Following Fluorescein

At 3 am on July 1^st, the RRS Discovery was holding position over a subsea canyon that slices into the southern edge of Rockall Trough. The dye release team from Exeter was huddled around a computer in the lab watching data stream in and communicating over radio with the winch driver. Almost two kilometers beneath us, InkBot was suspended 10 m from the sea floor releasing its payload. If the data stream in the lab indicated that water density surrounding InkBot had increased, they asked the winch to come up into slightly lighter water. If the reading decreased, they moved InkBot down into slightly denser water. The goal was to release the dye within a narrow density range so that we could track it later. Within a few minutes, 200 liters of concentrated fluorescein dye dissolved into the dark, cold, salty water. All of the dye had been injected into water within a few hundredths of a degree of 3.7°C. This would be our starting point for the next few days of dye chasing.

Tracking the fluorescein dye would be done with a fluorometer. InkBot has one built into its frame, so we kept it within the dye cloud for about an hour measuring the dye concentration and following the cloud slowly up the canyon. The better tracking tool, though, would be the fluorometer on the FastCTD. This instrument and its winch were designed to profile quickly while the ship is moving. With it, we could follow the dye patch as it moved up and down the canyon with the tides sampling it from one side to the other and from top to bottom as it grew in size and lessened in concentration.

The FastCTD suspended from the winch before deployment

As soon as InkBot was back onboard, the Scripps team was lowering the FastCTD from the boom on the back deck. We had analyzed some near-bottom current velocity data a few days before so we knew that the tide would move the dye up the canyon first and reverse six hours later. What we didn’t know was how much the dye would spread out or how fast.

Cartoon of dye moving up and down the canyon with the tide, growing in height and width over time

We wanted to be able to make fast decisions about where to search. Every time we chose a new direction to sample in, we would need to call up the bridge with updated waypoints of latitude and longitude. In an effort to make this communication step as seamless as possible, we designed a large grid of points along the axis of the canyon spaced 200 m apart. The grid points were lettered from west to east, A through S, and numbered from 0 to 12 in increments of 0.2, beginning at the deep part of the canyon and ending 12 km away where the water depth was 1000 m shallower. The bridge loaded all these waypoints into their navigation software and anytime we called up to say something like “head to L3.6, please!” we would be on our way within the minute. In grid-land, we had released the dye at J4.4, so once the FastCTD was in the water, we started moving in the direction of J12.0, up the canyon, and kept going until we found dye.

Left: Map of the canyon showing a green star at the dye release location, dark blue grid lines from 0 at the top to 12 at the bottom, our ship track in red and locations where we saw dye in green dots with sizes indicating relative dye concentration. Right: Prediction of tidal displacement of the water parcel into which we injected dye. Distances 0-12 km correspond to the same grid lines on the map. The red line shows our ship placement along the canyon of the axis over time and green dots show where we found dye, with the size correlated to the dye concentration detected at that time, as on the map.

It took about 3 hours. We first saw the dye about 2.5 km from the release site. It was close to where the tidal model predicted it should be and we noticed that it was still mostly contained in the 3.7 °C isotherm. The isotherm was domed upward as if that water parcel and the dye it contained had been squeezed by a flow convergence. By the time we reached the end of the dome, it was 7:00 am, approximately the time when the tidal prediction indicated the dye would be moving back down the slope. We turned around to chase it again.

First detection of dye by the FastCTD, colored in relative fluorescence units on a logarithmic scale. Temperature contours are plotted as labeled black lines.

A couple hours later, we caught it again. Dye was already seeping through into warmer waters; there were high concentrations at almost 4 degrees. The patch also looked a lot different than it did on the first pass: it had lost about half its height above the seafloor and it looked more elongated. Part of this was an illusion. During this transect, both the ship and the dye patch were moving in the same direction. We were traveling faster than it, but only slightly, so every time the FastCTD dove down to measure another profile, it was observing almost the same water mass as before and watching it evolve in its own reference frame.

Second detection of dye by the FastCTD, colored in relative fluorescence units on a logarithmic scale. Temperature contours are plotted as labeled black lines.

In total, we chased the dye for almost three days. By the end it was quite diffuse and difficult to identify the edges, so we spent the last 12 hours holding position in one spot collecting a nice time series. Now, there’s a lot of data to analyze and a lot of questions to answer. The dye in this experiment was a visual representative of bottom water in general. Does it allow us to see how bottom water mixes with the water above it? When, where, and how does that happen? Does it climb up the canyon walls or does mixing happen in the interior? Does it escape the canyon? Teasing apart these stories should keep us occupied for weeks and months to come. In the meantime, we’ve learned some things that will help inform the chemical tracer release that’s coming up soon. Namely, the bottom is not a quiescent place. Things change fast.