to collect any of them.

While not instantly intuitive, the explanation is pretty simple. Water is in-compressible and with a fixed volume in the trap itself, the subducting water parcel simply flows around the trap, just like adding salt water to an already saltwater filled up. You can pour all the salt water you want into the overflowing cup, it will never collect any more salt than it began with.

Experiment 0

The first experiment we’ll be running, with the help from an undergraduate student, will be to verify that our sediment traps do not collect subducting particles by adding non-sinking particles into the tank and then flow the water past the trap. We can conduct this simple test over a variety of flow conditions including speed, particle concentration, and even sediment trap angles. This will allow us to rigorously determine when our assumption is valid and when it is violated. As the title suggests, I would also like to use this tank as a platform for comparing computational fluid dynamic (CFD) models. Perhaps, I could model up a virtual replica and see if the CDF model recreates the observations (and under what conditions it does/doesn’t). It would certainly be useful to have the pair set (tank and model) for future studies.

As a secondary benefit, it will give us hands-on experience with working with the tank, the particles (cyanobacteria) and the associated measurements for designing and implementing future experiments.

The Tank

The tank we’ve now built is approximately 6 ft x 2ft x 2ft (h,w,l) and holds arround 600 L of water when completely full. Also note the two optically clear panels make the front and back. Hmm, that gives me a few ideas.


Back To Top