Satellites will address critical science priorities for quantifying ocean carbon

The ability to routinely quantify global carbon dioxide (CO 2 ) absorption by the oceans has become crucial: it provides a powerful constraint for establishing global and regional carbon (C) budgets, and enables identification of the ecological impacts and risks of this uptake on the marine environment. Advances in understanding, technology, and international coordination have made it possible to measure CO 2 absorption by the oceans to a greater degree of accuracy than is possible in terrestrial landscapes. These advances, combined with new satellite- based Earth observation capabilities, increasing public availability of data, and cloud computing, provide important opportunities for addressing critical knowledge gaps. Furthermore, Earth observation in synergy with in- situ monitoring can provide the large- scale ocean monitoring that is necessary to support policies to protect ocean ecosystems at risk, and motivate societal shifts toward meeting C emissions targets; however, sustained effort will be needed.

C limate-relevant gases, including carbon dioxide (CO 2 ) and oxygen (O 2 ), are continuously exchanged between the ocean and atmosphere, with considerable temporal and geographic variations. Collectively, ocean and terrestrial CO 2 uptake reduce the accumulation of anthropogenic CO 2 in the atmosphere and have acted as a brake on global warming. The oceans, which are the largest long-term net sink of carbon (C), have so far been responsible for the uptake of about 25% of total anthropogenic CO 2 (Le Quéré et al. 2018), and the magnitude of this uptake varies on seasonal, interannual, and probably decadal timescales (Watson et al. 2009;Wanninkhof et al. 2013;Gruber et al. 2019).
Concentrations of CO 2 in seawater have increased measurably over the past 35 years (Bates et al. 2014). This absorption (or "sink") of CO 2 into the oceans reduces atmospheric CO 2 concentrations but lowers ocean pH levels, fundamentally altering ocean chemistry (WebPanel 1) and environmental conditions for marine ecosystems (Raven et al. 2005). The gradual and long-term lowering of pH -termed ocean acidification -is measurable on decadal timescales, but short-term intra-annual episodic events, driven primarily by atmosphere and ocean interactions (eg upwelling of low-pH and potentially corrosive waters; Feely et al. 2008), can have catastrophic impacts on marine life and ecosystems. The areal extent and intensity of these events are increasing as the oceans absorb more CO 2 (Feely et al. 2008).
Estimates of the land C sink are usually calculated or verified by determining the differences between net anthropogenic emissions, atmospheric uptake, and ocean C uptake (Le Quéré et al. 2016), typically through the use of inversion modeling (eg Resplandy et al. 2018), or are estimated from dynamic global vegetation models (Le Quéré et al. 2018). Because all of these methods rely on an accurate estimate of the ocean C sink (that is, an estimate of its annual uptake, either as a major component of the calculation or as a necessary observational constraint), overlooking variations and trends in the ocean C sink can have a profound impact on the accuracy of estimates of the land C sink and the global C budget, and consequently on guidance for emissions targets.
Satellites will address critical science priorities for quantifying ocean carbon Jamie D Shutler 1 *, Rik Wanninkhof 2 , Philip D Nightingale 3 , David K Woolf 4 , Dorothee CE Bakker 5 , Andy Watson 6 , Ian Ashton 1 , Thomas Holding 1 , Bertrand Chapron 7 , Yves Quilfen 7 , Chris Fairall 8 , Ute Schuster 6 , Masakatsu Nakajima 9 , and Craig J Donlon 10 The ability to routinely quantify global carbon dioxide (CO 2 ) absorption by the oceans has become crucial: it provides a powerful constraint for establishing global and regional carbon (C) budgets, and enables identification of the ecological impacts and risks of this uptake on the marine environment. Advances in understanding, technology, and international coordination have made it possible to measure CO 2 absorption by the oceans to a greater degree of accuracy than is possible in terrestrial landscapes. These advances, combined with new satellite-based Earth observation capabilities, increasing public availability of data, and cloud computing, provide important opportunities for addressing critical knowledge gaps. Furthermore, Earth observation in synergy with in-situ monitoring can provide the large-scale ocean monitoring that is necessary to support policies to protect ocean ecosystems at risk, and motivate societal shifts toward meeting C emissions targets; however, sustained effort will be needed. In a nutshell: • The oceans cover >70% of the Earth's surface and are critical for food supply and maintaining global climate • Increasing carbon dioxide (CO 2 ) absorption alters ocean chemistry and ecology, affecting marine ecosystems over both the short and long term • Accurate estimates of CO 2 absorption by the world's oceans provide a powerful constraint on carbon (C) budgets, and are needed to inform policies to motivate societal shifts toward reducing C emissions • We review recent and foreseeable advances for studying oceanic CO 2 absorption, explain why satellite-based Earth observation is key to addressing existing knowledge gaps, and discuss how global monitoring is now both possible and necessary to support policy and conservation

REVIEWS
The solubility of CO 2 in seawater is highly sensitive to temperature (Woolf et al. 2016), which means that future warming of the upper ocean may diminish the ocean C sink. The capacity of the ocean to absorb additional CO 2 will also decline as a result of ocean acidification due to reductions in buffer capacity (Orr et al. 2005; WebPanel 1), and there is evidence that this is already occurring (Wanninkhof et al. 2013). In contrast, increased stratification and biological production could enhance the ocean C sink, while what impact the continual decline in polar ice will have remains unclear. The interplay between these counteracting processes makes future predictions challenging and warrants continued investigation.
Marine ecosystems provide food, energy, recreation, and mental and emotional well-being services to people worldwide. Future changes in the global ocean C sink may ease or strain human efforts to avoid dangerous climate change, and therefore measuring the oceanic C cycle must be a critical part of strategic and sustained long-term monitoring programs (Peters et al. 2017). The vastness of the world's oceans means that large-spatial-scale observations from space, combined with in-situ monitoring, are needed for identifying regions and ecosystems at risk of ocean acidification, advising emissions reductions, and evaluating progress toward the Paris Agreement and UN Framework Convention on Climate Change Sustainable Development Goals (Figure 1).
Major advances have been made over the past 5-10 years in scientific understanding (eg Wanninkhof et al. 2013;Woolf et al. 2016), instrumentation (eg WebTable 1), and international coordination (eg Bakker et al. 2016;Le Quéré et al. 2018), as well as through the exploitation of satellite-based Earth observation (eg Goddijn-Murphy et al. 2013;Landschützer et al. 2015). These advances, combined with the recent leap in Earth observation capabilities (eg Butler 2014), increased public access to satellite data, and cloud computing (eg Gorelick et al. 2017), provide important opportunities to fill existing knowledge gaps and enable strategic monitoring.
We explore the state-of-the-art techniques and recent breakthroughs in scientific understanding and highlight existing challenges. We then identify how Earth observation programs could enable direct observation of gas concentrations near the water surface, the processes governing its exchange, and the circulation and vertical transport of total CO 2 to further constrain the global C budget (Figure 2). We recommend the creation of a robust observational network that uses insitu and satellite-based Earth observations to quantify the atmosphere-ocean exchange of climate-relevant gases.

Current ability to monitor ocean carbon
First, we review current scientific abilities, focusing on the advances that have been made in the past 5-10 years.

Sea-surface exchange of carbon
The solubility of gaseous CO 2 is strongly temperature dependent (Woolf et al. 2016); cooling waters tend to absorb atmospheric CO 2 whereas warming waters release CO 2 . The in-water total CO 2 concentration can then be reduced via biological uptake by marine phytoplankton or increased via their respiration. On an annual basis, the net global pre-industrial flux of CO 2 across the atmosphere-ocean interface is thought to have been about 0.45-0.78 petagrams of C per year (Pg C yr −1 ; Wanninkhof et al. 2013;Resplandy et al. 2018) due to net riverine C transfer from land to ocean. The post-industrial net flux is the difference between how much C is absorbed by the ocean and how much is released, each of magnitude ~90 Pg C yr −1 . The sum of these two terms (post-industrial net flux and pre-industrial net flux) is the anthropogenic uptake of CO 2 . However, atmosphere-ocean fluxes of CO 2 cannot be measured directly over large spatial scales. The atmosphere-ocean gas exchange of CO 2 is typically calculated from the gas transfer velocity and the gaseous concentration difference at the atmosphere-ocean interface. Gas transfer velocities can be estimated in the field through bulk injection of gaseous tracers into the water column and measurement of the decline in tracer gas concentration over time. These methods provide estimates of gas transfer Research vessels, buoys, and transport ships all routinely collect measurements for monitoring ocean atmosphere-ocean gas exchange, but some oceans are not well sampled and much of the data collection is voluntary. (c) A coral reef viewed from space; these sensitive ecosystems can be negatively impacted by episodic ocean acidification events caused by upwelling (Feely et al. 2008 determined over a period of 0.5-4 days and ~100 km 2 of the ocean (eg Ho et al. 2011), and have shown that about 80% of the variance in gas transfer rates is explained by the wind speed (for wind speeds of 3-13 m s −1 ). The majority of published studies therefore describe gas transfer as a function of wind speed (eg Nightingale et al. 2000;Schuster et al. 2013), although the effect of wind on transfer is indirect, acting through the generation of wind waves, near-surface turbulence, and bubbles; other controls are also known to exist ( Figure 2, point 2). Fluxes can be measured directly by micro-meteorological (eddy covariance) techniques at hourly resolutions over ~1 km 2 (Butterworth and Miller 2016). The eddy covariance results have shown more variability between studies than the gas tracer results; some differences are attributable to experimental artifacts; however, other differences are due to processes that control fluxes but cannot be captured routinely by wind parameterizations (eg Butterworth and Miller 2016; Blomquist et al. 2017), or to region-specific complexities (eg those that occur in polar waters) (WebPanel 2). As such, scientific focus is now moving away from purely wind-based proxies for characterizing atmosphereocean gas exchange. Advancements in eddy covariance techniques now make it possible to collect (temporally and spatially) co-incident in-situ flux and in-situ remotely sensed measurements to improve our understanding of transfer processes, and also enable improved large-spatial-scale satellite-derived transfer. However, these opportunities have yet to be fully explored.

Ecosystem impacts
Uptake of anthropogenic CO 2 by the oceans is not without consequences. Gaseous CO 2 reacts with seawater to form carbonic acid (H 2 CO 3 ), a process involving the release of hydrogen (H + ) ions, some of which bind with existing carbonate (CO 3 2-) ions to form bicarbonate (HCO 3 -). As a result, the marine uptake of anthropogenic CO 2 is causing the oceans to become more acidic (increased H + and HCO 3 -, decreased pH and CO 3 2-; WebPanel 1). Ocean acidification will profoundly affect marine organisms worldwide (Pörtner et al. 2015) and over time reduce the capacity of the oceans to act as a sink for additional anthropogenic CO 2 (Orr et al. 2005). Calcifying organisms (which need CO 3 2to form their shells), including mollusks, crustaceans, corals, and some types of algae, are especially at risk, with far-reaching consequences for biodiversity, nutrient cycling, and habitat provision for socioeconomically important organisms (Brodie et al. 2014). However, some non-calcifying plants and microalgae are predicted to respond positively to acidification by increasing their photosynthetic and growth rates (Pörtner et al. 2015).
The net impacts of ocean acidification on marine systems and climate are not straightforward to predict, and therefore warrant strategic and large-spatial-scale monitoring.
Upwelling -the movement of deep water to the ocean surface -is a mostly wind-driven phenomenon occurring at various scales in the global oceans. Nearshore upwelling, composed of Ekman transport and pumping (resulting from wind friction on the surface water), occurs predominantly on the eastern boundaries of the ocean basins. Upwelling brings cold, nutrient-rich water to the surface, but the low pH of these waters can disrupt the food webs, reproduction, growth, and energy balances of marine communities (O'Donnell et al. 2013). Consequently, episodic upwelling events can negatively affect regional ecosystems, including the Arctic, sensitive tropical reef systems, and continental shelves important for wild fisheries (Feely et al. 2008) and aquaculture. Wind-driven upwelling is also considered key for driving ocean interior circulation (Toggweiler and Russell 2008), which transports and (via corresponding downwelling) locks away total inorganic CO 2 (CO 2aq + HCO 3 -+ CO 3 2-) for centuries. However, no direct methods currently exist to routinely monitor such global largespatial-scale water movements, with the exception of a handful of sites (eg www.bodc.ac.uk/rapidmoc).

Instrumentation and international collaboration
Automated instruments on ships of opportunity (volunteer merchant ships), moorings, autonomous vehicles, and floats (2) quantifying gas, momentum, and heat atmosphere-ocean exchange processes; (3) capturing near-surface gradients in the water; and (4) measuring internal circulation and surface transport.

REVIEWS
(WebTable 1) routinely measure surface-water gaseous CO 2 with high precision and accuracy. These measurements are then assembled into global datasets (Takahashi et al. 2014;Bakker et al. 2016). The Surface Ocean CO 2 Atlas (SOCAT) in particular is an international community effort with welldocumented protocols; each of its annual updates add over a million new data points, with a current data holding of 23.4 million measurements (Bakker et al. 2016). Concentrations of atmospheric CO 2 over the open ocean are less variable than those within the surface water. As a result, global atmospheric CO 2 is measured via 60 sustained marine boundary layer stations with weekly sampling (Conway et al. 1994). The individual measurements are freely available, and are also collated into the National Oceanic and Atmospheric Administration (NOAA) Greenhouse Gas Marine Boundary Layer Reference (MBLR, www.esrl.noaa.gov/gmd/ccgg/mbl/data.php). To date, all studies of the global ocean C sink utilize the zonally averaged MBLR data combined with surface-water gaseous CO 2 measurements (Panel 1; Figure 3c; eg Landschützer et al. 2015), but it is known that regions of meridional gradients (north-south gradients) caused by continental airflow are unlikely to be well resolved. To help assess the accuracy of the MBLR approach, atmospheric CO 2 measurements are now increasingly being collected in conjunction with the surface-water measurements.
However, the sparse nature of in-situ data collection means that atmospheric concentrations in many key regions remain poorly characterized.

Exploitation of satellite measurements
Remotely sensed satellite observations of sea-surface state, temperature, wind, salinity, ice, rain, and biology (WebTable 2) play critical roles in the calculation of large-spatial-scale oceanic gas fluxes, because in-situ parameterizations and sparse in-situ measurements need to be upscaled. On local scales, sea-surface temperature is a strong predictor of surface-water gaseous CO 2 values, and satellite-observed temperature is used to interpolate in time and space between these in-situ measurements. More complex methodologies are being usedincluding empirically based multiparameter methods and machine-learning approaches -that rely on satellite chlorophyll-a, sea-surface temperature, and global re-analysis datasets such as mixed layer depth (Figure 3b; eg Schuster et al. 2013). Finer spatial resolution remote-sensing observations have been particularly useful for studying processes and fluxes in heterogeneous areas that have commonly been omitted in globalscale flux calculations (eg within coastal oceans, upwelling zones, near the ice edge; Laruelle et al. 2017). A promising Panel 1. The importance of correct temperature handling for calculating atmosphere-ocean gas fluxes

Exploiting collated datasets of carbon dioxide (CO 2 ) measurements
Temperature is a key controller of the water solubility of CO 2 and, as such, all CO 2 measurements from water need to be paired with their respective temperature measurement. The temperature of seawater naturally varies over time, depth, and location, and different sampling techniques and vessels will sample water from differing depths. To reduce uncertainty, observations from different sources, locations, and times should therefore be corrected to a common temperature measurement dataset at a specified depth.

Accurate atmosphere-ocean gas flux calculations
The most accurate calculation of the atmosphere-ocean CO 2 gas fluxes requires the gaseous concentrations of CO 2 to be known at the top and bottom of a very small layer of water at the surface -the mass boundary layer -that is thinner than 1 mm (Woolf et al. 2016). Although these concentrations cannot be measured with current technology, they can be estimated using a combination of in-situ measurements taken at a few meters' depth, satellite observations of the water temperature, and knowledge of C kinetics and interfacial dynamics. Satellite remote-sensing technologies can observe the skin (top ~10-20 μm via thermal infrared) or sub-skin (~1 mm via microwave) temperatures; the skin temperature can also be inferred from the sub-skin (or vice versa) with knowledge of the local wind conditions (Donlon et al. 2002). Measurements of the skin and sub-skin temperatures allow an in-situ gaseous CO 2 concentration, collected from several meters below the surface, to be recalculated to the values at the top and bottom of the mass boundary layer.

Example method
The Surface Ocean CO 2 Atlas v4 dataset (Figure 3a) was re-analyzed (Goddijn-Murphy et al. 2015) to a common sea-surface sub-skin dataset (Banzon et al. 2016) and a skin dataset (sub-skin minus 0.17°C, based on Donlon et al. [2002]) to determine the in-situ gaseous CO 2 concentrations at the top and bottom of the mass boundary layer (Figure 3b). A spatially and temporally complete time series of global gaseous CO 2 concentrations was then calculated using a multilinear regression technique (Schuster et al. 2013). Atmosphere-ocean gas fluxes were calculated using the open-source FluxEngine toolbox , a wind-speed-based gas transfer velocity (Nightingale et al. 2000), a simple linear scaling of gas fluxes due to ice, and the ancillary datasets described in section 4a of Shutler et al. (2016) (Figure 3c). All calculations were consistent with the recommendations and guidelines of the rapid model of Woolf et al. (2016).

Implications of inconsistent temperature handling
The simple calculation provided in Woolf et al. (2016) suggests that ignoring the gaseous concentration differences across the mass boundary layer results in a 0.1-0.6 petagram C per year (Pg C yr −1 ) underestimate (bias) in the global ocean CO 2 sink. The analysis following the example method described in this panel gives a refined estimate of this bias of 0.37-0.44 Pg C yr −1 , with a 16-year average of 0.41 Pg C yr −1 . This equates to an 18-21% underestimate of the contemporary oceanic sink of 2.1 Pg C yr −1 (as provided by Le Quéré et al. [2018]). method of using remotely sensed salinity levels to determine regional alkalinity has been highlighted ; this approach currently lacks the accuracy to assess anthropogenic change but could provide a further constraint on surface inorganic C and help elucidate the impact and progression of ocean acidification. Satellite-based biological production and sea-surface temperature data have enabled preliminary identification of the drivers of variability in the ocean CO 2 sink (Henson et al. 2018). Current methods for calculating gas fluxes using Earth observation techniques do not always reflect the physics of the sea surface or the available satellite instrumentation. For example, gas transfer velocities are normally calculated using wind speed as a proxy for surface exchange, though both transfer processes and microwave wavelength returns are more closely related to sea-surface roughness and exchange. Direct measurements of surface properties from space (WebTable 2) provide opportunities for moving beyond simple proxies. Some studies have already begun to investigate this potential (eg Goddijn-Murphy et al. 2013;Pereira et al. 2018) but more advances are possible.

Increased satellite capability
We are currently in the midst of a revolution in satellite technology and availability as they become fully integrated into international society. Prior to 2009, individual (agencyowned) short life span, science-focused satellites were typically launched, whereas today, multiple long-term satellite programs exist (eg Aschbacher and Milagro-Pérez 2012) and commercial operators are promoting a plethora of satellite services. Examples of commercial activity include >150 optical CubeSats already in orbit for monitoring changes on land (Butler 2014) and the planned launch of 40 passive microwave CubeSats for monitoring weather (WebTable 2). This is occurring in parallel to the emergence of routine image and pattern recognition, artificial intelligence, and commercial availability of cloud computing, meaning that non-specialists can use a laptop computer to remotely access and analyze global satellite observations. For example, Google Earth Engine (Gorelick et al. 2017) now provides free access and computing for scientific analysis of global satellite observation datasets. Free and open access to these satellite data permits experimentation, removes barriers for research and exploitation, and could enable long-term monitoring. Thus, all components required for strategic monitoring of ocean C absorption already exist. Spacecraft operated by international agencies, such as Europe's Sentinel missions and NOAA's polar orbiting platforms, can provide the necessary long-term monthly measurements, while commercial satellites that deliver less accurate data could provide daily observations for identifying sudden or episodic (infrequent and short-term) changes (eg upwelling). Access to commercial data could be provided through an update of the existing International Charter for "space and major disasters", which already provides free access during episodic disasters (eg flooding). This update would acknowledge the "long-term man-made climate disaster".  Here, we identify key areas where satellite observations, in synergy with other methods, are likely to play a leading role in expanding scientific capability for studying and monitoring global ocean C (Figure 2).

Atmospheric CO 2 measurements near the water surface
Column-integrated CO 2 measurements obtained using remotesensing instruments with an accuracy of <1 part per million (ppm) over the ocean are now possible (Connor et al. 2016), and the ability to distinguish between lowermost tropospheric and upper tropospheric column CO 2 measurements means that CO 2 concentrations near the ocean surface can be separated from CO 2 transported over large distances (Kulawik et al. 2017). These lowermost tropospheric data could be used to identify temporal and spatial biases within the MBLR approach in oceanic regions influenced by continental airflow or point sources.
The accuracy of both approaches to resolve surface atmospheric CO 2 should be assessed using drone-observed or weatherballoon-observed vertical profiles (eg Corrigan et al. 2008).

Measuring near-surface concentrations
Satellite-based Earth observations focus on the top ~10 mm of the air-water interface and can be used to estimate nearsurface vertical gaseous concentration gradients that control gas fluxes. Increased knowledge of these near-surface concentrations is critical, given that concentration differences between air and water are relatively small and so any biases within these concentrations will have large impacts on calculated fluxes. The most accurate calculation of these air and water concentrations requires the temperature to be known on either side of a ~100-μm layer at the water surface. In-situ measurements of these two concentrations in the ocean are impractical, but new work has highlighted that they can be estimated through the combination of ship measurements, satellite observations, and knowledge of chemical kinetics and dynamics of the interface (Woolf et al. 2016). In Panel 1, we present the first global temporal analysis using this approach, and our analysis suggests that the oceanic sink of C is 18-21% larger than current estimates. This use of satellite observation also removes the impact of unknown biases within the in-situ measurements found in the large, collated datasets that document surface gaseous CO 2 . These biases, caused by different sampling conditions and equipment, were previously impossible to remove or characterize. Satellite-observed salinity data (eg Land et al. 2015) could be used to characterize the impact of near-surface salinity gradients on gas fluxes, which are likely to be important in regions with large river outflows and rainfall.

Direct measurements of multiple surface-exchange processes
Brightness temperature (radiance quantity of microwave energy leaving a surface) observations from passive microwave sensors provide a fundamental climate data record and can be used to infer rain, wind speed, sea state, foam, dielectric constant, salinity, and temperature (WebTable 2). The actual energy received by the sensor is a function of time and the surface area observed. This means that passive microwave measurements contain the signature of key processes controlling surface exchange (Figure 2, point 2), including the expansion and contraction of the surface ocean due to density changes and sea state, and therefore offer a unique and largely untapped resource for direct observations of multiple exchange processes. In-situ measurements of dissipation and emissivity coupled with measurements of gas exchange and eddy covariance fluxes should be used to identify robust, physically based relationships for describing gas exchange. Initial work should focus on the ~30-year global archive of passive microwave observations (WebTable 2) and SOCAT data. Due to the dynamic nature of the ocean, the large differences in time and space between field measurements and satellite observations will need to be accounted for.
Atmosphere-ocean fluxes of momentum, heat, and gas are interrelated but are typically studied in isolation. The wide range of current space observations offers the first solution for a more holistic and consistent approach to observing multiple exchange processes and properties. Gas exchange could be studied through the mean square slope to estimate surface expansion, wave breaking distributions, and wave energy dissipation. Combining momentum and heat flux methods in this way has already provided new knowledge in open ocean surface drag and heat process studies (Kudryavtsev et al. 2014), both of which are important for gas exchange. Such a holistic approach is likely to improve our understanding of how competing processes interact to influence gas exchange in polar waters (WebPanel 2). However, the study of compound events will require new satellite capabilities (eg concurrent high-resolution salinity, temperature, and wind-wave-current interactions).

Measuring internal ocean circulation and surface transport
The processes driving upwelling are well understood; information on the geographical location, the Earth's rotation, and local wind conditions is therefore sufficient to calculate Ekman flows. Knowledge of these properties has been applied to wind re-analysis data to infer the reason for changes in ocean C uptake in the Southern Ocean (Landschützer et al. 2015) but its application to satellite observations has yet to be fully exploited (Figure 4). The use of this theory in conjunction with satellite observations would constitute the first method for quantifying and monitoring large-spatial-scale threedimensional transport of momentum, heat, total CO 2 , and nutrients at critical locations. This would provide important constraints on global water mass budgets and global circulation, as well as providing the ability to characterize ecosystem health and identify regions most at risk from episodic acidification events. Daily passive microwave wind observations have been available since 1987, but new CubeSats will provide sub-daily observations of wind conditions (WebTable 2). This will make it possible to study how conditions at intense frontal regions affect gas exchange, as fronts themselves can greatly increase turbulence, which drives gas exchange (D'Asaro et al. 2011). Future satellite missions will expand these capabilities; the recently proposed Sea surface KInematics Multi-scale monitoring (SKIM; Ardhuin et al. 2018) and the Surface Water and Ocean Topography (SWOT; eg Oubanas et al. 2018) satellites will, by observing wind-wave-current interactions, enable the first direct spatial measurements of surface and vertical water transport. These satellites will provide the first measurements of vertical overturning (large-scale three-dimensional circulation) within equatorial regions and ocean basins (eg the Southern Ocean), as well as frontal turbulence and water movements within the marginal ice zone, across heterogeneous continental shelf boundaries and large river outflows.

Conclusions
Monitoring ocean C is crucial for identifying ecosystems most at risk of ocean acidification, closing the global C budget, informing policies to meet emissions targets, and tracking progress toward the goals of the historic Paris Climate Agreement developed by the 21st Conference of the Parties (COP21). Routine monitoring of the global oceans is now possible, but resources and impetus are required to help initiate long-term global ocean monitoring. Only satellites can provide the large-spatial-scale observations needed to efficiently and economically monitor the vast oceans, and these must now be used in synergy with in-situ observations. Such synergetic approaches are already leading to groundbreaking discoveries about the processes controlling gas fluxes. Our ability to observe the oceans and their ecosystems is increasing, and these advances, along with routine monitoring, will provide opportunities to fill existing knowledge gaps.