Global Perspectives on Observing Ocean Boundary Current Systems

Ocean boundary current systems are key components of the climate system, are home to highly productive ecosystems, and have numerous societal impacts. Establishment of a global network of boundary current observing systems is a critical part of ongoing development of the Global Ocean Observing System. The characteristics of boundary current systems are reviewed, focusing on scientific and societal motivations for sustained observing. Techniques currently used to observe boundary current systems are reviewed, followed by a census of the current state of boundary current observing systems globally. Next steps in the development of boundary current observing systems are considered, leading to several specific recommendations.


INTRODUCTION
Ocean boundary current systems are where society most frequently interacts with the ocean through fisheries, maritime transportation, oil and gas extraction, and recreation. These systems are home to intense and highly variable oceanic currents that redistribute mass, heat, salt, biogeochemical constituents, plankton, and pollution. Circulation patterns also influence the life history, foraging behavior, and abundance of many marine species (e.g., Mansfield et al., 2017). The coastal and open oceans are linked through boundary current systems where events such as coastal upwelling, sea level anomalies, primary productivity, fisheries, and weather are propagated between domains by various processes (e.g., eddies, Rossby waves, and advection). Boundary currents may be broadly categorized as either western boundary currents (WBCs; Imawaki et al., 2013) or eastern boundary currents (EBCs) based on their governing dynamics. In each ocean basin, WBCs play a prominent role in the climate system by redistributing heat from the equator toward the poles, while EBCs are some of the most biologically productive regions in the world and respond dramatically to climate variability (Chavez et al., 2008;Chavez and Messié, 2009).
In our changing climate, shifting hydrological cycles and weather patterns are expected to strongly impact oceanic boundary current processes. Observational evidence for such shifts is beginning to appear. Wu et al. (2012) noted enhanced warming of subtropical WBCs and their extensions during the 20th century, possibly linked to their poleward shift or intensification. Changes in the stability of WBCs have also been noted, with instabilities in the Gulf Stream shifting westward (Andres, 2016), increasing influence of warm core rings on shelf circulation , and a trend toward greater instability in the East Australian and Agulhas Currents (Sloyan and O'Kane, 2015;Beal and Elipot, 2016).
Oceanic ecosystems are being exposed to increasing pressure from major stressors including warming, deoxygenation, fishing, and acidification. EBCs in particular are projected to be strongly impacted by these stressors (Bakun et al., 2015). For instance, the Peru-Chile (Humboldt) Current system (see the section Peru-Chile Current System), a highly productive EBC and a regional source of greenhouse gases, is naturally affected by upwelling of offshore waters with low oxygen and pH onto the continental shelf (Helly and Levin, 2004) and by periodic El Niño Southern Oscillation (ENSO) events that change the water mass distributions, oxygenation, and productivity (Chavez et al., 2008;Gutiérrez, 2016;Graco et al., 2017); further, stress could have significant consequences for the regional ecosystem. Similarly, changes in the Gulf Stream under global warming are predicted to negatively impact fisheries in the Gulf of Maine and on the New England Shelf (Saba et al., 2016;Claret et al., 2018).
Sustained, interdisciplinary observations in boundary current regions are required for a comprehensive ocean observing system. For OceanObs'09, Send et al. (2010) proposed a global network of sustained monitoring arrays as part of the Global Ocean Observing System (GOOS). Send et al. (2010) broadly defined the properties to be observed as (1) the transports of mass, heat, and freshwater needed for monitoring the global climate in conjunction with basin-scale measurements and (2) local boundary-specific properties including eddy activity, changes in potential vorticity, air-sea interactions , ecosystem dynamics, and biogeochemistry. More recently, the 2017 GOOS workshop on "Implementation of Multi-Disciplinary Sustained Ocean Observations" (IMSOO; Palacz et al., 2017) focused, in part, on how to proceed with the development of a truly multidisciplinary boundary current observing system, building upon the more physical and climatefocused plans of Send et al. (2010). In particular, it was noted that observations that resolve along-boundary variability are needed in order to understand climate impacts on various societally relevant uses of boundary current systems (e.g., fisheries). The need to maintain a global perspective that targets all boundary current systems has been repeatedly recognized (Send et al., 2010;Palacz et al., 2017), particularly in developing nations where fisheries can be centrally important (Palacz et al., 2017). To that end, IMSOO planned to review established observing systems in the California Current System and East Australian Current in order to develop a blueprint for an adaptive, multidisciplinary observing system with relocatable subsystems to capture finer scales (Palacz et al., 2017).
Oceanic boundaries present a variety of challenges for sustained observing systems (Send et al., 2010). With strong flows in relatively shallow areas, spatial scales of O(1)-O(10) km, and temporal scales often shorter than a few days (e.g., He et al., 2010;Todd et al., 2013;Rudnick et al., 2017), the broad-scale (i.e., Argo and gridded satellite altimetry) and long-duration (e.g., HOTS, BATS, Station P, and CARIACO) measurements that constitute the observing system for the ocean interior are insufficient for boundary current systems. Multiple observing strategies are needed to measure the essential ocean variables (EOVs; Table 1) that can be used to understand and track the physical and biogeochemical processes of interest within boundary currents (Lindstrom et al., 2012). The optimal combination of observing methods will depend upon characteristics unique to each region. Send et al. (2010) noted that an additional challenge in observing boundary current systems is that there is no well-defined offshore "end" of a boundary current but rather a temporally and spatially variable transition to the interior. At the same time, oceanic boundaries generally lie within exclusive economic zones (EEZs), so the implementation of observing systems requires significant international cooperation.
The overarching purpose of this review is to examine the current state of the boundary current system component of GOOS, updating and building upon the OceanObs'09 review of Send et al. (2010). The section Scientific and Societal Needs considers the scientific and societal needs that comprehensive boundary current observing systems must fulfill. The section Observing Techniques reviews how various observing techniques are employed in boundary currents, highlighting key scientific advances from each platform. The section Current Status of Regional Boundary Current Observing Systems surveys the current state of boundary current observing systems globally. Table 2 provides a comprehensive collection of publications and datasets from the past decade, organized by region and platform. The section Future Outlook then considers the future development of boundary current observing systems. The section Summary Recommendations concludes with specific recommendations to promote development of a comprehensive global network of boundary current observing systems.

SCIENTIFIC AND SOCIETAL NEEDS
The Framework for Ocean Observing (Lindstrom et al., 2012), developed after OceanObs'09, recommended that ocean observing systems (1) be "fit for purpose" and driven by "scientific inquiry and societal needs"; (2) include physical, biogeochemical, and biological observations; (3) operate collaboratively based on established best practices; (4) balance innovation with stability; (5) promote alignment of independent user groups; (6) build on existing infrastructure as much as possible; and (7) provide maximum benefit to all users from each observation. Here, we present the scientific and societal needs that that should be met by comprehensive observing of oceanic boundary current systems, focusing on three broad categories: ecosystems and biogeochemistry (see the section Ecosystems and Biogeochemistry), weather and climate (see the section Climate and Weather), and connections between the shelves and deep ocean (see the section Shelf-Deep Ocean Connections).

Ecosystems and Biogeochemistry
Boundary current systems play an important role in carbon cycling through the physical and biological carbon pumps. WBCs are major sites of air-sea CO 2 exchange (e.g., Rodgers et al., 2008;Gorgues et al., 2010;Nakano et al., 2011) and have been shown to exhibit enhanced contemporary carbon uptake from the atmosphere (Takahashi et al., 2009;Landschützer et al., 2014). WBC CO 2 uptake is driven by a large pCO 2 disequilibrium with the overlying mid-latitude atmosphere, which is due to the rapid cooling of low Revelle factor waters advected from the tropics to midlatitudes. Since thick subtropical mode waters form during wintertime convection on the equatorward edges of the WBC extensions, the mode waters are key carbon sinks (e.g., Bates et al., 2002;Gruber et al., 2002;Ito and Follows, 2003;Levine et al., 2011;DeVries, 2014;Iudicone et al., 2016) and have been the target of detailed observational carbon studies (Andersson et al., 2013;Palevsky and Quay, 2017). However, it is still unclear how variability in the rate of mode water formation might impact ocean carbon uptake in these regions and what impacts these changes might have on the biological pump and higher trophic levels (e.g., fisheries). In the Kuroshio Extension region, there is evidence that the majority of carbon exported from the surface ocean during the spring and summer productive season is subsequently respired in the seasonal thermocline and ventilated back to the atmosphere during wintertime mode water formation (Palevsky et al., 2016;Fassbender et al., 2017b;Palevsky and Quay, 2017;Bushinsky and Emerson, 2018). The Southern Hemisphere WBCs are chronically undersampled, particularly during winter, leading to significant uncertainty in their contribution to the global ocean carbon sink.
Boundary current systems are highly productive regions (Chavez et al., 2008). The mechanisms of nutrient supply to surface waters that drive increased primary productivity differ among EBC and WBC systems, but their global contributions are similar (Chavez and Toggweiler, 1995). In EBC systems, the dominant source of nutrients is coastal upwelling (Chavez and Messié, 2009), while in WBC systems, geostrophic-and eddydriven upwelling predominates (Pelegrí and Csanady, 1991). Nutrient streams are important in the Gulf Stream (Pelegrí and Csanady, 1991;Pelegrí et al., 1996;Williams et al., 2006Williams et al., , 2011Palter and Lozier, 2008) and the Kuroshio (Guo et al., 2012(Guo et al., , 2013, transporting subsurface positive nitrate anomalies, which are delivered to the photic zone primarily by mesoscale and submesoscale processes (Nagai and Clayton, 2017;Honda et al., 2018;Yamamoto et al., 2018;. Nutrient cycles and drivers have not yet been studied in WBC systems of the Southern Hemisphere. Western boundary currents are also enriched in micronutrients (e.g., Fe, Zn, Cd, Co, and Ni) from land-sea exchanges.
They ultimately feed open ocean surface waters and, at lower   Steinfeldt et al., 2015;Capet et al., 2017;Klenz et al., 2018;Machu et al., 2019;Thomsen et al., 2019 latitudes, the equatorial undercurrent, where these micronutrients are critical in maintaining high levels of productivity. For instance, iron transported by boundary currents in the western Pacific feeds into the Pacific Equatorial Undercurrent, which then supplies iron to the eastern equatorial Pacific (e.g., Mackey et al., 2002;Ryan et al., 2006). In the North Atlantic, Gulf Stream rings supply iron to the subtropical gyre (e.g., Conway et al., 2018). Subpolar WBCs such as the Oyashio and Malvinas Currents are also likely to transport waters enriched in nutrients; wind-driven and shelfbreak upwelling then supplies nutrients to the euphotic layers, enhancing biological productivity (Matano and Palma, 2008;Ito et al., 2010;Valla and Piola, 2015). Locations at which subtropical and subpolar WBCs meet provide ideal environments for biological production, as warm subtropical waters converge with nutrient-rich subpolar waters (Brandini et al., 2000). The upwelling of deep, poorly ventilated water masses rich in inorganic nutrients and CO 2 and low in O 2 make EBCs areas of high air-sea fluxes, and the sensitivity of the upwelling process to climate variability contributes to large interannual and decadal scale changes in the magnitude of these fluxes (Friederich et al., 2002;Brady et al., 2019). EBCs also exhibit strong cross-shore gradients in fluxes; narrow strips of the nearshore ocean act as intense sources of CO 2 to the atmosphere, while the abundance of nutrients in these upwelled waters facilitates primary production that results in net uptake of CO 2 (Hales et al., 2005). The supply of poorly ventilated waters combined with high levels of organic-matter remineralization resulting from intense primary production in surface waters can trigger periods of anoxia and low pH in shelf waters (Feely et al., 2008;Zhang et al., 2010) with severe consequences for demersal and pelagic ecosystems (Chan et al., 2008;Monteiro et al., 2008;Bertrand et al., 2011).
Boundary currents play an important role in ocean ecosystems across all trophic levels. The intense levels of primary production associated with EBCs support rich ecosystems with relatively short food chains, and these systems provide at least 20% of the world's wild-caught fish despite covering less than 1% of the global ocean (Chavez and Messié, 2009). WBCs and EBCs are also oceanic regions where coastal and open ocean ecosystems are brought together and interact. Modeling studies have suggested that boundary currents are hotspots of microbial biodiversity (Barton et al., 2010;Clayton et al., 2013). This has been supported in the Kuroshio Extension by some in situ surveys (Clayton et al., 2014. At the other end of the trophic spectrum, recent work combining tag data and satellite altimetry data has shown that white sharks (Carcharodon carcharias) actively occupy warm-core anticyclonic eddies in the Gulf Stream (Gaube et al., 2018). The warmer waters in these mesoscale features allow the sharks to reduce the physiological costs of thermoregulation in cold water, thereby making prey more accessible and energetically more profitable. Similarly, the location of the Kuroshio axis and associated changes in water temperature have been shown to influence the behavior of juvenile Pacific bluefin tuna (Thunnus orientalis; Fujioka et al., 2018). In the Southern Benguela EBC upwelling system, the coastal, wind-driven upwelling along the southwest African coast supports planktonic food supplies for young pelagic fish, while the temperate Agulhas Bank shelf region provides suitable spawning habitat for large communities of fish including in particular anchovy and sardine (Hutchings et al., 2009a). Likewise, southern elephant seals feed along the intense fronts and eddies in the Brazil/Malvinas Confluence (Campagna et al., 2006). WBCs are also known to play an important role in the migration of other coastal and pelagic organisms, such as eels (Shinoda et al., 2011;Rypina et al., 2014) and salmon (Wagawa et al., 2016).
Marine heat waves (MHWs) are strongly linked with boundary current systems. For instance, the exceptional and devastating MHW event off Western Australia during summer of 2010/2011 was caused by a strengthening of the Leeuwin Current associated with La Niña conditions (Pearce and Feng, 2013;Feng et al., 2015), a 2014-2015 MHW had unprecedented impacts on the California Current System (Di Lorenzo and Mantua, 2016;Zaba and Rudnick, 2016), and an MHW in 2015-2016 impacted the Tasman Sea (Oliver et al., 2017). These discrete, prolonged periods of anomalously warm waters at particular locations (Hobday et al., 2016) can stress ecosystems, leading to increased mortality of marine species, closing of commercial and recreational fisheries, and coral bleaching (Cavole et al., 2016;Stuart-Smith et al., 2018). The addition of other stressors such as ocean acidification and deoxygenation, which are projected to increase in future warming scenarios, could amplify the ecosystem impacts of MHWs. Sustained physical and biogeochemical observations are necessary to improve forecasts of the frequency and magnitude of MHWs, as well as to assess the risk and vulnerability of marine ecosystems to extreme climate events (Frölicher and Laufkötter, 2018).

Climate and Weather
Boundary currents are an integral part of the global climate system as they redistribute heat and facilitate carbon uptake from the atmosphere (see the section Ecosystems and Biogeochemistry). In the Atlantic, boundary currents are key components of the Atlantic Meridional Overturning Circulation (AMOC; Frajka- Williams et al., 2019). Low-latitude WBCs that connect the subtropics to the equator at thermocline and intermediate levels are important contributors to the mass and heat budgets of the equatorial oceans, which influence climate modes such as ENSO (Lengaigne et al., 2012). Low-latitude WBCs are also suspected to contribute to the decadal modulation of the equatorial thermocline background state (e.g., Lee and Fukumori, 2003). Sustained monitoring of WBC transports would be particularly useful for climate and seasonal-to-decadal forecast centers (see Smith et al., 2019).
As climate change progresses, boundary current systems are likely to undergo further significant changes. Subtropical WBCs and their extensions are the fastest warming regions of the world ocean Yang et al., 2016). Climate model simulations have suggested that western boundary current extensions may move poleward under climate change (Saba et al., 2016). This poleward expansion of energetic WBCs may impact extreme temperatures and marine species migration (Johnson et al., 2011), as well as enhance eddy activity regionally (e.g., Oliver et al., 2015). While low-resolution climate models suggest strengthening and poleward migration of several of these currents under climate change, particularly in the Southern Hemisphere (Sen Gupta et al., 2012;Hu et al., 2015;Pontes et al., 2016), studies leveraging in situ velocity and satellite data suggest no significant increase in their transports since the early 1990s (Rossby et al., 2014;Beal and Elipot, 2016). This discrepancy motivates the collection of long-term measurements of baroclinic changes in boundary currents (i.e., subsurface temperature and salinity properties), as well as the vertical structure of the velocity, in order to understand and predict future changes.
In addition, ocean warming and a magnified hydrological cycle could drive significant changes in shelf ocean stratification, while changes to wind forcing will directly alter rates of upwelling. These ocean circulation processes, and meteorological forcing at the scales that impact upwelling, are poorly represented in climate models (Richter, 2015;Zuidema et al., 2016). Thus, we have little capability to predict how upwelling, winds, and other physical drivers of ocean property exchanges at the coastal/open ocean boundary will change in the future. The impact these changes will have on coastal ecosystems is simply unknown.
Detection and attribution of global sea level variability has improved considerably in the last decade (Cazenave et al., 2014;Marzeion et al., 2014). The location and strength of WBCs considerably influence the mean local sea level (Domingues et al., 2016;Archer et al., 2017b), possibly accounting for part of the mismatch between forecasts and observations of sea level at the coast (Ezer, 2015). Relationships between large-scale wind anomalies, basin-wide sea surface height (SSH), and WBCs (e.g., Boening et al., 2012;Volkov et al., 2019) suggest that observations of current strength and oceanic teleconnections can be used to improve seasonal to decadal coastal sea level forecasts, leading to improved assessments of impacts on infrastructure and groundwater quality (Slangen et al., 2014;Park and Sweet, 2015).
Boundary current systems influence synoptic and longer scale weather patterns. Large upper ocean heat content within WBCs can fuel development and intensification of tropical cyclones (Bright et al., 2002;Wu et al., 2008;Nguyen and Molinari, 2012;Galarneau et al., 2013). Strong sea surface temperature (SST) gradients across WBCs, particularly during winter months, destabilize the atmospheric boundary layer, fueling the midlatitude storm tracks and atmospheric blocking frequency, which in turn impact regional climate (Kelly et al., 2010;Nakamura, 2012;O'Reilly and Czaja, 2015;O'Reilly et al., 2016;Révelard et al., 2016;Ma et al., 2017). For instance, a weaker Gulf Stream SST front leads to a decrease in cold and dry spells over Europe (O'Reilly et al., 2016), while a sharper SST front in the Kuroshio Extension increases cyclogenesis and shifts the storm track northward, causing warming over eastern Asia and the western United States that can reduce snow cover by 4-6% (O'Reilly and Czaja, 2015;Révelard et al., 2016). Variability in the warm waters of the Agulhas influences summer rainfall over parts of South Africa (Jury et al., 1993;Nkwinkwa Njouodo et al., 2018). In EBC systems, SST minima are collocated with maxima in sea level pressure that are in turn associated with alongshore wind stress, wind stress curl, and cloud cover along the boundary (Sun et al., 2018), suggesting coupling with the full Hadley-Walker tropical atmospheric circulation, though the details of such coupling remain an open question.
Accurate weather and climate forecasting thus requires accurate representation of boundary current systems. However, most of the current ocean reanalyses used to initialize the monthly, seasonal, and decadal forecasts exhibit large errors in the boundary currents (Rouault et al., 2003;Valdivieso et al., 2017), hampering forecast performance. Coupled climate models, such as those used in the Intergovernmental Panel on Climate Change reports, also exhibit large deficiencies in boundary current regions (e.g., Siqueira and Kirtman, 2016;Zuidema et al., 2016), including warm SST biases in EBCs (e.g., Large and Danabasoglu, 2006). Current modeling and data assimilation capabilities are insufficient to fully represent boundary currents at the small spatial scales needed for forecasting. Subramanian et al. (2019) further consider how observing efforts, including within boundary currents, can contribute to improved subseasonal-toseasonal forecasting.

Shelf-Deep Ocean Connections
The coastal ocean and nearshore zones support a broad range of human activities in maritime industries and resource extraction, and the environmental health and productivity of these regions deliver important ecosystem services. As already noted, the proximity of energetic boundary currents in deep water adjacent to continental shelves mediates shelf-sea/deep-ocean exchange of properties. Along many coasts, this forcing can match or exceed local drivers of circulation such as tides, wind, and river inflows. Coastal ocean and shelf edge dynamics have immediate impacts on ecosystem function and productivity on weekly to seasonal timescales but can also drive multi-decadal changes in ecosystem structure through effects on habitat ranges and biodiversity, not only in coastal zones but also at basin scales.
While we have a broad understanding of the dynamics of upwelling in both WBC and EBC regimes, quantitative estimates of net shelf-sea/deep-ocean exchanges of freshwater and tracers integrated over extended along-shelf distances are few. Quantifying these exchanges is challenging where shelf-edge flow-bathymetry interactions foster variability at short length and timescales. Similarly, exchange flows are not always readily observable at the sea surface from satellite or shore-based remote sensing technologies (see the section Remote Sensing) because they are associated with bottom boundary layer flow driven by the boundary current encountering the seafloor or subduction at the sea surface due to boundary current detachment and mixing. Two efforts along the U.S. East Coast are striving to make such measurements using multi-platform observing arrays: the Processes driving Exchange At Cape Hatteras (PEACH) program and the Ocean Observatories Initiative (OOI; Smith et al., 2018;Trowbridge et al., 2019) Pioneer Array (see the section Northwestern Atlantic). Similarly, in situ and satellite remote sensing observations combined with high-resolution numerical simulations have provided insights into the shelfsea/deep-ocean exchanges near the confluence of the Brazil and Malvinas Currents Matano et al., 2014;Strub et al., 2015).
On narrow continental shelves adjacent to intense boundary currents, the impact of deep-ocean circulation on the shelf system is immediate, driving significant fluxes across the continental shelf edge through mesoscale and boundary layer dynamics. For example, mesoscale and submesoscale meandering of the Agulhas jet leads to strong episodic exchanges with shelf waters (Krug et al., 2017;Leber et al., 2017) that support high productivity over the eastern Agulhas Bank (Probyn et al., 1994) and may influence the well-known sardine run (Fréon et al., 2010). On broad continental shelves, bathymetric constraints on cross-isobath flow can hamper exchange at the shelf edge, trapping terrestrial inflows and establishing appreciable crossshelf buoyancy gradients that in turn sustain shelf-edge fronts (Fratantoni and Pickart, 2007;Howatt et al., 2018).
With changing climate, ocean warming and changes to the hydrological cycle could drive changes in vertical thermal stratification and across-shelf salinity stratification, altering ocean conditions at the inshore edge of boundary current systems (e.g., Gawarkiewicz et al., 2018) and potentially impacting across-shelf fluxes of nutrients and micro-nutrients that are important to sustaining coastal productivity (Fennel et al., 2006). Changes in watershed land use and global weather will alter the volume and characteristics of river flows discharged into the coastal zone. At continental shelf scales, key areas of uncertainty in the oceanographic response to climate variability and change include submesoscale processes and open oceanshelf exchange. Sustained observing efforts are needed that more fully capture the influence of boundary currents on exchanges with the coastal zone. Designing and deploying boundary current observing systems capable of operating across shelf and deep ocean regimes to deliver coherent views of the shelf-edge exchange is challenging.

OBSERVING TECHNIQUES
The highly variable and multi-scale characteristics of boundary currents necessitate an integrated observing system approach, in which high-resolution observations are nested within a backbone of observations over a broad area. Under the Framework for Ocean Observing (Lindstrom et al., 2012), design and implementation of ocean observing systems are focused around a set of EOVs that include physical, biogeochemical, and ecosystem parameters (Table 1) 1 . Design of an observing system for a particular region (e.g., a specific boundary current system) should proceed through a series of "readiness levels." In the concept phase, initial feasibility studies and peer review of proposed plans take place. Then, in the pilot phase, smallscale deployments are used to test and validate the proposed approach. Once the observing system reaches the mature phase, it is part of the sustained Global Ocean Observing System. No single observing platform can provide all of the necessary measurements (Table 1), so an optimal mix of observing platforms is needed. Determination of this mix of platforms will be specific to a particular boundary current system, taking into consideration the unique processes and scales at play in that system. Here, we briefly review how various observing platforms are currently being used in boundary current systems; Table 2 1 http://www.goosocean.org/eov refers to many other examples of these observing techniques being applied to boundary current systems.

Time Series
Time series measured from platforms fixed to the seafloor have long been and continue to be central to observing system design and implementation since they uniquely enable collection of long-term measurements at high temporal resolution (minutes to hours) at key locations. Traditional tall moorings (e.g., Johns et al., 2005) typically carry instruments on the mooring wire, within subsurface floats, and on surface buoys, if present; instruments are available to measure most physical EOVs and a growing number of biogeochemical and ecosystem EOVs ( Table 1). Moored surface buoys additionally carry suites of meteorological sensors on the buoy tower and sensors for biogeochemical and physical EOVs on the buoy bridle and mooring line just below the sea surface; these air-and seasurface measurements can be combined to estimate the air-sea exchanges of heat, moisture, CO 2 , and momentum . Inverted echo sounders (IESs) measure the time for sound pulses to travel from the bottom-mounted IES to the surface and back, which, in regions with good databases of hydrographic measurements, can provide full water column estimates of temperature, salinity, and density using the gravest empirical mode technique (Meinen and Watts, 2000). In the Florida Strait, a unique time series of volume transport has resulted from measuring the voltage induced in a submarine cable by seawater moving through the Earth's magnetic field (Larsen and Sanford, 1985;Baringer and Larsen, 2001;Meinen et al., 2010).
Dense, moored arrays of instruments remain the most effective way to return volume and property transport measurements with high temporal resolution. Subsurface moorings are more typical in WBCs due to the strong surface currents, although surface moorings have also been successfully deployed in the Gulf Stream (Weller et al., 2012) and Kuroshio Extension . Arrays of IESs can be used to infer geostrophic shear profiles and, with the addition of bottom pressure sensors (PIES) and near-bottom current measurements (CPIES), can provide estimates of the absolute geostrophic current Meinen et al., 2018). However, the high costs of building, deploying, and turning around such arrays makes them feasible only at a few key locations. Other observing assets are needed to provide spatially broad measurements.

Ship-Based Measurements
Measurements from both dedicated research vessels and ships of opportunity have been central to observing boundary current systems for decades. Research vessels can measure nearly every EOV (Table 1) through the full depth of the water column and are uniquely capable of collecting many types of samples (e.g., net tows and large-volume water samples). Ongoing sustained research vessel surveys of ocean boundary currents include the global GO-SHIP transects at 25-to 50-km resolution (Talley et al., 2016) and the California Cooperative Oceanic Fisheries Investigations (CalCOFI) surveys (McClatchie, 2014) in the California Current System (see the section California Current System). The servicing of boundary current mooring arrays, generally undertaken from research vessels, provides unique opportunities to undertake intensive process studies targeting key scientific questions. The primary limitations on research vessels' contribution to sustained boundary current observing are their high costs of operation (typically tens of thousands of dollars per day, excluding science personnel) and the infrequency of cruises.
The World Meteorological Organization (WMO) Voluntary Observing Ship (VOS) Program and Ship of Opportunity Program (SOOP) both make use of non-research vessels to collect observations globally, substantially augmenting the amount of ship-based observing. Both programs collect meteorological measurements with real-time observations benefiting weather forecasting, while SOOP additionally uses commercial ships to collect oceanographic measurements along frequently occupied trade routes in the global ocean. Oceanic measurements from SOOP include temperature profiles from expendable bathythermographs (XBTs) at 10-to 25-km resolution in boundary currents , surface temperature, salinity, plankton, and pCO 2 from flow-through systems, and, on specially equipped vessels, velocity profiles from hull-mounted ADCPs (e.g., M/V Oleander; Rossby et al., 2010). Several repeat transects across boundary currents have been maintained for multiple decades and so represent some of the longer datasets available (see the section Current Status of Regional Boundary Current Observing Systems). Fast-moving ships are often able to occupy transects directly across strong boundary currents in short periods of time, a feat not yet possible with other sampling platforms. However, subsurface measurements of variables other than temperature and velocity have remained elusive from ships of opportunity, and recovery of instruments deployed over the side is not practical on cargo vessels.

Autonomous Underwater Gliders
Autonomous underwater gliders (Rudnick, 2016b;Testor et al., 2019) routinely collect long-duration, high-resolution observations in a variety of boundary current systems globally (Todd et al., 2018b; Table 2). Gliders typically profile from the surface to 500-1,000 m, taking 3-6 h to complete a cycle from the surface to depth and back while covering 3-6 km horizontally through the water at a speed of about 0.25 m s −1 . During a mission lasting 3-6 months, a glider's survey track extends well over 2,000 km. Owing to the relatively slow speed of gliders, care must be taken when interpreting the observations, which contain both spatial and temporal variability (Rudnick and Cole, 2011). Sustained deployment of networks of gliders can provide observations with both high spatial resolution and year-round coverage (e.g., Figures 1A,B).
Realizable glider-based sampling plans in boundary currents vary primarily due to the strength of currents relative to a glider's speed. In EBCs and other boundary currents with relatively weak depth-average currents, gliders can occupy repeat survey lines. The California Underwater Glider Network (CUGN; Figure 1, left), which consists of three cross-shore transects off southern and central California that have been continuously occupied for more than a decade , exemplifies sustained glider observations in an EBC. In WBCs and other boundary currents where depth-average currents are significantly faster than a glider's speed through the water, gliders can be navigated so as to cross the observed flow as they are advected downstream, returning oblique transects. For example, multi-year surveys of the Gulf Stream (Figure 1, right; Todd et al., , 2018aTodd, 2017;Todd and Locke-Wynn, 2017) have now returned over 150 high-resolution transects across the WBC of the North Atlantic. Testor et al. (2019) further discuss efforts associated with the OceanGliders Boundary Ocean Observing Network (BOON).
Gliders can carry a variety of sensors (e.g., Figures 1C-F). Measurements of pressure, temperature (Figure 1D), conductivity, and depth-average currents are standard, enabling estimates of absolute geostrophic transport and other physical parameters at relevant scales in boundary currents. Measurements of biooptical (e.g., Niewiadomska et al., 2008;Henderikx Freitas et al., 2016) and bio-acoustic properties (e.g., Baumgartner and Fratantoni, 2008;Van Uffelen et al., 2017), dissolved oxygen (e.g., Figure 1E; Perry et al., 2008), nitrate, turbulent microstructure (St. Laurent and Merrifield, 2017), and velocity profiles ( Figure 1F;  are becoming increasingly common. The main constraints on sensors for gliders are the requirements for small-size, low-power consumption and multimonth stability. As sensor technology continues to mature, gliders will be well suited to carry sensors for additional EOVs, such as pH, in boundary currents.

Drifters
Surface Velocity Program (SVP) drifters drogued at 15-m depth (Niiler et al., 1995;Niiler, 2001;Centurioni, 2018) deployed as part of the Global Drifter Program (GDP) and the Global Surface Drifter Array (GSDA) are also important for understanding the structure and variability of boundary current systems. The GSDA archive dates back to February 1979 and includes over 32 million records of geographical location, 15-m depth velocity, and SST at 6-h resolution (e.g., Hansen and Poulain, 1996;Lumpkin and Pazos, 2007).
Drifter observations have been widely used in both EBCs and WBCs (see Table 2). Recently improved analysis techniques (e.g., Lumpkin, 2003;LaCasce, 2008;Koszalka and LaCasce, 2010;Laurindo et al., 2017) and expansion of the Lagrangian drifter array have allowed gridded, Eulerian statistics of nearsurface velocity to be produced at higher resolution, resulting in improved estimates of near-surface flow in boundary currents (e.g., Figure 2) at seasonal to interannual timescales (e.g., Niiler et al., 2003;Lumpkin and Johnson, 2013). Drifter observations in boundary currents offer opportunities for new analyses of long-term variability and trends (e.g., Johnson, 2001;Lumpkin and Johnson, 2013) and the dispersion of tracers and marine debris in the upper ocean (Lumpkin et al., 2012;van Sebille et al., 2015), which is driven by turbulence at scales from surface waves through the submesoscale to large-scale geostrophic eddies Lund et al., 2018).

Argo Floats
Over the past two decades, autonomous profiling Argo floats have become cost-effective and robust platforms. Over 3,700 active Argo floats provide global measurements of temperature, salinity, and pressure in the upper 2,000 m of the ocean, and some are also equipped with sensors measuring biogeochemical properties (Riser et   Though the Argo network was not designed to capture the details of boundary currents and lacks the resolution necessary to resolve narrow boundary currents, Argo data have nevertheless been used extensively in both WBCs and EBCs (see Table 2). Argo complements other boundary current observing efforts by providing collocated temperature and salinity measurements that are used to infer geostrophic shear from XBT temperature profiles, extending geostrophic shear from XBT and ocean glider data to 2,000 m, measuring reference velocities at parking depth (typically 1,000 m), and linking transport measurements of boundary currents to the ocean interior through basinwide integration (e.g., Zilberman et al., 2018). Following recommendations at OceanObs'09 , the Argo program is currently moving to double float density in WBC regions (Jayne et al., 2017). The Kuroshio ( Figure 3A) and Gulf Stream have historically been among the more densely populated sectors in the Argo array, while other boundary current regions (e.g., the Peru-Chile system, Figure 3B) lack the desired coverage.

Remote Sensing
Among the many oceanic variables that are routinely measured from satellites (Table 1), SSH, SST, and ocean color have been most used to study boundary current systems. Satellite measurements typically have resolutions of O(1)-O(10) km along the satellite track, with repeated measurements on daily to 10-day timescales at a given location. Boundary currents often have strong signatures in both SSH and SST, so satellite-derived gradients in these properties can approximate the strength and/or position of these currents (e.g., Imawaki et al., 2001), including variability on longer timescales (e.g., Qiu and Chen, 2005;Qiu et al., 2014;Andres, 2016). Synergy between dynamic height derived from temperature and salinity profiles and SSH can be exploited to produce synthetic reconstructions of boundary currents (van Sebille et al., 2010;Beal and Elipot, 2016;Majumder and Schmid, 2018;Zilberman et al., 2018), although these reconstructions crucially depend on assumptions about the nonsteric (barotropic and mass) variability. Weaknesses of SSH for observing boundary currents include reduced measurement quality within 40 km of the coast due to large uncertainties in the wet-tropospheric correction, unfiltered tides, and a lack of sufficient temporal and spatial resolution to capture the full spectrum of near-surface current variability observed by drifters (Poulain and Niiler, 1989;Centurioni and Niiler, 2003;Fratantoni and Richardson, 2006;Centurioni et al., 2008Centurioni et al., , 2009Maximenko et al., 2009). Products that combine SSH and drifter measurements have improved eddy kinetic energy and dynamic topography estimates (Maximenko et al., 2009;Lumpkin and Garzoli, 2011;Rio et al., 2014;Rio and Santoleri, 2018). Estimates of chlorophyll derived from satellite ocean color measurements provide information on biological productivity in boundary current systems worldwide (e.g., Messié and Chavez, 2015;Gómez-Letona et al., 2017). Because ocean color observations have higher resolution (O(1) km) than satellite altimetry measurements, they potentially provide insight into the rich fields of submesoscale instabilities that exist within  boundary current systems (Figure 4; Everett et al., 2014;Lee and Kim, 2018).
High-frequency (HF) radars (Paduan and Washburn, 2013) have been used effectively to monitor surface current variability of boundary currents (e.g., Kim et al., 2011;Archer et al., 2018). They directly map the total surface current within O(100) km of the coast at high resolution in time (∼1 h) and space (∼1 km) during long-term deployments (∼10 years). HF radar observations have proven useful for investigating both the mean surface velocity structure of boundary currents and associated submesoscale features that develop as boundary currents meander and shed eddies (Archer et al., 2018;Soh and Kim, 2018). Combining HF radar velocity estimates with satellitebased measurements of SST and ocean color (e.g., Figure 4) can provide a multidisciplinary view of surface circulation features at O(1)-km scales (e.g., . Some radar sites have been in continuous operation for more than a decade, offering opportunities to examine interannual to decadal variability of surface circulation. New radar sites can be installed and daisychained with existing sites, providing measurements of the alongshore evolution of boundary currents, as has been achieved along the West Coast of the United States .

CURRENT STATUS OF REGIONAL BOUNDARY CURRENT OBSERVING SYSTEMS
Existing observing systems for particular boundary currents are in various stages of development. Here, we review the current status of the observing systems currently operating in several EBCs and WBCs globally. The California Current System (see the section California Current System) is arguably the most well sampled boundary current in the world, offering hope that a fully integrated physical and biogeochemical system is achievable. Other boundary current systems, particularly in the Southern Hemisphere, are much less sampled. As was the case a decade ago (Send et al., 2010), biogeochemical and ecosystem EOVs (Table 1) remain much less well sampled than physical EOVs. Table 2 provides a more comprehensive collection of recent scientific results for each boundary current system as well as sources of publicly available observations.

California Current System
The California Current System is the EBC system of the subtropical North Pacific (Checkley and Barth, 2009). The equatorward flowing California Current carries relatively cool and fresh waters of subpolar origin, while the poleward California Undercurrent (Gay and Chereskin, 2009;Todd et al., 2011b) transports warmer saltier waters from the tropics along the continental margin. The California Current System is strongly influenced by the predominantly upwelling-favorable winds along the West Coast of North America.
Owing to the need to understand the collapse of the regional sardine fishery in the 1940s, there is a well-developed multidisciplinary observing system in the California Current System with a decades-long history of routine observations by the CalCOFI program (McClatchie, 2014, and references therein). Since 1949, CalCOFI has made regular (currently quarterly) measurements of physical, biological, and chemical properties at fixed stations along survey lines oriented perpendicular to the coast from research vessels (Figure 5)  the pelagic ecosystem in response to ENSO, the Pacific Decadal Oscillation, and secular trends. In the Northern California Current, the Newport Hydrographic Line (44 • 39.1'N) has been continuously occupied since 1961 (Huyer et al., 2007). Since 2005, autonomous underwater gliders have continuously surveyed along three of the CalCOFI lines as part of the CUGN (see the section Autonomous Underwater Gliders; Figures 1, 5) as well as along cross-shore transects as far north as the Washington Coast (Figure 5), returning measurements of physical properties and some biological proxies; the gliders complement the ship-based surveys by providing observations at higher spatial and temporal resolutions (e.g., Rudnick et al., 2017), albeit of a more limited set of properties. An array of PIES with end-point moorings off of Southern California monitors full-depth geostrophic transport; gliders routinely retrieve data from the PIES and transmit them to shore (Send et al., 2013). Since 2007, NOAA has led largescale coastal surveys along the U.S. West Coast every 2-4 years to determine the spatial distributions of carbon, oxygen, nutrient, biological, and hydrographic parameters (Feely et al., 2008. Starting in 2010, moored platforms throughout the California Current System established high-frequency time series of physical and biogeochemical parameters (Nam et al., 2011;Harris et al., 2013;Sutton et al., 2016). More recently, the OOI Endurance Array Trowbridge et al., 2019) has been deployed in the northern California Current System ( Figure 5); moorings on the shelf and continental slope provide high-resolution time series, while gliders provide high-spatialresolution observations between the mooring sites. A network of shore-based HF radars provides real-time surface currents within about 150 km of the coast along nearly the entire U.S. West Coast .

Peru-Chile Current System
The Peru-Chile Current System (or Humboldt Current System) is the EBC system of the subtropical South Pacific, extending from the equator to southern Chile (∼45 • S). It is characterized by a persistent stratus cloud deck, equatorward surface currents, strong wind-driven coastal upwelling, poleward undercurrents, and filaments and eddies that develop along the coasts of Peru and Chile (see Colas et al., 2012 and references therein). A subsurface oxygen minimum zone (e.g., Paulmier and Ruiz-Pino, 2009) results in upwelled waters being nutrient rich but low in oxygen (e.g., Silva et al., 2009;Pizarro et al., 2016). Due to its proximity to the equator, the Peru-Chile Current System is strongly influenced by equatorial variability through propagation of Kelvin and coastal trapped waves (Dewitte et al., 2012;Mosquera-Vásquez et al., 2013) and anomalous advection during strong El Niños (e.g., Colas et al., 2008).
The dramatic impacts of El Niño events on both weather and fisheries have driven monitoring of oceanographic properties and fish stock assessments along the Peruvian coast since the 1960s (Figure 6; Grados et al., 2018). Over the past decade, these surveys have taken place monthly along the northern Peruvian coast and at least twice per year along the entire Peruvian coast; shipboard ADCP surveys are conducted at least seasonally. Biweekly time series along the 100-m isobath between Paita (5 • S) and Ilo (17 • S), coastal tide-gauge stations, daily SST measurements at coastal laboratories, and a nearshore thermistor chain and bottom-mounted ADCP at 4 • 30'S ( Figure 6) allow monitoring of temperature and sea level anomalies and associated fluctuations in thermocline, oxycline, and nutricline depths. Measurements from the TAO/TRITON moored array and the Argo program ( Figure 3B) provide key broad-scale context to these coastal observations. Efforts are underway to improve monitoring of the Peru-Chile Current System. For instance, sustained glider surveys across the frontal region off northern Peru, where El Niño impacts are large, are planned to begin by 2020.

Leeuwin Current System and South Australian Current System
The boundary currents along the western and southern coasts of Australia have some unique features. The Leeuwin Current, which is the subtropical EBC of the southeastern Indian Ocean, is unusual in that it flows poleward along an eastern boundary, transporting warm, fresh tropical waters southward due to forcing by the Indonesian Throughflow and ocean atmosphere interactions in the Indian Ocean (Godfrey and Weaver, 1991); it is important for the upper ocean heat balance in the southeast Indian Ocean (Domingues et al., 2006). The Leeuwin Current hosts broad-scale downwelling (Furue et al., 2017;Liang et al., 2017) where eastward surface currents merge with the Leeuwin Current and then downwell into the Leeuwin Undercurrent at depths of 200-1,000 m. The equatorward Leeuwin Undercurrent carries waters of subantarctic origin along the western Australian coast (Woo and Pattiaratchi, 2008), leaving the coast near 22 • S to contribute to the lower limb of a zonal overturning (Furue et al., 2017) and the subtropical gyre (Schott et al., 2009). In winter, the Leeuwin Current merges with the southwestward-flowing Holloway Current off the northwest coast of Australia, the eastward-flowing South Australian Current off the south coast, and the southward-flowing Zeehan Current off the west coast of Tasmania to form the longest shelfbreak boundary current system in the world (Ridgway and Condie, 2004;D'Adamo et al., 2009;Ridgway and Godfrey, 2015). Along the continental slope south of Australia, the westward flowing Flinders Current results from the collision of the equatorward deep ocean Sverdrup transport with the deep shelf slope of the Great Australian Bight (Middleton and Cirano, 2002;Middleton and Bye, 2007) and is a unique northern boundary current.
Coastal sea level observations at Fremantle have long been used as a proxy for the strength of the Leeuwin Current (Feng et al., 2003). Since 2008, the Australian Integrated Marine Observing System (IMOS; Hill et al., 2010) has been monitoring the shelf component of the Leeuwin Current near 32 • S using shelf moorings , gliders, and HF radars (Figure 7). Short-term deployments (2012Short-term deployments ( -2014 have also been carried out off the northwest coast of Australia (Ridgway and Godfrey, 2015). XBT surveys from ships of opportunity in and out of Fremantle, though not targeted for the Leeuwin Current, have taken place since the mid-1980s (Wijffels et al., 2008). IMOS makes ongoing observations of the South Australian Current system with dedicated moorings and glider missions monitoring the Flinders Current (Figure 7).

Benguela Current System
The Benguela Current Large Marine Ecosystem is the eastern boundary upwelling system of the South Atlantic. The equatorward Benguela Current is unique in that it is bounded by warm currents at its northern and southern edges, the Angola Current to the north and the Agulhas Current (see the section Agulhas Current) to the south. Coastal upwelling is linked to the seasonal position of the South Atlantic high pressure system, resulting in a number of upwelling cells along the southern African coast with divergent seasonality; the strongest year-round upwelling occurs off Lüderitz (∼26 • S), effectively dividing the Benguela Current System into northern and southern subsystems. The northern Benguela upwelling system is highly productive (e.g., Louw et al., 2016) but also prone to hypoxia over the continental shelf that is modulated by a seasonal poleward undercurrent bringing low-oxygen waters to the shelf in summer and fall and an equatorward undercurrent that brings oxygenated water in winter and spring (Duncombe Rae, 2005;Mohrholz et al., 2008;Monteiro et al., 2008). The southern Benguela upwelling system experiences intense, pulsed upwelling in summer and quiescence in winter (Shannon and Nelson, 1996;Weeks et al., 2006;Hutchings et al., 2009c), although the direction of net Ekman transport appears to be offshore year-round (Carr and Kearns, 2003). This region also experiences hypoxia (and occasionally anoxia) in inshore waters, particularly in the region of St. Helena Bay (Pitcher and Probyn, 2011;Pitcher et al., 2014), but low-oxygen events are driven solely by bacterial respiration of organic matter from surface waters (Monteiro and van der Plas, 2006) and can result in mass mortalities of commercial fish stocks and rock lobster (e.g., Cockcroft et al., 2000Cockcroft et al., , 2008Van der Lingen et al., 2006).
In the southern Benguela Current System, monthly ship-based sampling of fisheries-relevant parameters took place through the 1950s and 1960s and then intermittently until 1988, after which surveys of fisheries, hydrographic, and chemical properties have been conducted two to three times per year (Figure 8; Moloney et al., 2004). Since 2012, quarterly surveys as part of the Integrated Ecosystem Program have additionally monitored the carbonate system. Various multifunctional moorings have been deployed over the years, including a buoy for oxygen and temperature and a harmful algal bloom detection system in the vicinity of St. Helena Bay (see Hutchings et al., 2009b). The Namibian Ministry of Fisheries and Marine Resources conducts regular monitoring of hydrographic conditions and commercial fish resources in Namibian waters of the northern Benguela (Figure 8); regular shipboard oceanographic monitoring began in 1999 with sampling frequency varying from two to eight occupations annually along most lines and up to twice per month off Lüderitz during the lobster fishing season. Long-term, though intermittent, moored observations have been collected at 23 • S, 14'03 • E, and coastal stations are maintained along the Namibian coast (Figure 8).

Canary Current System
The Canary Current Large Marine Ecosystem extends from the northern tip of the Iberian Peninsula (43 • N) to south of Senegal (12 • N), corresponding to the extent of the northeasterly trade winds in the northeastern Atlantic. Upwelling occurs yearround with meridional shifts in the trade winds leading to seasonality in the latitudinal range of upwelling, particularly in the south (Benazzouz et al., 2014;Faye et al., 2015), where strong intraseasonal to longer timescale variability is driven by internal or remotely forced pulsations of the trade winds, passages of African easterly waves, and oceanic coastally trapped waves (Polo et al., 2008;Diakhaté et al., 2016;Oettli et al., 2016). The ecosystem is broadly divided by the Strait of Gibraltar into the Iberian and the Northwest African areas, though strong subregional differences are observed due to variability in factors including coastal configuration, oxygen concentration, nutrient fertilization, and productivity (Arístegui et al., 2009). The continental shelf in the Canary Current System is the most extensive of any EBC, and persistent circulation features are associated with the topography of the shelf. Large filaments of coastal upwelled water stretch offshore from the numerous capes and promontories (e.g., Cape Guir and Cape Blanc), transporting waters rich in organic matter into the oligotrophic subtropical gyre (Álvarez-Salgado et al., 2007;Lovechio et al., 2018). The Canary Archipelago interrupts the equatorward flow of the Canary Current, leading to strong mesoscale variability downstream of the islands (Arístegui et al., 1994). Island eddies and upwelling filaments interact to exchange water properties, resulting in an efficient route for transporting organic matter to the open ocean (Arístegui et al., 1997;Barton et al., 1998). As a major upwelling area, the Canary Current System is highly productive and the focus of intensive fisheries. Interannual and decadal variability in fisheries landings and distributions of small pelagic fishes has been related to environmental changes associated with the North Atlantic Oscillation and, to a lesser extent, ENSO in the southern part of the region (see reviews in Arístegui et al., 2006;Arístegui et al., 2009).
There have been numerous process-oriented field programs in Canary Current System in the past 20 years, including the Coastal Transition Zone (CTZ; Barton and Arístegui, 2004) and Canary Islands Azores Gibraltar Observations (CANIGO; Parrilla et al., 2002) programs. However, physical, biogeochemical, and ecosystem monitoring efforts have been less routine than have other EBCs (Figure 9). The European Station for Time series in the Ocean Canary Islands (ESTOC) has completed over 20 years of continuous meteorological and surface and mid-water physical and biogeochemical monitoring. The Cape Verde Ocean and Atmospheric Observatories (CVOO/CVAO) are a deep water mooring and an atmospheric station that have been deployed since 2006 in a region critical for climate and greenhouse gas studies and for investigating dust impacts on marine ecosystems.
Both ESTOC and CVOO/CVAO are part of the European open ocean fixed point observatories (FixO3). An additional mooring has been recording oceanographic properties and particle fluxes with sediment traps off Cape Blanc continuously since 2003 (Nowald et al., 2015). Long-term measurements of coastal oceanic and atmospheric properties from buoys off Morocco and Senegal have begun during the last 4 years. Ship-based hydrographic and biogeochemical sampling has taken place twice per year since 2006 at the latitude of the Canary Archipelago as part of the RAPROCAN program (Figure 9), which aims to monitor the Canary Current and maintain the ESTOC mooring. Gliders have periodically surveyed between the African coast and the Cape Verde Islands (Figure 9; Karstensen et al., 2017;Kolodziejczyk et al., 2018).

Northwestern Atlantic
The Gulf Stream comprises the upper limb of the AMOC in the North Atlantic subtropical gyre, carrying warm, saline waters from the tropics to higher latitudes. It flows along the eastern seaboard of the United States before separating from the continental margin near Cape Hatteras. The Labrador Current is the WBC of the subpolar gyre. The North Atlantic Deep Western Boundary Current is a deep limb of the AMOC that carries cold water masses from the tail of the Grand Banks of Newfoundland equatorward (Pickart and Watts, 1990). It encounters the Gulf Stream at the tail of the Grand Banks and again at Cape Hatteras, where a portion is entrained into the abyssal interior FIGURE 9 | Map of the boundary current observing effort for the Canary Current System (see the section Canary Current System) with details as in Figure 5. (Pickart and Smethie, 1993;Bower and Hunt, 2000a,b) while the rest continues to flow equatorward along the western boundary and into the Southern Hemisphere (Southwestern Atlantic). The strength of the Deep Western Boundary Current may influence the latitude at which the Gulf Stream detaches from the continental margin (Thompson and Schmitz, 1989). Along the edge of the adjacent Middle Atlantic Bight shelf, a persistent shelfbreak front and associated shelfbreak jet (Linder and Gawarkiewicz, 1998) transport waters equatorward with secondary frontal circulation leading to upwelling and elevated primary productivity (Marra et al., 1990). The shelfbreak jet continues southward until just north of Cape Hatteras, where it turns offshore as it encounters the much stronger Gulf Stream (Gawarkiewicz and Linder, 2006).
The boundary current observing system for the subtropical northwest Atlantic (Figure 10) is anchored by decades-long measurements at several fixed locations along the boundary. In the Florida Strait near 27 • N, cable-based measurements of Gulf Stream transport and quarterly to bi-monthly ship-based sampling have been ongoing since 1982 as part of the Western Boundary Time Series (WBTS; Baringer and Larsen, 2001;Meinen et al., 2010). Far to the northeast, where the Gulf Stream has separated from the continental margin, XBT, shipboard ADCP, and surface temperature and salinity measurements are obtained twice weekly from M/V Oleander, a cargo ship running between New Jersey and Bermuda (Rossby et al., , 2014Wang et al., 2010). The AX10 XBT line between New York and Puerto Rico crosses the Gulf Stream just upstream of the Oleander line and conducts high-resolution sampling within the boundary current (e.g., Domingues et al., 2018). Since 2015, gliders have been used to routinely survey across the Gulf Stream between Florida and Massachusetts (Figure 1; Todd, 2017;Todd and Locke-Wynn, 2017;Todd et al., 2018b), providing subsurface observations that fill the gap between the WBTS and Oleander and AX10 lines. Two moored arrays with complementary repeat hydrographic sampling have focused on the Deep Western Boundary Current for a decade or more. The RAPID-MOCHA array of subsurface moorings and PIES near 26.5 • N has been in place since 2004 with hydrographic stations reoccupied about every 9 months (Meinen et al., 2013). Farther north, the Line W array of subsurface moorings was in place from 2004 to 2014 with repeat ship-based sampling every 6-12 months (Toole et al., 2017). The OOI Pioneer Array south of New England Trowbridge et al., 2019) and the PEACH array near Cape Hatteras use a mixture of moorings, gliders (e.g., Gawarkiewicz et al., 2018), and land-based remote sensing (e.g., Haines et al., 2017) to characterize the dynamics of the shelfbreak jet and exchange between the shelf and deep ocean in the vicinity of the Gulf Stream and its eddies. In the subpolar northwestern Atlantic at 53 • N, transport of the Labrador Current has been monitored since 1997 using a combination of moored and shipboard observations (Zantopp et al., 2017).

Northwestern Pacific
In the Northwestern Pacific, bifurcation of the westward North Equatorial current between 11 • and 13 • N along the Philippine coast Rudnick et al., 2015b) forms the poleward Kuroshio and the equatorward Mindanao Current. The Kuroshio becomes a more coherent jet as it flows along the Taiwanese coast (e.g., Centurioni et al., 2004), into the East China Sea, and along the southern Japanese coast before separating from the continental margin near 35 • N to form the Kuroshio Extension, an eastward, meandering jet in the open North Pacific. The Mindanao carries waters from the North Pacific southward to feed tropical circulation in both the Pacific and Indian Oceans (Schönau et al., 2015). The Oyashio is the western boundary current of the North Pacific subpolar gyre and converges with the Kuroshio to the east of Japan. This convergence region has rich frontal structure as various water masses meet and are modified and is a key area for fisheries (Yasuda, 2003).
The Japan Meteorological Agency (JMA) has carried out repeat hydrographic surveys two to five times annually at the PN line in the East China Sea since 1971 (Aoyama et al., 2008;Figure 11) and at the TK line south of Kyushu since 1987 (Oka and Kawabe, 2003) to monitor physical and biogeochemical EOVs in the Kuroshio. JMA has also monitored the Ryukyu Current system (Ichikawa et al., 2004) flowing south of the Ryukyu Islands at the OK line southeast of Okinawa, which is connected to a zonal section along 24 • N. Furthermore, the JMA has maintained physical and biogeochemical surveys along 137 • E across the western North Pacific to monitor major currents of the subtropical and tropical gyres including the Kuroshio (Nakano et al., 2015;Oka et al., 2018). Monthly fisheries surveys and hydrographic stations along the A-line off Hokkaido have been occupied since 1987 (Kuroda et al., 2015) with collocated long-term moorings (Kono and Kawasaki, 1997). JAMSTEC has sampled hydrographic stations K2 (47 • N, 160 • E) and KNOT (44 • N, 155 • E) in the subpolar north Pacific at least annually since 1997 (Wakita et al., 2010). The Kuroshio Extension Observatory (KEO; Cronin et al., 2015) is a surface mooring that has been located in the subtropical recirculation gyre south of the Kuroshio Extension at 32.3 • N, 144.6 • E (Figure 11) since 2004. KEO monitors air-sea exchanges of heat, moisture, momentum, and CO 2 ; sea surface temperature, salinity, and ocean acidification; and upper ocean temperature, salinity, and currents below the surface buoy. Since 2014, a sediment trap mooring has been located at KEO (Honda et al., 2018). More recently, the CLIVAR Northwestern Pacific Ocean Circulation and Climate Experiment (NPOCE) has deployed an array of subsurface moorings, some with real-time data transmission, across the western Pacific, South China Sea, and Indonesian seas (Figure 11) that cover the major currents in these regions (e.g., Hu et al., 2013Hu et al., , 2015Hu et al., , 2016Zhang et al., 2014;Chen et al., 2015;Wang et al., 2017). Multiple XBT transects cross boundary currents within the region (see Goni et al., 2019). Gliders have been deployed for extended periods in the Kuroshio and Mindanao (Figure 11), generally sampling obliquely across the boundary currents as they were advected downstream (e.g., Rainville et al., 2013;Schönau and Rudnick, 2017).

Southwestern Pacific
The East Australian Current is the subtropical WBC of the South Pacific; it is a strong, meandering current with large poleward heat transport  that separates from the continental margin between 30 • S and 32 • S to join a dynamic eddy field (Cetina Heredia et al., 2014) in the Tasman Sea. The low-latitude WBC system of the South Pacific originates as the equatorward Gulf of Papua Current along the northeast coast of Australia, which then flows through the Solomon Sea as the New Guinea Coastal Undercurrent before feeding into the equatorial current system. This is a major contributor to the mass and heat budget of the tropical Pacific, acting as a conveyor belt for micronutrients from the western continental margins to the eastern Equatorial Pacific upwelling region. These low-latitude WBCs split into numerous branches around topographic obstacles and flow through narrow passages, presenting challenges for sustained observing. The sustained observing system for the East Australian Current and its extension (Figure 7) currently consists of highdensity XBT transects (PX05, PX06, PX30, and PX34; Goni et al., 2019); Argo floats; a deep moored array at approximately 27 • S; HF radar sites near 32 • S and 30 • S; a regional array of shelf moorings (including biogeochemical and biological sensors) at 30 • S, 34 • S, and 36 • S; and numerous glider deployments from northern Australia (11 • S) to the Tasman Sea (42 • S) (Roughan and Morris, 2011;Roughan et al., 2013Roughan et al., , 2015. These observational platforms complement each other well, providing a distributed boundary current observational system for the East Australian Current that has been shown to constrain ocean models well (Kerry et al., 2018). Additional sustained measurements are needed to characterize the seasonal changes in the transports of mass, heat, and freshwater in the East Australian Current and its eddy field. Effective monitoring strategies would be to deploy moored arrays in key regions; to increase Argo float and drifter density in the WBC region; and to implement glider sampling along existing high-density XBT lines within the East Australian Current, its eddy field, and recirculation.
In the low-latitude WBC system, long-term, sustained observations of the heat and mass transport through the southern entrance of the Solomon Sea have been provided by gliders since 2007  and an array of PIES since 2012 (Figure 7). Concurrent, short-term process studies including mooring deployments have been conducted as part of the CLIVAR-SPICE program (Ganachaud et al., 2014). Future monitoring efforts should integrate measurements across platforms, with the existing measurements in the southern entrance complemented by observations at the northern exits of the Solomon Sea (e.g., moorings, HF radars, and glider transects) to resolve the partitioning of the flow joining the equator (see Smith et al., 2019).

Agulhas Current
The Agulhas Current is the poleward WBC of the subtropical South Indian Ocean (Lutjeharms, 2006). It flows as a fast (>1.5 m s −1 ), deep-reaching (>3,000 m) jet along the continental slope of southeast Africa (Beal and Bryden, 1999;Beal et al., 2015). Near 40 • S, the Agulhas flows into the open ocean, where it retroflects under the influence of the large-scale wind stress curl to flow eastward into the Indian Ocean as the Agulhas Return Current (de Ruijter et al., 1999). Leakage of warm, salty Agulhas waters into the South Atlantic by rings, eddies, and filaments (Boebel et al., 2003;Richardson, 2007) is thought to influence the AMOC on timescales from decades to millennia .
In 2010, the Agulhas Current Time-series experiment (ACT) established a moored array to measure the volume transport of the Agulhas Current along a satellite altimeter ground-track (#96) near 34 • S (Figure 8) for a period of 3 years. The array consisted of seven full-depth current meter moorings and four CPIES that captured the breadth and depth of the Agulhas jet, including during offshore meander events . Following ACT, a consortium of South African, U.S., and Dutch scientists deployed the Agulhas System Climate Array (ASCA) in 2016 for long-term monitoring of the Agulhas Current as part of GOOS. ASCA augmented the original ACT array design with conductivity-temperature recorders to measure the heat and freshwater fluxes. The long-term success of ASCA was dependent on an ambitious plan of capacity building and resource sharing among nations, and, owing to a number of challenges, this plan was not fulfilled, and the array was pulled out of the water in 2018, following a 2-year deployment. In 2015, the Shelf Agulhas Glider Experiment (SAGE) demonstrated the feasibility of operating autonomous robotic platforms to sample the shelf regions of the Agulhas Current (Krug et al., 2017). Since SAGE, growing regional interest in monitoring with autonomous platforms led to formation of a South African multi-institutional scientific consortium named Gliders in the Agulhas (GINA). GINA conducted two additional glider missions in 2017 and 2018 and is working toward the development of a sustained glider observing system for the Agulhas Current coastal and shelf regions. The influence of the Agulhas leakage on the AMOC has been monitored since 2013 by an array of CPIES and tall moorings as part of the SAMBA line at 34.5 • S (Figure 8; Ansorge et al., 2014). Thus far, no sustained ecological or biogeochemical measurements have been made in the Agulhas, though the addition of oxygen sensors to SAMBA moorings is planned.

Southwestern Atlantic
In the South Atlantic, the WBC system consists of the poleward Brazil Current and the equatorward North Brazil Undercurrent, both of which originate from the bifurcation of the South Equatorial Current between 10 • S and 20 • S (e.g., da Silveira et al., 1994;Rodrigues et al., 2007), and the equatorward Malvinas current in the subpolar gyre. The Brazil Current and Malvinas Current both separate from the South American continental margin between 35 • S and 40 • S to flow eastward at the Brazil-Malvinas confluence (Olson et al., 1988). The North Brazil Undercurrent constitutes a bottleneck for the interhemispheric mean flow of the upper limb of the AMOC as it transports warm waters of South Atlantic origin across the equator (e.g., Schott et al., 1998;Zhang et al., 2011;Rühs et al., 2015). The Deep Western Boundary Current carries much of the lower limb of the AMOC off the coast of South America (Schott et al., 2005;Meinen et al., 2013).
For more than a decade, high-density XBT transects  have been occupied near 22 • S and 34 • S (AX97 and AX18) across South Atlantic WBCs (Figure 12; Dong et al., 2015;Lima et al., 2016). Near 11 • S, an array of four tall moorings and two PIES has measured transport of the North Brazil Current since 2013 (Figure 12; Hummels et al., 2015). At 34.5 • S (Figure 12), an array of PIES, CPIES, and a bottom-mounted ADCP has monitored the Brazil Current and Deep Western Boundary Current (Meinen et al., 2013(Meinen et al., , 2017 in conjunction with periodic hydrographic surveys (Valla et al., 2018). A series of yearlong deployments of current meter arrays along 41 • S since 1993 (Figure 12; Vivier and Provost, 1999;Spadone and Provost, 2009;Paniagua et al., 2018), in conjunction with satellite altimetry, has allowed for production of a 24-year transport time series for the Malvinas Current (Artana et al., 2018a).

FUTURE OUTLOOK
We recommend establishing and maintaining a global network of boundary current observing systems. Each distinct observing system will need to be tailored to the unique aspects of that particular boundary current system and also follow best practices established by the international community. Such a network of regional boundary current observing systems is a crucial element of GOOS. To date, boundary current observing systems in different regions and countries have developed largely independently. Development and maintenance of a global network of boundary current observing systems that is fit for purpose would benefit from the standards outlined in the Framework for Ocean Observing (Lindstrom et al., 2012). In particular, application of the Framework across boundary current observing systems should foster communication and data sharing; contribute to capacity building, particularly in developing countries; encourage confidence and support from funding agencies; and promote international collaboration and scientific and technological innovation.
Boundary currents typically lie within the EEZs of coastal states, so development and maintenance of boundary current observing systems will require the cooperation and support of appropriate governing authorities. Considering the difficulty of obtaining international funding for observations in national waters, there is a need for a community of regional boundary observers. Moreover, many boundary currents span multiple countries, so the observing system for a single boundary current system is likely to require collaboration and coordination between several countries. The advective nature of boundary currents may even require that mobile or drifting assets deployed within one country's EEZ be recovered within another EEZ. Sharing of measurements taken within EEZs, particularly those that have economic impacts such as some biogeochemical measurements, remains a challenge. By moving toward international collaboration in the design and implementation FIGURE 12 | Map of the boundary current observing system for the Southwestern Atlantic (see the section Southwestern Atlantic) with details as in Figure 5.
of boundary current systems as suggested by the Framework for Ocean Observing, there is hope for building the high-level governance structure needed to surmount the challenges posed by boundary currents falling within EEZs. The Large Marine Ecosystems effort has identified distinct boundary regions that cross international borders and has gained international traction through the Global Environment Facility and the International Union for Conservation of Nature; leveraging this effort to facilitate international cooperation and governance for sustained boundary current observations may be fruitful.
For any particular boundary current system, a complete observing system will require a combination of currently available observing platforms (see the section Observing Techniques), as well as future platforms, to optimally measure EOVs at necessary spatial and temporal resolutions to address relevant scientific and societal needs. Through the Framework process, specific observing platforms, sampling choices, and instruments would be matched to the relevant questions. Drifting and mobile assets that provide spatially resolved measurements at the expense of temporal resolution will need to be combined with moored assets that provide high-frequency measurements at key locations and land-or satellite-based remote sensing that provides spatially broad measurements of sea surface properties. Such integrated arrays, as are already in place in the California Current System, at the Ocean Observatories Initiative (OOI; Smith et al., 2018;Trowbridge et al., 2019) Endurance and Pioneer Arrays, and along the Australian coasts as part of the Integrated Marine Observing System, offer critical opportunities for intercalibration between instruments on fixed and mobile assets; such intercalibration is particularly important for biogeochemical sensors (e.g., Palevsky and Nicholson, 2018). Since similar needs arise in most boundary current systems, the Framework process should provide a means for determining the extent to which the same observing strategies should be applied to address similar needs in different systems. Additional studies that compare different sampling techniques in a given boundary current system could provide guidance on the strengths and limitations of each technique and how to better exploit their complementarity.
While the discussion of observing platforms in the section Observing Techniques focused on mature observing platforms with proven records of sustained operation in boundary currents, there is no doubt that recently developed observing platforms and sensing technology will become integral parts of future boundary current observing systems. For instance, more fast-moving autonomous underwater vehicles (AUVs) and autonomous surface vehicles (ASVs) will be deployed to conduct adaptive and targeted sampling in response to real-time needs. Propellerdriven AUVs have thus far seen limited use in boundary currents. Though able to carry large instrument payloads and move much faster (1-2 m s −1 ) than gliders, propeller-driven AUVs have been limited by battery endurance to missions typically lasting hours to days; improvements in battery technology and autonomous charging are expected to make propeller-driven vehicles capable of long-duration sampling in the near future. Fast-moving, longendurance ASVs (e.g., Saildrones and WaveGliders) are poised to become key platforms for making measurements near the air-sea interface, including meteorological measurements, pCO 2 , subsurface currents, and plankton biomass. Due to the use of renewable energy, these ASVs generally carry a larger number of sensors and have longer duration than other autonomous platforms (e.g., Zhang et al., 2017). Planned high-resolution, satellite-based altimetry measurements (e.g., SWOT), smaller and dramatically cheaper satellites (e.g., Cubesats), and, potentially, geostationary satellites positioned over boundary regions offer the prospect of dramatically increased spatial and temporal resolution of surface properties.
At some locations, boundary currents have been continuously observed for many years using various techniques. For instance, the CalCOFI program has maintained quarterly ship-based stations for more than 65 years (McClatchie, 2014), the WBTS has made cable-and ship-based measurements in the Florida Strait for more than 35 years (see the section Northwestern Atlantic), and hydrographic sampling has occurred monthly along the inside edge of the East Australian Current since the 1940s (Lynch et al., 2014) and is now an integral part of the East Australian Current observing system (Roughan and Morris, 2011). Long-term measurements like these are invaluable for capturing decadal variability and secular trends. Sites at which decades-long measurements exist should be maintained and serve as anchors for comprehensive boundary current observing systems. These long-term measurement sites at the boundaries also serve as points at which the boundary current observing systems are linked to the basin-scale ocean observing system. Since 2004, the WBTS has been integrated with the U.K.-U.S. RAPID-MOCHA program that measures meridional transport at 26.5 • N in the North Atlantic, while several long-standing, cross-Pacific XBT transects intersect the U.S. West Coast within the CalCOFI domain .
Existing boundary current observing systems are largely focused on measuring physical processes, with biogeochemical and ecosystem processes only beginning to gain traction, largely due to the advent of new sensors. The California Current System (see the section California Current System) and Benguela (see the section Benguela Current System) are exceptions, having had sustained observations of EOVs relevant to physics, biogeochemistry, and biology and ecosystems for over 65 and 30 years, respectively. However, these ship-intensive models are unlikely to be suited to all boundary current systems due to a wide range of factors (e.g., cost, proximity to the coast, existing infrastructure, and available manpower). Although the methods for measuring many of the EOVs needed to monitor biogeochemical and, to a greater extent, ecosystem processes are time-intensive and require a platform for collecting water, new sensors are being developed to reliably measure a range of biogeochemical and biological EOVs. Many of these sensors have been successfully deployed on BGC-Argo floats as part of the SOCCOM project . Increasing the measurements of biological and ecological EOVs should be prioritized if we are to understand, monitor, and predict the physical-biological connections and processes that support marine-based industries and activities and, importantly, seafood security.
Providing publicly available data in a timely manner is a key attribute of any ocean observing system. These observations should be provided in formats that are discoverable, accessible, and readily subset, following conventions agreed upon by the community (see Wilkinson et al. (2016) for a set of general principles for management of scientific data). Many platforms already provide observations in near real-time through a variety of services. Transmission of data through the Global Telecommunications System is particularly important if those observations are to be used in operational numerical modeling. Advances in real-time data collection from sub-surface moorings (e.g., Send et al., 2013) will be critical to providing boundary current observations in a timely manner for forecasting and prediction. Widespread dissemination of comprehensive boundary current observations can foster synergies with other disciplines, including the geophysics (tsunamis and earthquakes), physics, meteorological (e.g., tropical and extratropical cyclone forecasting; Domingues et al., 2019), and fisheries communities.
In addition to providing raw observations, there is a need for providing synthesized data products that are tailored to user needs. Integration of complementary data types can yield useful metrics. Further advances in data analysis techniques and statistical methods should aid in using multi-platform data to increase temporal and/or spatial resolution of metrics. The Southern California Temperature Index  is an example of such a data product.
Boundary current observations play a key role in constraining ocean models (e.g., Todd and Locke-Wynn, 2017), while models play a complementary role by filling gaps between sparse observations in a dynamically consistent manner (e.g., Todd et al., 2011bTodd et al., , 2012Gopalakrishnan et al., 2013). Increased availability of boundary current observations, particularly in regions that are currently poorly sampled, should lead to continued improvements in regional models and predictive tools. At the same time, higher resolution climate models that can resolve boundary currents are becoming more plentiful and should begin to rely on high-resolution boundary current observations as constraints. One specific goal would be to reduce climate models' warm SST biases within EBCs; continuation and expansion of long-term measurements in EBCs as well as focused process studies to study upper ocean and atmospheric dynamics in EBCs would contribute to this goal. Observation impact studies derived from data assimilating models provide guidance on the value of a range of observation types in resolving boundary current transport, as well as for constraining the eddy field in ocean reanalyses (e.g., Kerry et al., 2016Kerry et al., , 2018. It remains an open question how best to integrate models with interdisciplinary (e.g., biogeochemical) observations to study ecosystem dynamics, though advances are being made in the assimilation of biological parameters (e.g., Song et al., 2012). Observing System Simulation Experiments tailored for boundary current systems can also provide insight to the type, spatial distribution, and frequency of observations required to improve numerical simulations of boundary current dynamics (Hoffman and Atlas, 2016). Targeted observations can reduce biases in the initialization of models used to forecast extreme weather events and support local decision making (Halliwell et al., 2017).
Downscaling coarse resolution climate model predictions through the application of higher resolution regional and coastal models is now common and has shown promise but still faces research challenges. Furthermore, a significant amount of physical, biogeochemical, and biological response on the continental shelf is due to episodic oceanic and atmospheric events at timescales of variability that are absent from coarse models and cannot be recovered locally. To be valid globally, the veracity of downscaled models needs to be appraised by supporting observations of shelf edge fluxes in a diversity of circulation regimes.
Funding sustained observing efforts is a significant challenge. Portions of the observing system that have proven their readiness for long-term deployment have been discontinued after one or more short-term funding cycles. For instance, it is currently not clear how ship-time and funding challenges will be met for a reestablishment of ASCA (see the section Agulhas Current) in the future. In the typical 3-to-4-year cycles of scientific funding, early years (e.g., pilot phases) of observing efforts are readily fundable based on the promise of quick scientific results. Observing efforts that have endured for a decade or longer can leverage their long histories and clear relevance to decadal variability or secular trends to secure continued funding. The middle years, roughly years 4 through 10 as programs transition from pilot to mature components of the GOOS, are particularly difficult to fund.
The provision of robust three-dimensional and time-varying ocean circulation estimates in boundary current systems, resolving scales of a few kilometers, is seemingly within reach through advances in data-assimilative ocean models and rapid developments in observations platforms and sensors. However, the development of integrated observing systems that deliver the scope of observations required and the models capable of fully utilizing those observations is challenging. Success will require coordinated international collaborations, bringing together the expertise of the ocean modeling and observational communities. Establishment of an Ocean Boundary Task Team would provide a mechanism for the exchange of information regarding observing and model strategies, sensor developments, analysis techniques to combine data from the various observing platforms, and model development and application. The Task Team would also enable capacity building, encourage timely and appropriate transfer of knowledge, and provide a mechanism to instigate multinational observing systems with shared goals amongst participating nations. Endorsement of the Task Team by IOC/WMO or similar international organization is critical due to interests of multiple coastal state EEZs and the resulting complex governance needs.

SUMMARY RECOMMENDATIONS
The following actions are recommended to promote development of a comprehensive global network of boundary current observing systems in the next decade: (1) Maintain existing long-term (i.e., multi-year) observational records; (2) Expand the use of mobile, autonomous platforms (e.g., gliders, AUVs, and ASVs) to provide continuous, highresolution, broad-scale monitoring of EOVs; (3) Deploy moored platforms at key locations to measure high-frequency variability; (4) Continue and expand the provisioning of real-time observations and encourage post-processed data to be made publicly available as quickly as possible; data should be provided in readily discoverable formats that can easily be subset; (5) Continue development and expand deployment of sensors for ecological and biogeochemical EOVs; (6) Establish an Ocean Boundary Task Team to foster international community development and end-user engagement and to guide evolution of observing systems as user requirements change; (7) Expand collaborations between observational efforts, modeling efforts, and societal users to meet stakeholder and end-user needs; and (8) Increase focus on exchange between continental shelves and the deep ocean boundary currents to develop observing systems that span the continuum from the land to the deep ocean.

AUTHOR CONTRIBUTIONS
RT led the manuscript. Other lead authors (FC, SCl, SCr, MGo, MGr, XL, JS, and NZ) helped to conceive the manuscript and participated in all the stages of development. All authors provided input and/or edited the text.

ACKNOWLEDGMENTS
We gratefully acknowledge the efforts of all parties involved in collecting, analyzing, and disseminating the wide-ranging observations discussed here. We thank the OCB Program, US CLIVAR, MBARI, and OMIX for supporting the Ocean Carbon Hot Spots Workshop, which facilitated the discussion of carbon cycle research in Western Boundary Current systems. We also gratefully acknowledge J. Hildebrandt (WHOI) for the assistance in editing the final manuscript. This is PMEL contribution # 4890.