Observing the Biological Carbon Pump with Optical and Imaging Sensors
Meg L. Estapa and Emmanuel Boss
The biological carbon pump starts with the fixation of CO2 into organic matter by phytoplankton in the surface ocean (Volk and Hoeffert, 1985). Most of this material is cycled through the food web and respired back to CO2, but a portion is transferred into deep water, resulting in a net flux of carbon from the atmosphere into the deep ocean that is globally estimated at 5 to >12 PgC per year (Boyd and Trull, 2007; Henson et al., 2011; Siegel et al., 2014). The estimate has a large uncertainty because observations of the vertical, biological carbon flux in the global ocean are scarce, particularly in the upper 1,000 m where rapid flux attenuation occurs. Processes that contribute to the biological carbon pump include the direct sinking of phytoplankton cells, aggregates, and zooplankton fecal matter; the subduction of suspended particulate organic carbon (POC) and dissolved organic carbon (DOC), and active transport by vertically migrating zooplankton (Ducklow et al., 2001; Siegel et al., 2014). Open questions include identification of specific biological mechanisms that drive carbon export and how these vary spatially and temporally; the interaction between physical processes and export of biologically derived carbon; the importance of particle size and density (including content of ballast minerals such as biogenic silica and particulate inorganic carbon) to export efficiency; and the development of process-based rather than statistical models that will enable us to predict future behavior of the biological pump under changing climate conditions.
Biological carbon fluxes can change on time scales of days to weeks, and can be spatially patchy on scales smaller than 10 km (Estapa et al., 2015). Measurements made in a Lagrangian frame aboard autonomous platforms have therefore featured heavily in key studies since the last ALPS workshop in 2003 (Rudnick and Perry, 2003). A review chapter by Stemmann et al. (2012) broadly summarizes developments in biogeochemical sensors on autonomous platforms; here we focus specifically on progress in measurements of the biological pump.
Measurement of sinking or subducting particle fluxes requires a sensor-platform combination that can detect the small flux of sinking particles against the much larger background stock of suspended particles. Typically, particle detection is carried out with bulk bio-optical sensors (e.g., backscatter, turbidity, fluorescence, beam attenuation) or imaging sensors (e.g., cameras, Laser Optical Particle Counter [LOPC], P-Cam). The more mature, bulk bio-optical sensors are easily integrated onto standard profiling float and glider platforms, have low power requirements and data volumes, but are not always specific to the sinking fraction of particles; imaging sensors are still maturing and have higher power and data requirements but provide information on particle size and transparency and can better elucidate specific mechanisms of the biological pump. However, while particles carrying carbon into the deep ocean have been observed to range from 10 μm (Durkin et al., 2015) all the way up to several centimeters (e.g., Bochdansky et al., 2016), no single imaging or particle counting sensor covers this entire size range. Another issue is that most sensor optical sampling volumes are too small to capture some of the largest, rarest particles. Finally, the present lack of a sensor for DOC that is suitable for deployment on autonomous platforms restricts carbon flux measurements to the particle-mediated export pathways listed above.
Sensor-platform combinations for measuring sinking particle flux have tended to fall into two categories: (1) those that physically collect sinking particles, either temporarily for imaging, or for sample return to a ship, and (2) those that repeatedly collect optical or image profiles of large (assumed sinking) particles in the water column and then use a deduced or assumed particle sinking rate to derive fluxes. Both approaches have advantages and drawbacks that are detailed in the following section, which covers significant developments since 2003.
Advances Since 2003
Direct Particle Interception Techniques
Semi-Autonomous Sediment Traps. The collection of sinking, upper-ocean particle samples from an untethered, quasi-Lagrangian platform is advantageous even disregarding the other benefits of platform autonomy, because of biases from hydrodynamic effects associated with surface tethered sediment traps (Buesseler et al., 2007). Standard profiling floats have been modified independently by two groups to carry sediment traps for ship-supported sample collection. Both designs—the Neutrally-Buoyant Sediment Trap (NBST; based around a SOLO float and designed at Woods Hole Oceanographic Institution; Valdes and Price, 2000) and PELAGRA (based around an APEX float and designed at the National Oceanography Centre, Southampton; Lampitt et al., 2008), have featured prominently in recent biological carbon pump process studies. Both platforms have more recently been modified to carry bulk optical sensors and camera systems, which are described separately in sections below. In this respect they serve as an important intercalibration link between completely autonomous, sensor-based approaches and traditional sediment trap and 234Th tracer-based observations that are still the primary tools of the longest-running time-series programs (Estapa et al., 2017).
Transmissometer as “Optical Sediment Trap”. The first truly autonomous measurements of sinking carbon flux were made by using a vertically mounted transmissometer aboard a profiling float to physically collect sinking particles on the upward-looking optical window during the drift phase of the float’s mission cycle (Bishop et al., 2004; Bishop and Wood, 2009; Estapa et al., 2013, 2017; Figure 1). This method has the advantages of not requiring a particle sinking-rate assumption to be made, and utilizing commercially available, mature sensor technology with relatively low power and data transmission requirements. It is best suited to use in areas where calibration samples (for instance, versus a neutrally buoyant sediment trap) can be collected, and in the upper few hundred meters of the water column where ambient turbulence is sufficient to carry sinking particles into the transmissometer sensing volume (Estapa et al., 2017).
Imaging Sediment Traps. Building further upon the concept of optical detection of physically intercepted, sinking particles is a class of new devices that are best described as imaging sediment traps. Observations from one such device, the Carbon Flux Explorer (CFE), are presented by Bishop and Wood (2009) and Bishop et al. (2016), and illustrate the wealth of information about sinking particle size and origin that is gained through use of imaging sensors. The CFE consists of an imaging trap mounted aboard a profiling SOLO float; power and data are self-contained but at the time of this writing, physical platform collection is required to retrieve data post-deployment.
Indirect Techniques Requiring Estimates of Settling Velocity
Optical Spike Flux. Profiles of bulk optical properties collected at a fast sampling rate often contain many spikes, which have for some time been interpreted as arising from large particles passing through the optical detection volume (Bishop, 1999; Gardner, 2000; Bishop and Wood, 2008). By filtering optical profiles of fluorescence and backscattering to separate the baseline signal from this “spike” signal, Briggs et al. (2011, 2013) were able to estimate the relative vertical distribution of large particles from autonomous float and glider observations during the 2008 North Atlantic Bloom Experiment. In that study, the export flux of large aggregates occurred as distinct pulses during the study period and so the increasing penetration depth of the large particle spikes was used to deduce the particle sinking rate and estimate the particulate carbon flux. This method also has the advantage of using only low power, commercially mature sensors, although some means of estimating the particle sinking rate and converting the bulk optical properties to carbon are required. The profile repeat interval and the sensor sampling rate must also be relatively fast in order to implement this method.
Fluxes Derived from Changes in the Vertical Distribution of Particles Over Time. Optical or imaging sensors aboard autonomous profiling platforms can be used to estimate the change in the vertical distribution of particles over time down to some reference depth, and therefore derive a flux estimate. In this method, the particle sinking speed must again be derived from observations, and the water column must not experience appreciable shear during the measurement period. The optical or imaging sensor properties determine the type(s) of sinking particles that can be observed. Recent papers illustrate different applications of the method. Dall’Olmo and Mork (2014) and Dall’Olmo et al. (2016) utilized bulk optical backscattering sensors to show how the spring/summer shoaling of the mixed layer in part drives the seasonal export cycle (the “mixed layer pump” described by Gardner et al.,1995). As optical backscattering is mainly sensitive to particles <20 μm, the authors surmised that the observed flux signal was due to small, sinking particles or to large particles disintegrating at depth. Jackson et al. (2015) used the SOLOPC sensor/platform combination in a similar manner to derive sinking rates of larger particles sensed by the LOPC, which counts particles in the water column using a sheet of adjacent laser beams and allows discrimination of particle sizes ranging from 90 μm to 3,500 μm.
Fluxes Derived from Particle Size Distributions and Modeled Settling Velocities. Imaging and particle sizing sensors capable of resolving water column particle size distributions can be used to estimate carbon fluxes if an accurate, modeled particle settling velocity spectrum is available. Most examples in the literature that estimate particulate carbon fluxes using this type of technique rely on ship-based image profiles of a device such as the Underwater Video Profiler (e.g., Guidi et al., 2007, 2016; McDonnell and Buesseler, 2010, 2012) or holographic sensors (such as Sequoia Scientific’s LISST-HOLO or the 4Deep holographic microscope). One of the first applications used particle size distributions from SOLOPC profiles and settling velocities predicted via Stokes’ Law to estimate carbon fluxes due to particles >90 μm in diameter (Jackson and Checkley, 2011). Ongoing efforts to adapt and integrate imaging sensors onto profiling floats also include onboard image processing to allow fully autonomous operations. These include the GUARD1 system (Corgnati et al., 2016) and the Octopus sensor (a miniaturized, low-power version of the Underwater Vision Profiler), which is being integrated into the NKE float platform (Mar Picheral, pers. comm.). The main drawbacks of these particle imaging methods are the requirement for an accurate estimate of the particle settling velocity size spectrum, and the current lack of an imaging sensor capable of resolving the entire, relevant particle size range (from 10 μm up to tens of millimeters).
The benefits of making particle flux measurements from autonomous platforms will include broader spatiotemporal coverage, better links to satellite remote-sensing observations, and higher-resolution measurements of a patchy set of processes. However, such measurements are not yet widespread. One of the main challenges is that bulk optical properties and particle imagery must be translated into geochemical (usually carbon) flux units, and the accuracy of flux estimates is only as good as the calibration. Sinking particles range through six orders of magnitude in size, which currently requires a multi-sensor approach; particles responsible for carbon export also have a broad range in composition, fractal dimension, and pigmentation. These factors will continue to make the site-specific calibration of particulate flux sensors a requirement in studies going forward. Further complicating the need for calibration is the lack of a standard method for direct measurements of carbon flux given the issues with many types of sediment traps (Buesseler et al., 2007), and the three-dimensional, time-dependent nature of 234Th derived measurements of flux (e.g., Buesseler et al., 2009).
Sensor developments that would improve autonomous observations of biological pump processes include a sensor for dissolved organic carbon, and a particle imaging sensor with a large sensing volume (to detect rare, large particles) and that is capable of resolving the full size range of sinking particles. In general, imaging sensors will require greater capabilities for built-in, onboard data reduction so that parameterized observations can be transmitted via satellite, minimizing the risk of data loss in the event a platform cannot be recovered.
The incorporation of all but the simplest particle flux observational techniques into large-scale autonomous sample programs such as Bio-Argo is currently precluded by the available power and communications budgets of float platforms. At present, the only method described above that could be easily managed within the proposed US Biogeochemical Argo framework is the derivation of flux from changes in the vertical distribution of particles with time, assuming particle distribution is measured with a low power, commercially available sensor such as a backscattering sensor. Binned profiles every one to two days to 1,000 m would be sufficient for this technique. Utilization of the “optical spike flux” method would require sampling at very high vertical resolution, and implementation of the “optical sediment trap” technique would require measurements to be made during the “drift” phase at a depth shallower than 1,000 m. Both of these methods could be implemented on a large scale (perhaps on a subset of floats in a globally distributed program) using currently available platforms and technology. All of the other methods described above require the collection and transmission of large amounts of image data using sensors with high power requirements and are thus better suited at present to medium-length deployments or ship-supported process studies.
Bishop, J.K.B. 1999. Transmissometer measurement of POC. Deep Sea Research Part I 46:353–369, https://doi.org/10.1016/S0967-0637(98)00069-7.
Bishop, J.K.B., M.B. Fong, and T.J. Wood. 2016. Robotic observations of high wintertime carbon export in California coastal waters. Biogeosciences 13:3,109–3,129, https://doi.org/10.5194/bg-13-3109-2016.
Bishop, J.K.B., and T.J. Wood. 2008. Particulate matter chemistry and dynamics in the twilight zone at VERTIGO ALOHA and K2 sites. Deep Sea Research Part I 55:1,684–1,706, https://doi.org/10.1016/j.dsr.2008.07.012.
Bishop, J.K.B., and T.J. Wood. 2009. Year-round observations of carbon biomass and flux variability in the Southern Ocean. Global Biogeochemical Cycles 23, GB2019, https://doi.org/10.1029/2008GB003206.
Bishop, J.K.B., T.J. Wood, R.A. Davis, and J.T. Sherman. 2004. Robotic observations of enhanced carbon biomass and export at 55°S during SOFeX. Science 304:417–420, https://doi.org/10.1126/science.1087717.
Bochdansky, A.B., M.S. Clouse, and G.J. Herndl. 2016. Dragon kings of the deep sea: Marine particles deviate markedly from the common number-size spectrum. Scientific Reports 6:22633, https://doi.org/10.1038/srep22633.
Boyd, P.W., and T.W. Trull. 2007. Understanding the export of biogenic particles in oceanic waters: Is there consensus? Progress in Oceanography 72:276–312, https://doi.org/10.1016/j.pocean.2006.10.007.
Briggs, N., M.J. Perry, I. Cetinić, C. Lee. E. D’Asaro, A.M. Gray, and E. Rehm. 2011. High-resolution observations of aggregate flux during a sub-polar North Atlantic spring bloom. Deep Sea Research Part I 58:1,031–1,039, https://doi.org/10.1016/j.dsr.2011.07.007.
Briggs, N.T., W.H. Slade, E. Boss, and M.J. Perry. 2013. Method for estimating mean particle size from high-frequency fluctuations in beam attenuation or scattering measurements. Applied Optics 52:6,710–6,725, https://doi.org/10.1364/AO.52.006710.
Buesseler, K.O., A.N. Antia, M. Chen, S.W. Fowler, W.D. Gardner, Ö. Gustafsson, K. Harada, A.F. Michaels, M. Rutgers v. d. Loeff, M. Sarin, and others. 2007. An assessment of the use of sediment traps for estimating upper ocean particle fluxes. Journal of Marine Research 65:345–416.
Buesseler, K.O., S. Pike, K. Maiti, C.H. Lamborg, D.A. Siegel, and T.W. Trull. 2009. Thorium-234 as a tracer of spatial, temporal and vertical variability in particle flux in the North Pacific. Deep Sea Research Part I 56:1,143–1,167, https://doi.org/10.1016/j.dsr.2009.04.001.
Corgnati, L., S. Marini, L. Mazzei, E. Ottaviani, S. Aliani, A. Conversi, and A. Griffa. 2016. Looking inside the ocean: Toward an autonomous imaging system for monitoring gelatinous zooplankton. Sensors 16, 2124, https://doi.org/10.3390/s16122124.
Dall’Olmo, G., J. Dingle, L. Polimene, R.J.W. Brewin, and H. Claustre. 2016. Substantial energy input to the mesopelagic ecosystem from the seasonal mixed-layer pump. Nature Geoscience 9:820–823, https://doi.org/10.1038/ngeo2818.
Dall’Olmo, G., and K.A. Mork. 2014. Carbon export by small particles in the Norwegian Sea. Geophysical Research Letters 41:2,921–2,927, https://doi.org/10.1002/2014GL059244.
Ducklow, H.W., D.K. Steinberg, and K.O. Buesseler. 2001. Upper ocean carbon export and the biological pump. Oceanography 14(4):50–58, https://doi.org/10.5670/oceanog.2001.06.
Durkin, C.A., M.L. Estapa, and K.O. Buesseler. 2015. Observations of carbon export by small sinking particles in the upper mesopelagic. Marine Chemistry 175:72–81, https://doi.org/10.1016/j.marchem.2015.02.011.
Estapa, M.L., K. Buesseler, E. Boss, and G. Gerbi. 2013. Autonomous, high-resolution observations of particle flux in the oligotrophic ocean. Biogeosciences 10:5,517–5,531, https://doi.org/10.5194/bg-10-5517-2013.
Estapa, M., C. Durkin, K. Buesseler, R. Johnson, and M. Feen. 2017. Carbon flux from bio-optical profiling floats: Calibrating transmissometers for use as optical sediment traps. Deep Sea Research Part I 120:100–111, https://doi.org/10.1016/j.dsr.2016.12.003.
Estapa, M.L., D.A. Siegel, K.O. Buesseler, R.H.R. Stanley, M.W. Lomas, and N.B. Nelson. 2015. Decoupling of net community and export production on submesoscales in the Sargasso Sea. Global Biogeochemical Cycles 29:1,266–1,282, https://doi.org/10.1002/2014GB004913.
Gardner, W.D. 2000. Sediment trap sampling in surface waters. Pp. 240–281 in The Changing Ocean Carbon Cycle: A Midterm Synthesis of the Joint Global Ocean Flux Study. R.B. Hanson, H.W. Ducklow, and J.G. Field, eds, Cambridge University Press.
Gardner, W.D., S.P. Chung, M.J. Richardson, and I.D. Walsh. 1995. The oceanic mixed-layer pump. Deep Sea Research Part II 42:757–775, https://doi.org/10.1016/0967-0645(95)00037-Q.
Guidi, L., S. Chaffron, L. Bittner, D. Eveillard, A. Larhlimi, S. Roux, Y. Darzi, S. Audic, L. Berline, J.R. Brum, and others. 2016. Plankton networks driving carbon export in the oligotrophic ocean. Nature 532:465–470, https://doi.org/10.1038/nature16942.
Guidi, L., L. Stemmann, L. Legendre, M. Picheral, L. Prieur, and G. Gorsky. 2007. Vertical distribution of aggregates (>110 um) and mesoscale activity in the northeastern Atlantic: Effects on the deep vertical export of surface carbon. Limnology and Oceanography 52:7–18, https://doi.org/10.4319/lo.2007.52.1.0007.
Henson, S.A., R. Sanders, E. Madsen, P. J. Morris, F. Le Moigne, and G.D. Quartly. 2011. A reduced estimate of the strength of the ocean’s biological carbon pump. Geophysical Research Letters 38, L04606, https://doi.org/10.1029/2011GL046735.
Jackson, G.A., and D.M. Checkley Jr. 2011. Particle size distributions in the upper 100 m water column and their implications for animal feeding in the plankton. Deep Sea Research Part I 58:283–297, https://doi.org/10.1016/j.dsr.2010.12.008.
Jackson, G.A., D.M. Checkley Jr., and M. Dagg. 2015. Settling of particles in the upper 100 m of the ocean detected with autonomous profiling floats off California. Deep Sea Research Part I 99:75–86, https://doi.org/10.1016/j.dsr.2015.02.001.
Lampitt, R.S., B. Boorman, L. Browna, M. Lucas, I. Salter, R. Sanders, K. Saw, S. Seeyave, S.J. Thomalla, and R. Turnewitsch. 2008. Particle export from the euphotic zone: Estimates using a novel drifting sediment trap, 234Th and new production. Deep Sea Research Part I 55:1,484–1,502, https://doi.org/10.1016/j.dsr.2008.07.002.
McDonnell, A.M.P., and K.O. Buesseler. 2010. Variability in the average sinking velocity of marine particles. Limnology and Oceanography 55:2,085–2,096, https://doi.org/10.4319/lo.2010.55.5.2085.
McDonnell, A.M.P., and K.O. Buesseler. 2012. A new method for the estimation of sinking particle fluxes from measurements of the particle size distribution, average sinking velocity, and carbon content. Limnology and Oceanography Methods 10:329–346, https://doi.org/10.4319/lom.2012.10.329.
Rudnick, D.L., and M.J. Perry, eds. 2003. ALPS: Autonomous and Lagrangian Platforms and Sensors, Workshop Report. 64 pp, https://geo-prose.com/pdfs/alps_report.pdf.
Siegel, D.A., K.O. Buesseler, S.C. Doney, S.F. Sailley, M.J. Behrenfeld, and P.W. Boyd. 2014. Global assessment of ocean carbon export by combining satellite observations and food-web models. Global Biogeochemical Cycles 28:181–196, https://doi.org/10.1002/2013GB004743.
Stemmann, L., H. Claustre, and F. D’Ortenzio. 2012. Integrated observation system for pelagic ecosystems and biogeochemical cycles in the oceans. Pp. 261–278 in Sensors for Ecology: Towards Integrated Knowledge of Ecosystems. J.-F. Le Galliard, J.-M. Guarini, and F. Gaill, eds, Institut Écologie et environnement, CNRS, France.
Valdes, J.R., and J.F.A. Price. 2000. A neutrally buoyant, upper ocean sediment trap. Journal of Atmospheric and Oceanic Technology 17:62–68, https://doi.org/10.1175/1520-0426(2000)017<0062:ANBUOS>2.0.CO;2.
Volk, T., and M.I. Hoffert. 1985. Ocean carbon pumps: Analysis of relative strengths and efficiencies in ocean-driven atmospheric CO2 changes. Pp. 99–110 in The Carbon Cycle and Atmospheric CO: Natural Variations Archean to Present. E.T. Sundquist and W.S. Broecker, eds, American Geophysical Union, Washington, DC.
Meg L. Estapa, Skidmore College, Saratoga Springs, NY, USA, email@example.com
Emmanuel Boss, University of Maine, Orono, ME, USA, firstname.lastname@example.org