Light-scattering features of turbidity-causing particles in interconnected reservoir basins and a connecting stream

water research 43 (2009) 2280–2292

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Light-scattering features of turbidity-causing particles in interconnected reservoir basins and a connecting stream Feng Penga, Steven W. Efflera,*, Donald C. Piersonb, David G. Smithc,1 a

Upstate Freshwater Institute, P.O. Box 506, Syracuse, NY 13214, USA New York City Department of Environmental Protection, 71 Smith Avenue, Kingston, NY 12401, USA c New York City Department of Environmental Protection, 465 Columbus Avenue, Valhalla, NY 10595, USA b

article info

abstract

Article history:

Light-scattering features of minerogenic particles in interconnected reservoir basins and

Received 8 August 2008

a connecting stream in the watershed of New York City’s water supply system, where these

Received in revised form

particles dominate scattering, were characterized by scanning electron microscopy inter-

18 December 2008

faced with automated X-ray microanalysis and image analysis (SAX). SAX provided

Accepted 16 February 2009

information on composition (in terms of elemental X-rays), shapes, number concentration,

Published online 25 February 2009

size distribution, and projected area concentration (PAVm) of particle populations. Mie theory calculations based on SAX results were used to estimate the scattering coefficient

Keywords:

and the mean scattering efficiency at a wavelength of 660 nm [bm(660) and ].

Drinking water turbidity

Throughout the study system, nonspherical clay mineral particles in the 1–10 mm size

Runoff

range dominated PAVm, light scattering and its surrogate, nephelometric turbidity (Tn).

Light scattering

Patterns of particle size contributions to bm(660) (and Tn) remained relatively invariant over

Suspended minerogenic particles

a wide range of Tn (more than 200-fold difference). The median size for these contributions was most often w2.5 mm. The credibility of the SAX characterizations of the light-scattering features of the minerogenic particles and the calculations based on Mie theory for the study system was supported by (1) the strength of the Tn–PAVm relationship, (2) the reasonable closure between Tn measurements and calculated values of bm(660), and (3) the closeness of to the limiting value of 2 for polydispersed particle populations. Upstream sources of turbidity-causing particles within the study system were demonstrated to have highly similar light-scattering features. This indicates similar potencies for the particle populations from these sources for turbidity impacts in downstream waters and supports the direct incorporation of Tn measurements into loading calculations to evaluate relative contributions of these inputs with respect to such impacts. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Optical attributes are important features of water quality and the functioning of aquatic ecosystems. Cloudy or turbid water associated with elevated levels of light scattering is a general concern with respect to aesthetic impairments, and more

specifically for water supply lakes and reservoirs, because of regulatory requirements to deliver water that meets turbidity standards. The intensity of the light-scattering process is quantified by the magnitude of the scattering coefficient, b (in m1), one of the four inherent optical properties of water (Kirk, 1994). Turbidity (Tn) is measured as the light scattered from

* Corresponding author. Tel.: þ1 315 431 4962; fax: þ1 315 431 4969. E-mail address: [email protected] (S.W. Effler). 1 Now retired. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.02.018

water research 43 (2009) 2280–2292

a beam within a rather wide angle centered on 90 (Kirk, 1994) with a nephelometer (nephelometric turbidity units, NTU). Despite limitations of this metric (e.g., Davies-Colley et al., 2003; Effler et al., 2006), Tn remains the measure of choice by the regulatory community to control related features of water quality in water supplies in the United States. Measurements of Tn are known to be well correlated and linearly related to b (Davies-Colley et al., 2003; Kirk, 1994) and other surrogates of b (Effler et al., 2006). Inorganic, or minerogenic, particles make substantial, and in some cases dominant, contributions to light scattering in many freshwaters (Davies-Colley et al., 2003; Kirk, 1985; Peng and Effler, 2005, 2007). For example, increases in minerogenic particle concentrations have been an important cause of the degradation of visual clarity in Lake Tahoe (Swift et al., 2006). The magnitude of b (and therefore Tn) depends on four features of a particle population: number concentration (N ), particle size distribution (PSD), composition, and particle shape (Kirk, 1994; Stramski et al., 2004). Detailed characterizations of these features of a particle population can be used to directly calculate b at a specified wavelength of light (l) according to,

bðlÞ ¼

N X

Qb;i ðmi ; l; di ÞPAi

(1)

i¼1

where Qb,i is the scattering efficiency of particle i with projected area PAi. The value of Qb,i depends on the particle’s relative (to water) refractive index (mi ¼ ni  in0i , where ni and n0i are the real and imaginary parts of the complex index of refraction of particle i, respectively), its size (di), and wavelength (l). Values of n and n0 depend on particle composition (Woz´niak and Stramski, 2004), and for minerogenic particles, those of n0 (associated with light absorption) are negligible compared to n. Successful application of the above expression, supported by specification of the light scattering attributes of the individual particles of a population, has been described as solving the forward problem in optics studies (Peng and Effler, 2007; Sullivan et al., 2005). The primary basis of evaluation of success in the implementation of the forward method is the extent of closure of the estimates of b(l) with bulk measures of this attribute (Green et al., 2003; Peng and Effler, 2007). The forward method represents a potentially powerful vehicle to partition b (and Tn) into contributions of various particle size and chemical classes. This is fundamental information related to the origins and behavior of these particles and for understanding their effects in the underwater light field. Individual particle analysis (IPA) techniques that provide both compositional and morphological information on suspended particles are necessary for the implementation of the forward method (Eq. (1)). Recently, characterization of minerogenic particles with an IPA technique, by scanning electron microscopy (SEM) interfaced with automated image and X-ray analyses (SAX), and integration of the results into Mie theory calculations of b have demonstrated reasonable levels of closure and consistency with bulk measures of scattering for a limited number of fresh water systems (Peng and Effler, 2007; Peng et al., 2007). Water quality managers widely seek to partition contributions of constituents or attributes of concern into distinct

2281

sources in complex ecosystems with multiple inputs. Acquisition of the appropriate information for selected mass constituents (e.g., nutrients, suspended solids) is straightforward, based on the development of mass loadings from paired monitoring of input flows and concentrations of the constituents of interest. However, adoption of such an approach to partition contributions of different sources to downstream Tn is potentially more problematic because this metric is optics, rather than mass, based. Use of total suspended solids (TSS) as a surrogate is superficially attractive. However, such an approach is fundamentally flawed in many systems because of differences in the particle size dependency of TSS and b (i.e., Tn) (Effler et al., 2008a). This is commonly manifested in weak correlations between TSS and Tn, systematic differences in the relationship among systems, and variations in time and space within individual systems (Davies-Colley et al., 2003). Incorporation of Tn instead in loading calculations represents a potentially viable alternative. However, given the dependencies of b on multiple attributes of the particle population (Eq. (1)), there is potential for substantial differences in scattering (turbidity) characteristics within individual sources and among multiple sources. For example, certain combinations of n (chemical composition), size (PSD), and shape could result in the same concentration of particles from different sources yielding different values of b (i.e., Tn). This introduces the possibility of potential differences in ‘‘potency’’ amongst multiple turbidity sources. Two of the potential ‘‘potency’’ factors, chemical composition and PSD, are embedded within values of the average scattering efficiency factor, , the ratio of b to the total geometric cross-section concentration of particles. Moreover, greater contributions to b (i.e., Tn) from smaller particles in an upstream source of turbidity compared to another could result in greater persistence of turbidity downstream from that source. Here we use SAX to describe the light-scattering features of minerogenic particles in interconnected reservoir basins and an intervening stream within the watershed of New York City’s water supply system, where these particles dominate b (and Tn). Implementation of the forward method is advanced through integration of SAX results into Mie theory calculations of Qb,i and b (Eq. (1)). These calculated features of light scattering form the basis to resolve contributions of different sizes and chemical classes to scattering and are used to address the relative potency issue of different sources of turbidity in the study system. The credibility of these implementations of the forward method, which combine SAX results and Mie theory calculations, is evaluated through comparisons to bulk measurements of Tn and consistency of with literature values. The findings are considered in the context of the implications of partitioning turbidity inputs in this and other multiple source systems with simple loading calculations that directly use Tn data.

2.

Study system

New York City’s (NYC) drinking water supply consists of three upland supply areas, the Croton, the Catskill, and the Delaware watersheds. The total watershed area, which includes 19 reservoirs, is approximately 5055 km2. This water supply does

2282

water research 43 (2009) 2280–2292

not receive filtration treatment, though a filtration plant is under construction for the Croton System, which presently supplies w10% of the City’s water. The remainder of the water supply operates under a Filtration Avoidance Determination that requires aggressive protection of the source water, including maintenance of low Tn levels. Kensico Reservoir, located (long. 73 460 W; lat. 41 040 2600 N) w50 km north of Manhattan (Fig. 1), is the terminal reservoir for the Catskill and Delaware systems. Water from these watersheds reaches Kensico via aqueducts with the same names (Fig. 1), supplying approximately 4.9  106 m3 per day (only 2% of the water enters from the small local watershed). Kensico Reservoir has a surface area of 9.05 km2, a mean depth of 13.1 m, a volume of 119  106 m3, and an average completely mixed flushing rate of 15 times per year. Water leaving this reservoir is delivered directly to the City’s distribution system and thus must meet the federal Surface Water Treatment Rule standard for Tn (not greater than 5 NTU on entry to the distribution system). The serial configuration of the upstream reservoirs (see Fig. 1) effectively reduces the turbidity-causing particle concentrations through sedimentation (Effler et al., 2001), with yet further deposition opportunities offered within Kensico Reservoir. Water quality within Kensico Reservoir is good with respect to the Tn limit the vast majority of the time. However, very high Tn can enter from the Catskill Aqueduct (w147 km long) from the elevated levels that develop in Ashokan Reservoir (Fig. 1) after particularly large runoff events in the Catskill system. To avoid violations of the Tn standard in the Kensico Reservoir withdrawal following these rare runoff events, alum is added to the Catskill Aqueduct (Fig. 1), 3.9 km upstream of Kensico Reservoir, to promote coagulation and sedimentation of the turbidity-causing particles in a small area of the reservoir in the immediate vicinity of the aqueduct. Alum treatment was required on only three occasions over the 1985–2004 interval. However, three alum treatments were needed during 2005, causing increased concerns with respect to the origins and character of the turbidity from the Catskill System. The Catskill System includes Schoharie Reservoir (42 230 3000 N; 74 270 W) and Ashokan Reservoir (41 560 1800 N; 74 130 0900 W) (Fig. 1). Schoharie Reservoir has a surface area of 4.6 km2, a volume of 79  106 m3, and mean and maximum depths of 17 and 41 m, respectively. This impoundment would flush w10 times per year if completely mixed. The reservoir’s primary tributary is Schoharie Creek, which drains 75% of its watershed. Schoharie Reservoir is an upstream reservoir in the NYC system and water withdrawn from it travels through a 29-km tunnel, and then 20 km through a connecting Esopus Creek, before reaching the west basin of Ashokan Reservoir (Fig. 1). Esopus Creek is a shallow stream that acts as a conduit for particle transport; deposition and resuspension are not noteworthy factors with respect to the overall turbidity conveyed. Ashokan Reservoir has two basins, the west and the east, which are separated by a dividing weir (Fig. 1). The west basin has a full capacity and surface area of 179  106 m3 and 12.8 km2, respectively, while the larger east basin exceeds these dimensions by factors of 1.7 and 1.6, respectively. The west and east basins have average completely-mixed flushing rates of 5.4 and 3.1 times per year, respectively. The Esopus Creek

a

b

Fig. 1 – Map of study system (a), and sampling sites (b): upstream of the Schoharie Reservoir discharge in Esopus Creek (site 1), the water column of Schoharie Reservoir proximate to the withdrawal (site 2) and within the withdrawal facility of the reservoir (site 3), Esopus Creek downstream of the Schoharie Reservoir discharge and before its entry into the west basin of Ashokan Reservoir (Coldbrook, site 4), two sites located in the eastern part of the west basin of Ashokan (sites 5 and 6), east basin of Ashokan proximate to the intake (site 7), and a site in the Catskill aqueduct arm of Kensico (site 8).

water research 43 (2009) 2280–2292

inflow, which carries a combination of the stream’s natural flow plus the Schoharie Reservoir withdrawal, is the dominant input to Ashokan Reservoir. Water withdrawn from Ashokan is the source water for the Catskill Aqueduct input to Kensico (Fig. 1). Usually this is taken from the east basin. The west basin functions as a settling basin for turbidity-causing particles, particularly after runoff events, to protect the east basin and Catskill Aqueduct from high Tn levels (Effler et al., 1998). Both Schoharie Reservoir (Effler et al., 2006) and the west basin of Ashokan Reservoir (Effler et al., 1998) routinely suffer from conspicuous increases in Tn following runoff events. These impacts spread into the east basin of Ashokan only for the largest runoff events. The causes of the increases in Tn are largely associated with external loadings received during the events from the respective primary tributaries (Gelda and Effler, 2007). Drawdown, occurring commonly in both reservoirs (particularly in Schoharie), can augment turbidity loading through promotion of sediment resuspension (Effler and Matthews, 2004). Stream beds and banks, composed of lacustrine sediments, are important sources of turbiditycausing particles in both Schoharie Creek (Smith, 2002) and Esopus Creek. The flow input from the Schoharie Reservoir withdrawal can be the dominant hydrologic component in Esopus Creek during summer low flow intervals, but this input is modest compared to the watershed runoff during storm events. Managers are particularly concerned with relative concentrations and character of turbidity-causing particles in the inputs from Schoharie Reservoir versus Esopus Creek itself, both of which are delivered in this stream’s inflow into Ashokan Reservoir, in order to identify the appropriate focus for rehabilitation to limit the export of turbidity to Kensico Reservoir. Wide variations in the relationships between Tn and TSS have been observed in both the Schoharie Reservoir withdrawal (Fig. 2a) and Esopus Creek (Fig. 2b), based on long-term monitoring [New York City Department of Environmental Protection (NYCDEP), unpublished data] over a wide range of runoff conditions. Application of such relationships, along with corresponding source flow rates and TSS observations, to assess respective contributions of turbidity would necessarily be accompanied by substantial uncertainty. Moreover, the issue of potential differences in the potency of turbidity from the different sources could not be addressed with such data sets.

3.

Methods

3.1.

IPA/SAX protocols

Water samples were pressure filtered through 0.4-mm pore sized polycarbonate membrane filters (GE Osmonics); turbidity readings were used to guide the selections of sampling volumes to avoid particle overlapping (particle mass loading was less than 5 mg cm2). Atteia et al. (1998) determined PSDs by SEM of natural aquatic particles separated by cascade filtrations and demonstrated that particles smaller than the respective pore sizes of the membrane filters were retained; for example, the peak frequency sizes on the 2- and 0.8-mm pore sized filter were approximately w0.8 and 0.2 mm,

2283

a

b

Fig. 2 – Tn–TSS relationships from long-term NYCDEP monitoring at (a) Shandaken Tunnel discharge (1989–2006), and (b) Coldbrook (site 4) of Esopus Creek (1995–2005).

respectively. Particles deposited on filters were vacuum coated with carbon before being subject to SAX. IPA was performed by an Aspex PSEM 2000 system (Aspex Corp., Delmont, PA) controlled by an AFA (Automated Feature Analysis) software. The instrument is calibrated per annum with a magnification reference standard (Geller ˚ nalytical Laboratory, MRS-3) in the 25–30,000 magniMicroA fication range. The electron beam current was approximately 0.8 nA with an accelerating voltage of 20 keV. Magnification of 2000 was used, resulting in a field dimension of 80 mm  80 mm. AFA directs the electron beam to raster the field area with a 0.3-mm step size (user defined); a particle is detected when a pixel’s brightness level (based on backscattered electrons) exceeds the set threshold. The digital control then switches the beam from search to measuring mode with the highest available pixel dimension (2048  2048; i.e., 0.04 mm/pixel). AFA adopts a ‘‘rotating chord algorithm’’; it draws a horizontal chord from the point of feature detection to the last, above-threshold pixel and then forms a vertical chord through the center pixel of the first chord. This process of successively drawing and bisecting chords is iterated to locate the approximate centroid of a feature. Morphological measurements are based on the chords drawn through the centroid and series of triangles formed by the centroid and the chords. The shape, or ‘‘nonsphericity’’, of a particle is represented by an aspect ratio (ASP), defined as the ratio of the length of the longest chord to that of its orthogonal chord (ASP of a spherical particle is 1). The representativeness of the ASP value is expected to diminish for the smaller particles as the numbers of pixels enclosed in a feature outline decrease. PA is the sum of the areas of the triangles defined by the centroid

2284

water research 43 (2009) 2280–2292

and series of chords. The value of d is calculated as area equivalent diameter. X-rays of sodium and higher atomic number elements (total of 16 elements) were acquired for 3 s live time by an energy dispersive Sirius Si(Li) detector (e2v Scientific Instruments). The X-ray spectrometer is calibrated through an instrument routine by collecting X-rays on copper. Though Xrays of light elements (carbon, oxygen) from filter substrate or carbon coating are also emitted, these were not counted and stored. Gross X-ray counts were typically 1000–2000 for inorganic particles. Casuccio et al. (2004) have demonstrated that inorganic particles of w0.3–0.4 mm are readily distinguishable from polycarbonate substrate even when background carbon counts are dominant. The signal-to-noise ratios of major elements (Al and Si) were generally greater than w20. Particles were compositionally classified into one of seven generic types based on normalized elemental X-ray counts (Peng and Effler, 2007): clay minerals, quartz, silica-rich, iron/manganese, miscellaneous (other minerogenic), diatom, and organics. The organics class, characterized by low X-ray counts (<750 for 3 s acquisition), under-represents the contributions of organic particles (Peng et al., 2007). At least 1500 particles were characterized by SAX for each sample. The total PA per unit volume of water (PAV, in m1) is the summation of the individual PAi values (extrapolated according to the analyzed fraction of the particle deposition area on the filter) divided by the volume of sample filtered. PAV, a central feature in the bulk scattering effects of a particle population (Peng and Effler, 2007), can be partitioned into various particle type contributions. The minerogenic PAV (PAVm) excludes contributions by the organics and diatom classes.

3.2.

PSDs and calculation

Example PSDs are presented in the generally accepted format, in terms of the size distribution function, F(d ), defined by FðdÞ ¼

dN Dd

(2)

where dN is the particle number concentration in the size range (Dd ) of d  1=2Dd. Thirty-three logarithmically spaced size bins were defined for the 0.2–40.5 mm range (Peng and Effler, 2007), resulting in F(d ) observations in equally spaced size intervals when plotted on the log-log scales. The lower boundary of the size distributions was extended to 0.2 mm. Analyses of samples collected on both 0.2- and 0.4-mm pore sized filters indicated that particles in that size interval are only slightly under-represented by the 0.4 mm; this underestimation, however, does not have a noteworthy effect on the estimates of b (subsequently). Several functional representations of PSDs (Risovic´ and Martinis, 1995) have been used for marine particles and three were previously evaluated for Schoharie Reservoir (Peng and Effler, 2007). Two of these are considered here, the Junge (or hyperbolic) model [F(d ) ¼ Cdj, where C and j are constants] and the ‘‘B’’-component of the two-component model, 2C-B [FðdÞ ¼ CB dmB expðbB dgB Þ, four constants]. Estimates of bulk bm values at l ¼ 660 nm were computed according to the forward method (Eq. (1)) from SAX results for minerogenic particles; b varies only modestly over the visible

wavelengths (400–700 nm), and light attenuation at 660 nm is mainly due to particulate scattering. Values of Qb,i were calculated based on Mie theory for homogeneous spheres (Bohren and Huffman, 1983). Values of ni were specified according to the SAX results based on the listing for various minerals of Woz´niak and Stramski (2004), and a uniform 0.001 was assumed for n0i . The mean scattering efficiency of the minerogenic particle populations at 660 nm () was calculated as the ratio of bm(660) to PAVm, and is presented as a metric of the turbidity potency of the various particle populations. As only results of bm calculated for 660 nm are presented, the wavelength reference is suppressed hereafter unless otherwise noted. The contributions of various particle sizes to overall bm were calculated and represented in both cumulative and percent distribution formats.

3.3.

Sampling strategy

Thirty samples were collected from eight sites in 2005, extending from Schoharie Reservoir through the Catskill arm of Kensico Reservoir (Fig. 1), for paired SAX and Tn (Hach 2100 AN) (Clesceri et al., 1998) analyses. Seven of the sites are within the Catskill system and one in Kensico Reservoir. The sites included the water column of Schoharie Reservoir proximate to the withdrawal (site 2) and within the withdrawal facility of the reservoir (site 3), Esopus Creek upstream of the Schoharie Reservoir input from the tunnel (site 1) and downstream of the tunnel discharge (site 4), two sites located in the eastern part of the west basin of Ashokan (sites 5 and 6), a site in the east basin of Ashokan proximate to the intake (site 7) for the Catskill aqueduct, and a site in the Catskill aqueduct arm of Kensico (site 8) (Fig. 1). The sites were chosen to support partitioning and contrast of Schoharie Reservoir and Esopus inputs, and depict spatial patterns and potential transformations within Ashokan Reservoir. The sampling reflected a wide range of conditions with respect to stream flow (runoff events), operation of the Schoharie Reservoir withdrawal (i.e., discharge to Esopus Creek), and extent of drawdown (a steady increase of 3 m over the June–July period). For example, the tunnel discharge from Schoharie Reservoir was not operated for most of April in 2005 because of the high Tn levels in the reservoir, thereby eliminating the effects of this input in downstream portions of the system during that interval. The runoff event of early April 2005 was a conspicuous feature of the study year (peak flow w500 m3 s1 on 3 April, 10–20 m3 s1 for base flow conditions). This was a rare runoff event, estimated to have a return interval for the observed peak flow of w25 years. The persistent high Tn levels in the withdrawal of the east basin of Ashokan Reservoir following this event required alum treatment for 77 days to avoid high levels in Kensico Reservoir.

4.

Results

4.1. Minerogenic particle composition and morphometries Chemical composition results for minerogenic particles are presented in the form of percent (%) contributions of the particle

2285

water research 43 (2009) 2280–2292

lower quartz contributions than in downstream portions of the study system. However, these features of composition were more uniform throughout the study system in early April when the highest PAVm levels prevailed (Table 1). The results depict relatively uniform chemical composition of minerogenic particle populations throughout the system, and establish the terrigenous origins of this material. The minerogenic particle populations generally deviated from sphericity in all the samples analyzed, with mean ASP values that ranged from 1.78 to 2.88; 25 of the samples had values within the much narrower range of 1.78–2.17 (Table 1). However, the shapes were heterogeneous for all the samples,

types to overall PAV and minerogenic PAV (PAVm) (Table 1). The chemical classification scheme performed well in representing the minerogenic particles for the study sites, as more than 95% of the PAVm was associated with the specified particle types (i.e., not in miscellaneous group). Minerogenic particles dominated the measured PAV in all samples over the very wide range of PAV (i.e., Tn) encountered, representing 91–99% of the total (Table 1). The clay mineral class was the dominant component, representing, on average, 78% of PAV and 81% of PAVm. Quartz was the second most important type in general. The Schoharie Reservoir summertime particle populations tended to be somewhat more enriched in clay minerals and had somewhat

Table 1 – Sample information and analysis results. System

Schoharie Reservoir

Date Tn PAV Site IDa/ (2005) (NTU) (m1) Depth (m) Clay Quartz

Sirich

Fe/ Mn

ASP (std. dev.)d

d50 (mm)

Misc Organics Diatom

7 Apr 7 Apr 21 Jun 6 Jul 19 Jul 2 Aug 17 Aug 30 Aug 14 Sep

440 434 7.2 4.6 2.2 4.3 3.2 6.9 27.8

85.7 84.4 1.06 0.71 0.37 1.49 0.85 1.50 6.50

69.8 74.7 87.9 80.4 90.2 92.1 93.6 87.9 84.2

16.7 14.8 5.0 10.9 1.7 3.5 1.8 4.8 6.9

5.0 3.8 2.3 3.1 3.1 1.0 2.0 1.6 1.7

1.0 1.7 0.5 0.9 0.6 0.3 0.6 0.9 1.4

2.4 2.6 1.3 2.4 2.3 0.9 0.9 1.3 1.8

2.6 1.0 2.4 1.2 1.3 0.4 0.5 0.9 1.2

2.5 1.6 0.7 1.2 0.8 1.9 0.7 2.6 3.0

1.89 (0.84) 1.88 (0.94) 2.27 (2.98) 2.01 (2.27) 2.58 (2.15) 2.43 (2.45) 2.40 (3.59) 2.17 (2.64) 2.12 (2.31)

2.43 2.54 5.21 3.95 6.06 6.51 5.76 5.04 5.13

2.38 2.35 1.84 2.22 2.12 2.09 2.12 2.16 2.15

11 Janb,e 11 Janb,e 22 Febc,f 22 Febc,f 13 Aprb 19 Julc 11 Janb 22 Febc 13 Aprb 19 Julc

1.5

0.32

71.0

11.1

3.8

1.0

10.4

2.7

0.1

2.00 (1.23)

2.33

2.37

1.6

0.30

70.4

12.8

4.8

1.6

3.6

1.8

4.9

1.92 (0.97)

2.55

2.32

2.3

0.47

76.0

10.2

6.7

0.9

1.8

3.5

1.0

1.93 (1.20)

2.53

2.31

1.7

0.41

76.6

11.9

3.9

1.0

3.9

2.7

0.0

1.97 (0.89)

2.35

2.36

76.1 18.8 2.6 5.6 66.7 23

18.8 3.54 0.41 0.74 19.2 4.17

76.6 71.3 73.0 75.8 70.5 82.9

15.2 7.0 15.7 12.5 15.3 6.5

1.9 7.2 4.4 3.5 4.2 6.3

0.7 1.3 1.2 1.9 0.7 0.5

2.4 4.2 3.1 2.7 3.7 1.6

2.4 1.8 0.7 2.8 2.7 1.7

0.8 7.4 2.0 0.9 2.9 0.5

1.82 (0.93) 1.78 (0.89) 1.90 (0.91) 1.86 (1.46) 1.88 (1.19) 1.83 (1.3)

2.74 2.88 2.46 2.24 2.66 2.86

2.36 2.33 2.37 2.34 2.33 2.30

Ashokan W. basin

5/2 5 Aprb 5/0 26 Aprb 5/0 7 Julc 5/0 17 Augc 6/3 4 Aprb 6/6 4 Aprb 6/32 4 Aprb

231 154 3.7 1.7 94.2 242 464

64.1 22.4 0.40 0.30 38.8 72.2 138.0

75.9 70.3 75.9 75.2 77.1 73.8 78.5

14.5 13.2 9.9 9.8 14.4 18.0 12.7

3.5 6.2 7.5 7.1 4.2 4.5 3.0

0.8 1.1 0.6 1.0 1.0 0.8 1.8

2.5 3.3 2.9 1.1 1.8 1.6 2.3

1.0 5.8 1.5 3.9 0.4 0.1 1.0

1.8 0.1 1.7 1.9 1.3 1.2 0.8

1.90 (0.98) 1.84 (0.80) 1.99 (1.50) 2.49 (4.16) 1.97 (1.28) 1.79 (1.0) 2.08 (1.19)

2.80 2.00 2.91 2.75 2.39 2.76 1.81

2.34 2.40 2.23 2.32 2.36 2.36 2.44

Ashokan E. basin

7/2 7/0

4 Apr 7 Jul

31.8 2.57

6.72 0.45

67.1 84.1

20.7 4.8

4.5 5.0

1.5 0.4

3.5 1.9

0.8 3.8

2.0 0.1

1.83 (0.72) 2.14 (1.75)

2.61 3.09

2.37 2.23

Kensico

8/10 8/10

6 Apr 17 May

21.5 1.57

6.10 0.34

74.1 78.3

17.1 5.4

3.0 6.2

0.8 1.8

2.0 2.0

1.4 3.2

1.7 3.1

1.82 (1.02) 2.88 (3.06)

2.84 2.79

2.35 2.29

Esopus Creek

3/– 2/15 2/10 2/10 2/10 2/10 2/10 2/10 2/10

PAV type composition (%)

1/– 1/– 1/– 1/– 1/– 1/– 4/– 4/– 4/– 4/–

a b c d e f

Site identifications are marked in Fig. 1. Shandaken Tunnel was turned off. Shandaken Tunnel was turned on. Values (mean and standard deviation in parentheses) are for minerogenic particles. These are sampling duplicates. These are sampling duplicates.

2286

water research 43 (2009) 2280–2292

as indicated by the relatively large standard deviation values for ASP (Table 1). No systematic dependence of this representation of shape on size (d ) was observed. The similarities of ASP values for the various samples and sites suggest that there were no major differences in particle shapes among the study sites. Microscopic observations also indicated particle shapes were diverse in this system and ‘‘plate-like’’ configurations were common for clay minerals (Peng and Effler, 2007).

4.2.

PSDs of minerogenic particles

PSDs of minerogenic particles are presented for two samples (Fig. 3) that are generally characteristic of the patterns observed for the samples from this interconnected study system. One of the samples was from site 2 in Schoharie Reservoir (see Fig. 1), soon after (7 April) the major runoff event; the other was from site 4 in Esopus Creek nearly a week later (13 April, Shandaken Tunnel was not operating). Values of Tn were high for both samples, 434 NTU in the reservoir, and 67 NTU in the stream. Frequency peaks were observed at sizes of w0.4 mm for these samples, rather than a monotonic increase in N as d decreased. Values of N were higher throughout the reported range for the reservoir sample, but particularly over the submicron to 3 mm range (Fig. 3). The Junge model, even when constrained to sizes 1.25 mm, consistent with results from earlier (e.g., Bader, 1970) particle sizing instrumentation, provided a poor representation of the PSDs for this study system (Fig. 3). In contrast, the 2C-B model described these PSDs well except for sizes 0.4 mm (Fig. 3).

4.3.

were observed in both Schoharie Reservoir and the west basin of Ashokan Reservoir after the major runoff event of early April. The maxima for sites 1 and 4 in Esopus Creek included here were 76 and 67 NTU, respectively; however, Tn levels peaked at substantially higher values (NYCDEP, unpublished data) than those captured by sampling for SAX. The impact was diminished in the east basin of Ashokan Reservoir and Kensico Reservoir for the April runoff event. PAVm was a strong linear predictor of Tn (Fig. 4a) and the calculated minerogenic scattering coefficient, bm (Fig. 4b). Variations in PAVm explained 94% of the very wide differences in Tn observations ( p < 0.001), and 98% of the variations in this surrogate of scattering for the lower Tn levels of <90 NTU (Fig. 4a). The near linearity in the bm versus PAVm relationship

a

b

Scattering and PAV

Wide variations in scattering, as represented by Tn values, were observed at all the sites. Levels of Tn <5 NTU occurred during low runoff intervals (Table 1), and values>400 NTU

c

Fig. 3 – Examples of minerogenic PSDs and function fits. Sample from site 2 was collected on 7 April (Tn 434 NTU), and that from site 4 on 13 April (Tn 66.7 NTU). Fitted functions are Junge to particles ‡1.25 mm (C [ 1.29 3 1010 LL1 mmL1, j [ 3.3) and 2C-B to particles ‡0.9 mm (CB [ 5.11 3 1017 LL1 mmL1, mB [ 3.88, bB [ 17.21, gB [ 0.286).

Fig. 4 – Evaluations of relationships (with linear regression results) between turbidity and SAX-based particle determinants: (a) Tn versus PAVm, (b) bm(660) versus PAVm, and (c) Tn versus bm(660).

water research 43 (2009) 2280–2292

(Fig. 4b) reflects the uniformity of (mean of 2.28, coefficient of variation of 5.4%). The estimates of bm, based on SAX results, were equally strong predictors of the variations in Tn (Fig. 4c). This is consistent with the performance of PAVm (Fig. 4a) as a predictor and the uniformity of .

4.4.

Size contributions of minerogenic particles to bm

The contributions of various sizes of minerogenic particles to calculated bm are depicted for samples collected following the early April runoff event in both cumulative (Fig. 5a) and noncumulative (Fig. 5b) formats, in the context of percent of the total. The upstream (sites 1–4) and downstream (sites 5–8) conditions are separately illustrated for clarity in different formats. Similar size dependencies of bm were demonstrated for these samples across the bounds of the study system. According to these calculated patterns, more than 95% of bm, and thereby its surrogates (e.g., Tn), were associated with particles in the size range 1–10 mm. A peak contribution was made by particles slightly larger than 2 mm, and a secondary

a

maximum by particles somewhat larger than 4 mm (Fig. 5b). This secondary peak was primarily a result of a secondary maximum in the dependence of Qb,i on d at w4–5 mm. A valuable simplifying statistic to represent the size dependency characteristics of bm (and Tn) is d50, the fiftieth percentile (i.e., median) size with respect to light scattering (Fig. 5a, inset). The similarity in these patterns prevailed (i.e., d50 range 2.4–2.9 mm) despite nearly 20-fold differences in Tn observed for these sites for the early April samples (Fig. 5c). No withdrawal from Schoharie Reservoir was being made (i.e., no inflow to Esopus Creek from the Shandaken Tunnel) at the time of sample collection at Esopus Creek, thereby eliminating the effects of this source on site 4 observations. However, shifts to higher d50 values (Table 1) were subsequently observed in Schoharie Reservoir in the summer, when the Shandaken Tunnel was on and the reservoir surface was being drawn down (3 m over the June–July interval). The d50 levels were relatively uniform temporally in the stream at both sites 1 and 4. Moreover, the higher Schoharie Reservoir d50 values of summer were not strongly manifested downstream at site 4 on 19 July (Table 1). Summertime shifts of d50, similar to those observed at Schoharie Reservoir, were not found in the west basin of Ashokan Reservoir, which remained relatively full (by comparison) during that interval. However, a modest shift to increased contributions by smaller particles was observed in this basin by late April, nearly four weeks following the major runoff events; the d50 value decreased to 2.0 mm (Table 1).

5.

b

c

Fig. 5 – Size distribution pattern of bm(660) presented as percentage contributions by size classes for samples collected during early April runoff event: (a) cumulative, sites 1–4, with the concept of median particle size of scattering (d50) illustrated, (b) noncumulative, sites 5–8, and (c) d50 (bar) and Tn (open symbol) values.

2287

Discussion

5.1. SAX characterizations and forward method performance The representativeness of the SAX characterizations, in supporting the implementation of the forward method and evaluation of turbidity in these interconnected reservoirs and Esopus Creek, is demonstrated through at least three performance features. First, PAVm has been shown to be a strong linear predictor of Tn (Fig. 4a). This substantiates the position (Treweek and Morgan, 1980), rarely demonstrated with individual particle data from natural systems until recently (Peng and Effler, 2007), that scattering (in this case, mostly by clay minerals) is proportional to the summed particle projected area (Eq. (1)). Moreover, this supports the position that minerogenic particles dominate light scattering and Tn levels throughout this interconnected study system. Second, the reported strong relationship between Tn and calculated values of bm (Fig. 4c) provides a reasonable level of closure for this implementation of the forward method that relied on the combination of SAX results and Mie theory calculations. The slope of this relationship (Fig. 4c) is within the range of values reported elsewhere, based on other estimative or measurement protocols for b(l) (Effler, 1988; Effler et al., 1998; Vant and Davies-Colley, 1984). Certain refinements in the estimates of overall b(l) could be made by adding a minor component for organic particles (e.g., forward method calculations for diatoms and organic particles). Moreover, the minor adjustments of bm(l) estimates for the artifact of nonspherical particles

2288

water research 43 (2009) 2280–2292

lying flat on filters in SAX characterizations (Peng and Effler, 2007; Peng et al., 2007) could be included. However, these refinements are not warranted here, as the primary limitation for a more rigorous closure analysis is the acknowledged imperfect performance of Tn as a measure of b(l) (DaviesColley et al., 2003; Kirk, 1994). Finally, values of were consistent with expectations for natural heterogeneous particle populations, being only slightly more than 2, the limiting value approached by polydispersed particle populations (Bricaud et al., 1983; Jonasz, 1987). The SAX characterizations of shapes (Table 1) indicate that the Mie theory stipulation of particle sphericity was not met for the minerogenic particles of the study. Similar ASP values were reported for Schoharie Reservoir, based on sampling conducted in 2002 (Peng and Effler, 2007), as well as for more chemically diverse minerogenic particle populations in several lakes in New York (Peng et al., 2007). Nonsphericity of most particles in marine waters is generally recognized (Gordon and Du, 2001; Jonasz, 1987). Though characterizations of such attributes for freshwaters have been limited (Peng and Effler, 2007; Peng et al., 2007), a similar situation likely prevails, and is further supported by the observations of this study (Table 1). Optical modeling analyses for marine systems have most often (e.g., Babin et al., 2003; Stramski and Woz´niak, 2005; Sullivan et al., 2005) simulated particle scattering through Mie theory calculations for homogeneous spheres as a reasonable approximation. Deviations from sphericity cause shifts in the volume scattering function, particularly in backscattering directions (Mishchenko and Travis, 1994). However, the effects on overall scattering (i.e., integrated through all angles) are substantially less (Volten et al., 1998). More complex, and computationally demanding, frameworks that accommodate some of the effects of particle shape (Mishchenko and Travis, 1994) have their own sources of uncertainty (Stramski and Woz´niak, 2005), including deviations of ecosystem particles from idealized shapes (Gordon and Du, 2001). Mie theory calculations for homogeneous spheres presently represent a viable, if not the only reasonable, framework to make forward method estimates of b(l) based on individual particle information for thousands of particles that differ widely in size and shape. The performance of Mie theory-based estimates of bm (Fig. 4c) supports the use of this computational framework here and the representativeness of the SAX observations, and indicates the extent of deviations from sphericity that prevailed for these minerogenic particles was not an important influence on b. Accordingly, differences in shape(s) (i.e., ASP) (Table 1) among the sites and samples of the study system were apparently not an important influence on variations in Tn. It follows that the increasing uncertainty in ASP values for smaller particles is not a noteworthy issue with respect to the representativeness of b estimates presented here. The level of closure achieved here augments the evolving evidence, albeit for a limited number of systems (Peng and Effler, 2007; Peng et al., 2007), that SAX characterizations combined with Mie theory calculations can be used to obtain representative estimates of both Qb,i(l) and bm(l). While demonstrations to date have mostly been for systems where clay minerals are dominant, similar success has been reported for cases where calcite (usually formed autochthonously) was a prominent component (Peng et al., 2007).

5.2.

Implications of and d50 values

The mean scattering efficiency of the minerogenic particles of the individual samples, , is a quantification of their light scattering and turbidity potencies. The uniformity of the potencies of turbidity throughout this system of interconnected reservoirs and the intervening stream is reflected by the similarity of the values (Table 1). This is consistent with the generally uniform composition of the particle populations (Table 1), as well as the similar n values of the constituents; for example, 1.167 for clay minerals (kaolinite) and 1.148 for quartz (Woz´niak and Stramski, 2004). It further reflects the similarity of PSDs, such as represented by d50 (Table 1). It is noteworthy that the reported shifts to higher d50 values in Schoharie Reservoir in summer and lower d50 values in Ashokan Reservoir in late April did not result in substantive changes in (Table 1). The shift in d50 in Schoharie Reservoir (Table 1) was presumably associated with inputs of larger particles from sediment resuspension that accompanied drawdown of the reservoir surface (Effler and Matthews, 2004; Peng and Effler, 2007). This may result in lower persistence of this input relative to Esopus Creek inputs in downstream systems during drawdown intervals due to higher settling velocities. This effect is not important in a management context, as it is manifested in dry weather and low Tn conditions. Moreover, the extensive drawdown observed in 2005 is not perennial (Gelda and Effler, 2007). The shift to lower d50 values in the west basin of Ashokan Reservoir (Table 1) may reflect the operation of particle sorting processes, such as the preferential loss of larger particles through deposition.

5.3. Minerogenic particles and advancing the forward method for light scattering Certain features of the minerogenic particle populations documented for this study system are consistent with those reported elsewhere. For example, in studying a small flood event (peak flow rate 67 m3 s1) in the Magela Creek of northern Australia, Hart et al. (1993) separated the collected suspended matter into suspended particulate matter (operationally defined by diameters >1 mm) and coarse colloidal matter (nominally 0.1–1 mm), and reported that the coarse fraction was mostly inorganic nature and the finer organic (determined by loss on ignition). Similar application of the IPA technique revealed that the mineral component of both fractions was dominated by Fe-rich aluminosilicates followed by quartz particles (estimated weight 72–82% and 10%, respectively). Though no quantitative results were reported for the morphology of the particulate material studied by Hart et al. (1993), micrographs clearly showed the irregular shapes of particles. The ASP results and microscopic observations in this study are qualitatively consistent with the ‘‘plate-like’’ structure generally attributed to clay mineral particles. There is little question that this attribute contributes to their persistence in water columns (Davies-Colley et al., 2003). SAX is well suited to support implementation of the forward method because it provides a robust representation of light-scattering features of minerogenic particle populations, including N, PSDs, particle composition, and particle shape.

water research 43 (2009) 2280–2292

Results for the first three of these features are well matched to support Mie theory calculations and representative forward method predictions of bm. Though such IPA techniques (notably electron beam probing X-ray microanalysis) have been widely used in aquatic particle analysis (e.g., Hart et al., 1993; Lartiges et al., 2001; Xhoffer et al., 1992), its integration into the field of optics study to model light-scattering effects of particles has emerged only recently (Peng and Effler, 2007; Peng et al., 2007). As evidenced from the success of the Mie theory calculations, the shape information was not critical in this case; however, this should not be interpreted as a necessarily ubiquitous condition for freshwater minerogenic particles. The shape information remains potentially valuable in characterizing other features of particle behavior such as settling, and could eventually support the future application of more complex predictive algorithms for light scattering that accommodate particle shape (Mishchenko and Travis, 1994). Feedback from Mie theory calculations in the form of sensitivity analyses (Peng and Effler, 2007; Peng et al., 2007) demonstrates the adequacy of SAX capabilities with respect to the size range represented (0.4–10 mm) and chemical characterization. This size range includes the particles responsible for light scattering, primarily >1 mm. A number of studies have focused on smaller particle sizes within the lower portion of this size range, and in some cases extended the lower limit (Gallegos and Menzel, 1987; Lartiges et al., 2001; Perret et al., 1994). These studies were concerned with the potential role of small particles in the transport of contaminants instead of optical implications. For example, the exclusion of the larger (e.g., >w4 mm) particles in these studies, often for technological reasons and different research objectives, limited the extent to which these approaches can be applied to the light scattering issue. The shapes of the PSDs reported here for minerogenic particles, similar to those reported for SAX results previously (e.g., Peng and Effler, 2007; Peng et al., 2007), deviated substantially from hyperbolic shapes. Ceronio and Haarhoff (2005) have documented similar inadequacy of hyperbolic function in modeling submicron particles in potable waters; however, the extent to which these observations are broadly representative is still uncertain. Earlier electronic particle counters that had higher size thresholds of detection (1–2 mm) also supported hyperbolic PSD patterns in marine waters (Bader, 1970). The Junge function has been invoked as representative in a number of marine optics studies (e.g., Babin et al., 2003; Stramski et al., 2004). However, deviations from the hyperbolic pattern have also been widely reported (Jonasz and Fournier, 1996; Risovic´ and Martinis, 1995). Patterns of PSDs in various aquatic systems deserve extensive research attention through application of multiple particle technologies. The particle composition attribute interacts with PSD in influencing scattering (Kirk, 1994). The lack of chemical partitioning, particularly differentiation of organic versus inorganic particles by counting and sizing devices, represents a substantial limitation related to specification of the refractive index. In several systems where both minerogenic and organic scattering components were estimated (through Mie calculations based on SAX results and empirical bio-optical model based on phytoplankton concentration, respectively), we demonstrated reasonably good closure between the sum of the two components and field measurement of particulate

2289

b(650) (Peng et al., 2007). Approaches to resolve both minerogenic and organic components of b(l) to advance the forward method include: (1) heavy metal staining techniques of organic particles in applications of SAX to estimate the organic component, and (2) paired employment of SAX (for minerogenic particles) and a particle counter that covers the particle size range important to light scattering. The second approach depends on a residual calculation (organic ¼ total  minerogenic) and thus the internal consistency of SAX and counter characterizations. The consistency requirement represents a substantial challenge, given the differences in underlying principles upon which the various particle characterization technologies rely; for example, particle counters based on electric resistance (e.g., Bader, 1970) or light scattering (e.g., Lartiges et al., 2001), microscopic image analyses based on two-dimensional projection (e.g., Atteia et al., 1998; this study), and sedimentation field-flow fractionation based on particle density (e.g., Gimbert et al., 2005). SAX can reliably resolve fundamentally different generic particle types such as quartz, clay minerals (even some specific types such as illite, kaolinite, or smectite), calcite, and iron-rich particles. This information is valuable concerning the sources and origins of particles contributing to light scattering. Various common inorganic constituents in freshwaters have similar values of n (Woz´niak and Stramski, 2004) and thus related uncertainties in chemical characterization (e.g., the specific clay minerals) have negligible effect on estimates of bm(l) (Peng et al., 2007). The partitioning of minerogenic scattering into the major inorganic components is well supported by their relative contributions to PAV (Table 1).

5.4.

Findings in the context of turbidity loading

The variability documented in Tn–TSS relationships for this system (Fig. 2), based on long-term monitoring, is not unusual (Davies-Colley et al., 2003). While a statistically significant linearity is manifested, substantial deviations from best fit relationships occur. The origins of this variability are of particular interest within the context of quantifying the transport and downstream delivery of turbidity for the study system and many other impacted systems. The SAX results of this study provide insights concerning this issue. While particles 10 mm did not make noteworthy contributions to Tn values (Fig. 5a and b), these sizes could have made major contributions to particle mass, as demonstrated for the primary tributary for Schoharie Reservoir (Effler et al., 2008a). These differences in size dependency of optical and gravimetric features of particle populations are an important, if not the dominant, source of variability in Tn–TSS relationships. There are numerous possible drivers of such variability. For example, there can be systematic shifts in contributions by larger particle sizes during runoff events associated with elevated levels of turbulence. However, the effects on the smaller sized turbidity-causing portion of the particle populations have heretofore not been assessed. The light-scattering features of the particle populations of the study system remained relatively invariant over a wider range of driving conditions and Tn levels. These observations suggest that variability in Tn–TSS relationships within the study system (Fig. 2) is largely attributable to variations in features that

2290

water research 43 (2009) 2280–2292

regulate TSS, such as variations in the content of larger (e.g., >10 mm) particles. This is likely the case in many systems, though additional studies are recommended to demonstrate the ubiquity of these observations. The use of TSS and Tn as surrogate metrics of each other has a long history in water quality studies, despite the fundamental differences in the optical and gravimetric features of particle populations. Particle loadings, a concern for downstream turbidity, have most often been represented by the mass loading rate of TSS (i.e., product of TSS concentration and flow, with units of mass/time). The results presented here suggest a more representative loading with respect to impacts on downstream light scattering would use Tn observations directly rather than TSS. The product of Tn and flow is not rigorously a mass loading rate (i.e., units of NTU m3 d1), described elsewhere as a ‘‘quasi’’-load (Effler et al., 2008a; Ward and Principe, 2002). The use of turbidity loads would represent a more reliable quantification of the relative light scattering impacts of the multiple sources within this study system, by avoiding the inherent uncertainty of the Tn–TSS relationships. This approach is particularly supported for this system where the light-scattering features of the particle populations remain relatively uniform. Application of the turbidity loading concept is appropriate in other systems where turbidity and potential multiple sources are issues. Additional support for the approach is provided by the rigorously additive character of the components and sources of b(l) (Davies-Colley et al., 2003; Kirk, 1994). Moreover, Tn is an intensive, rather than extensive, property that does not depend on the system size. As such, the Tn of a mixture of two volumes of water can be computed by volume averaging, as applied by Davies-Colley et al. (2003) to estimate values for rivers formed by converging streams. Consistent with these considerations, tributary turbidity loads were used to drive a mechanistic turbidity model for Schoharie Reservoir (Gelda and Effler, 2007).

6.

Conclusions and implications

Nonspherical clay mineral particles in the 1–10 mm size range are the dominant form of light scattering and turbidity in these interconnected reservoirs and the intervening Esopus Creek over large ranges of runoff and turbidity conditions. PSDs of these particles deviated strongly from hyperbolic (Junge) distributions; for example, measured concentrations of submicron particles were much lower than those represented by a best fit Junge function. Patterns of particle size contributions to scattering and turbidity remained relatively uniform over a wide range of Tn values throughout the study system, particularly following a major runoff event, with d50 values (see Fig. 5a) of w2.5 mm. Modest variations in d50 values were observed within Schoharie Reservoir during summer drawdown and in the west basin of Ashokan Reservoir nearly a month after the major runoff event, probably associated with sediment resuspension and in-reservoir particle size sorting processes, respectively. The credibility of the SAX characterizations of the lightscattering features of the minerogenic particles and of subsequent (Mie theory) estimates of bm and for the study system was supported by the following: (1) PAVm was a strong linear predictor of Tn, (2) reasonable closure was

achieved between measurements of Tn and calculated values of bm, and (3) calculated values of approached the limiting value of 2 for polydispersed particle populations. This performance supports the implementation of programs of coupled particle analysis by SAX and Mie theory calculations to quantitatively evaluate water quality issues driven, or substantively influenced, by light scattering, such as clarity as measured with a Secchi disc (Davies-Colley et al., 2003). The performance of the Mie theory calculations indicates that particle shape is not a primary regulator of scattering and turbidity within the study system. The mean scattering efficiency (), by quantifying the effects of particle composition and PSD, represents an appropriate basis to compare potency of different sources of turbidity. The sources within this study system (Esopus Creek and Schoharie Reservoir) were uniform in that regard. The direct use of Tn measurements in calculations of turbidity loads is recommended, instead of indirectly through TSS measurements, which relies on uncertain and variable Tn–TSS relationships. The capability to assess the light scattering attributes of a large number of inorganic particles over relatively short analytical times with SAX offers the opportunity to advance partitioning of b(l) and bm(l). This has value for lakes and reservoirs, as well as Case 2 marine waters, where minerogenic particles are important in regulating optical characteristics. SAX data provide valuable insights to guide development and parameterization of mechanistic water quality models for turbidity (Gelda and Effler, 2007) and related optics metrics of water quality (Effler et al., 2008b). One such model is presently being used to evaluate management and operational alternatives to abate turbidity problems in the New York City water supply reservoirs (Gelda and Effler, 2007). SAX measurements could have particular value for systems where rehabilitation actions are under evaluation to achieve improvements in water clarity and minerogenic particles are important. Many systems can be expected to demonstrate more complex patterns than those reported here with respect to composition of particle types and sizes, associated with multiple sources and drivers. The chemical characterization capabilities are particularly powerful in such cases in identifying the origins of minerogenic particles and their impact on scattering and related optical metrics of water quality such as clarity. Such information is critical to establish the feasibility of, and to identify the proper targets for, reaching management goals for clarity (Effler et al., 2008b). Additional SAX characterizations, with widely different minerogenic particle populations, are recommended for a broad array of aquatic ecosystems, in concert with appropriate optical measurements, to expand the demonstration of the robustness of the technology and advance the partitioning of light scattering and dependent optical measures in these systems.

Acknowledgments Funding for this research was provided by the New York City Department of Environmental Protection. Sampling was conducted by B.A. Wagner and M.E. Spada. This is contribution 258 of the Upstate Freshwater Institute.

water research 43 (2009) 2280–2292

references

Atteia, O., Perret, D., Adatte, T., Kozel, R., Rossi, P., 1998. Characterization of natural colloids from a river and spring in a karstic basin. Environ. Geol 34 (4), 257–269. Babin, M., Morel, A., Fournier-Sicre, V., Fell, F., Stramski, D., 2003. Light scattering properties of marine particles in coastal and open ocean waters as related to the particle mass concentration. Limnol. Oceanogr 48 (2), 843–859. Bader, H., 1970. The hyperbolic distribution of particle sizes. J. Geophys. Res 75, 2822–2830. Bohren, C.F., Huffman, D.R., 1983. Absorption and Scattering of Light by Small Particles. Wiley, New York. Bricaud, A., Morel, A., Prieur, L., 1983. Optical efficiency factors of some phytoplankters. Limnol. Oceanogr 28 (5), 816–832. Casuccio, G.S., Schlaegle, S.F., Lersch, T.L., Huffman, G.P., Chen, Y., Shah, N., 2004. Measurement of fine particulate matter using electron microscopy techniques. Fuel Process. Tech 85 (6–7), 763–779. Ceronio, A.D., Haarhoff, J., 2005. An improvement on the power law for the description of particle size distributions in potable water treatment. Water Res 39 (2–3), 305–313. Clesceri, L.S., Greenberg, A.E., Eaton, A.D. (Eds.), 1998. Standard Methods for the Examination of Water and Wastewater, American Public Health Association, American Water Works Association, Water Environment Federation, Washington, DC. Davies-Colley, R.J., Vant, W.N., Smith, D.G., 2003. Colour and Clarity of Natural Waters: Science and Management of Optical Water Quality. Blackburn Press, Caldwell, NJ. Effler, S.W., 1988. Secchi disc transparency and turbidity. J. Environ. Eng 114 (6), 1436–1447. Effler, S.W., Matthews, D.A., 2004. Sediment resuspension and drawdown in a water supply reservoir. J. Am. Water Resour. Assoc 40 (1), 251–264. Effler, S.W., Perkins, M.G., Ohrazda, N., Brooks, C.M., Wagner, B.A., Johnson, D.L., Peng, F., Bennett, A., 1998. Turbidity and particle signatures imparted by runoff events in Ashokan Reservoir, NY. Lake Reserv. Manage 14 (2), 254–265. Effler, S.W., Matthews, C.M., Matthews, D.A., 2001. Patterns of gross deposition in reservoirs enriched in inorganic tripton. Can. J. Fish. Aquat. Sci 58 (11), 2177–2188. Effler, S.W., Prestigiacomo, A.R., Peng, F., Bulygina, K.B., Smith, D.G., 2006. Resolution of turbidity patterns from runoff events in a water supply reservoir, and the advantage of in situ beam attenuation measurements. Lake Reserv. Manage 22 (1), 79–93. Effler, S.W., O’Donnell, D.M., Matthews, D.A., Perkins, M.G., O’Donnell, S.M., Gelda, R.K., Prestigiacomo, A.R., Peng, F., Smith, D.G., Bader, A.P., Mayfield, J.D., 2008a. Insights for the structure of a reservoir turbidity model from monitoring and process studies. Lake Reserv. Manage 24 (1), 69–86. Effler, S.W., Gelda, R., Perkins, M., Peng, F., Hairston Jr., N.G., Kearns, C.M., 2008b. Patterns and modeling of the long-term optics record of Onondaga Lake, New York. Fundam. Appl. Limnol 172 (3), 217–237. Gallegos, C.L., Menzel, R.G., 1987. Submicron size distributions of inorganic suspended solids in turbid waters by photon correlation spectroscopy. Water Resour. Res 23 (4), 596–602. Gelda, R.K., Effler, S.W., 2007. Modeling turbidity in a water supply reservoir: Advancements and issues. J. Environ. Engrg 133 (2), 139–148. Gimbert, L.J., Haygarth, P.M., Beckett, R., Worsfold, P.J., 2005. Comparison of centrifugation and filtration techniques for the size fractionation of colloidal material in soil suspensions using sedimentation field-flow fractionation. Environ. Sci. Technol 39 (6), 1731–1735.

2291

Gordon, H.R., Du, T., 2001. Light scattering by nonspherical particles: Application to coccoliths detached from Emiliania huxleyi. Limnol. Oceanogr 46 (6), 1438–1454. Green, R.E., Sosik, H.M., Olson, R.J., 2003. Contributions of phytoplankton and other particles to inherent optical properties in New England continental shelf waters. Limnol. Oceanogr 48 (6), 2377–2391. Hart, B.T., Douglas, G.B., Beckett, R., Put, A.V., Grieken, R.E.V., 1993. Characterization of colloidal and particulate matter transported by the Magela Creek system, Northern Australia. Hydrol. Process 7 (1), 105–118. Jonasz, M., 1987. Nonsphericity of suspended marine particles and its influence on light scattering. Limnol. Oceanogr 32 (5), 1059–1065. Jonasz, M., Fournier, G., 1996. Approximation of the size distribution of marine particles by a sum of log-normal functions. Limnol. Oceanogr 41, 744–754. Kirk, J.T.O., 1985. Effects of suspensoids (turbidity) on penetration of solar radiation in aquatic ecosystems. Hydrobiologia 125, 195–208. Kirk, J.T.O., 1994. Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press, UK. Lartiges, B.S., Deneux-Mustin, S., Villemin, G., Mustin, C., Barres, O., Chamerois, M., Gerard, B., Babut, M., 2001. Composition, structure and size distribution of suspended particulates from the Rhine River. Water Res 35 (3), 808–816. Mishchenko, M.I., Travis, L.D., 1994. Light scattering by polydispersions of randomly oriented spheroids with sizes comparable to wavelengths of observation. Appl. Opt 33 (30), 7206–7225. Peng, F., Effler, S.W., 2005. Inorganic tripton in the Finger Lakes of New York: importance to optical characteristics. Hydrobiologia 543, 259–277. Peng, F., Effler, S.W., 2007. Suspended minerogenic particles in a reservoir: Light-scattering features from individual particle analysis. Limnol. Oceanogr 52 (1), 204–216. Peng, F., Effler, S.W., O’Donnell, D.M., Perkins, M.G., Weidemann, A.D., 2007. Role of minerogenic particles in light scattering in lakes and a river in central New York. Appl. Opt 46 (26), 6577–6594. Perret, D., Newman, M.E., Negre, J.-C., Chen, Y., Buffle, J., 1994. Submicron particles in the rhine riverdI. Physico-chemical characterization. Water Res 28 (1), 91–106. Risovic´, D., Martinis, M., 1995. A comparative analysis of seaparticle-size distribution models. Fizika B 4 (2), 111–120. Smith, D.G., 2002. Turbidity in Catskill Watershed, New York City Department of Environmental Protection, Division of Drinking Water Quality Control. Bureau of Water Supply, Valhalla, NY. Stramski, D., Boss, E., Bogucki, D., Voss, K.J., 2004. The role of seawater constituents in light backscattering in the ocean. Progr. Oceanogr 61, 27–56. Stramski, D., Woz´niak, S.B., 2005. On the role of colloidal particles in light scattering in the ocean. Limnol. Oceanogr 50 (5), 1581–1591. Sullivan, J.M., Twardowski, M.S., Donaghay, P.L., Freeman, S.A., 2005. Use of optical scattering to discriminate particle types in coastal waters. Appl. Opt 44 (9), 1667–1680. Swift, T.J., Joaquim, P.-L., Schladow, S.G., John, E.R., Alan, D.J., Charles, R.G., 2006. Water clarity modeling in Lake Tahoe: Linking suspended matter characteristics to Secchi depth. Aquat. Sci 68 (1), 1–15. Treweek, G.P., Morgan, J.J., 1980. Prediction of suspension turbidities from aggregate size distribution. In: Kevanaugh, M.C., Leckie, J.O. (Eds.), Particulates in Water: Characterization, Fate, Effects and Removal. American Chemical Society, Division of Environmental Chemistry, Washington, DC, pp. 329–352.

2292

water research 43 (2009) 2280–2292

Vant, W.N., Davies-Colley, R.J., 1984. Factors affecting clarity of New Zealand lakes. N.Z.J. Mar. Freshwater Res 18 (3), 367–377. Volten, H., de Haan, J.F., Hovenier, J.W., Schreurs, R., Vassen, W., Dekker, A.G., Hoogenboom, H.J., Charlton, F., Wouts, R., 1998. Laboratory measurements of angular distributions of light scattered by phytoplankton and silt. Limnol. Oceanogr 43 (6), 1180–1197. Ward, C.O., Principe, C.M.A., 2002. 2001 Watershed Water Quality Annual Report. New York City Department of Environmental

Protection, Division of Drinking Water Quality Control. Bureau of Water Supply, Valhalla, NY, p. 110. Woz´niak, S.B., Stramski, D., 2004. Modeling the optical properties of mineral particles suspended in seawater and their influence on ocean reflectance and chlorophyll estimation from remote sensing algorithms. Appl. Opt 43 (17), 3489–3503. Xhoffer, C., Wouters, L., Van Grieken, R., 1992. Characterization of individual particles in the North Sea surface microlayer and underlying seawater: comparison with atmospheric particles. Environ. Sci. Technol 26 (11), 2151–2162.