Underwater optical environment in the coastal waters of British Columbia , Canada

We describe the underwater light field of the Strait of Georgia in spring and summer, using apparent optical properties (reflectance, attenuation coefficient of downwelling irradiance, the average cosine of downwelling irradiance, and the attenuation of scalar irradiance). Both the attenuation and reflectance of photosynthetically available radiation (PAR; 400–700 nm) are highest in the turbid waters of the Fraser River plume, due to scattering by mainly inorganic particles and absorption by coloured dissolved organic matter, phytoplankton, and other organic particles. Light is most diffuse in the surface waters of the plume and least diffuse at depth and away from the plume. Throughout the Strait, blue and red wavelengths are attenuated most rapidly resulting in a green peak of reflectance, the portion of the electromagnetic spectrum that penetrates the most deeply. PAR is attenuated to 1% of its surface intensity within 6–22 m in the spring and 4–23 m in the summer. For red and blue light, the depth of 1% penetration is never deeper than 9 m. All of the visible radiation, with the exception of some green light, is absorbed within the outflowing layer (15–30 m) that is exported from the Strait with the estuarine circulation. The rapid extinction of light helps to explain the very shallow distribution of phytoplankton.


Introduction
The Strait of Georgia (SoG), a highly productive, estuarine coastal sea off the west coast of Canada, supports several commercial fisheries, including that for the iconic sockeye salmon.Its productivity is governed by physical, chemical, and biological forcing, including the penetration of sunlight into the water column (Peña et al. 2016).The underwater light field is of fundamental importance to phytoplankton dynamics and ultimately to fisheries.In addition to the external forcing (length of day, sun angle, cloud cover, and surface-reflected light), the quantity and quality of the available light for photosynthesis depends on the attenuation processes, specifically on the scattering and absorption of light within the water column (Kirk 1994).
In the SoG, these attenuation processes are strongly influenced by the Fraser River (Loos and Costa 2010).The largest river on Canada's west coast (3660 m 3 •s −1 ) (Water Office 2017), the Fraser discharges particles and organic matter into the central SoG in a highly turbid surface plume, particularly during its May/June freshet (Milliman 1980;Kostaschuk et al. 1995;Luternauer et al. 1998; Barrie and Currie 2000;Stecko and Bendell-Young 2000).However, a thin, turbid layer persists at the surface of the SoG throughout the year (Johannessen et al. 2005).The Fraser plume, especially in the spring and summer, also presents well-defined inherent optical properties (IOP) with high scattering and absorption as a result of high concentrations of inorganic particulates and dissolved organic matter.The concentrations of suspended particles and chromophoric dissolved organic matter (CDOM) are inversely correlated with salinity, generally decreasing away from the river toward the western and northern parts of the SoG.Chlorophyll a (chl a) concentration is patchy, and its spatial distribution varies, but its absorption is more dominant in the northern waters (Loos and Costa 2010).
To date, little research has addressed the underwater visible light field or how light is attenuated within the euphotic zone of the waters of the SoG.Here we describe the underwater light environment of the SoG in spring and summer, when the biogeochemistry is the most dynamic.The underwater light environment is described based on its apparent optical properties (AOP), specifically, reflectance, attenuation coefficients of downwelling scalar irradiance, and the average cosine of downwelling irradiance, a measure of the diffusivity of light.We combine field and laboratory data with a radiative transfer model to define how the underwater light field varies spatially and with depth during spring and summer conditions.These data are unique for this region, and are relevant for defining light availability for primary production and the heat budget in the ocean.For instance, the Regional Ocean Modeling System parameterization of the downwelling flux of photosynthetically available radiation (PAR, 400-700 nm) is based on the vertical attenuation of downwelling light, which is defined according to the Jerlov water classification (Mobley et al. 2015).

Study area
The SoG exhibits estuarine circulation characterized by lower salinity seaward surface flow driven by the discharge of the Fraser and other rivers, and a deep return flow of more saline and nutrient-rich waters from the Pacific Ocean into the Strait (Li et al. 2000).Vigorous tidal mixing occurs in Haro Strait and Boundary Pass (Fig. 1), where deeper waters from the Pacific Ocean are mixed with surface waters (Masson and Cummins 2004).The waters of the SoG within the euphotic zone are influenced by the discharge of the Fraser River to different degrees depending on the season, surface currents, and tides.Thus, the salinity structure in the SoG creates the stratification that is always present in these waters (Tully and Dodimead 1957;Waldichuk 1957).
High concentrations of suspended particles and dissolved organic matter in the surface layer are generally observed, particularly in the Fraser River plume (Pharo and Barnes 1976;Kostaschuk and Luternauer 1989;Kostaschuk et al. 1993;Luternauer et al. 1998;Kostaschuk 2002).Primary productivity is high (∼280 gC•m −2 •year −1 ) and mainly limited by light, as for most of the year nutrients are supplied in excess by inflow of upwelled water from the Pacific Ocean (Thomas and Grill 1977;Stockner et al. 1979;Harrison et al. 1983Harrison et al. , 1994;;St. John and Pond 1992;Yin et al. 1995Yin et al. , 1997;;Mackas and Harrison 1997;Yin and Harrison 2000).

Data collection
Optical data (AOP and IOP) and water samples were collected at 11 stations in April 2006 and at 27 stations in July 2006 (Fig. 1), aboard the MSV John Strickland.Stations were chosen to span the optical variability of the waters of the SoG under the influence of the Fraser River, including stations used in previous research (Pawlowicz et al. 2003;Collins 2005;Johannessen et al. 2006).The methods and results for the IOP measurements were described in detail by Loos and Costa (2010) and will not be discussed further in this paper.
Spectral downwelling plane irradiance, E d (z,λ), and profiles of spectral upwelling radiance, L u (z,λ), were measured with Satlantic Minispec OCR-3000 sensors (400-800 nm; 1 nm spectral resolution) installed on a free-falling profiler.Above-water spectral downwelling plane irradiance, E s (0+,λ), was collected using a Satlantic Minispec OCR-3000 installed at the top of a 6 m long pole on the upper deck of the MSV Strickland to avoid shadowing.Underwater radiance reflectance, R r (z,λ), was computed using irradiance and radiance from the surface to the 1% penetration depth, Z 1% , as follows (Mobley 1995): The data were processed and binned to 1 m depth intervals.The reflectance presented in eq. ( 1) is also known as the "remote sensing ratio" (Mobley 2017) as L u (z,λ) and E d (z,λ) were measured in the water column (0−) for any depth.This reflectance is different from the commonly used "remote sensing reflectance", which is measured above the water (0+) at the water-air interface and uses water leaving radiance (the total upward radiance minus the surface-reflected sky radiance), L w (λ), instead of L u (z,λ).The relationship between the two is usually assumed to be: L w (λ) ≈ 0.544 L u (z,λ) (Mobley 1999;Doxaran et al. 2004).Our reflectances were collected under water, so there are differences in magnitude between our underwater reflectances and the remote sensing reflectances due to the different refraction indices of the two media (water and air).We were only interested in what was happening within the water column.Assuming E d is the same for both measurements (realistically they will be different simply due to the logistics of acquiring these data in the field), the remote sensing reflectances at the water-air interface (0+) will be lower than our remote sensing ratios.The results we have presented here are primarily for comparison with the underwater reflectances.
The diffuse attenuation coefficient for spectral downwelling plane irradiance, K Ed (z,λ), was calculated as follows (Mobley 1995): Z 1% of PAR was determined from the ratio of in-water spectral downwelling plane irradiance, E d (z,PAR), to above-water downwelling plane irradiance, E s (0+,PAR).Z 1% is approximately equivalent to the depth of the euphotic zone, Z eu (Kirk 1994;Behrenfeld and Falkowski 1997).
Seabird SBE 37SI conductivity-temperature-depth data were also acquired simultaneously with the optical data and binned to 1 m depth intervals.

HydroLight radiative transfer modelling
Numerical modelling of the optical properties of the waters of the SoG was performed with HydroLight 4.3 (Sequoia Scientific, Inc.) to calculate scalar irradiance, E o (z,λ), and the average cosine of downwelling irradiance (μ(z,λ)).E o (z,λ) accounts for incoming light over all angles (Mobley 2001), whereas E d (z,λ) is limited to light propagating downward and E u (z,λ) to light propagating upward only.E o (z,λ) better describes the irradiance available for photosynthesis from all directions (Bergmann et al. 2004).
HydroLight carries out radiative transfer calculations through invariant imbedding techniques that require IOP and environmental conditions as inputs (Mobley 1994).The HydroLight model was first validated for the study area using quantities that were measured in the field (E d (z,λ), L u (z,λ), K Ed (z,λ), and R r (z,λ); Table 1).The entire set of field data was entered into HydroLight's model ABACBB to solve the radiative transfer equations.The ABACBB model separated the optical constituents into pure water, particulate matter (detritus and phytoplankton), and CDOM.The input dataset consisted of measured optical properties (c t′ (z,λ), a t′ (z,λ), a c′ (z,λ)), reported by Loos and Costa (2010) (spectral range 400-700 nm), as well as pure water absorption values (Pope and Fry 1997), chl a concentration (from HPLC measurements (Loos and Costa 2010)), wind speed (collected with the vessel's anemometer), cloud cover (assessed visually), air pressure (collected with the vessel's onboard hygrometer), measured above-water spectral downwelling plane irradiance, E d (0+,λ), and date and time of field sampling.To ensure that the HydroLight output would closely match possible in situ conditions, it was necessary to input backscattering ratios, b b′ (z,λ)/b t′ (z,λ).However, as b b′ (z,λ) had not been measured in situ, those ratios were chosen from the HydroLight library and input as Fournier-Forand (FF) scattering phase functions (Bergmann et al. 2004).FF values were optimized for each station (values ranging from 0.004 to 0.028) by examining the closure between in situ and modelled R r (z,λ).All simulations were performed from the surface to Z 1% .Modelled E d (z,λ), L u (z,λ), K Ed (z,λ), and R r (z,λ) were then compared with in situ measured data to assess the performance of the simulation by analyzing the coefficient of determination, r 2 , and the slope of the line of best fit (Table 1).

Hydrographic data and water optical constituents
The average discharge of the Fraser River was 2578 m 3 •s −1 in April 2006, and 3933 m 3 •s −1 in July (high but not peak discharge, bracketing the May-June freshet).The surface water of the SoG was  stratified in both April and July, with a well-defined halocline and thermocline, both of which were strongest near the Fraser River and weakened westward and away from the plume.The hydrographic and biogeophysical properties of the OMs were previously defined by Loos and Costa (2010).

Optical dynamics
The in situ and modelled optical data allowed for a description of the light climate in the SoG waters in spring and summer.Generally, the quality and quantity of the in-water light field differed depending on how light was attenuated as a result of different optical constituents.
The reflectance spectra varied in magnitude and shape among OMs and at different times of the year.However, there was a clear transition from waters dominated by the Fraser plume (OM1), through a mixture of plume and Strait waters (OM2), into northern and deeper waters (OM3) (Fig. 2).R r (z,λ) was highest in OM1 (0.021 sr −1 in April and 0.040 sr −1 in July), and lowest in OM3 (deeper water) (Fig. 2).The strong reflectance in OM1 is a consequence of the high scattering caused by suspended particles and the low absorption-to-scattering ratio for these waters that was observed by Loos and Costa (2010).
The shape of the reflectance spectra varied among water masses and between the spring and summer conditions.There was a clear transition from waters dominated by the Fraser plume (OM1), through a mixture of plume and Strait waters (OM2), into northern and deeper waters (OM3) (Fig. 2).Generally, the reflectance was lowest at 400-450 nm (purple-blue), and peaked at 520-640 nm (yellow-green).The yellow-green peak was less pronounced in OM2 and OM3 than in OM1, consistent with the observation that in OM2 and OM3 absorption by CDOM contributed more to the total attenuation of blue light than did absorption and scattering by suspended particles (Loos and Costa 2010).
The reflectance minimum at 675 nm and small peak at 685 nm (due to absorption and fluorescence of chl a, respectively) were less pronounced in OM1 (Fig. 2).This was a consequence of the absorption by CDOM (1.0 m −1 in April and 0.53 m −1 in July) and increased scattering due to high concentrations of suspended particles (6.9 mg•L −1 in April, 9.9 mg•L −1 in July) in these waters (Loos and Costa 2010), which overwhelmed the effects of chlorophyll.
The high turbidity of the OM1 waters played a strong role in the attenuation coefficient of downwelling irradiance, K Ed (z,λ), and follows Jerlov's classification of waters with the highest turbidity.Jerlov's water body classification scheme is based on the vertical attenuation of downwelling irradiance.In all, there are five typical open ocean spectra (I, IA, IB, II, and III) and nine typical coastal spectra (1-9), with turbidity increasing with class number (Jerlov 1976).In April, in situ K Ed (z,λ) in OM1 was slightly higher than Jerlov's most turbid classification for coastal waters (Fig. 3).In July, K Ed (z,λ) in OM1 was similar to that of Jerlov's Type 9 in the short wavelengths but exceeded Type 9 above 500 nm.Vertical profiles of E d (z,PAR)/E s (0+,PAR) further corroborate the fact that light was attenuated with depth most rapidly in the turbid waters of the Fraser River plume (Fig. 4).The magnitude of K Ed (z,λ) decreased with distance from the Fraser River; OM3 waters were the clearest of all three OMs, particularly in July, when K Ed (z,λ) in OM3 was similar to Jerlov Type 1.In April, OM3 waters were somewhere between Jerlov types 3 and 5 (Fig. 3).

Scalar irradiance and the average cosine
For brevity, optical results will be discussed for four key wavelengths in the blue (411 nm), green (530 nm), and red (650 and 675 nm) parts of the electromagnetic spectrum to show the dynamics of light.HydroLight output predicted all the measured variables with p < 0.05.The p-value was used here simply to demonstrate the significance of the calculated values.In our case, we assumed that p < 0.05 indicated that the modelled results were significant and can be trusted.The relationships were stronger for E d (z,λ), L u (z,λ), and K Ed (z,λ), which are less sensitive to the geometry of the incoming light (Bukata et al. 1995) than for R r (z,λ) (Table 1).
HydroLight output resulted in low values of μ(z,λ) (∼0.7 at 411 nm) for OM1 waters, indicating a very diffuse light field (Fig. 5), whereas μ(z,λ) was higher in OM2 and OM3 waters.Light was less diffuse away from the plume and in deep water, as indicated by the increase in μ(z,λ) to a maximum of 0.9 at 411 nm in OM3.The high light diffusivity of OM1 is associated with high scattering caused by inorganic suspended particles.The association of in-water light diffusivity with inorganic or organic dominant matter was further explained based on data from Loos and Costa (2010) and according to the ratio of backscattering to particulate scattering, b b′ (z,λ)/b p′ (z,λ), and parameter B of McKee and Cunningham (2005), which considers the modelled backscattering coefficients and in situ particulate scattering coefficients (eq.( 3)).For OM1 waters, b b′ (z,λ)/b p′ (z,λ) values were the highest, decreasing away from the plume and with depth.A high OM1 b b′ /b p′ ratio (e.g., 0.012 at 530 nm) indicates a high proportion of inorganic particles in the suspended load, because of the high index of refraction of such particles (Twardowski et al. 2001;Boss et al. 2004) and is consistent with the high proportion of inorganic particles in river plume waters (Johannessen et al. 2003).OM2 and OM3 waters had low b b′ /b p′ indicating a higher proportion of organic particles in these waters (Table 2).The relationship between B wd and b p′ (z,675)/a t′ (z,675) indicated that scattering in OM1 was less variable and had lower wavelength dependence (B wd ≈ 1.0 in April and July) than that in OM2 and OM3 waters due to the higher concentration of suspended particles in OM1.The greater wavelength dependence of the scattering coefficient in OM2 is attributed to the organic nature of the particulate, namely phytoplankton.
Modelled normalized scalar irradiance, nE o (z,λ), obtained from the ratio E o (z,λ)/E o (0+,λ), was the lowest in OM1 and indicated that blue and red wavelengths were quickly attenuated within the first 5 m to below 5% of the surface intensity (Figs. 6a,6c).Below this depth, available light was predominantly in the green wavelengths (Fig. 6b).This is also supported by the low K Ed (z,500-600 nm) in the green spectrum (Fig. 4).The modelled attenuation coefficient of scalar   Note: μ(z,λ), average cosine of downwelling irradiance; E d (z,λ), spectral downwelling plane irradiance; L u (z,λ), spectral upwelling radiance; K Ed (z,λ) downwelling irradiance attenuation coefficient; R r (z,λ), underwater radiance reflectance; b b′ (z,λ), particulate backscattering; b p′ (z,λ) particulate scattering; nE o (z,λ), normalized scalar irradiance; 0-, in the water.3) in spring and summer was similar to values reported by Stockner et al. (1979) and Harrison et al. (1991) in the waters of the SoG during winter and spring.Virtually none of the light that enters the SoG passes all the way through the surface estuarine outflow layer.The depth of separation between the outflowing and inflowing layers has been modelled at 50 m (Pawlowicz et al. 2007) and 30 m (Riche and Pawlowicz 2014), and the depth of the Fraser River plume at 15 m (Johannessen et al. 2006;Masson 2006).The more highly depth-resolved profiles shown here (Figs. 2, 7) suggest that most of the light field is attenuated by the fresh water contained within the uppermost 7 m in spring and summer.

Loos et al.
By 15 m, only green light remains (Fig. 7), and even then only in areas away from the Fraser River plume (S4, S5, and S6 group of stations in Fig. 1).By 30 m, all of the downwelling irradiance has been  (Loos and Costa 2010) or reflected back out of it (Figs. 3,7).Figure 2 shows the high reflectance for the OM1 waters.Consequently, all of the light is attenuated within the outflowing surface layer, which has a residence time of approximately 10 d in the Strait (Pawlowicz et al. 2007).This is consistent with the observation by Pawlowicz et al. (2007) that the SoG is a small net exporter of heat, based on the absorption of shortwave (UV + visible) radiation.
Z 1% was 6.0-22.0m in spring and 4.0-23.0m in the summer.The depth of 1% penetration of blue radiation at one of the chlorophyll absorbance peaks (443 nm) was always shallower than 9.0 m and often shallower than 5.0 m.Light at all wavelengths was more rapidly attenuated in the summer than in the spring within the river plume, because of the high scattering by inorganic particles and high absorption by CDOM (Loos and Costa 2010).However, away from the plume, blue wavelengths were more rapidly attenuated in the spring, because of the absorption by phytoplankton (Loos and Costa 2010).
The effects of the attenuation of light can be seen in the depth distribution of phytoplankton, which occurs over the top 30 m in spring, but only over 12 m in summer (Peña et al. 2016).Primary productivity is always limited by light in the southern Strait, so productivity is actually higher in that region in the summer, when more light is available, than in the spring (Peña et al. 2016).In the central Strait, which includes waters under the strong influence of the Fraser plume, phytoplankton are thought to be usually limited by light and occasionally by nutrients (Peña et al. 2016); although, it can be difficult to differentiate the effects of stratification (nutrient limitation) from those of turbidity (light limitation), as the two factors co-occur in the plume.In fact, primary production is low in the plume in summer (Peña et al. 2016), because both light (due to high attenuation and therefore lower E d (z,λ)) (Fig. 7) and nutrients are in short supply.In the summer, because of their shallow distribution, some phytoplankton are exported from the Strait (Peña et al. 2016) with the illuminated layer of water and absorbed radiation; although, most suspended particles are retained within the Strait (Johannessen et al. 2003).

Conclusions
The results presented here described the in-water light environment in the SoG, based on in situ observations and radiative transfer modeling.In the SoG, the euphotic zone is shallow due to high turbidity by particles and high absorption by CDOM.In both spring and summer, light generally penetrates <15 m into the water column.The light is nearly all attenuated within the outflowing water.Only a small amount of green light passes through the upper layer into the inflowing layer beneath; this likely limits primary production to the uppermost layer.
The waters under the influence of the turbid Fraser River plume showed the highest variability and the most rapid attenuation of light with depth.The rapid attenuation was mostly due to high backscattering by suspended inorganic particles as well as to absorption by CDOM.Reflectance was also highest in the plume waters.A peak at 685 nm, indicating chlorophyll fluorescence, was visible in most reflectance profiles, except in the summer in turbid waters strongly influenced by the Fraser plume.Furthermore, red and blue light were attenuated to <1% of their surface intensity within the uppermost 5 m, whereas green and yellow light persisted to about 20 m.This was the result of the high K Ed (z,400∼520 nm).Therefore, K Ed (z,λ) decreased with depth as the concentrations of the absorbing and scattering constituents decreased, with the lowest values (clearest waters) in deeper waters.The very turbid Fraser plume water was characterized by a very diffuse underwater light field, as indicated by the low average cosine (μ ≈ 0.7 at 411 nm), which increased with depth and with distance from the Fraser River to a maximum in OM3 (μ ≈ 0.9 at 411 nm).
The surface waters of the SoG are highly stratified.Although ship-based studies seldom resolve the surface layer at more than 5-10 m vertical resolution, the results presented here demonstrate that

Fig. 1 .
Fig. 1.Study area and sampling stations in the Strait of Georgia, British Columbia, Canada.North and South Arms of the Fraser River are also indicated.April stations are represented as black dots; all stations were sampled in July.

Fig. 2 .
Fig. 2. Means of radiance reflectance, R r (z,λ), for the first metre of each optical water mass (OM) in (a) spring and (b) summer 2006.Note the different y-axis scales in plots (a) and (b).

Fig. 5 .Fig. 4 .
Fig.5.Average cosine at 411 nm for all three optical water masses (OMs) in July.Diffusivity increases with decreasing average cosines.OM1 had the lowest average cosines because of their high attenuation due to inorganic particulate scattering.

FACETSFig. 6 .
Fig. 6.Ratio between in-water downwelling scalar irradiance and above-water downwelling scalar irradiance in spring and summer at (a) 411 nm, (b) 530 nm, (c) 650 nm, and (d) 675 nm for all stations; (e) ratio between photosynthetically available radiation (PAR) in-water downwelling scalar irradiance and abovewater downwelling scalar irradiance in April and July for all stations.
Loos et al.FACETS | 2017 | 2: 872-891 | DOI: 10.1139/facets-2017-0074 885 facetsjournal.comsalinity,temperature, turbidity, and irradiance change rapidly with depth over the top 10 m.The strong stratification has consequences for phytoplankton distribution and the energy budget of the system.More highly depth-resolved measurements are necessary to study these processes accurately in the SoG and in other river-dominated coastal margins.

Table 1 .
Relationships between measured and modelled variables (r 2 and slope of the line of best fit).

Table 3 .
Modelled and in situ radiometric quantities and apparent optical properties.
Loos et al.