Testing SOCCOM floats and sensors

June 15, 2015

We’ve had a busy week here at MBARI. The focus has been test deployments of two profiling floats offshore of Monterey Bay with a film crew from Climate Central (Ted Blanco and Greta Shum) recording the whole process, including drone flights during the float recovery.
Fig. 1 Greta Shum ready to go deploy floats on the R/V Paragon.

One of the keys to a successful SOCCOM program is reliable float and sensor operation. And that is a challenge. The SOCCOM network is a robotic system that must operate for years after deployment with no direct, human interaction. If there is a bug in the software (an array that overflows after a few profiles), or the wrong version of firmware code loaded, or an electronic design that fails relatively soon after deployment, then we are in deep trouble. And in the history of profiling floats, all of those things have happened. What a disaster it would be to deploy a year’s worth of floats, only to see them fail due to an avoidable problem. The only solution to that is to test, test, and test, and then do some more testing.
Fig. 2 An Apex float end cap showing the DuraFET pH sensor (lower grey cylinder) and ISUS nitrate optics (upper grey cylinder) and the ISUS light source (copper rectangle) and photodiode array (far right blue box). A yellow handheld multimeter for scale.

And testing has been our goal this week. The micro-controller used for both the ISUS nitrate sensor and the DuraFET pH sensor has gone out of production and Luke Coletti has designed a new, dual-controller that will run both sensors (and save some money since we only need one controller). But we don’t just put that new controller in floats and start deploying. First, the controller and sensors were tested extensively at MBARI. This identified a few problems on the board that were corrected, more testing was done and we were finally satisfied.

A set of electronic boards and sensors were then sent to the University of Washington where they were assembled into a float and it was placed in a float simulator. The simulator allows the whole system to be exercised through a variety of conditions. This includes the equivalent of years of profiles, but also tests designed to stress the system. What happens to the interaction between the float and sensor if the float rises faster than the sensor can sample, what happens if float stays at one depth, what happens when power sags, do all the commands to the sensor operate as the float controller expects. Early tests identified one problem that could leave the sensor microcontroller in an undefined state if sensor power was cycled rapidly. This necessitated a modification to the controller board. But that also required all of the testing to be restarted. The revised controller successfully passed through all of these conditions after several months of testing and several simulated years of float operation. Were we done. No.

The next step in testing then involved building two complete floats and sensors, ballasting and pressure testing the whole assemblies, and then shipping to MBARI, for a test deployment in the ocean. Dana Swift from UW came down with the floats to oversee the process. This is what we were doing this week.

The sensors were first tested in the 1.5 million liter test tank here at MBARI, where Dana, Hans Jannasch and Luke verified that they were operating as expected. Ginger Elrod and Carole Sakamoto verified that the pH and nitrate sensors were returning acceptable measurements when compared to their observations (pH =8.035 and nitrate = 18.4 µM in the tank). Float and sensor operations all looked good.

Fig. 4 The Paragon loaded up with the testing crew (left to right; Greta, Hans Jannasch, Ted Blanco, Dana Swift and Ken J.) and floats, and ready for sea. Photo credit Ted Blanco.

Then we loaded the two floats on the R/V Paragon with Gene Massion at the helm and took them to sea. The floats were deployed in 1200 m water depth, about 20 kilometers west of Moss Landing. Of course, things never quite go to plan for an exercise like this. We’d hoped to go to sea on Wed., June 3. But the weather didn’t cooperate. A small front came through with 10 m/s (20 knot) winds and seas up to 3 m on Wednesday at NOAA buoy 46042. That’s pretty rough for a 10 m long boat. The forecast was for moderating winds on Thursday, so we delayed. Naturally, the weather didn’t moderate and we went to sea on Thursday with seas up to 3.5 m at the NOAA buoy. Despite the weather, the deployments were both successful. The floats executed 7 vertical profiles down to 1100 m depth, and all data was transmitted to shore (look for floats 9265MtyBay and 9274MtyBay on FloatViz). The weather finally did moderate and we had a beautiful trip on Saturday to recover floats (successful). And we saw killer whales, a pod of 12 humpback whales, albatross and lots of otters and sea lions.
Fig. 5 The Paragon backing up to float 9274. The next land west is Japan. Those floats are hard to see. Photo credit Ted Blanco.

Is testing done. No. One of the floats (9274) will now be shipped to Hawaii and deployed on a HOT cruise, which occur about monthly. That will give us 5 or 6 more months of float time in the ocean to finally verify that there are no unexpected problems before we start deploying this version of the float in the Southern Ocean. During this time, all the hardware and software versions are locked and no changes are allowed so no problems are inadvertently introduced. If a change is made, all the testing restarts.

That’s what it takes to develop a robust Southern Ocean observing program. Check back later for the Climate Central video story of sensor testing.