Optical properties of size and chemical fractions of suspended particulate matter in littoral waters of Quebec

Mass-specific absorption (aSPM) and scattering (bSPM) coefficients of suspended particulate matter (SPM) were measured for different size (0.2-0.4 μm, 0.4-0.7 μm, 0.7-10 μm, and >10 μm) and chemical (organicvs mineral-rich) fractions in surface waters (i.e., 0-5 m depth) of the Saint Lawrence Estuary and Saguenay Fjords (SLE-SF) during spring of 2013. For the spectral range 400-700 nm, scattering cross sections for particulate inorganic matter were commonly larger 10 with respect to those measured in other littoral environments. This phenomenon was attributed the lower water turbidity and associated decrease on mean particle size of SLE-SF surface waters with respect to other river-influenced regions (e.g., Gironde River). aSPM values in our study area were relatively high in locations having iron-enriched particulates. Lastly, correlation analysis suggests that particle composition (size distribution) has a larger impact on aSPM (bSPM) variability.

Remote sensing allows mapping of SPM in littoral environments where the spatial and temporal variability of suspended particulates is relatively high.Indeed, synoptic measurements derived from spaceborne ocean color sensors are commonly applied for estimating C SPM based on visible (i.e., wavelength,  = 400-700 nm) (Miller and McKnee, 2004;Montes-Hugo and Mohammadpour, 2012) and NIR-SWIR (near-and short-wave infrared) ( = 700-3,000 nm) (Doxaran et al., 2002) 25 spectral bands.Despite this progress, there is still a lack of understanding regarding how SPM microphysical characteristics (e.g., particle chemical composition and size distribution) relate to mass-specific inherent optical properties (IOPs).This Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.
knowledge is essential for deriving more accurate remote sensing algorithms for estimating C SPM and developing new optical inversions for retrieving second-order attributes of SPM (i.e., chemical composition, size distribution).
The remote sensing of particle size and/or composition in coastal and oceanic waters has been attempted based on four main methodologies: (1) analysis of spectral changes of IOPs (Loisel et al., 2006), (2) empirical relationships between massspecific IOPs and biogeo-physical characteristics of SPM (e.g.mean diameter of particulates) (Bowers et al., 2009), (3) 5 optical inversions of different volume scattering functions (Zhang et al., 2014), and (4) changes on water leaving polarized reflectance (Loisel et al., 2008).
The Saint Lawrence Estuary (SLE) and the Saguenay Fjords (SF) constitute a large sub-Arctic system characterized by relatively high concentrations of chromophoric dissolved organic matter (CDOM) (Nieke et al., 1997).The accurate monitoring of C SPM and SPM characteristics in these waters is crucial for understanding regional climate effects on coastal 10 erosion (Bernatchez and Dubois, 2004) and occurrence of harmful algae blooms (Fauchot et al. 2008).Despite this need, there is a lack of information regarding how optical properties are linked to particle second-order attributes and what is the spatial variability of mass-specific IOPs of SPM.For this reason, our contribution has two main objectives: (1) to characterize the mass-normalized IOPs for size and chemical fractions of SPM in different locations of the SLE-SF and during spring conditions, and (2) to establish relationships between mass-specific optical properties of SPM, 'bulk' particle 15 characteristics related to size distribution and mineral content, and optical proxies within the visible and near-infrared spectral range (i.e.,  = 700-1,000 nm).
This study is organized in three sections.In the first section, mass-normalized spectral absorption and scattering coefficients for size and chemical SPM fractions are calculated for different optical environments of the SLE-SF that are characterized by a variable CDOM contribution to light attenuation and distinct particle assemblages.In the second section, the response of 20 mass-normalized absorption and scattering coefficients of SPM fractions to variations in particle size distribution and mineral-content are investigated.Lastly in the third section, covariations between optical proxies and microphysical properties of SPM are examined.

Study area 25
The SLE can be divided in two main regions having contrasting biological productivity and bathymetry: the upper (UE) and the lower (LE) estuary (Levasseur et al., 1984).Non-algal particulates (NAP) and CDOM dominate the underwater light attenuation of UE waters (Nieke et al., 1997).This is in part related to the inflow of CDOM-rich and NAP-rich waters coming from the St. Lawrence River (Tremblay and Gagné, 2007).Unlike NAP and CDOM, contribution of phytoplankton to IOPs increases towards the mouth of the SLE (Montes-Hugo and Mohammadpour, 2012).Historical studies performed 30 during summer of 1975 suggest that size distribution of SPM differs between UE, LE and SF regions (Poulet et al., 1986).

Field surveys
Discrete water samples for biogeochemical and optical measurements were obtained in 23 locations distributed throughout the SLE (n =18) and SF (n = 5) regions (Fig. 1).Samples corresponding to a sampling depth of 0-2 m were collected during 10 June 3-9 of 2013 by using an oceanographic rosette equipped with Niskin bottles (volume = 12 L).For each sampling location, mass of different size fractions of SPM, IOPs for different SPM size fractions, and particle size distribution spectra were measured.

Biogeochemical analysis
The concentration of SPM and particulate inorganic matter (C PIM ) in g m -3 was measured gravimetrically with a precision of 15 15% and 25%, respectively (Mohammadpour et al., 2015).Size fractionation of SPM was done after sequentially filtering the original samples through pre-weighted membranes having a diameter of 47 mm and a pore size of 10 µm (Whatman, polycarbonate), 0.7 µm (GF/F, Whatman, glass fiber), 0.4 µm (Whatman, polycarbonate), and 0.2 µm (Nucleopore, polycarbonate).The contribution of size fraction i to the total mass of SPM (F SPM i , i = 0.2-0.4µm, 0.4-0.7 µm, 0.7-10 µm, and >10 µm) was computed by normalizing their weight by the total weight of unfractioned samples that were retained on 20 0.2 µm membranes.The inorganic fraction of SPM (i.e., particulate inorganic matter or PIM) was obtained after removing the organic fraction (i.e., particulate organic matter or POM) of the original sample by combustion at 450°C for 6 h.Due to the dehydration of clays, this procedure may introduce an additional uncertainty of -10% and +10% on particulate inorganic (PIM) and organic matter (POM), respectively (Barillé-Boyer et al., 2003;Stavn et al., 2009).The contribution PIM and POM to SPM mass is F SPM j where j superscript symbolizes PIM or POM, respectively.25

Optical measurements
Discrete water samples for CDOM absorption coefficient (a CDOM ) determinations were done in the lab following protocols suggested by Müller and Horn (1990).CDOM is defined here as the fraction or dissolved organic matter passing trough a membrane with a nominal pore size of 0.2 m.Total absorption (a) and beam attenuation (c) coefficient measurements of four size-fractioned water samples (0.2 -0.4 µm, 0.4 -0.7 µm, 0.7 -10 µm, and > 10 μm) were performed onboard using an 30 absorption-beam attenuation meter (ac-s, WetLabs).Optical measurements were corrected by applying a flat baseline at a Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.reference wavelength of 715 nm (Bricaud and Stramski, 1990).This is a first order correction for scattering effects on nonwater absorption coefficient estimates.Thus, the calculation of particulate absorption coefficients in this study is expected to have a bias with respect to true values measured using absorption-meter instruments that are less influenced by particulate scattering (e.g., point-source integrating-cavity absorption meters) (Röttgers et al., 2014).Spectral scattering coefficient measurements (b) were derived by subtracting a from c at each wavelength.The particle size spectra within the size range 3-5 170 µm were measured on 'bulk' (i.e., without size fractionation) samples and by using a red laser (wavelength = 670 nm) diffractometer (LISST-100X, type B, Sequoia Scientifics) (Agrawal et al. 1991).

Optical proxies of particle microphysical characteristics
Optical composite parameters directly related to remote sensing reflectance (R rs ) (Table 1) were constructed based on inwater IOPs determinations.Unfortunately, no remote sensing reflectance measurements were available during this study.10 Spectral values of a and b can be linked to the irradiance ratio measured just below the water surface (R(0 -)) (Morel and Prieur, 1977): where f is a coefficient that varies with atmospheric (e.g., solar zenith angle) and water (e.g., single scattering albedo) 15 parameters (Morel and Gentilli, 1996), b b eff is the total (i.e., water + particulates) backscattering efficiency (i.e., b b /b) where b b is the total backscattering coefficient).The magnitude of  depends on refraction and internal reflection of photons at the air-water interface.For a nadir-looking sensor, the Q n ( 0 )is defined as the ratio between upwelling irradiance and upwelling radiance just beneath the sea surface and as a function of the solar zenith angle ( 0 ).From equations ( 1) and ( 2), three biogeo-optical indices (BOI) were proposed for estimating changes in 'bulk' chemical composition (superscript comp) and 20 size distribution (superscript size1 and size2) of SPM: where a SPM is the particulate absorption coefficient, F is the polynomial function g + g 2 , where g = b() (b()+ a()) -1 .Notice 25 that F resembles Gordon's formulation of R rs for nadir-view geometry and optically deep water et al., 1988), where f 1  0.0949 I and f 2  0.0794 I, I  t 2 n -2 or the air-sea interface divergence factor (t is the airsea transmittance and n is the refractive index of seawater). 1 ,  2 ,  3 ,  4 ,  5 and  6 correspond to wavelengths 443, 488, 555, 570, 670 and 675 nm, respectively.Values of a SPM were derived by subtracting the contributions of CDOM and seawater to a.The absorption coefficient (a w ) and scattering (b w ) coefficient of seawater were computed at in situ salinity 30 Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.and temperature by using empirical parameterizations suggested by Pope and Fry (1997) and Zhang et al. (2009), respectively.
The equation ( 3) was suggested based on empirical relationships between a SPM ( 6 )/a SPM ( 4 ) and POC/C SPM ratios, where POC is the particulate organic carbon concentration (Wozniak et al., 2010).BOI size1 and BOI size2 indices for particle size distribution were based on Carder et al. (2004) andD'Sa et al. (2007) published R rs band ratios, respectively.These ratios are 5 correlated to the spectral slope of particulate backscattering.In general, BOI comp values are expected to increase as SPM becomes richer in POC.Likewise, BOI size1 and BOI size2 are anticipated to decrease as particulates become larger or water contribution to backscattering increases at relatively low water turbidities.

Optical cross sections and mass-normalized IOPs
Spectral values of mass-specific absorption ( a j ) and scattering ( b j ) cross sections for mineral and organic fractions of SPM 10 were estimated from multiple regression analysis (Sokal et al., 1995).The superscript j indicates inorganic (PIM) u organic (POM) particulate matter.For the case of size fractions of SPM, a mass-normalized optical property was calculated for particulate absorption and scattering coefficients: where m is the mass in g m -3 for each size class i.

Statistical analysis
The influence of particle size and chemical composition variations on a SPM , b SPM ,  a ,  b , a i * , and b i * was investigated using the non-parametric Spearman rank correlation coefficient ( s ) (Spearman, 1904).This metrics was also applied to examine the response of a i * and b i * values to changes on the exponent of the power-law distribution of particle size distribution () or 20 the slope of log-transformed number of particulates per unit of volume as a function of their size range (Junge, 1963).Values of  were computed based on linear regression models where dependent and independent variables are randomly selected (i.e., type II parameterization).Although particle size distribution in natural waters may not follow a Junge-type slope, its use here was justified since our main interest was to have a first-order assessment of size effects of particulates on IOPs variability.The sensitivity of BOI comp , BOI size1 , and BOI size2 to variations of different chemical and SPM size fractions was 25 quantified based on the magnitude of  s .Lastly, potential functionalities between mass-normalized IOPs and BOI indices were examined for different study areas based on linear regression analysis model type II.

Spatial variability of SPM fractions
In terms of particle size distribution, contrasting areas in the SLE-SF were identified.In UE, particulates having a diameter larger than 10 µm had in average a contribution of 11% to the total SPM mass (Table 2).This proportion was lower in the LE (up to 9%) and SF (up to 6%) sub-regions.The largest mass contribution of smallest-sized particulates (i.e., diameter < 5 0.4 µm) was calculated in the lower estuary.Lastly, the intermediate size class 0.7-10 µm was the fraction having the maximum contribution to SPM in the SF (76.5% in average).In general, the Junge slope calculations suggested the presence of relatively larger particulates in the LE with respect to UE and SF sub-regions.Indeed, the arithmetic average and range of  for LE, UE and SF locations were 1.67 and 0.9-2.4,2.4 and 2.3-2.4,and 2.4 and 2.1-2.6,respectively.The uncertainty of  calculations varied between 8 and 90% with smaller errors in the LE.Unlike particle size distribution, chemical composition 10 of SPM was less variable (F SPM PIM range = 20 -87 %).In average for each sub-region under investigation, the mass of suspended particulates was always dominated by inorganic matter (arithmetic average of F SPM PIM = 0.58, 0.62 and 0.70 for SF, UE and LE, respectively, Table 2).The largest variability of mineral content of SPM was characteristic of waters with relatively shallow depths and a greater contribution of freshwater discharge by the St Lawrence River (e.g., sampling locations 12 and 13 in the UE, Fig. 1).15

Relationships between SPM fractions and IOPs
In general, size and chemical composition of SPM were important second-order attributes affecting the scattering coefficient of suspended particulates.In general, b SPM response to changes on SPM size fractions and chemical composition ( s up to 0.71 and 0.59, t up to 21.17 and 15.35, Student-t test, respectively) was greater with respect to that associated to a SPM ( s up to 0.53 and 0.21, t up to 13.13 and 4.51, respectively, Student-t test) (Table 3).The larger influence of particle size 20 distribution on b SPM compared to a SPM values was supported by correlations between  and IOPs ( s up to 0.50, t up to 12.12, Student-t test;  s up to 0.33, t up to 7.34, Student-t test) (Table A1).Unlike particle size, the impact of SPM chemical composition on a SPM was principally manifested at relatively short wavelengths (i.e.,  = 440-556 nm,  s up to 0.21, t up to 4.51, Student-t test, Table 3).Indeed, the highest correlations between SPM size fractions and a SPM values were computed in the red-NIR spectral regions (e.g.,  s up to 0.41, t up to 9.44, Student-t test).25

Mass-specific optical properties of SPM
The variation of mass-normalized scattering and absorption coefficients of SPM for different size and chemical fractions are shown in Fig. 2. Full spectral variation of regional averaged a SPM * and b SPM * values are depicted in Fig. A1 (Appendix A).In general, sub-regional averages of mass-normalized IOPs of particulates with different size ranges were higher with respect to optical cross sections of chemical fractions (up to 2 and 3 orders of magnitude for a and b, respectively).For a wavelength of 30 , respectively (Fig. 2b).
For the spectral range 440-556 nm, mass-normalized absorption coefficients of SPM tended to be higher for particulates within the lower size range (i.e., 0.2-0.4m) (Fig. 2a, left-axis).Also, this trend appeared to be reversed at longer 5 wavelengths.Unlike mass-normalized absorption coefficients of size fractions, mass-specific cross sections of chemical fractions showed only differences within the red and near-IR wavelengths (Fig. 2a, right-axis).For the whole study area, the arithmetic average of mass-normalized scattering coefficients for the size fraction 0.2-0.4µm were larger with respect to that associated to the size fraction >10 µm (Fig. 2b, left-axis).At a wavelength of 440 nm, the mass-specific scattering cross sections for PIM were substantially higher (1.060  0.206 m 2 g -1 ) than those corresponding to POM (0.359  0.123 m 2 g -1 ) 10 (Fig. 2b, right-axis).
In general, the magnitude of the mass-normalized absorption coefficient at 440 nm and computed for different size and chemical fractions was higher in UE-SF with respect to LE locations (Fig. 3a).Notice that absorption or scattering cross sections for chemical SPM fractions are not shown in UE locations given the insufficient number of samples to perform a multiple regression analysis.In Saguenay Fjord waters, the maximum a SPM * (440) values (up to 4.6 m 2 g -1 ) were associated 15 with the largest size fraction of SPM (Fig. 3, left-axis).Unlike size fractions, no substantial sub-regional differences were detected for  a PIM (440) and  a POM (440) values (P > 0.05, t up to 0.42, Student-t test) (Fig. 3, right-axis).In general,  and F SPM PIM correlations with mass-normalized IOPs suggest that particle chemical composition has a larger influence on a i * (440) ( s up to 0.50, t up to 12.12, Student-t test) with respect to particle size ( s up to 0.32, t up to 6.85, Student-t test) (Table 4).Unlike mass-specific absorption coefficients calculated at a wavelength of 440 nm, mass-specific scattering 20 coefficients computed at 550 nm and for different size and chemical fractions of SPM presented smaller variations among spatial domains (Fig. 3b).Only for the intermediate size fraction 0.7-10 m, the regional average of b i * (550) in UE-SF (0.432-0.501 m 2 g -1 ) was larger with respect to that computed in LE waters (0.136  0.027 m 2 g -1 ) (Fig. 3b, left-axis).Unlike a i * (440), b i * (550) variability was less influenced by changes on particle composition ( s up to 0.42, t up to 9.72, Student-t test) (Table 4).Conversely, the impact of changing particle dimensions, as inferred from  s correlations, was greater for 25 b i * (550) ( s up to 0.37, t up to 8.36 Student-t test) with respect to a i * (440) ( s up to 0.33, t up to 7.34 Student-t test) values.

Optical proxies
Correlations between individual samples of size-based fractions of SPM and optical proxies of particle size and chemical composition are presented in Table 5.In general, it was found that BOI size1 was a more selective biogeo-optical indicator for retrieving second-order properties of SPM than BOI size2 and BOI comp .Indeed, BOI size2 was also dependent on particle 30

Uncertainty of optical properties
Inherent optical properties in this study were derived from an ac-s instrument.Thus, large errors on absorption coefficients may be anticipated in relatively turbid waters if original measurements are not corrected by scattering effects (Boss et al., 2009;McKee et al., 2013).These effects are mainly attributed the acceptance angle of the transmissometer and the multiple scattering of photons.The acceptance angle of the ac-s instrument is ~0.9° and much larger than that corresponding to the 15 LISST-100X diffractometer (~0.027°).Thus, a larger underestimation on c magnitude is expected in ac-s with respect to LISST-100X measurements due to a larger contribution of forward-scattered photons arriving to the detector of the former optical instrument.Further comparisons of c(532) measurements derived here by ac-s and LISST-100X showed that c values as derived from ac-s were 23-84% lower with respect to those determinations based on LISST-100X.This is consistent with Boss et al. (2009) who reported that uncorrected Wet Labs ac-9 attenuation values are approximately 50%-80% of equivalent 20 LISST attenuation data.Unfortunately, c deviations due to acceptance angle variations were not corrected in this study due to the lack of true c values as obtained by using an integrating cavity absorption meter (e.g., PSICAM) (Röttgers et al., 2005).Notice that these errors are much greater with respect to the optical variability associated to each sample determination as computed from ac-s measurements (e.g., < 1% at  = 532 nm).
In this investigation, the 'flat' baseline correction was selected for correcting residual scattering in absorption coefficient 25 estimates as derived from ac-s measurements.This technique was chosen due to the lack of PSICAM measurements or critical ancillary optical information (e.g., particle backscattering efficiency) to tune up a Monte Carlo scattering correction approach (McKee et al., 2008).The 'flat' scattering correction approach is expected to provide a fair correction of a values in oceanic waters (up to 15% underestimation at wavelengths shorter than 600 nm, see Fig. 8b, McKnee et al., 2013) but may result in large deviations (up to 100% decrease in the NIR) of a values in relatively turbid waters (e.g., a > 0.2 m -1 ) such as 30 the Baltic/North Sea.Also, this issue is present when the proportional correction method of Zaneveld et al. (1994)  Unlike the 'flat' baseline, the scattering residual of the proportional method is spectrally dependent but still relying in one reference wavelength in the NIR spectral range.Approximations justifying the use of the 'flat' (i.e., zero absorption signal in the NIR) and 'proportional' (i.e., wavelength-dependent scattering phase function) method are still in debate (McKnee et al., 2013).Lastly, the Monte Carlo correction method (McKee et al., 2008) has in general better agreement (error <10%) with true a values as derived from an integrating cavity absorption meter.However, this approach may also have major 5 uncertainties due to assumptions regarding IOPs (e.g., particulate backscattering ratio and volume scattering function) and changes on scattering efficiency by the inner wall of the reflective tube due to aging (McKnee et al., 2013).Thus in conclusion, the resulting particle-related IOPs and mass-specific optical coefficients obtained in the SLE-SF waters may present large errors (i.e., > 50%) with respect to true values and at wavelengths longer than 550 nm.This bias is anticipated to be maximum (minimum) in UE (LE) locations.10

Spatial patterns of SPM microphysical characteristics
A striking finding in this study was the important weight contribution of relatively large particulates (i.e., >10 µm) in UE waters.This phenomenon was likely attributed to the active resuspension of sediments associated with vertical mixing produced by tidal currents and winds (Yeats, 1988).Conversely, this effect was secondary in relatively deep waters of SF and LE where large and heavy particulates are rapidly removed from the water column and deposited along submarine 15 canyons (Gagné et al., 2009).
Although chemical composition of SPM size fractions was not analyzed in this study, additional correlations between total F SPM PIM and SPM size fractions values suggest that smallest particulates were richer in inorganic matter ( s = 0.27, t up to 5.89, Student-t test, Table A2).Also, the opposite was true for the largest particulates ( s = -0.27,t up to -5.89, Student-t test).This finding confirms previous studies showing that relatively small (~2 μm) particulates in the SLE are mainly 20 composed by minerals (Yeats, 1988;Gagné et al., 2009).
In this contribution, a large proportion of particulates with a diameter above 50 m and lower  values were typically found in LE locations.This regional variation in SPM size distribution was attributed to the major influence of large-sized particulates derived from phytoplankton as  was strongly correlated with chlorophyll a concentration ( s = -0.45,t up to -10.58, Student-t test, Table A3).These results also support historical observations made during July and August and showing 25 a greater proportion of relatively large particulates (i.e., > 5 and < 50 µm) over the LE locations (Chanut and Poulet, 1979).

Spatial variability of mass-specific optical coefficients
In this study, a SPM * measurements in the visible and near-IR range were in the upper range or higher than those reported in the literature for temperate coastal waters (e.g., Mobile Bay, River of La Plata, Elbe Estuary, Gironde Estuary) (Stavn and Richter, 2008;Doxaran et al., 2009;Dogliotti et al., 2015) (Table 6).In general, lowest a SPM * values commonly 30 corresponded with samples obtained in very turbid environments (i.e., > 100 g m -3 , Gironde River, La Plata River) (Dogliotti Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License. et al., 2015;Doxaran et al., 2009).Notice that part of this decrease can be attributed to an incomplete removal of multiple scattering effects.One mechanism explaining the general decrease of a SPM * in very turbid waters is related to packaging effects (Morel, 1974;Zhang et al., 2014).At higher turbidities, particulates become dominated by larger size distributions, thus as mean diameter of particles increases, the scattering efficiency of SPM decreases.In SF waters, the magnitude of a >10m Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.

Optical proxies of particle size and composition
The response of three optical composite variables (BOI size1 , BOI size2 and BOI comp ) to changes on size and composition of different particle assemblages was evaluated based on correlation analysis.In general, BOI size1 was the most selective optical index for tracking variations on particle micro-physical properties.Indeed, BOI comp (BOI size2 ) was also substantially affected by size distribution (chemical composition) of SPM.The lack of specificity of BOI comp may respond to the use of a spectral 5 range where phytoplankton has a maximum light absorption peak (i.e.,  = 675 nm).As phytoplankton cells become larger (e.g., above 20 µm), the total chlorophyll a concentration of phytoplankton cells increases (Montes-Hugo et al., 2008).As result, the magnitude of a SPM at a wavelength of 675 nm is expected to increase affecting positively BOI comp .Lastly, BOI size1 and BOI size2 response was mainly associated with variability of small-sized and intermediate-sized SPM fractions, respectively.This selectivity is particularly interesting as both indexes may be combined for developing more robust metrics 10 for estimating SPM size spectra distributions in littoral waters.

Conclusions
The measure of optical cross sections of SPM is essential for developing optical inversions and improve our understanding regarding the origin of optical signatures in remote sensing studies and map biogeo-chemical components in surface waters.
In this contribution, we presented for the first time, mass-specific scattering and absorption coefficients of size fractioned 15 SPM in estuarine waters of the Saint Lawrence River and a major SLE tributary, the Saguenay Fjord.
Despite the intrinsic variability of weight-normalized IOPs due to variations of particle micro-physical attributes, the following trends were observed: 1. the mass-specific absorption coefficient of SPM was preferentially influenced by changes in particle chemical composition, 2. particle size had a larger impact on b SPM * than a SPM * , and 3. optical proxies of SPM size distribution BOI size1 was more specific than optical proxy related to particle chemical composition (i.e., BOI comp ).These 20 relationships are anticipated to be useful in the context of predicting mass-specific IOPs based on satellite remote sensing measurements.
This investigation was supported by the Natural Sciences and Engineering Research Council of Canada, Individual Discovery grant, project title: "Optical remote Sensing models of suspended Particulate matter in the St. Lawrence Estuary 25 "(OSPLE), awarded to Dr. Martin Montes Hugo.Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.

Table 5 .
Particle size and chemical composition effects on optical proxies.Statistic confidence levels of  s values are Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.

Figure captions Figure 1 .
Figure captions Figure 1.Study area.UE (green triangles), LE (blue rectangles) and SF (red circles).GSL is the Gulf of St. Lawrence.

Figure 2 . 5 Figure 3 .
Figure 2. Spectral variation of mass-normalized optical coefficients of SPM.(a) particulate absorption at  = 440 nm, (b) particulate scattering at  = 550 nm.Each bar corresponds to the arithmetic average over the whole study area; uncertainty bars symbolize  2 standard errors.5

Table 2 . Summary of biogeochemical variables during June 2013. Acronyms UE, SF, LE are defined in Table 1. N is the number of samples per sub-region.
Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-159Manuscript under review for journal Biogeosciences Discussion started: 1 June 2017 c Author(s) 2017.CC BY 3.0 License.