oBseRVINg BIogeoChemICal CyCles at gloBal sCales WIth PRoFIlINg Floats aNd glIdeRs PRosPeCts FoR a gloBal aRRay

Chemical and biological sensor technologies have advanced rapidly in the past five years. Sensors that require low power and operate for multiple years are now available for oxygen, nitrate, and a variety of bio-optical properties that serve as proxies for important components of the carbon cycle (e.g., particulate organic carbon). These sensors have all been deployed successfully for long periods, in some cases more than three years, on platforms such as profiling floats or gliders. Technologies for pH, pCO2, and particulate inorganic carbon are maturing rapidly as well. These sensors could serve as the enabling technology for a global biogeochemical observing system that might operate on a scale comparable to the current Argo array. Here, we review the scientific motivation and the prospects for a global observing system for ocean biogeochemistry.

are now available for oxygen, nitrate, and a variety of bio-optical properties that serve as proxies for important components of the carbon cycle (e.g., particulate organic carbon).These sensors have all been deployed successfully for long periods, in some cases more than three years, on platforms such as profiling floats or gliders.
Technologies for pH, pCO 2 , and particulate inorganic carbon are maturing rapidly as well.These sensors could serve as the enabling technology for a global biogeochemical observing system that might operate on a scale comparable to the current Argo array.
Here, we review the scientific motivation and the prospects for a global observing system for ocean biogeochemistry.are now available for oxygen, nitrate, and a variety of bio-optical properties that serve as proxies for important components of the carbon cycle (e.g., particulate organic carbon).These sensors have all been deployed successfully for long periods, in some cases more than three years, on platforms such as profiling floats or gliders.
Technologies for pH, pCO 2 , and particulate inorganic carbon are maturing rapidly as well.These sensors could serve as the enabling technology for a global biogeochemical observing system that might operate on a scale comparable to the current Argo array.
Here, we review the scientific motivation and the prospects for a global observing system for ocean biogeochemistry.et al., 2004, 2009).

autoNomous aNd lagR aNgIaN PlatFoRms aNd seNsoRs
Observing the ocean over broad scales requires high-endurance and relatively low-cost sensors and platforms.Prospects for such capabilities were reviewed at the "ALPS: Autonomous and Lagrangian Platforms and Sensors Workshop" in 2003 (Perry and Rudnick, 2003a, b).In the six years since that workshop, the field has matured rapidly.We briefly review these developments below.

Platforms
In 2003, profiling floats were a relatively mature technology and gliders were rapidly developing.Since that time, the Argo array of profiling floats (Roemmich et al., 2004(Roemmich et al., , 2009) ) has reached its design goal of 3000 floats in the ocean, while sensors Oxygen sensors are now being deployed on floats for multiyear periods with little or no drift in sensor response (Körtzinger et al., 2004;Riser and Johnson, 2008).These long-term deployments are possible because these platforms spend much of their time at depth, in cold and dark waters.
Consequently, biofouling is less of an issue than when sensors are permanently fixed in the upper ocean, for example, on moorings.Figure 1 shows the remarkable stability attained with an oxygen sensor deployed for nearly two years on a profiling float (Tengberg et al., 2006).
As of June 2009, more than 200 floats had been deployed with oxygen sensors, with about 150 currently active.The amount of oxygen data that is being delivered to national data centers from profiling floats is already nearly double that coming from ships.Gruber et al. (2007) address in detail the feasibility of deploying large numbers of oxygen sensors on profiling floats.Oxygen sensors have been deployed on gliders, also with excellent results (Figure 2; Nicholson et al., 2008;Perry et al., 2008).
Nitrate sensors based on direct optical detection (Johnson and Coletti, 2002) are now deployed on profiling floats, and they have operated for more than 500 days.The power budget implies    (Riser and Johnson, 2008) and Seaglider measurements (Nicholson et al., 2008;Perry et al., 2008) demonstrate the potential to quantify net community production (NCP)  Performing optical measurements at well-identified time periods of the day (e.g., sunrise, sunset) allows NCP to be estimated (Claustre et al., 2008).
The vertical flux of POC, which is a critical component of the biological carbon pump, can be determined by several methods.Qualitative changes in carbon export have been documented using transmissometers deployed on profiling floats (Bishop et al., 2004;Bishop and Wood, 2009) (Körtzinger et al., 2008) to obtain a full three-dimensional picture.
Although small in size, OMZs thus have global-scale implications.However, the processes that maintain OMZ and control their dynamics are not well understood.There is great concern that these zones will enlarge in the future due to ocean deoxygenation, continuing a trend that has already been observed (Stramma et al., 2008).If these lowoxygen water masses were to extend into coastal regions, the impact on coastal water quality and coastal marine life would be severe.are key players in global biogeochemical cycles.They contribute significantly to the ocean uptake of anthropogenic CO 2 (Sabine et al., 2004).They are the most significant source of oxygen to intermediate depths, and they modulate the strength of the globally significant Peru-Chile oxygen minimum zone (Meissner et al., 2005).Subantarctic mode water is the main source of nutrients to the ocean's thermocline, and it supports a large fraction of biological production north of 30°S (Sarmiento et al., 2004).
The nutrient properties of SAMW, which are set through biogeochemical processes in the subantarctic zone, are thus critical in determining ocean productivity.
There is evidence for freshening of intermediate waters and warming of mode waters over the past 10 to 20 years, with an overall increase in mode water production, consistent with models of anthropogenic climate change (Banks et al., 2000).The nutrient content of SAMW, set by biological processes within its formation region, is also sensitive to climate change (Sarmiento et al., 2004).
Determining the biological processes that set the nutrient and oxygen properties of To date, there are relatively few efforts to assimilate biogeochemical data into models (e.g., Gregg, 2008), and achieving an integrated system will require substantial effort in this area.Some of the proposed pilot studies would represent ideal test beds for making rapid progress.Brasseur et al., 2009).
Finally, an integrated data management system must be available for a biogeochemical observing system.Most of the data collected by profiling floats are delivered to a single data management system, which makes it relatively easy to access (Figure 5).Glider data are not handled in this manner, and this situation must be remedied.
With these efforts, a novel biogeochemical observing system could be developed, permitting us to address some key scientific questions in an innovative manner.We envision two strategies to move forward with the implementation beyond the regionalscale experiments.For some biogeochemical sensors (e.g., oxygen; Gruber " " R e g u l a R I s s u e F e at u R e B y K e N N e t h s .J o h N s o N , W I l l I a m m .B e R e l s o N , e m m a N u e l s .B o s s , Z a N N a C h a s e , h e R V é C l a u s t R e , s t e V e N R .e m e R s o N , N I C o l a s g R u B e R , a R N e K ö R t Z I N g e R , m a R y J a N e P e R R y, a N d s t e P h e N C .R I s e R aBstR aCt.Chemical and biological sensor technologies have advanced rapidly in the past five years.Sensors that require low power and operate for multiple years u l a R I s s u e F e at u R e aBstR aCt.Chemical and biological sensor technologies have advanced rapidly in the past five years.Sensors that require low power and operate for multiple years B y K e N N e t h s .J o h N s o N , W I l l I a m m .B e R e l s o N , e m m a N u e l s .B o s s , Z a N N a C h a s e , h e R V é C l a u s t R e , s t e V e N R .e m e R s o N , N I C o l a s g R u B e R , a R N e K ö R t Z I N g e R , m a R y J a N e P e R R y, a N d s t e P h e N C .R I s e R broad areas.Satellite observations have helped to overcome this limitation, but the suite of observable biogeochemical parameters is constrained to ocean color during cloud-free periods in the upper one-fifth of the euphotic zone.The result is that we have only limited understanding of how ocean biogeochemistry is changing in response to natural climate variations such as El Niño or to anthropogenic climate changes driven by the accumulation of greenhouse gases in the atmosphere.Observing ocean biogeochemistry at the proper spatial and temporal scales required to address driving scientific questions will require a revolution.Such a revolution in the observing technology is at hand (Johnson et al., 2007; see also the September 2008 Limnology and Oceanography special issue on Autonomous and Lagrangian Platforms and Sensors, and various white papers prepared for Ocean Obs 09, http://www.oceanobs09.net).Rapid progress is being made in the development of a suite of chemical and biological sensors that are low power, small in size, precise and accurate, and have long-term stability.Within the past five years, sensor technologies for oxygen, chlorophyll, particles, and nitrate have been refined.These sensors are now capable of deployment on long-endurance missions on autonomous platforms such as profiling floats and gliders.They represent the enabling technology for a global biogeochemical observing system that might operate on a scale comparable to the physically oriented Argo array (Roemmich more than 100 gliders(Rudnick et al.,    INtRoduCtIoNThe challenge of understanding the ocean's role in the global carbon cycle and its response to a changing environment requires an expanded scale of observation in both space and time.Traditional ocean sampling methods rarely achieve measurements on the spatial and temporal scales required to answer questions regarding such phenomena as the magnitude of the uptake of anthropogenic carbon dioxide and the severity of resulting ocean acidification(Sabine et al., 2004), ventilation of the thermocline in mode-water-formation areas and associated formation of preformed nutrients (e.g.,Sarmiento et al., 2004), rate and spatial distribution of ocean deoxygenation(Stramma et al., 2008), transport of nutrients to the euphotic zone in nutrient-starved, stratified ocean regions (e.g.,Jenkins and Doney, 2003), and mechanisms controlling the ocean's biological carbon pump (e.g.,Ducklow et al., 2001Ducklow et al.,  et al., 2009)).Ship-based sampling cannot resolve mesoscale and higher-frequency processes (e.g., eddies, tropical storms) and gives little sense of change over 2004) are now exploring the ocean.Both profiling floats and gliders cycle vertically in the ocean by changing their buoyancy.Platform density is changed by pumping oil stored in an internal reservoir into an external bladder to increase volume at constant mass (to ascend) or by deflating the bladder (to descend).Gliders can be programmed to sample along a predetermined path due to their ability to change internal weight distribution.Coupled with external wings that provide lift, the change in weight distribution allows gliders to move forward at speeds near 0.5 m s -1 as they cycle vertically.Floats drift with the currents at parking depths of (typically) 1000 m before ascending to the surface at programmed time intervals (hours to days).At the surface, both floats and gliders transmit their data to orbiting satellites using either Service Argos (most floats) or the Iridium system (some floats, all gliders).Many floats in the Argo Program have operated for five years or more, while some types of gliders have survived for nearly one year, including extended operations under ice.
Kenneth S. Johnson (johnson@mbari.org) is Senior Scientist, Monterey Bay Aquarium Research Institute, Moss Landing, CA, USA.William M. Berelson is Professor, Department of Earth Sciences, University of Southern California, Los Angeles, CA, USA.Emmanuel S. Boss is Associate Professor of Oceanography, University of Maine, Orono, ME, USA.Zanna Chase is Assistant Professor, College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR, USA.Hervé Claustre is Senior Scientist, Laboratoire d'Océanographie de Villefranche, Villefranche-sur-mer, France.Steven R. Emerson is Professor of Oceanography, University of Washington, Seattle, WA, USA.Nicolas Gruber is Professor of Environmental Physics, Institute of Biogeochemistry and Pollutant Dynamics, Eidgenössische Technische Hochschule (ETH), Zürich, Switzerland.Arne Körtzinger is Professor of Chemical Oceanography, Leibniz Institute of Marine Sciences, Kiel, Germany.Mary Jane Perry is Professor of Oceanography, Darling Marine Center, University of Maine, Walpole, ME, USA.Stephen C. Riser is Professor, School of Oceanography, University of Washington, Seattle, WA, USA.

Figure 1 .
Figure1.oxygen and density data collected over 580 days from a free-drifting argo float off the east coast of Canada in the labrador Current, with oxygen readings as the float passed 1800-m depth.The average oxygen readings during the whole period were 295.0 ± 0.7 µmol l -1 .FromTengberg et al. (2006) . Particle load is the main driver of water turbidity or transparency in the open ocean.Turbidity can be quantified by the measurement of the backscattering coefficient (backscattering meter), while transparency is measured by the particle attenuation coefficient (transmissometer).In open ocean waters, particulate organic carbon (POC) is the main source of particles, and both optical measurements can be converted to a concentration of POC with reasonable accuracy (e.g., Bishop and Wood, 2009).Boss et al. (2008) succeeded in observing chlorophyll fluorescence (proxy for phytoplankton) and light scattering (proxy for POC) in the upper 1000 m of the North Atlantic for three years (Figure 3).These observations resolved the annual cycle with no apparent sensor drift at depth, and they revealed remarkable events driven by mesoscale processes.Gliders are routinely deployed with bio-optical and chemical sensors for repeat observations along transects in the coastal and open ocean.Many of these time series have been sustained for years that they can operate over four years with 60 nitrate measurements from 1000 m to the surface at a cycle time of five days.Detection limits are on the order of 0.5 to 1 µM.Although not sufficient to measure euphotic zone nitrate concentrations in many regions, they can resolve annual cycles in mesotrophic, bloom-forming regions such as the North Atlantic (D' Asaro et al., 2008) or in high nitrate, low chlorophyll (HNLC) regions (Figure 4) and measure nitrate transport driven by mesoscale events.These nitrate sensors are also being adapted to gliders.Other biogeochemical sensors have undergone preliminary deployments on floats or gliders.Measurements with gas tension devices (GTD; McNeil et al., 2006) can be combined with oxygen concentrations to determine the partial

Figure 2 .
Figure 2. (a) objective analysis-mapped o 2 -saturation anomaly (%) and (b) temperature (°C) averaged over a 50-km x 50-km box in the upper 200 m near the hawaii ocean time-series (hot) station.The black line indicates mixed-layer depth and the dashed black line is the depth of σ θ = 24.30.FromNicholson et al. (2008) Figure3.Chlorophyll estimated with a fluorometer deployed on a profiling float in the labrador sea region of the North atlantic.Fluorescence data were converted to chlorophyll concentration using calibrations with both lab data and satellite-based chlorophyll (seeBoss et al., 2008, for details).Figure adapted from data reported byBoss et al. (2008) using sensors.Euphotic zone studies of oxygen evolution are presently limited to observing the concentration changes below the mixed layer.The sensors do not yet have sufficient accuracy to quantify changes resulting from gas exchange with the atmosphere that are driven by the small difference between actual surface oxygen concentration and the values that would occur at atmospheric equilibrium.This level of accuracy should, however, be achievable using in situ calibration (see below).In HNLC regions, NCP can be assessed from the seasonal change in nitrate concentration and a suitable mixing model (Macready and Quay, 2001).Optical nitrate sensors now deployed on profiling floats are currently resolving these cycles at locations in the subarctic North Pacific (Figure 4) and in the Southern Ocean.These observations do

Figure 4 .
Figure 4. temperature, nitrate, and oxygen measured at ocean station Papa in the North Pacific (50°N, 145°W) by apex profiling float 5143, which is equipped with an Isus nitrate sensor and aanderaa oxygen optode.data downloaded from www.mbari.org/chemsensor/floatviz.htm.
. Particles accumulate on the optics while the float is parked at constant depth, and light attenuation by these particles is measured after some time interval with the float at constant depth.The particles are then washed from the optics with a pump and the cycle can be repeated.This sensor can provide a high-resolution record of relative changes in particle flux, but an absolute calibration is not yet possible.Carbon flux from the euphotic zone is directly related to the oxygenutilization rate in the mesopelagic region immediately below.This value has been determined by measuring the seasonal change in oxygen concentration from a series of Argo floats in the South Pacific (Martz et al., 2008).Because changes are a small fraction of the O 2 standing crop in this region and horizontal advection is an important and unknown variable, the method is most sensitive in the shallow mesopelagic.Many of the capabilities discussed above were used in a demonstration experiment to study the North Atlantic spring bloom during 2008 (D' Asaro et al., 2008).This experiment occurred near the location of the 60°N Joint Global Flux Study (JGOFS) North Atlantic Bloom Experiment conducted from 1989-1991.Although this site had been well studied, preliminary results demonstrated a novel perspective on processes during the bloom.The 2008 experiment used an array of gliders and Lagrangian mixed-layer floats equipped with bio-optical and chemical (oxygen and nitrate) sensors to monitor the temporal evolution of the spring bloom.The work was also designed to integrate satellite observations and biogeochemical models with in situ sensor data.It provided quantitative assessments of net ecosystem production from oxygen evolution and nitrate drawdown, as well as the fate of the organic matter produced.It demonstrates how to extend information gained from Eulerian observatories One of the goals of the Scoping Workshop was to consider potential topics for these next-generation experiments of multiyear endurance over large ocean regions.A summary of these discussions is presented here as examples of multiyear float and glider experiments focused on questions that have global-scale implications.They are not, however, intended to represent the full complement of research issues that can be addressed using in situ sensors coupled with glider and float technology.In addition, longer-term follow-on studies to the North Atlantic experiment mentioned above were discussed.Quantifying the Role of Biology in the drawdown of Co 2 at the subtropical-subpolar Boundary in the North Pacific An important oceanographic problem that can be successfully addressed using remote sensors is the study of net biological oxygen production at the boundary between the subarctic and subtropical gyres in the Pacific Ocean.The seasonal drawdown of pCO 2 at this boundary creates a carbon sink from atmosphere to ocean that is roughly equal to the the ChalleNge oF uNdeRstaNdINg the oCeaN's Role IN the gloBal CaRBoN CyCle aNd Its ResPoNse to a ChaNgINg eNVIRoNmeNt ReQuIRes aN exPaNded sCale oF oBseRVatIoN IN Both sPaCe aNd tIme.processes to the pCO 2 drawdown, and the mechanisms responsible for these observations, are not clear.Because of the dramatic seasonality of the pCO 2 drawdown, understanding its origin requires observations over complete annual cycles, a prospect difficult to achieve from ships.A study using floats and gliders would be capable of resolving the annual cycle of NCP and carbon export across the Pacific in the region between the subarctic and subtropical gyres.consumed.This significant feedback loop may regulate ocean productivity and climate over millennial time scales SAMW and detecting (and predicting) interannual and longer-time-scale changes in these properties are important priorities for studies of ocean biogeochemistry.These questions are well suited for addressing by sustained observations using floats or gliders.the Path FoRWaRd Over 200 floats in the Argo Program have been equipped with dissolved O 2 sensors (Figure 5a), and many gliders are now equipped to measure oxygen.Dozens of floats and gliders are equipped with bio-optics, primarily chlorophyll fluorometers and particle backscatter sensors (Figure 5b).Funding is now the CoRe oF a gloBal BIogeoChemICal oBseRVINg system Is BeINg deVeloPed, alBeIt IN a laRgely ad hoC FashIoN." " available to increase the number of floats with nitrate sensors from the present set of four (Figure 5b) to 40.The core of a global biogeochemical observing system is being developed, albeit in a largely ad hoc fashion.Moving forward with integrated, regional-scale experiments using autonomous platforms and multiple biogeochemical sensors as a model for a global observing system requires three significant developments.First, we must demonstrate that sensors can generate climate-researchquality data.This standard means that results from sensors on different platforms must be intercomparable.For example, oxygen measurements should be made with a precision and accuracy of better than ± 0.5% in order to avoid significant biases in biological rate estimates due to atmospheric oxygen exchange.They also must be highly accurate in order to be useful for determining seasonal to long-term trends in the ocean's oxygen content.Oxygen sensors can meet this requirement in terms of their precision (Figure 3), but very often factory calibrations do not produce accuracy at this level (e.g., Körtzinger et al., 2005; Riser and Johnson, 2008), limiting the opportunity to compare data across platforms.There are opportunities to make significant advances in calibration, but these will require both laboratory development and field testing.For example, optode O 2 sensors can be recalibrated using atmospheric pO 2 measurements when the floats surface (Körtzinger et al., 2005).Delayed-mode calibration of oxygen sensors may also be possible, as is now done for Argo salinity sensors, by using existing and newly developed oxygen climatologies at depth.Such standardized procedures need to be developed and implemented at data centers for oxygen and other biogeochemical properties.Second, we must demonstrate the operation of completely integrated observing systems that combine in situ sensors deployed on long-endurance platforms with satellite sensors and dataassimilating, biogeochemical-ecological models.In particular, data assimilation into models is essential to interpret the large amounts of data in their spatial and temporal context, and to turn concentration measurements into quantitative rate estimates over large regions of the ocean.

Figure 5 .
Figure 5. (a) all argo floats reporting from June 9-16, 2009.(b) all argo and argo-equivalent floats equipped with oxygen sensors (red circles), bio-optical sensors (green diamonds), and nitrate sensors (yellow triangles) and reporting from June 9-16, 2009.data were downloaded from the argo Coriolis data Center (http://www.coriolis.eu.org/cdc/argo.htm),except the nitrate float locations, which were downloaded from www.mbari.org/chemsensor/floatviz.htm.The census of argo oxygen floats misses a number of active systems that are currently profiling under antarctic ice and will not report until the next austral summer.
et al., 2007), a collaboration could be set up to add the sensors directly to the existing Argo platforms.Such collaboration requires funding to offset the full marginal costs of adding sensors.More complex sensor arrays might best be deployed as a mission separate from Argo, especially when they require different sampling intervals.Future sensor developments, including pH, pCO 2 , and possibly acoustic sensors for zooplankton, would enrich the system.These future developments are not essential, however, to begin regionalscale experimentation with a system that could provide significant and quantitative information on the carbon cycle and related biogeochemical properties in the ocean.In particular, the existing and proven optical and chemical sensor set would be sufficient in many areas to document temporal changes in the biological carbon pump and seasonal to long-term changes in oceanic oxygen.…a NoVel BIogeoChemICal oBseRVINg system Could Be deVeloPed, PeRmIttINg us to addRess some Key sCIeNtIFIC QuestIoNs IN aN INNoVatIVe maNNeR.