Improvements to profiling float technology in support of Equatorial Pacific biogeochemical studies

PIs: Stephen C. Riser, School of Oceanography, University of Washington; Kenneth Johnson, Monterey Bay Aquarium Research Institute (MBARI); and Thomas Mitchell, SeaBird Scientific

Description

In recent years additional sensors have been added to the basic CTD observations made on Argo floats. Many floats have O2, NO3, and chlorophyll and backscatter (FLBB) sensors that have performed well throughout the world ocean (Johnson et al., 2017). More recently, pH sensors have been added to a number of floats, providing an unprecedented view of the subsurface oceanic carbon cycle and biological pump (the Deep Durafet sensor that has been installed on profiling floats was an awardee in the 2015 XPrize competition). Many of these floats have been deployed as part of the NSF-sponsored Southern Ocean Carbon and Climate Observations and Modeling (SOCCOM) program.

The success of BGC profiling floats has led to ambitious plans for their future use. The National Science Foundation has recently funded a $53 million program, GO-BGC, that will deploy 500 BGC floats in the world ocean as part of an anticipated array of 1000 BGC floats. This leads immediately to a major question: what will be the source of these floats? Most of the floats successfully deployed in SOCCOM have been produced in the float laboratory at the University of Washington, fabricated from APEX components manufactured by Teledyne/Webb Research Corporation. Some of the sensors used on the floats (O2, FLBB) have been purchased commercially, with the others (NO3, pH) produced in-house at MBARI. The BGC floats produced by the UW/MBARI consortium have performed well, with over 50% lasting at least 5 years. However, the UW float lab’s present production of ∼40 floats per year cannot be substantially increased, nor can the production of NO3 and pH sensors at MBARI (licensed versions of NO3 and pH are also produced commercially by Sea-Bird). Thus, for global programs involving BGC-equipped floats, where 200-250 floats will eventually need to be deployed each year in order to sustain scientific coverage, another source for reliable floats and sensors (likely a wholly commercial source) must be found. We have identified SeaBird Scientific as the most promising source of these commercial BGC floats, and this project, funded as part of the National Ocean Partnership Program (NOPP), is designed to improve the quality of the commercially available SeaBird floats over the next few year so that they will be available to the scientific community when large quantities of BGC floats will be needed in a few years.

This is a 3-year project that began late in 2019. The specific goals of the project are (1) to produce a modified version of the commercially-available SeaBird dissolved O2 sensor that can be collect air calibration samples and be used on both SeaBird and other commercial float varieties; (2) design a fabricate a new, larger hull for the present version of the BGC-Navis, SeaBird’s commercial BGC float model (this hull will allow for more buoyancy and also allow the float to carry more battery packs, for longer life); (3) improve the quality of the SeaBird Deep Durafet pH sensor, which is produced at SeaBird via license from MBARI; and (4) add a downwelling radiometer sensor to BGCNavis
float.

The plan for this work is to carry out tasks (1) and (3) during years 1 and 2 of the grant and tasks (2) and (4) during years 2 and 3 of grant. At the present time task 1 is well underway, with a prototype of an improved O2 sensor designed, built, and deployed on a profiling float near Hawaii. Task 3 is also in progress, with laboratory tests at SeaBird underway. Task 2 is now in a design phase, with several hull designs being evaluated and field tests of a prototype due by the end of 2021. Task 4 is also underway, with tests of the SeaBird OCR504 downwelling radiometer sensor planned soon on UW-produced APEX floats and the addition of the sensor to BGC-Navis floats later in 2021.

Prototype SeaBird Navis float with 3 dissolved O2 sensors, prior to being deployed at Hawaii. We are collecting data from all 3 sensors and comparing the new prototype, the SBE83, to the performance of the known and highly used SBE63 and Aanderaa 4330 sensors. The float was launched in early November of 2020 near the island of Hawaii.

The trajectory of the float through early March of 2021.

The trajectory of the float through early March of 2021.

Accomplishments

SeaBird Scientific has provided a commercial O2 sensor, the SBE63, for a number of years. This sensor has been evaluated by the oceanographic community and found to be as good as any sensor available for long-term use on profiling floats. To date, the sensor has been used only on SeaBird Navis floats. The SBE63 has the drawback that it must be mounted inside the pumped fluid circuit of the CTD unit; this allows the sensor to sample the same water as seen by the temperature and conductivity sensors, but because of the mounting it is not possible to collect O2 samples in air while the float is transmitting data on the sea surface. It has been shown that collecting in-air measurements in this manner and using them to correct instrument drift in the O2 sensor is vital to producing data
of the quality required in biogeochemical studies of the ocean. A number of papers have been published showing how to do this, using known values for atmospheric O2 concentration estimated from NCEP observations wherever the float surfaces.

In order to make such in-air measurements, the SBE83 has to be moved outside of the CTD fluid circuit and mounted on the float endcap. SeaBird has now made this change in a prototype sensor, known as the SBE83, and the new sensor has been mounted on a specially-designed prototype float for testing in the ocean. The idea was to put 3 separate O2 sensors on a single prototype float: the first, the new SBE83; the second, a SBE63 mounted inside the fluid circuit; and third, an Aanderaa 4330 Optode sensor mounted externally near the SBE83. The 4330 and SBE63 are known to work well, so the idea was to test the new SBE83 against them in order to examine the stability of the SBE83 over time. This test float was deployed near Hawaii in November of 2020 and has been collecting profiles at intervals from 2 to 5 days since then, as shown in the accompanying figure.

Laboratory tests of an improved pH sensor, compensated for pressure, are now underway, with the first deployment on prototype floats planned for later this year. Analogous deployments will be done with the SeaBird OCR504 radiometer sensor. The design work for an improved hull is now underway. In general, this project is on track to meet its goals over its 3-year timeline.

An additional part of this program is to derive a biogeochemical climatology for the Equatorial Pacific that can be used to help derive parameters such as pCO2, total alkalinity, and DIC from the standard biogeochemical parameters measured by profiling floats. Dr. Brendan Carter and his colleagues at NOAA have made considerable progress on this problem and are preparing a paper based on their results.

Lessons Learned

SeaBird has successfully engineered an improved O2 sensor that will likely be used extensively on commercially available BGC profiling floats in the future. The availability of these floats is crucial to large programs such as GOBGC. It is our hope that the other goals of this NOPP project will be met in a timely manner; the design and testing of a new float hull is the most difficult job remaining and is underway. The partnership of UW, MBARI, and SeaBird has generally worked well. We have had weekly teleconferences and have communicated well, even in the COVID era.

Publications

Carter, B. and coauthors (2021) Locally interpolated regressions version 3: updated algorithms for global seawater property estimation. In preparation.

Data

At this time, only one float, an engineering prototype, has been deployed, as shown in the figure above. These data are not yet being made available on public websites, as they consist of very preliminary observations and the sensor design is still undergoing refinement. When floats with the final design are deployed, sometime in the fall of 2021, the data will be made available through the Argo data portal.

The initial results of the prototype deployment show that the SBE83 behaves well and is well-correlated with the measurements from the Aanderaa 4330. Both sensors show measure low of the air O2 value estimated by NCEP by 1-2%, not atypical of float-based O2 sensors. During the first few months the float surfaced at all times of the day, leading to some variations in the measured O2 values, since all O2 sensors of this type are subject to contamination by sunlight. This was one factor that this deployment was designed to test.

The initial results of the prototype deployment show that the SBE83 behaves well and is well-correlated with the measurements from the Aanderaa 4330. Both sensors show measure low of the air O2 value estimated by NCEP by 1-2%, not atypical of float-based O2 sensors. During the first few months the float surfaced at all times of the day, leading to some variations in the measured O2 values, since all O2 sensors of this type are subject to contamination by sunlight. This was one factor that this deployment was designed to test.

Initial results of the comparison of the 4330 and SBE83 sensors (both mounted externally) show that that the SBE83 is more precise than the 4330; this can be seen by noting that the Gain in the SBE83 is consistently lower than the Gain of the 4330. Here the Gain is defined as the scale factor necessary to make the sensor agree with the NCEP atmospheric O2 results. As can be seen, the Gain for the SBE83 is typically 1% lower than that for the 4330. Both sensors exhibit a Gain dependent on the time of day when the surface data are collected.

Initial results of the comparison of the 4330 and SBE83 sensors (both mounted externally) show that that the SBE83 is more precise than the 4330; this can be seen by noting that the Gain in the SBE83 is consistently lower than the Gain of the 4330. Here the Gain is defined as the scale factor necessary to make the sensor agree with the NCEP atmospheric O2 results. As can be seen, the Gain for the SBE83 is typically 1% lower than that for the 4330. Both sensors exhibit a Gain dependent on the time of day when the surface data are collected.