Temporal and Spatial Dynamics of Physical and Biological Properties along the Endurance Array of the California Current Ecosystem

The coastal margin of the Pacific Northwest of the United States is a highly dynamic and productive region. Here, we use satellite, high-frequency mooring, and glider estimates of biologically relevant physical and optical variables to characterize seasonal patterns and latitudinal and cross-shore gradients in particle concentrations between the Washington and Oregon shelves. Consistent with prior research, we find that the Columbia River exerts a strong seasonal influence on the Washington shelf, but smaller coastal rivers and resuspension processes also appear important in determining particle distributions nearshore during winter across the full study region. We find fluorescencebased measurements of chlorophyll to be similar in magnitude across the two shelves over the time period examined, although the much weaker wind stresses off Washington indicate that processes other than upwelling are important determinants of chlorophyll changes in those areas, as previously suggested. These in situ observations contrast with the overall differences observed from satellite data, which consistently show higher chlorophyll concentrations off the Washington coast. This research suggests that latitudinal differences in chromophoric dissolved organic matter may be a partial explanation for perceived trends in satellite-derived chlorophyll. The observations presented are nascent; maturation of temporal and spatial coverage of OOI data sets will be necessary to more conclusively link physical forcing and biogeochemical responses. By Fernanda Henderikx Freitas,


INTRODUCTION
Embedded within the California Current System, the coastal ocean along the Pacific Northwest (PNW) of the continental United States has recently become highly instrumented with the installation of an array of moored and autonomous assets via the National Science Foundationsupported Ocean Observatories Initiative (OOI). The oceanography of this area is complex. Strong temporal and spatial variability in the physics and biogeochemistry of the upper water column is largely the result of wind-driven upwelling and downwelling processes, as well as the freshwater influence of the fourth largest river in North America and other numerous small mountainous rivers. Monitoring changes in the ecological properties of these heterogeneous regions is challenging and requires both synoptic surveys and high-resolution Eulerian time series.
Large-scale upwelling along ocean margins occurs mainly in eastern boundary regimes (e.g., Peru/Chile, Southeast Atlantic, PNW) and is a highly seasonal phenomenon driven by shifts in wind strength and direction. In the PNW, prevailing northerly winds lead to intense upwelling periods when cold, dense, and nutrient-rich waters advected onto the continental shelf reach the euphotic zone (e.g., Huyer, 1983), resulting in enhanced phytoplankton production (e.g., Dickson and Wheeler, 1995) and the formation of biogenic particles (e.g., Wetz and Wheeler, 2003). Wind forcing also drives fluctuations in surface currents, water properties, and sea level, with typical scales of variability on the order of three to 10 days (Hickey and Banas, 2003). Additionally, upwelling-favorable winds drive crossshelf transport that actively pushes the freshwater plume of the Columbia River southward and offshore of the Oregon shelf in the spring and summer months (Barnes et al., 1972;Hickey et al., 2005;Saldías et al., 2016).
Much less is known about the biogeochemistry of non-upwelling periods, even though these conditions are sustained throughout much of the year.
For example, downwelling-favorable winds and moisture-laden storms prevail along the PNW in fall and winter months, which leads to large fluvial inputs of biogeochemically relevant, land-derived constituents to the coastal ocean, including freshwater (i.e., buoyancy), dissolved inorganic nutrients, and dissolved and particulate organic matter (e.g., Wheatcroft et al., 2010;Goñi et al., 2013). During these periods, the colder waters of the Columbia River are diverted northward over the Washington shelf and slope, altering regional current patterns (Hickey et al., 1998).
Productivity in the PNW has traditionally been assessed from shipboard surveys and satellite data. Studies comparing the productivity patterns of the Washington and Oregon shelves have concluded that absolute chlorophyll (chl) concentrations, a proxy for productivity, as well as the cross-shelf extent of chl signals, are seasonally larger for the Washington shelf despite significantly weaker upwelling winds relative to Oregon Banas, 2003, 2008;Davis et al., 2014). Ware and Thomson (2005) reported similar patterns of increased concentrations of chl, zooplankton, and fish to the north of the California Current System. In these studies, the seeming paradox was attributed to key geographical differences between the two locations, namely (1) the continental shelf is broader in Washington, thus potentially leading to increased residence times for nutrient and phytoplankton-rich waters in the region, (2) the Columbia River, which flows north into the Washington shelf during much of the year, is a potential source of limiting nutrients including iron and silicate, (3) proximity to the Strait of Juan de Fuca in British Columbia, Canada, likely allows for the advection of nutrient-rich waters onto the Washington shelf, and (4) the larger number of submarine canyons off the Washington shelf potentially allows for deeper, nutrient-rich waters to reach areas closer to the surface over the seasons and sustain productivity during weak upwelling periods.
In the following, we summarize longterm satellite estimates of particle distributions for the Washington and Oregon shelves, and raise the importance of considering the contribution of chromophoric dissolved organic matter (CDOM) in the interpretation of productivity patterns from satellite data. Then, we describe latitudinal gradients in temperature, salinity, pigment fluorescence, CDOM, and particle backscattering from high-frequency OOI observations during upwelling and downwelling periods of 2015 and 2016. Finally, we discuss the potential roles of wind-driven upwelling, resuspension processes, and riverine discharge in driving particle distributions across the Washington and Oregon shelves.

METHODS AND DATA SOURCES OOI Endurance Array
The OOI Endurance Array (http:// oceanobservatories.org/ array/coastalendurance) provides observations of cross-shelf and along-shelf variability in oceanographic properties off the Oregon and Washington coasts via a set of moorings and autonomous glider surveys (see map in Smith et al., 2018, in this issue). Two cross-shelf moored array lines, the Oregon (OR; Newport Line, ~44.6°N, inshore of 125°W) and Washington (WA; Grays Harbor Line, ~47°N, inshore of 125°W) lines each include inshore, midshelf, and offshore moorings at bottom depths of approximately 25, 80, and 600 m. In this work we used in vivo fluorescence-based estimates of chlorophyll (chl, mg m -3 ) and fluorescent-based CDOM (370 ± 20 nm /460 ± 120 nm excitation/ emission wavelengths, ppb), volume scattering coefficients at 700 nm (β (124°), m -1 sr -1 ), water temperature, and salinity data sets from the OR and WA OOI surface mooring sensor packages, nominally sampled at 7 m depth as determined by the OOI mooring design. All available recovered and telemetered data from the January 2015 to March 2017 data period were acquired to improve temporal coverage; no QA/QC flags had been previously applied to these data. Instances of discrepancies between recovered and telemetered data were observed, in which case the recovered data sets were selected. Personal communication with OOI data managers indicated that these discrepancies were due to "time-of-day" telemetry errors.
Chlorophyll and CDOM data were obtained in scaled units based on OOI pre-deployment calibration coefficients, and only nighttime chl data are used in the analysis to avoid the effects of quenching of chl fluorescence during daylight hours (Müller et al., 2001). Volume scattering coefficients were converted to particulate backscattering coefficients (b bp , m -1 ) using the relationship b bp = 2π χβ p (θ), where χ is a dimensionless factor equal to 1.076 based on Sullivan et al. (2013), and the particulate volume scattering β p is obtained by subtracting the volume scattering of seawater of Zhang et al. (2009) from β. Particulate backscattering is considered a proxy for suspended particles, including phytoplankton, detritus, and sediments (Jonasz and Fournier, 2011). The CDOM instrument configuration is specific for estimation of humic-like dissolved organic matter, and thus is a proxy for land-derived materials transported by rivers (Coble, 1996). Data presented here as daily averages were smoothed using a one-hour moving average filter on log-transformed observations to reduce the effect of highfrequency variability. The minimum values of chl, b bp , and CDOM per mooring were subtracted from each time series to facilitate mooring-to-mooring comparisons, given that OOI biological sensors are not intercalibrated.
Despite the long-term sampling efforts, several sections of data were excluded from the analysis due to low quality of the retrieved outputs. These included raw volume scattering count data between October 2014 and April 2015 for both inshore moorings that were characterized by a repetitive pattern of progressive signal increase up to the upper limit of detection, followed by abrupt drops to zero values. Similar issues (e.g., maxedout values possibly associated with biofouling) affected several sections of data from these moorings between January and May 2016. Additionally, data density is often more complete for the OR line than the WA time series, which yields large gaps in the comparative data set. For example, only 66 days of data are available for the offshore WA mooring in 2015, compared to 268 days for the offshore OR mooring.
Nearly coincident glider observations from the WA (glider/deployments g386/2, g384/3, and g312/4) and OR (g384/1, g320/2, and g382/2) long Endurance Lines (sampling between 128°W and 125°W) were obtained for the period 2015-2016, totaling about 40 days of data at each line. From these data, we calculated freshwater content (FWC; m), which represents the thickness of freshwater in the water column, as follows: where S is salinity, S 0 is the reference salinity of 32.5 for the Columbia River plume (Barnes et al., 1972), and d is the plume thickness according to the depth of S 0 . Additional glider-derived optical data (chl, CDOM, and b bp ) from the short Endurance Lines (sampling between 125°W and 124.3°W) off WA and OR were obtained (g326/1, g247/4, g381/1 and g311/1). Mission configurations consisted of repetitive cross-shore transects, with each deployment lasting between one and two months. Few simultaneous glider deployments were available for WA and OR, or at the times of moored observations. Thus, mean cross-shore sections per deployment were acquired as examples of cross-shore distributions of bio-optical properties during different times of the year at the two lines. A deep offset correction was applied to all glider optical data in order to improve comparisons between data from different gliders (Schmechtig et al., 2014).
Wind stress, significant wave heights, and river discharge measurements were used to inform potential controls on the observed physical and bio-optical observations. Wind and wave data were obtained from NOAA's National Data Buoy Center (NDBC) buoys 46041 and 46050 (http://www.ndbc.noaa.gov). Hourly Columbia River discharge data were obtained from US Geological Survey gauge 14105700 (https://nwis.waterdata. usgs.gov). Monthly discharge totals for 21 small mountainous rivers along the OR-WA coastline were obtained from USGS gauges ( Figure 1a). All correlation coefficients (r) shown in this study are considered significant if within the 95% confidence interval.

Satellite Observations of Phytoplankton and Particle Distributions
Monthly satellite retrievals of chlorophyll (chl sat ), absorption coefficients of non-algal materials and CDOM at 443 nm (a cdom (443)), and remote sensing reflectances at 555 nm (R rs (555)) from 2003 to 2016 were obtained from NASA MODIS Aqua Level 3 data products (https://oceancolor.gsfc.nasa.gov). Here, chl sat (NASA's OCx algorithm) is considered a proxy for the presence of phytoplankton (e.g., Cullen, 1982). As a proxy for the concentration of CDOM in coastal waters, a cdom (443) was estimated from the Garver-Siegel-Maritorena bio-optical algorithm (Maritorena et al., 2002). R rs (555) is, to first order, a proxy for the influence of low-salinity and sedimentrich river plumes in the coastal PNW (e.g., Thomas and Weatherbee, 2006;Saldías et al., 2016). R rs (555) is correlated with chl sat and presumably plankton biomass in clear oceanic waters not influenced by strong land-ocean interactions.

Latitudinal Gradients in Satellite and Glider Properties and the Impact of Riverine Inputs
The variability in the magnitude and seasonality of river discharge along the Washington-Oregon margin is illustrated in Figure 1a. The dominant and year-round influence of the Columbia River is clear, with peak discharge in winter and spring as a result of precipitation and melt of the continental snow pack. Columbia River flows are largely regulated by dams, which dampen seasonal discharges and keep summer flows artificially high because of releases from hydroelectric operations. The smaller mountainous rivers that line the PNW margin show wintertime peaks in discharge following seasonal precipitation patterns, with very low flows in the summer. Between November and March, the combined discharge of the 21 gauged small mountainous rivers equals about 30%-50% of the discharge of the Columbia River over the same period, while in the summer months the small rivers are equivalent to less than 10% of the Columbia River discharge. The geographical locations of the WA and OR lines relative to major river inputs help explain some of the patterns discussed in this study. For instance, the WA line is under direct influence of the Columbia River, while the OR line is influenced by a few small mountainous rivers (e.g., Alsea and Yaquina) and is considerably far from the mouths of more important rivers to the south (e.g., Umpqua and Rogue).
Temporal variability of R rs (555) inshore of 125°W (the cross-shore extent of the moored observations) show marked latitudinal differences between Washington and Oregon ( Figure 1b). As expected from a relationship between low-salinity river plumes and light scattering (Pak et al., 1970;Palacios et al., 2009), good agreement is observed between the timing of R rs (555) peaks (Figure 1b) and total discharge (Figure 1a). The role of the Columbia River in advecting waters northward along the Washington coast in the first few months of the year, and to the south and outer edges of the OR shelf in the summer, is evident in the R rs (555) fields (Figures 1b and 2a,b). A secondary role for phytoplankton in contributing to increased R rs (555) signals in the summer cannot be disregarded. The crossshore extent of the offshore advection of the plume during different times of the year is illustrated from glider observations (Figure 3), which reveal times when areas offshore of the OR line experience larger volumes of freshwater compared to the WA line as a result of Columbia River advection.
Discharge from mountainous coastal rivers as a result of high rainfall from "atmospheric rivers" of moisture (e.g., Warner et al., 2012), particularly to the south of the OR line, also correspond to increases in R rs (555) in winter months (Figure 1a,b). Nonetheless, the observed R rs (555) maxima inshore during fall/ winter (Figure 2a (555), (c) chl sat , and (d) a cdom (443) from the MODIS-Aqua sensor for coastal waters inshore of 125°W. Note that the discharge data are discrete measurements, and the interpolation along the latitude axis is for visualization purposes only. The vertical magenta lines correspond to January 1.
but also increased particle resuspension processes following storm events, which in this region are marked by coherent changes in winds and surface wave heights (a proxy for mixing; Kniskern et al., 2011; see also Figure 6k-l). Indeed, long-period surface gravity waves have been shown to transport suspended sediments across the shelf and modulate changes in particulate backscattering (i.e., suspended materials) inshore of the 100 m isobath in the absence of strong riverine inputs (Drake and Cacchione, 1985;Henderikx Freitas et al., 2017). Notably, the distance between the coast and the 100 m isobath varies along the Washington-Oregon coastline (i.e., more than 35 km at and north of the WA line, ~32 km at the OR line due to the presence of the Stonewall Bank, and less than 15 km north and south of OR). These differences in shelf configuration may partially explain the overall lower R rs (555) values between 43°N and 46°N in Figure 1b, because offshore waters make up a larger portion of the area east of 125°W in some of those regions. Temporal variability of chl sat is also large (Figure 1c), with a higher magnitude and broader cross-shelf extent of chl sat along the WA line compared to the OR line (Figure 2c,d). The coefficient of variation of the annual maximum of chl sat at ~125°W of the OR and WA lines is 34% (range = 3.7-12.9 mg m -3 ) and 70% (range = 8.5-56.8 mg m -3 ), respectively. The maximum in chl sat occurs between June and September at both locations, with OR showing a more distinct seasonal cycle than WA, where relatively high values of chl sat appear to persist for longer periods during the upwelling season (Figure 2c,d). A matchup analysis between in situ chl observations from the OOI moorings and chl sat estimates shows that the satellite data overestimate in situ chl observations closest to the coast (Figure 4a). This is because chl sat algorithms rely on the use of information in the blue region of the reflectance spectrum, which in coastal areas is highly affected by the presence of absorbing aerosols, suspended sediments, and CDOM (Gordon, 1997;Dierssen, 2010). Satellite retrievals of a cdom (443) show overall larger signals in WA compared to OR (Figure 1d), and similar seasonal cycle patterns to those observed in chl sat (Figure 2e,f). The enhanced cross-shore values in both chl sat and a cdom (443) at WA during lower productivity months such as December (Figure 2e) are intriguing, and suggest a role for CDOM contamination in the chl sat retrievals. The notion that phytoplankton, CDOM, and sediments all contribute to chl sat signals implies that the seemingly high chl sat concentrations at WA relative to OR may at least partially be a consequence of the proximity of WA to CDOM-rich sources such as the Columbia River plume and the Grays Harbor outflow (e.g., Palacios et al., 2009Palacios et al., , 2012. Cross-shore glider estimates of CDOM (Figure 3 and 5) suggest a stronger crossshore gradient in CDOM concentrations off the WA shelf, even at times when larger volumes of freshwater are registered at the OR shelf (i.e., June 2016 in Figure 3o). Certainly, transport, dilution, and other biogeochemical processes that affect CDOM variability need to be considered when comparing distributions at the two sites. Additionally, the large bandwidth of the CDOM fluorometer is tuned to detect CDOM in turbid waters, and thus not suitable for examining subtle changes in this property in clearer ocean conditions further from the coast. Examples of well-resolved coastal glider observations in Figure 5 point to key differences between the cross-shore distribution of bio-optical properties at WA and OR. First, the differences in the bathymetry along the two Endurance Lines may explain the presence of bottom nepheloid layers at WA (as denoted by elevated b bp and near zero chl values along the slope), which suggests that mixing processes may be more important to determining cross-shelf b bp variability at WA than at OR (at least within the sampled region). Second, CDOM distributions over the shelf are notably different at the two locations (particularly in Figure 5b, where observations at the two locations were obtained simultaneously), with patterns in WA suggesting that the nearshore region acts as a source of CDOM. Additional data collected simultaneously along the WA and OR lines would be necessary to determine if these example features hold throughout different seasons and years.

Latitudinal Gradients in Bio-Optical Properties from High-Frequency Moorings
High-frequency measurements from surface moorings expand the observations to the inner shelf regions of WA and OR, and provide temporally resolved observations that allow characterization of the latitudinal differences and interrelationships between bio-optical parameters over the 2015-2016 period ( Figure 6). During the observed period, chl maxima at times occur at the mid-shelf sites rather than inshore, possibly following offshore advection of the upwelling front, whereas minima are generally found further from the coast (Figure 6a,b). In contrast, a more apparent cross-shore gradient is depicted in the b bp observations, with elevated values generally occurring closest to shore (Figure 6c,d). Periods of synchronous changes in chl across the shelf are observed in a few instances (e.g., May and July 2016 at WA), but overall differences among the moorings denote the patchy nature of chl distributions across the shelf. The similarities and differences between chl and b bp spatial and temporal distributions indicate variability in the types of materials dominating optical properties at the different moorings. Biogenic particles typical of productive pelagic oceanic environments have large chl and low-to-moderate b bp values, whereas waters dominated by minerogenic particles have high b bp and low chl concentrations (Henderikx Freitas et al., 2016). Thus, the relationships between b bp and chl in Figure 4b for samples collected between May and September at both lines indicate that nearshore b bp estimates are more likely Both long Endurance Line profiles (128°W to 125°W) and short Endurance Line profiles (125°W to 124.3°W) were used whenever available. Plots were made using smoothing lengths of 0.25 degrees (longitude) and 10 m (depth). White bars denote missing data.
to be affected by non-phytoplankton materials (even in periods when discharge is presumably low), whereas optical properties of offshore samples are more likely to be dominated by changes in phytoplankton concentrations. These observations contextualize the relationships between chl and b bp with depth and distance from shore previously shown in the glider sections (Figure 5b). The chl variations along the OR and WA lines are comparable over the course of the upwelling season, with mid-shelf averages of chl for the May-September months of 2015 and 2016 reaching 3.34 ± 2.58 (one standard deviation) mg m -3 at WA and 4.61 ± 2.65 mg m -3 at OR. However, differences between OR and WA chl concentrations can exceed 20 mg m -3 at any given time during the more productive months of the years (e.g., August 2016; Figure 6a,b). Nonetheless, these values are considerably different from the average seasonal concentrations suggested from satellite estimates (Figures 1c and 2c,d). Nearshore b bp patterns at WA and OR are also comparable in timing and magnitude, but differences are apparent in the mid-shelf, where WA waters show elevated backscattering compared to OR in the winter months (Figure 6c,d). The similarities nearshore are expected because both WA and OR are under the influence of riverine inputs, but mid-shelf differences may point to the broader cross-shore extent of the Columbia River plume off WA compared to coastal river plumes created by smaller rivers to the south off OR. Bathymetry differences between the two lines may also explain these mid-shelf discrepancies, as illustrated from glider observations ( Figure 5). The temporal distribution of CDOM somewhat agrees with that of b bp when concentrations are high, particularly during winter, whereas during the remaining periods values oscillate above a background that is larger at WA during 2015, and relatively similar at WA and OR in 2016 (Figure 6e,f). The consistently lower CDOM concentrations at the mid-shelf OR mooring compared to offshore OR between October 2015 and March 2016 (Figure 6f) are suspicious, and may indicate drift or instrument issues.
In contrast with the optical properties, consistent cross-shore gradients in temperature  are observed across WA and OR, with nearshore waters presenting the coolest overall observations. Cross-shore gradients in temperature are strongest in spring-summer, consistent with the expected effects of upwelling and solar heating, and less pronounced in fall-winter, following stronger mixing processes. Nearshore and mid-shelf temperatures are overall cooler during the upwelling season in OR, as expected given the stronger wind conditions in OR compared to WA (wind stress is on average 1.5 times stronger on the Oregon shelf compared to the Washington shelf during the 2015-2016 period; Figure 6l). Salinity patterns in OR present marked seasonal changes in nearshore values (Figure 6i), showing the most saline waters in spring-summer and the freshest values in fall-winter. Salinity values are freshest overall along the WA line, as expected given its proximity to the Columbia River ( Figure 6j). Low-salinity peaks in December 2015 and January 2016 at both WA and OR inshore moorings (Figure 6i,j) agree well with CDOM maxima in Figure 6e,f. It is important to note that although the surface mooring observations cannot vertically resolve variations of physical and bio-optical properties, surface chl values have been shown to be well correlated with depth-integrated chl at the OR line (McKibben, 2016). However, we cannot exclude the importance of subsurface features such as thin layers of phytoplankton or bottom and intermediate nepheloid layers that are not sampled by the moorings (e.g., Figure 5). Additionally, the role of upwelling in lifting isopycnals toward the coast (and therefore changing the vertical distribution of chl across the shore) as suggested between May and June 2016 in OR (Figure 5a), undoubtedly complicates interpretation of moored data that are measured at the same depth (7 m) across the shelf. Further analyses should be expanded to include OOI Profiler Mooring data in order to resolve these vertical variations. Concurrent profiler data, however, were not available for the time period of this study.

Processes Controlling Observed Variability in In Situ Optical Properties
Establishing causal links between physical and biogeochemical changes in the surface ocean is not trivial. Cumulative alongshore wind stress, a metric that represents energy input into the upwelling system over the course of each season, is plotted in Figure 7 for WA and OR for periods of 2015 and 2016. Cumulative chlorophyll values over time, which are commonly used to assess bloom initiation times (e.g., Brody et al., 2013) and to summarize and highlight major changes in local chl values (e.g., McQuatters-Gollop et al., 2008), are also plotted in Figure 7 as cumulative chl anomaly relative to the respective time-series minimum (e.g.,, rapid and large increases in chl lead to increasing slopes, and small or null changes lead to flat slopes). The importance of upwelling events in driving increases in mid-shelf chl is clear at both locations, as abrupt increases in cumulative wind stress quantities are generally accompanied by steeper slopes of cumulative chl values (e.g., June 2015 and July-August 2016). In contrast, periods of flat or negative cumulative wind stress slopes (e.g., July 2015 for both WA and OR and . Temporal evolution of physical and optical variables at 7 m depth across the moorings of the WA and OR Endurance Lines (from daily averages). swh = significant wave height from the National Data Buoy Center (NDBC) buoy 46041. CRD = Columbia River Discharge. ws = alongshore wind stress convoluted with eight-day exponential decay as in Austin and Barth (2002). Wind stress measurements are from NDBC buoys 46041 (WA) and 46050 (OR). SWH at WA and OR buoys are well correlated with an r 2 = 0.90 for the 2005-2016 period.
August-September 2016 for WA) show overall flat cumulative chl anomalies. The late summer increases in cumulative chl in OR compared to WA during both years analyzed (also apparent in Figure 6a,b) may be related to the longer duration of the upwelling season at OR compared to WA. Nonetheless, the elevated chl values at WA despite much weaker upwellingfavorable winds indicate that other processes must contribute to the chl signals off WA, as reported by previous authors (e.g., Hickey and Banas, 2008). However, the latitudinal differences between chl estimates at WA and OR may not be as strong as the differences suggested from satellite data, given the relative similarity in the timing and magnitude of the variability of in situ chl estimates at the two lines (Figures 6a,b and 7). Although the moored observations alone are not sufficient to assess whether the apparent similarities in chl distribution (and the observed higher chl concentrations in OR in the summer) hold over multiple seasons and years, we hypothesize based on the evidence presented here that CDOM may be an important factor affecting the latitudinal differences observed from satellite-derived chl estimates (Figure 2). This is particularly relevant when interpreting chl sat signals off the Washington coastline, given the strong interference of the CDOM-rich Columbia River plume at those locations (Palacios et al., 2009(Palacios et al., , 2012. The OOI b bp record shows the importance of biological land-ocean interactions and resuspension processes in controlling b bp distributions. Salinity and b bp are inversely correlated at all WA moorings (r = −0.16, −0.33, and −0.53 offshore, at the mid-shelf, and inshore, respectively) and at the inshore OR mooring (r = −0.57). This indicates that allochthonous sources partially control b bp variability particularly at the inshore moorings. However, positive (but weak) correlations are observed at the mid-shelf and offshore OR moorings (r = 0.29 and r = 0.21, respectively), indicating the added importance of pelagic phytoplankton in characterizing the bulk b bp further offshore at OR, as shown in Figure 4b. Evidence for the role of mixing or resuspension processes in affecting the variability in b bp is observed between November 2015 and January 2016, when increases in nearshore b bp at WA and OR occur during a period of downwelling-favorable winds (Figure 6l), increased significant surface wave height conditions, and relatively low discharge ( Figure 6k). Indeed, positive relationships are observed between b bp and significant wave heights at the WA and OR inshore moorings (r = 0.61 and r = 0.51, respectively; correlations are the strongest at zero lag), whereas the relationships with discharge are weaker at the OR inshore mooring (r = 0.16), and, surprisingly, not significant at WA. The lack of strong relationships with discharge may be related to the relatively low discharge levels observed in 2015 compared to the climatology (see Figure 1). Nonetheless, these observations indicate that river discharge alone does not explain nearshore particle distributions. No significant correlations are observed between wave heights and b bp at any of the mid-shelf or offshore moorings, as expected given the constraint that increased depths impose on the particle resuspension processes.

CONCLUSIONS
Satellite observations of chl along the Washington-Oregon latitudinal range are consistent with the existing paradigm of higher annual magnitude and larger cross-shore extent of chl signals off the Washington shelf, despite reduced upwelling strength compared to Oregon. However, in situ bio-optical data for the time period of OOI data availability show relative similarity between the magnitude of the cross-shore chl distributions at WA and OR, and increased presence of suspended sediments and CDOM off WA. These observations, although temporally limited, indicate potential contamination of satellite retrievals of chl due to CDOM and suspended materials in the water column, particularly off the WA shelf, that should caution further attribution of chl sat signals to differences in production. Nonetheless, the considerable differences in wind stress between WA and OR indicate that processes other than upwelling are important for determining chl distributions at WA, as previously suggested. Glider-derived data show the strong seasonal influence of the Columbia River plume as a source of freshwater to the shelf off WA and to OR offshore waters, in agreement with known shifts in wind forcing. Full water column depth analysis using profiler and additional glider data that temporally align with each other and with mooring data during key seasonal events would be needed to ascribe latitudinal gradients in chl and particle distributions to variations in productivity. As the OOI time series matures, we will be better poised to discriminate between riverine impacts on particle loading and plankton productivity, and to more conclusively address the factors that lead to biogenic carbon production and export in the Northern California Current ecosystem.