THE EFFECT OF PULP MILL EFFLUENT ON FINE-GRAINED SEDIMENT MORPHOLOGY AND STORAGE IN THE FRASER RIVER AT PRINCE GEORGE, B.C. by Simon Biickert B.Sc., University of Victoria, 1996 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF MASTER OF SCIENCE m ENVIRONMENTAL SCIENCE © Simon Biickert, 1999 UNiVERSITY OF N BRITISH COtU~~~~ERN lEBRAAY Prince George, BC THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA September 1999 All rights reserved. This work may not be reproduced in whole or in part, by photocopy or other means, without the permission of the author. lll The Effect of Pulp Mill Effluent on Fine-grained Sediment Morphology and Storage in the Fraser River at Prince George, B.C. Abstract A growing body of literature is examining the degree of and effects of the flocculation of fine riverine sediments. It has been shown that flocculation changes the morphology of the fine riverine sediment in the water column. The implications of these changes are as yet incompletely understood, but assumed to cause increased sediment deposition. Some recent studies have investigated the effect of pulp mill effluent as a flocculant, but have not satisfactorily ascertained the relationship between the effluent and flocculation. In this study, measurements of the fine sediment morphology upstream and downstream of a pulp mill were compared to measure the effect of pulp mill effluent as a flocculant. The observations were repeated weekly throughout the annual cycle of temperature, sediment concentration and discharge, allowing the effect of the pulp mill effluent to be put into the context of the natural variation of flocculation of river sediments. To measure the actual fine sediment deposition rates with and without the presence of the pulp mill effluent, a series of sediment traps were deployed during low and medium flow conditions. The size fractions trapped in them provided insight into the natural level of deposition and the amount of enhancement associated with the effluent. The data show that the natural size of floes in the Fraser River, as represented by differences in the effective and absolute particle size distribution, is small when compared to results from measurements in other rivers. Floc sizes are larger under low-flow conditions, especially under winter ice. The annual IV range of floc sizes of the natural t ~ 10 pm) was much larger than the increase measured due to the presence of pulp mill t ~ 1 pm). No statistically significant increase in the rate of fine sediment accumulation in the sediment traps was noted due to the influence of the pulp mill effluent, indicating that, at least in the near-field plume (300m), it has no effect on deposition. v Table of Contents Title Page i Approval 11 Abstract 111 List of Figures IX List of Tables Xll Acknowledgements xiii Chapter 1: 1.0 Introduction 1 Rationale 1 1.1 Study objectives 3 1.2 Project overview 3 Literature Review 7 Chapter 2: 2.0 Background 7 2.1 Floc formation mechanisms 8 2.2 2.1.1 Aggregation/ disaggregation mechanisms 9 2.1.2 Attraction/repulsion ........................................ 10 Floes in fluvial systems ........................................ 12 Case Study: Weldwood Pulp Mill on the Athabasca River... 14 Case Study: Laboratory analysis of Northwood Pulp Mill effluent as a flocculant ........................................ 15 2.3 Effluent and flocculation 17 2.4 Particle size measurement techniques 2.5 2.4.1 Camera methods 19 19 2.4.2 Settling chambers 20 2.4.3 Micropore filter 22 2.4.4 Laser diffraction/backscatter 22 Summary Chapter 3: 24 Suspended fine sediment morphology under the influence of pulp mill effluent 3.0 Methodology 3.1 Baseline data ........................................ 26 ......................................................................... . 26 26 Vl 3.2 3.3 3.2.1 Position relative to diffuser 27 3.2.2 Site characteristics 28 In-situ measurements 3.3.1 3.4 27 Sample site locations .............................................................. . 29 Velocity profile 3.3.2 Temperature/Conductivity 30 3.3.3 Bacteria culture 31 Water samples 32 3.4.1 ......................................................................... . Sampling technique 3.4.2 Sample filtering and processing 3.4.2.1 3.4.2.2 ....................................... . 32 ....................................... . 33 Glass fibre (GF) filter: suspended sediment quantification ....................................... . 3.4.2.3 34 Micropore filter: effective particle size 34 Distribution 3.4.2.3.1 BioQuant particle analysis 3.4.2.3.2 Fractal analysis 35 37 Results 3.5.1 33 (BIO) filter: absolute particle size distribution 3.5 29 38 Variability of test conditions 38 3.5.1.1 River discharge 39 3.5.1.2 Effluent discharge 41 3.5.1.3 Total suspended sediments (TSS) 42 3.5.1.4 Organic: inorganic ratio in suspended sediments 42 3.5.1.5 Temperature 42 ....................................... . 3.5.2 Effluent characterization 43 3.5.2.1 Conductivity 43 3.5.2.2 Bacterial concentrations 45 3.5.3 Variation in sediment characteristics over time 46 3.5.3.1 Absolute particle size distribution (APSD). 46 3.5.3.2 Effective particle size distribution (EPSD) .. 50 3.5.4 Variation in EPSD among sites 3.5.4.1 Sample differences - Komogorov-Smirnov Z 56 56 Vll 3.5.4.2 Site differences: bootstrapping 58 3.5.4.3 Fractal analysis 62 3.5.4.4 Resolution effect 63 65 Discussion 3.6 3.6.1 Effluent plume influence ....................................... . 3.6.1.1 Conductivity and temperature 65 3.6.1.2 Bacterial concentration 66 3.6.1.3 Total suspended sediment concentration .... 66 3.6.2 Variation of APSD with environmental conditions 67 3.6.2.1 Source slope 67 3.6.2.2 Modal particle size ...................................... . 68 3.6.3 Variation of ESPD with environmental conditions 69 3.6.4 Variation of ESPD among sites 71 Chapter 4: Fine sediment settling and storage under the influence of pulp mill effluent ........................................ 75 .................................................................. .. ...... 75 4.0 Methodology 4.1 Sample site locations 4.1.1 75 Position relative to diffuser 76 4.1.2 Site characteristics 4.2 65 Trap design 4.2.1 ............. ...... ..................................... ........... ....... Infiltration bags 4.2.1.1 4.3 4.3.2 77 77 ........ 79 ........................................ 79 Laboratory procedure- sealable tubes ........................................ 81 4.2.2.1.1 Organic analysis 4.2.2.1.2 Inorganic grain size analysis 81 4.2.2.1.3 Settling rate analysis 82 Results 4.3.1 ........................................ Laboratory procedure- Infiltration bags 4.2.2 Sealable tubes 4.2.2.1 76 81 84 Variability of test conditions 84 4.3.1.1 River discharge 84 4.3.1.2 Effluent discharge 85 Effluent characterization 85 viii 4.3.3 4.4 4.3.2.1 Conductivity 86 4.3.2.2 Bacterial concentrations ...................................... . 89 Trapped sediment ....................................... . 90 4.3.3.1 Quantity ....................................... . 90 4.3.3.2 Breakdown by size class 91 4.3.3.3 Absolute particle size distribution 93 4.3.3.4 Organic matter content 96 4.3.3.5 Settling velocity and density 98 101 Discussion 4.4.1 Effluent plume influence 101 4.4.1.1 Conductivity and temperature 101 4.4.1.2 Bacterial concentration 102 4.4.2 Trap effectiveness 103 4.4.3 Particle quantity and coarse sizing 104 4.4.4 Sediment characteristics 105 4.4.4.1 APSD 105 4.4.4.2 Organic matter content ....................................... . 106 4.4.4.3 Settling rate and density ...................................... . 107 Chapter 5: Conclusions Recommendations for further work 110 112 References 115 Appendix A- Analysis of Northwood effluent 120 IX List of Figures 1.1 British Columbia, Canada, showing Prince George 1.2 Annotated air photo showing sample sites. North is at top. Scale is approximately 1:20000 3.1 Greyscale view of Micropore filter through BioQuant system 3.2 Black and white image with user-defined particle/non-particle areas 3.3 Particle edges defined and measured by BioQuant 3.4 Environmental variation observed in the Fraser River over the sample period in 1997. Measurements taken upstream of Northwood Pulp Mill at WSC station at Shelley, B.C. Points on discharge graph indicate measured shear stress at the upstream (grey square) and near-field downstream (black diamond) sites 3.5 Effluent output by Northwood Pulp Mill as reported by Northwood Inc. and Fraser River discharge as measured at WSC station at Shelley, B.C. 3.6 Record of relative conductivity upstream and downstream of the effluent plume Feb.- Dec. 1997 3.7 Coulter spectra of representative times of the year. m =source slope, Q = river discharge (m3 ·s-1). Modal size is circled and labelled. Source slope is measured from the size classes between 1 and 10 pm 3.8 Source slopes measured during different flow conditions. Mean values of ice covered data set= 0.022±0.02 while the ice free means= 0.047±0.02 3.9 Modal particle sizes measured during different flow conditions. Mean values of ice covered data set= 23±10 pm while the ice free means= 15±7 pm X 3.10 Effective particle size distributions at representative times of the year 3.11 Effective particle size distributions under changing environmental conditions 3.12 D50, D 84, and D99 of effective particle size distributions Gan.- Dec., 1997) 3.13 D 84 particle size varying with discharge in the Fraser River 3.14 Relationship between measured shear velocity and river discharge 3.15 Bootstrapped EPSD data, boxplots of D84 measurements around October, 1997 river discharge maximum 3.16 Effective Particle Size Distributions for identical filters using different levels of magnification 4.1 Infiltration bag installation and recovery from Lisle and Eads (1991) 4.2 Fraser River discharge at Shelley, B.C., during period of measurements 4.3 Short-term record of conductivity taken July 22, 1998 from 9 am to 2 pm 4.4 Short-term record of conductivity taken Oct. 19, 1998 from 10 am to 2 pm 4.5 Boxplot of bacterial concentrations in river samples and in raw effluent 4.6 Mass of sediment caught by the infiltration bags per day vs. the total suspended sediments 4.7 Breakdown of sediment trapped in infiltration bags at control site. 4.8 Breakdown of sediment trapped in infiltration bags at effluent-influenced site 4.9 Absolute particle size distribution of sediment yielded during winter sampling by tube-type traps. Mode identified as 28pm. Site is marked UP (upstream) or DN (downstream) XI 4.10 Absolute particle size distribution of sediment yielded during winter sampling by infiltration bags. Mode identified as 37 p.m. Site is marked UP (upstream) or DN (downstream) 4.11 Organic content of 1997 sediment samples by date, with organic content of suspended sediment for reference 4.12 Organic content of 1998 sediment samples by date, with organic content of suspended sediment for reference 4.13 Organic content of samples from tube-type traps by site and year 4.14 Settling velocity distributions of sediment samples taken in 1997 and 1998 4.15 Density distributions of sediment samples taken in 1997 and 1998 4.16 Density of particles versus diameter; 1997 results 4.17 Density of particles versus diameter; 1998 results Xll List of Tables 3.1 Bacterial concentrations- February 1998 water sampling 3.2 Kolmogorov- Smimov test results indicating a deviation in the EPSD from the base case (Feb 07) 3.3 Kolmogorov- Smirnov test results of the far-field downstream site with control site. Effluent dilution factor is presented. Grey shading indicates a statistically significant result 3.4 Analysis of site, date and filter factors on D 84 of EPSD 3.5 Environmental conditions at the control site at each week depicted in figure 3.13 3.6 Sample fractal dimensions at different times of the year 3.7 Fractal dimensions at the control site using higher magnification (lOx objective) 4.1 Average concentration of ions in Northwood effluent and resulting effective Na+ concentration, from Evans (1996) 4.2 Selected results of settling analysis Xlll Acknowledgements I would like the opportunity to thank a number of people without whom this work would not have happened. • To my academic supervisor, Ellen Petticrew, who has provided a level of support which has been extraordinary. Her feedback has been instrumental in the successful completion of the project. I take it as a complement that her evil red pen remains half full. • To Michael Church, who provided a grant in aid of research to Ellen Petticrew, which came from money supplied by the Fraser River Action Plan. • To the members of my committee, Michael Church and Lito Arocena, who provided a wealth of knowledge, experience and guidance. • To the whole crew who had to sit through the Northwood safety video (in some cases more than once). Barb Strobl, Chris Cena, Jen McConnachie, Chris Spicer, Tauqeer Waqar, and Tammy Biickert provided months of gripefree field and lab assistance. • To Jen McConnachie and Chris Cena, who collected all of the 1998 field data. • To Christine Breed, who donated her expertise on culturing bacteria. • To Trent Hoover, who tested my tube trap design in Joe Ackerman's laboratory flume • To Chris Spicer, who provided friendly rivalry and plenty of games of Magic, Myth and Marathon. • To Peter Gabriel and Kate Bush, whose duet "Don't Give Up" was my anthem • To Apple Computers, who make the tools that got the job done • To my wife Tammy, who has been loving and understanding and supported me in this inhuman endeavour 1 Chapter 1 1.0 Introduction Rationale It has been shown in the recent literature that fine sediment transported in river systems does not move solely as primary particles, but in aggregates of two or more particles (Droppo et al., 1996; Droppo et al., 1998a; Church and Krishnappan, 1998; Petticrew and Droppo, in review). This riverine work was based upon the methods and findings of oceanographic studies of "marine snow" (Logan and Wilkinson, 1990; Kranck et al., 1993; Milligan and Hill, 1998). Research on floc processes in freshwater systems was delayed relative to the marine investigations as riverine environments were thought to be too energetic, riverine shear levels being considered excessive for both floc formation and the maintenance of aggregation. As well, ionic concentration, thought to be a determining factor in the formation of bonds between inorganic constituent particles, is relatively low in freshwater. Yet sediments measured in freshwater systems are shown to be aggregated in systems exhibiting a wide range of size and energy (Droppo and Ongley, 1994; Petticrew, 1996; Droppo et al., 1998; Phillips and Walling, 1998). The maximum measured floes sizes are small (102 p.m) compared to marine floes (103 p.m), which is likely an indication of the different conditions (shear stresses, ionic concentration, biological material) in the two environments. As a floc grows in size, with an increasing number of constituent particles, its density diminishes and its structure becomes more fragile, but its mass increases, resulting in greater observed settling velocities relative to those of its constituent particles (Namer and Ganczarczyk, 1993). Shear forces disaggregate 2 floes which are not strong enough to remain intact (Milligan and Hill, 1998), resulting in greater numbers of smaller, stronger floes. So while riverine environments are noted to generate sediment floes the turbulent environment often results in smaller, potentially stronger floes than the marine snow observed in the open ocean. The factors which contribute to the creation of the freshwater floc are varied, with no clear causal relationships shown beyond the confines of the laboratory. Results from experiments indicate that the total suspended solid (SS) concentration and the ionic concentration of the suspension medium will increase the number of interparticle collisions (Lick and Huang, 1993; Lick et al., 1993) and improve their chances of remaining in contact, respectively. Biochemical studies have shown that bacteria play a large role in freshwater floc stability (Van Loosdrecht et al., 1987a,b; Droppo et al., 1997; Bura et al., 1998). Fluid dynamics predict that increasing shear stresses placed on the floes both increases the frequency of interparticle collisions and causes breakup by breaking the floc structure (Lick and Huang, 1993; Lick et al., 1993; Milligan and Hill, 1998). An investigation of a natural river system which exhibits variability in all of these factors could provide information on the relative role of the individual factors in floc development. Concern regarding the flocculation of fine sediments in freshwater systems stems from the use of these systems for waste disposal. Both organic and inorganic toxins are released regularly into river channels as a cheap and effective means of disposal. Fine sediments, defined as particles< 63pm, have been shown to chemically adsorb a wide range of the waste materials released into rivers (Ongley et al., 1992; Droppo et al., 1998b; Finlayson et al., 1998). 3 Predictions of dilution and transportation are based on calculations involving disaggregated fine sediment, which is hydrodynamically different than aggregated sediment (Droppo et al., 1998b). Settling rates of flocculated sediment are much higher than those of their much smaller and lighter constituent particles. Understanding the implications of these changed particle transport dynamics is very important for modeling the fate of contaminants. 1.1 Study objectives The objectives of this study were to determine if: 1) the sediments of the upper Fraser River are flocculated; 2) this flocculation varied under the changing riverine conditions of an annual period; 3) the release of pulp mill effluent affected flocculation; 4) an increase in sediment accumulation rate was observed under the influence of the pulp mill effluent. 1.2 Project overview This research is based on repeated measurements of water and sediment conditions in the Fraser River at Prince George, British Columbia, Canada (Figure 1.1). Sample sites were selected upstream and downstream from the effluent diffuser from the Northwood pulp mill (Figure 1.2). At these locations, water t~ grab samples and other in-stream measurements were taken weekly from January to December, 1997. As well, sediment traps were placed near these sites in the river bed in late 1997 and mid 1998. 4 Three sites were selected based on geographic constraints. Northwood Pulp and Timber Ltd., releases their effluent into a short, straight reach of a meandering Fraser River. The upstream (control) site and the near-field downstream (effect) site were situated so that there was maximum similarity in water conditions and bed conditions. This resulted in the near-field site being relatively close to the effluent release, and the presence of the effluent was less consistent than was desired. This led to the establishment of a second effect site in the downstream far-field. This site was situated at a bend in the Fraser River and as such one could not claim flow or bed similarity, but it was regularly bathed by a water-effluent mix. - - - --, I I I I I i I -i Figure 1.1- British Columbia, Canada, showing Prince George. Water samples were filtered for bulk sediment characteristics (suspended sediment concentration and organic matter concentration), constituent particle 5 size distributions, and aggregated particle size distributions. Sediment trap samples were also analyzed for constituent particle sizes. Fine sediments collected in the trap samples were settled in a laboratory settling tube to quantify differences in gravel-stored fine sediment morphology (size and density). Figure 1.2 -Annotated air photo showing sample sites. North is at top. Scale is approximately 1:20000 Chapter two presents a literature review, while chapters three and four represent the results of two separate studies. Chapter three, "Suspended fine sediment morphology under the influence of pulp mill effluent", concentrates on 6 the in-stream particle size distributions, both absolute and effective, under varying water conditions. Chapter four, "Fine sediment settling and storage under the influence of pulp mill effluent", investigates the settling and storage of fine suspended sediments in a gravel-bed river, both with and without the presence of pulp mill effluent. The final chapter summarizes the findings and makes recommendations for further study. 7 Chapter 2: 2.0 Literature Review Background Traditional theory of riverine sediment transport holds that there are ~ methods of moving erosional products t ~ suspension, saltation, and rolling/ sliding over the bed.. :. The method which predominates depends on the size of the sediment particle and the energy of the stream (Dunne and Leopold, 1978). Silt-sized and smaller (<63 pm) particles, once suspended in the water column, can remain there readily by in-column turbulence (Knighton, -- 1984). Only once the energy of the stream decreases to negligible amounts (in still water) do the fine particles settle (Dunne and Leopold, 1978). The smallest particles, the silts (2.0- 63 pm) and clays (<2.0 pm), are the extreme example of this, being the lightest and having the lowest settling velocity. The ability for even low energy flow to maintain these size classes in suspension results in a supply-limited sediment flux rather than a transport-limited flux (Knighton, 1984). The amount of silts and clays (henceforth referred to as fine particles) moving in the channel is directly proportional to the amounts of these sediments being introduced to the channel. Knighton (1984) states that in-channel residence times of fine sediments can be measured in hours or days, rather than in 100s or 1000s of years, as with pebble and cobble-sized clasts. However, the fine grained sediments introduced and/ or stored in the channel do not move as do larger grains. These sediments are cohesive. This attribute manifests itself in two ways. First, the critical shear stress for erosion ~ cohesive fine sediments is higher than for individual non-cohesive particles' of the same size, meaning once these sediments have settled they require more 8 shear to re-entrain than their size would indicate. Second, flocculation, or the aggregation of fines into larger particles called floes, occurs. Flocculation has historically been studied predominantly in saline environments (Kranck, 1979; Kranck, 1980; Kranck et al., 1993, Kilps et al., 1994; Milligan and Hill, 1998). Research was initiated following observations of greater than expected rates of settling under turbid conditions in estuarine environments (Kranck, 1979). Field studies showed that the fine sediment was settling more quickly than expected because it was aggregating into large floes, which have a much greater settling velocity than their constituent particles. 2.1 Floc formation mechanisms The mechanisms by which floes form are not fully understood (Ongley et al. , 1992). Exactly how flocculation occurs, the rate at which it occurs, the strength of the floes and the factors which influence formation are topics of current investigation. However, there are three main factors which have been noted to regulate the formation and size of floes: the frequency of interparticle collisions, the shear stresses placed on the floes, and the balance of repulsive and attractive forces between particles (Partheniades, 1993). The recognized environmental factors which affect these three factors are suspended sediment (SS) mineralogy and concentration, organic matter supply, temperature, ionic concentration and shear velocity (Petticrew and Biickert, 1998). 9 2.1.1 Aggregation/disaggregation mechanisms In order for two particles or smaller floes to join and become a larger floc, they must collide. There are three mechanisms by which this is possible: Brownian motion, differential settling and shear-induced collisions. Under quiescent conditions, interparticle collisions among sub-micron sized material are possible only by means of Brownian motion, the random movement of the clay grains by the collision of water molecules. For larger particles and aggregates, this motion is minimal, and does not play a significant role (Partheniades, 1993). In environments with lower shear values, such as lacustrine environments away from shore, differential settling is dominant (Lick et al., 1993). Larger particles settling more quickly overtake and collide with smaller particles, creating the opportunity for aggregation. As larger aggregates form, settling velocities increase, increasing the settling rate differential between the larger and smaller bodies. The focus of most flocculation theory is on collisions caused by turbulent shear (Lick et al., 1993; Hill and Nowell, 1995). Shear turbulence is representative of more energetic environments. This energy can both create and destroy floes . Collisions between particles, which can result in aggregation, appears to be a stochastic process. Collisions between a developed floc and either a single grain or another floc can result in no change, aggregation into a larger floc, or destruction of the floc (Lick et al., 1993). Shear stress itself can break up large floes, which tend to be ephemeral and low-density (Tambo and Watanabe, 1979a; Tambo and Watanabe, 1979b; Namer and Ganczarczyk, 1993). 10 Partheniades (1993), states that flocculation within fluvial systems, which are energetic environments with high shear stresses imposed by water flowing over the channel boundaries, is almost entirely caused by turbulence. Laboratory results have shown that, given a constant shear stress, the particle size distribution in the water column will achieve a steady state wherein the breakup and creation of floes are balanced. Increasing the shear stress will increase the rate of flocculation and shorten the time to steady-state, but will decrease the median particle size (Krishnappan et al., 1994; Krishnappan and Lawrence, in press). Of critical importance is the elevated level of shear in the zone immediately above the bed. Floes entering this zone that do not have the necessary structural strength will be broken up and recycled into the water column to potentially re-aggregate (Droppo et al., 1998b). 2.1.2 Attraction/repulsion The repulsive forces between particles are caused by the double layer electrostatic effect caused by the surface electronegativity of clay-sized particles. These are overcome to create floes by two main processes. The first is the suppression of the electrostatic double-layer by cations within the water. The second is the action by bacteria on the particle surfaces. Which of these processes dominate depends largely on the characteristics of the water. The characteristics of seawater greatly differ from those of river or lake water. In the marine environment, there is a high concentration of hydrated cations (salts), which act as flocculators. Alternatively, the low ionic concentration of river water means that ionic chemical processes are probably less important than biologic factors in freshwater systems (Ongley et al., 1992). The role of bacteria is supported by 11 laboratory evidence, which shows increasing flocculation as the temperature of a freshwater test suspension was raised from near o·c to 19·c, allowing the growth of bacteria (Lau, 1990a). The first process, ionic suppression, is a result of the mineral structure of the clay particles in the fine sediment fraction. Layer silicate minerals (clays) and iron and aluminum oxides in solutions with neutral or high pH have a negative surface charge caused by broken Fe-OH and Al-OH complexes. This attracts a layer of cations, which give the particles a mutually repulsive electrostatic layer. Suppression of this layer reduces the range at which the repulsive forces between clay particles can act. This allows the longer-range Vander Waal's forces to act, attracting colloid particles to one another and, in doing so, aiding flocculation (McBride, 1994; Evans, 1996). Marine floes in the turbid estuarine environment seem to be created in response to ionic suppression, due to the abundance of sodium ions. The second process, biological activity, creates floes by acting as "glue". There are two processes by which bacteria can increase the mutual attraction between particles in suspension: exuding extracelluar polymeric fibre (Bar-Or and Shilo, 1987; Droppo and Ongley, 199Gb; Droppo et al., 1996; Liss et al., 1996; Droppo et al., 1997), and hydrophobic surface adhesion (van Loosdrecht et al., 1987a,b). The polymeric fibre is a macromolecule with a high charge density, which short-circuits the electrostatic double-layer. This allows the fibre to bridge between particles, bringing them together into floes (Bar-Or and Shilo, 1987; Bura et al., 1998). Hydrophobic surface adhesion is a natural function of bacterial growth and nutrient uptake. The action of bacteria is spurred by the particles in the water being a source of nutrients (nitrogen, phosphorous and carbon). The 12 bacteria adhere to the particles in the water by increasing their surface hydrophobicity. Under normal conditions, the particles in the water can have a large percentage of their surface area covered by these adhesive bacteria. This would increase the probability of two particles remaining stuck to each other after a collision. 2.2 Floes in fluvial systems Rivers and streams are a relatively new arena for studying floes, although Skvortsov made observations as early as 1959 (Phillips and Walling, 1995a). Marine floes are noted to be delicate structures (Kranck et al., 1993), and therefore unlikely to survive in a river environment. However, studies have shown that floes do indeed form in rivers (Ongley et al., 1992; Kranck et al., 1993; Phillips and Walling, 1995a; De Boer, 1996; Petticrew and Biickert, 1998), and in fact play a major role in the transport of fine sediment in the river system. In some southwestern Ontario rivers, it has been observed that over 90% of the fine sediment by volume is transported in floes with 3 or more primary grains (Droppo and Ongley, 1990a). However, these aggregates only make up approximately 11% of the total suspended particles by number. Flocculation plays a large part in the movement of fine sediments in river systems and challenges some of the previously assumed characteristics of riverine movement of wastes. Effluent release rates are based on the concentrations measured downstream, with the assumption that dilution is the only mechanism by which the concentration of the effluent is being reduced. The river is seen to be a conveyor of the effluent, with the receptor of the waste being the water body at the mouth of the river. In the case of the Fraser River, that 13 receptor is Georgia Strait. This approach to waste management does not include the potential for the wastes to affect their own transport vector. As the bulk of effluent entering the Fraser River from the Northwood pulp mill outfall is organic (Evans, 1996; Nylund, Northwood, Inc., 1997 pers. comm.) (see Section 3.5.1) and with high ionic concentrations (Nylund, 1997 pers. comm.), pulp effluent has the potential to affect flocculation processes, rather than remain an inert substance. Effluent discharge rates are based on the dilution of the effluent into the river as a finite solute in a finite solvent. The concentrations measured downstream of the 'mixing zone' are directly related to the dispersal of the released effluent into the river (Lau, 199Gb). However, if the biosolids which constitute pulp mill effluent are flocculation inducers, there is a problem with the dispersal model of pollutant removal. Metals and other pollutants are preferentially adsorbed by fine sediments, due to their surface charges and due to the high surface area available for reaction (McBride, 1994). Additionally, the resident bacteria on the floes have a high contaminant binding efficiency (Rao et al., 1988). Increased aggregation from the introduction of pulp mill biosolids would create larger floes than would normally exist, potentially increasing their settling velocity. If the effluent is affecting or comprising part of riverine floes, an increased settling velocity could result in deposition of the effluent along with any associated contaminants within the river system (Ongley et al., 1992; Droppo et al., 1998). Flocculated sediments can be deposited in the backwaters and sloughs and on the floodplains of rivers (Arakel et al., 1995; Asselman & Middelkoop, 1995). Aggregated fine sediment structures have been observed in the gravels of biologically active headwater streams (Petticrew, 1996; Petticrew, 1998). With a 14 sufficiently large and stable floc, deposition can occur in the interstitial spaces in the gravel bed. Given this possibility for increased storage of fine sediments in or on the riverbed, pulp mill effluent may require different treatment before it is released into the river system. Case Study: Weldwood Pulp Mill on the Athabasca River A study by Krishnappan et al. (1994) was conducted in February and September 1993 to evaluate the low-flow influence of pulp mill effluent on the fine sediment transport balance in the Athabasca River near Hinton, Alberta, Canada. The study used two methods of quantifying the flocculating effect of the pulp-mill effluent: 1) particle-size analysis using modified Malvern laser diffraction equipment both in situ and in the laboratory and 2) sediment flux calculations at river cross-sections both upstream and downstream of Hinton, Alberta and the Weldwood Pulp Mill (Krishnappan et al., 1994). The times of measurement, February and September, were chosen for the low river discharge at these times, when the ratio of effluent to river discharge would be at its t t ~ 4%). The higher concentration of effluent should have provided a larger effect, which would be more easily quantified. The river discharge in February was 27.7 m 3 ·s-1, and in September was 149 m 3 ·s-1 • The summer peak flow was not published. The Fraser River at the Northwood Pulp Mill has a midwinter base flow of 150 m 3·s-1, which indicates that it is a much larger system than the Athabasca River at Hinton. Particle size distributions showed a significantly larger median instream particle size as compared to constituent or sonified samples removed from the river (50±5 pm instream vs. 30±5 pm dispersed). This evidence of flocculated 15 sediments was observed at the Entrance cross-section (upstream of the pulp mill) and at the Obed cross-section (20 km downstream of the pulp mill). There was no significant difference in the effective particle size distributions observed at the Entrance and Obed cross-sections. Results from total suspended sediment measurements and calculated particle transport rates indicated significant deposition between the Entrance and Obed cross-sections. The slope of the channel downstream of the Entrance crosssection was steeper than that upstream, which would tend to discourage deposition between the Entrance and Obed cross-sections (Krishnappan et al., 1994). The particle transport rate measured in February 1993 at the Entrance cross-section was calculated to be 63.7 metric tonnes per day, and the rate at the Obed cross-section was 20 metric tonnes per day, a reduction of 70%. This implies that between the pulp mill and the downstream cross-section, fully 70% of the previously suspended sediment had been deposited within the Athabasca River's channel. Microscopy indicated that the organic fibres from the pulp mill effluent were forming aggregates with the inorganic grains in the river. This suggests a causal relationship between pulp mill effluent and altered physical transport characteristics of the fine sediment in the Athabasca River system. Case Study: Laboratory analysis of Northwood Pulp Mill effluent as a flocculant A study by Evans (1996) was designed to directly measure flocculation effects caused by the addition of pulp mill effluent to sediment-water mixes under laboratory conditions. Laboratory measurements of instantaneous 16 turbidity and settling tube suspended sediment concentration were used to evaluate the flocculation effect caused by introducing various concentrations of pulp mill effluent to samples of Fraser River water and sediment/water suspensions. The first turbidity experiment used Fraser River water I sediment mixture samples (turbidity 69 NTU), which were combined in different proportions with Northwood Pulp Mill effluent (5 NTU). The solutions were mixed and then the resultant turbidity was compared against the predicted turbidity value. Turbidity was up to 19% percent lower than predicted in mixtures with 50% water I sediment and 50% effluent but, on average, the measurements showed an increase in turbidity of 1.6%, which was attributed to experimental error. A regular decrease in turbidity would have been attributed to a flocculation and settling effect caused by the effluent on the sediment (Evans, 1996). The second turbidity experiment used suspensions of sediment and distilled water. The sediments used were illite (a type of clay) and trapped Fraser River sediments. The Fraser River sediments had been trapped for use in a previous study using plexiglass tubes positioned at approximately mid-depth in the Fraser River. Post-hoc analysis of the Fraser River sediments revealed a depletion of particles < 10 pm as compared to the distribution of suspended particles found in the Fraser River. This fact was used to explain why the Fraser River sediments did not display any turbidity reduction with the addition of pulp mill effluent, when the illite did. The results indicated that the flocculation effect was relevant only to smaller discrete particles (Evans, 1996). Settling tube experiments were conducted to monitor the settling of the suspended sediment. The most dramatic results occurred when an illite 17 suspension was mixed with pulp mill effluent, generating a five-fold increase in the observed settling rates. The flocculation effect of the pulp mill effluent on both the Fraser sediments and illite occurred within one minute of mixing, yet the settling rates observed were small, even when enhanced by the flocculation. The sizes of the floes were not measured directly, only extrapolated from settling rates. The short time frame for the effect agrees with the modeling work by Vine (1996) which indicated that the mixing and flocculation of the effluent should occur within 30 em of the effluent release point in the Fraser River. 2.3 Effluent and flocculation The link between pulp mill effluent and increased flocculation and sedimentation has yet to be strongly supported. The work by Krishnappan et al. (1994) and Evans (1996), show indirectly that flocculation is occurring in the presence of effluent. Krishnappan et al. (1994) do not demonstrate enhanced flocculation at their effect site, but do show a massive deposition effect, confounded by the presence of the town of Hinton (pop. 10,000), which also lies on the Athabasca River between the control and effect sites. Evans (1996) used turbidity as his measure of flocculation, which is an indirect technique and has problems when a mixture of particle sizes is evaluated (Evans, 1996). The results of his settling experiments were significant, yet it is unclear how the results from a quiescent settling tube compare to the energetic environment of the Fraser River. One theory as to why Evans' (1996) work did not support the effluent/ flocculation relationship conclusively is the high concentration of 18 sodium (300- 450 ppm) (Nylund, 1997, pers. comm.) in Northwood's effluent. This concentration is a result of the pulping process, and also serves to assist Northwood's effluent treatment design, which is a series of aerated destabilization basins. Sodium, when in high concentration, yet below the critical coagulation concentration (575- 3450 ppm) and with relatively low concentrations of divalent cations such as magnesium and calcium, does not act as a flocculator, but as a dispersant (McBride, 1994). More recent work by Church and Krishnappan (in press) and Krishnappan and Lawrence (in press), which focused on the Fraser River from Prince George to its mouth, indicated that the suspended sediment upstream of the Northwood effluent diffuser was entirely disaggregated. The median particle size measured in situ was identical ~ 12 p.m) to the median particle size after sanification, which breaks up all aggregates into constituent particles. It was concluded that the aggregation (median particle size ~ p.m) seen 200m downstream of the effluent outfall was entirely due to the change in organic and ionic chemistry caused by the introduction of the effluent (Krishnappan and Lawrence in press). It is pointed out that the Fraser River catchment upstream of Northwood is sparsely populated, and is host to no major industrial development and very little human impact, resulting in little or no introduction of anthropogenic organic, bacteria-rich materials (Krishnappan and Lawrence, in press). This observation is compared to the Nechako River system, which has population centres with corresponding sewage treatment outfalls, as well as agriculture in the river floodplain. Where the more nutrient-rich Nechako empties into the Fraser, the floc sizes are noted to have a median size of 50 p.m (Krishnappan and Lawrence, in press). 19 2.4 Particle size measurement techniques One of the difficulties of measuring flocculation is that standard particle sizing methodologies are based on techniques which destroy (either intentionally or not) the aggregates found in suspension (Kranck and Milligan, 1983; Phillips and Walling, 1995a; Milligan and Hill, 1998; Krishnappan and Lawrence, in press). Increased shear due to sampling (grab bottles, pumping) has been noted to break up floc structures (Milligan and Hill, 1998). An approach used in the grain size analysis of fine sediment involves the removal of all organic material and sanification of samples prior to analysis in a Coulter Multisizer (Kranck and Milligan, 1983). This is an acceptable technique to measure the inorganic constituent particle size distribution. The resultant constituent particle size spectra provide a necessary baseline for calculating the amount of flocculation in the in situ particle size distribution. In order to examine the in situ structure of flocculated suspensions, a non-destructive technique is required. 2.4.1 Camera Methods Perhaps the most straightforward solution to the problem of observing floc structure is to take their picture. Ideally, photography is completely noninvasive, as it requires only the remote sensing of the floes in situ. For measuring marine and riverine floes, a submerged plankton camera can be used as in Milligan (1996) and Petticrew (1996). In systems with large median particle sizes ~ pm, Kranck et al., 1993) and low sediment concentration, non-microscopic photogrammetry is suitable. Using this method, 20 a camera is encased so that it points through a glass pane across a gap of water at a collimated flash. When the shutter is opened, the flash operates, backlighting the floes between the flash and the glass. The result is an image with silhouetted floes showing black against a white background, which can be analyzed with appropriate software or manually. This method is non-invasive, and the aperture between the flash and camera, through which the water and particles pass, is wide enough to inhibit shearing of floes in low flow. This approach has some limitations, which include the difficulty of aligning the camera so that the gap is parallel to flow in a fluvial system, and the minimum resolution of the camera. When the water passes through the gap between the camera and the flash, flow is accelerated, increasing shear stress at given flows. This has implications for floc breakup, which need to be addressed. The minimum resolution for digital analysis using this method is 32-43 p.m/pixel, due to the film, Kodak Photo CD digitization and lens aperture (Milligan, 1996). This minimum size is as large as or larger than most measured floes sizes in river systems (Lick et al., 1993; Droppo and Ongley, 1994; Krishnappan et al., 1994; De Boer, 1997; Droppo et al., 1998; Krishnappan and Lawrence, in press), and would provide little useful information for these particle sizes. 2.4.2 Settling chambers Another method used (Droppo and Ongley, 1994; Droppo et al., 1996, 1997, 1998a,b) is to capture of the particles within a settling chamber. These plexiglass tubes, modified from plankton settling chambers, allow the settling of suspended material into a 3 mL reservoir plate with removable circular 21 microscope slides (Droppo et al., 1996). The settling chamber (a cylinder, diameter ~ em) is immersed parallel to flow with both ends open. The chamber is blocked off at both ends, one end being blocked by a reservoir chamber. The plankton chamber is then brought to the vertical, and the contents of the tube are allowed to settle onto a slide in the bottom of the reservoir chamber. The amount of water in the chamber can be varied by altering the length of the tube, allowing a range of 10 - 100 mL sample sizes to account for changes in suspended sediment concentration. The slide can then be analyzed using an inverted microscope such as the Zeiss Axiovert 100. A variation on this technique includes the stabilization of the 3 mL sample within the plankton chamber reservoir using aragose. This stabilizes the floc structure, allowing multiple microscopic techniques without disruption of the floes (Droppo et al., 1996). Proper handling of a plankton chamber is required to ensure a representative field sample. Hours of quiescent conditions are necessary to settle the floes properly, and then care must be taken not to disturb the reservoirs once the tube has been removed. This is problematic in cold weather, when the contents of the plankton chamber can freeze solid before the sampling is completed, and in locations where the sampling site is a long distance from the laboratory. 22 2.4.3 Micropore filter Floes can also be counted and sized after they have been settled onto a filter and are viewed through a microscope. A water sample is drawn through a filter of known pore size (0.45 p.m) under low suction, leaving the floes in the water column on the filter's surface (De Boer, 1997). The filter can then be observed under a microscope and the floc outlines digitized and sized. Acquiring the water sample can be as in the settling chamber method described above, where a cylinder of known volume is placed vertically on a filter and the water drawn through. Alternately, the sample can be procured via a bottle sample, returned to the lab, and filtered under controlled conditions. Filtering using the settling chamber immediately upon procuring the sample solves the problem of waiting for the floes to settle, but retains the inability to work in cold environments, where the pumping mechanism and the filtration apparatus can freeze solid. Using bottle samples has all-weather capability, but introduces a level of error from the extra handling, transportation and resampling (Phillips and Walling, 1995a). 2.4.4 Laser diffraction/backscatter The methodology which appears to best characterize the in situ floc structure is laser diffraction/backscatter sizing. A probe, connected to a portable computer, is submersed in the water flow. The particles flowing past the apparatus' aperture are sized, giving particle counts in real-time (Krishnappan et al., 1994; Phillips and Walling, 1995b; Krishnappan and Lawrence, in press). Two separate methods based on the laser technology are used: 1) diffraction and 2) backscatter. 23 Diffraction-based measurements as in Krishnappan et al. (1994) and Krishnappan and Lawrence (in press) use a Malvern particle-size analyzer, modified for field use, though it still is bulky at over 60 kg (Krishnappan, 1996, pers. comm.). It must be operated using a hoist to lower it into the water. The Malvern operates by passing a laser through the water I sediment suspension, and measuring the diffraction caused by particles passing through the lowintensity beam (2 mW). The diffraction pattern is converted to particle sizes via the Fraunhofer diffraction theory (Krishnappan et al., 1994). As with the plankton camera, the Malvern can have flows generated in the aperture which disrupts floes. Backscatter-based measurements operate with an oscillating lens which sweeps the focus point of a laser across a measured water volume. Reflected signals are converted to chord lengths of individual particles (Phillips and Walling, 1995b). Where the Malvern operates by projecting its beam perpendicular to the flow (usually vertically through the flow), a laser backscatter device, such as the Par-Tee 200/300 Particle Size Analyzer used by Phillips and Walling (1995b), projects opposite to the flow direction, eliminating the majority of interference to flow that the probe may cause. The probe itself is small, being connected to the laser emitter, electrical generator and computer via a single, 10 metre data/ fibre optic cable. This is much more easily deployed than the Malvern. This technique, relying on reflecting the beam off of the particle surfaces, requires much higher energy output, 250,000 W · cm-2 • Both techniques provide real-time particle size distributions that are truly in situ: the sediment is not disrupted prior to measurement. Additionally, both laser methods have the advantage of being able to measure both the aggregated 24 size distribution and the disaggregated size distribution. Other methods of particle sizing require one method (microscopy) to measure the floes, and then another (Coulter Multisizer) to measure the constituent grains. Both the Malvern and the Par-Tee 200/300 can be used to measure both. However, both systems also suffer when compared to other techniques on the basis of cost and ease of use. Both systems require a significant investment of capital in the equipment, costing thousands or tens of thousands of dollars. That, coupled with the team of researchers required to deploy the Malvern, makes its use for measurement a costly proposition. The Par-Tee succeeds in being less costly and easier to deploy, but as a new technique, requires calibration of its results to known, standard methods (Phillips and Walling, 1995b). 2.5 Summary Earlier research has shown that the relationship between flocculation and the addition of pulp mill effluent to a sediment/water suspension is uncertain. The chemical and biological constituents of pulp mill effluent should enhance flocculation, yet the results, especially those from Evans (1996), are inconclusive. The shear levels in the Fraser River may be too high and the organic content of the suspended sediment too low for substantial flocculation, but the work by Krishnappan et al. (1994) implies that introducing pulp-mill effluent into an energetic stream such as the Athabasca River can cause significant deposition under low flow conditions. The measurement of the in situ particle size distribution can be accomplished using several different methodologies, yet no one method is without drawbacks. Choice of method has to be based upon cost, suitability to 25 field conditions, quality of results, and, within resource limits, best adapted to answer the particular questions asked. 26 Chapter 3 Suspended fine sediment morphology under the influence of pulp mill effluent 3.0 Methodology This study was designed to determine if the effluent being released from the Northwood Pulp Mill is influencing the flocculation of the fine sediment load in the Fraser River north of Prince George, British Columbia. This effect was measured by collecting a representative sample of the river water influenced by the effluent diffuser, and comparing the morphology of this suspended sediment to a similar sample taken outside of the pulp mill effluent. This work was undertaken between January and December, 1997, with field sampling on the Fraser River both upstream and downstream of the Northwood Pulp Mill effluent diffuser. 3.1 Baseline data In order to study the morphology of the fine sediments moving downstream in the Fraser River, two separate baselines were required. The first was the morphology of the sediment prior to the influence of the effluent. This was established by replicating the measurements of the water column upstream of the diffuser. The second baseline was the background variation of the sediment morphology over the annual cycle of environmental conditions within the Fraser River. Temperature, conductivity, discharge and shear velocity all vary over the annual period and can influence flocculation. Therefore, these factors need to be considered in an evaluation of the influence of pulp mill effluent on flocculation. This baseline was established by sampling over an 27 annual period every seven days starting in the second week of January, 1997 and continuing until the last week of December, 1997. 3.2 Sample site locations The primary criteria for the selection of sample sites on the Fraser River were the location relative to the diffuser and similarity of river bed conditions. Using these criteria, two sites were designated, a control site ~ upstream of the diffuser, and t metres ~ 300 metres downstream of the diffuser. Samples were taken from the bank, in approximately 1.2 metres of water. In late May 1997 a third site was added, as there was some concern that the downstream site was not being influenced by the effluent plume on a regular basis (Figure 1.2). The third site, located well into the first major bend in the Fraser River downstream of the effluent release, was chosen strictly due to its position relative to the observed path of the effluent plume, not for any bed similarities with the previous sites. In comparison with the near-field downstream site, this new farfield downstream site was positioned in two metres of water and was on riprap, rather than a gently sloping bed of mixed gravel. 3.2.1 Position Relative to Diffuser The effluent diffuser from the Northwood pulp mill has six nozzles, positioned in pairs, approximately 25 meters from the river bank. The upstream sample site, 50 meters upstream of the diffuser was chosen as close as possible to the diffuser to minimize the possibility of having any confounding effects introduced in the distance between the sample sites. 28 As can be seen in the air photo (Figure 1.2), the plume released from the diffuser remains a coherent, definable body until well downstream of the sites selected for this experiment. The plume broadens downstream, influencing the water at the river's right bank. As the sampling of the Fraser River would be conducted from or near the bank, the near-field downstream location was placed at the point where air photos displayed the influence of the effluent at the bank. The far-field downstream site was located at the first accessible point beyond the near-field site. The Fraser River bends sharply south below the nearfield site, creating a cut bank on the plume-influenced side of the river. Fortunately, the British Columbia Railway intersects the Fraser River after leaving the Northwood Pulp Mill site, providing access. Below this point, access was limited for several kilometers. 3.2.2 Site characteristics Location relative to the diffuser provided the rough positioning of the sample sites, while the fine scale characteristics of the sites determined their final location. The most restricted positioning was the near-field downstream site. Relative to the diffuser, it was located at the junction of the effluent plume and the bank. However, there were geomorphic concerns, as immediately downstream of this point, the geometry of the river course changed, widening to create a slow-moving, depositional area characterized by fine sediments rather than gravels. This set of conditions was unmatched elsewhere in the study reach, so it provided a firm downstream boundary for the near-field site. 29 The upstream control site was positioned so that it would physically resemble the near-field downstream site. It was similar in shoreline slope, bed composition (D84 z 2 em), and flow conditions. The position of the far-field site was also restricted, this time by access. The construction of the railway which provided the access to the site required the dumping of a large amount of riprap to stave off erosion of the bank. There were no sites at all which were accessible this distance from the diffuser which were physically similar to the near-field or control sites. However, conductivity measurements and aerial evidence did show that this site was constantly bathed by the effluent plume. 3.3 In situ measurements To evaluate the changes in the flow regime and effluent influence at the sample sites, velocity profiles were taken at the upstream and near-field sites, and conductivity, temperature and bacteria cultures were taken at all three sites. 3.3.1 Velocity profile Velocity was measured using a calibrated Swoffer rotating propeller current meter for six ice-free months, while a Marsh-McBirney electromagnetic current meter was used for evaluation between late August and early October. There were no variations noted in the results from using the two different current meters. In order to calculate shear velocity, the flow velocity at the sites was measured at 10 em intervals from the bed to the surface at the point of sampling. There was considerable variation in the current due to eddies in the river. To compensate for this, the velocity was recorded as a maximum and 30 minimum measured velocity in meters per second over a period of 1.5 minutes at each point on the profile. 3.3.2 Temperature I Conductivity The effluent plume from the Northwood pulp mill is best described as a heated ~ C in excess of river temperature) organic-rich fluid with a high ionic concentration (predominantly sodium ions). Given this, the best tracking mechanism for plume movement was temperature and/ or conductivity. By using the control site to measure the background temperature and conductivity in the Fraser River at the sample time, measurements at the near-field and farfield sample sites would signal the presence or absence of the effluent plume. Temperature and conductivity were measured for the first five months of 1997 using a Corning Checkmate field conductivity meter, accurate to ±0.5% of the range, or 9 pS·cm·1 • This was replaced by a Hach C0150 meter, also accurate to ±0.5% of the range (9 pS·cm-1), due to the ease of use and reliability of the Hach unit. The measurement was taken by immersing the measurement probe into the water and waiting for the digital readout to stabilize. The precision of the temperature measurements was ±0.1 OC and the precision of the conductivity was ±1pS. In 1998, after the field work period described in section 3.0, YSI 6920 conductivity data loggers were placed in the Fraser River at the near-field and control sites to test whether or not there was a periodicity to the presence of the effluent at the near-field site. Immersion times were ~ hours in mid-summer and fall flow conditions (904 and 714 m 3 ·s·1 respectively), with conductivity and temperature measured at one second intervals. 31 3.3.3 Bacteria culture Measurements of bacterial concentration were not done initially due to the number of other measurements being taken. It was included as an a posteriori means of determining the presence of the effluent plume at the downstream sample sites. The effluent, as a nutrient-rich, heated liquid was hypothesized to be an excellent environment for bacterial culture, which would result in increased bacteria populations at the downstream sites when compared to the upstream site. The first analysis was undertaken in February, 1998 and subsequent bacteria cultures were performed on four separate dates in May-July, 1998. Growth of the bacterial colonies was performed as per Harley and Prescott (1996) and supported by Gonzalez et al. (1996). Sampling in the February period involved using Nalgene bottles to capture one litre quantities of the river water at the upstream and far-field downstream sites, and one litre of the effluent directly from the final settling pond just before the release into the Fraser River. These samples were returned immediately to the University of Northern British Columbia in an insulated container. The May-July samples were different, taken from the fine sediments in the riverbed as sediments are a good medium for growth of bacteria. The bacterial concentration in the sediment reflects a longer-term cumulative bacterial effect of the plume, rather than taking an instantaneous sample of the water column (perhaps missing an eddy of the plume). The bacterial counts were determined in grams of wet sediment weight rather than in millilitres of water sample. This increased the counts of bacteria relative to the earlier water samples but still 32 allowed comparison of upstream and downstream effects. The effluent was still taken as a liquid, rather than sampling the sediment from the settling pond. 3.4 Water samples Measuring other key environmental variables such as total suspended sediment and sediment organic content, as well as the quantification of the aggregation of the suspended sediment was most easily accomplished in the laboratory. Samples of water were removed from the Fraser River and taken to the laboratory at the University of Northern British Columbia, a 20 minute drive, where they were filtered and analyzed. 3.4.1 Sampling technique River water samples were captured using multiple one litre, wide mouth Nalgene bottles. The bottles were filled by hand, with one person wading out to a depth of 1.25 m, inverting a bottle, immersing it to a depth of approximately 60 em (halfway to the bed), then slowly returning the bottle to an upright position, allowing the bottle to fill. The lid was then screwed on, and the bottle was returned to the surface and transferred to the river bank. There was no air in the sample bottles. In order to preserve the sample structure and composition during their return to the university, they were carefully stored and carried in a Coleman brand insulated cooler. Due to the variation in the Fraser River's flow over the annual cycle, some modifications did occur during the weekly sampling. During the first three months of 1997, it was necessary to use an ice auger to drill a hole for sampling. The sample bottles were then passed through the hole in the 33 ice and filled as per above. 1997 was also a year of high water on the Fraser River. The peak flow in June 1997 was an approximately one in 25 year event (G. Davidson, Ministry of Environment, Lands and Parks (MoELP), pers. comm.) resulting in the need for sampling from a canoe at the upstream and near-field sites, and using a five meter pole at the far-field site. In each case, the samples were obtained as near as possible to the standard technique, mentioned above. 3.4.2 Sample filtering and processing The sample filtration provided sediment which was measured in three different ways. To this end, each of the sites' water samples collected from the Fraser River were filtered using three different methods. 3.4.2.1 Glass Fibre (GF) filter: suspended sediment quantification To measure the total ~ sediment (TSS) concentration in the river - water, as well as the organic matter (OM) content of the sediment, up to three ------ litres of water was required during periods of low total suspended sediments. Following the American Public Health Association (APHA)'s (1995) guidelines for measuring total sediment concentration and organic I inorganic ratio, preashed and pre-weighed glass fibre filters with a nominal 8p.m pore size were saturated with sediment, dried, and weighed to w-sgrams. The weight differential between the first and second measurements indicates the amount of sediment contained in the filtered volume of water. The sediment-laden filters were then ashed in a muffle furnace at 55o·c for an hour to removed the volatile organic content. Reweighing the filter allowed the determination of the amount of organic material the sample contained. 34 3.4.2.2 (BIO) filter: Absolute Particle Size Distribution Coulter Counter™ analysis of the filtered inorganic sediment provided the constituent or absolute particle size distribution for comparison to the effective particle size distribution. Coulter analysis sizes and counts individual mineral grains; this is accomplished by first removing the organics in a lowtemperature asher and further dispersion by ultrasonic treatment (Kranck and Milligan, 1983). This procedure required an ashable, Millipore acetate filter, which was completely removed by low-temperature ~ C ashing, leaving only the inorganic component of the sediment sample. These 8 pm pore size filters were covered in sediment then dried in preparation for ashing. Ashing was conducted at the Bedford Institute of Oceanography (BIO) in Dartmouth, Nova Scotia. The samples were then returned to the University of Northern British Columbia for inorganic particle size analysis using a Coulter Electronics Multisizer™. The Coulter Multisizer™ is an electroresistance particle sizer, the operation and calibration of which is described in Kranck and Milligan (1983). 3.4.2.3 Micropore filter: Effective Particle Size Distribution The effective particle size distribution (EPSD) was measured optically, using a method similar to that of De Boer (1997). The effective particle size is the aggregated particle size, which is more representative of the sediment state in situ. Smaller volumes of water (8 to 90 mL, depending on TSS concentration) were passed through 0.45 pm filters at low levels of suction (80 kPa). This left a population of sediment aggregates on the surface of the filter. The volume of water passed through the filter was varied depending on the sediment concentration of the sample, ensuring that a minimum of 105 particles were 35 available to measure yet few enough to minimize overlap of individual particles. The wet filter was then placed on a slide and dried to obtain a uniform filter light transmissivity. Once dried, the slide was positioned on the microscope connected to the BioQuant OS/2 image analysis system. 3.4.2.3.1 BioQuant particle analysis The underlying concept of the BioQuant image analysis system is the automatic identification of particle edges based on the intensity of light/ dark in the digital image collected from the microscope. In the black and white image returned (Figure 3.1), each pixel had a value between 0 (black) and 255 (white). BioQuant required the user to specify the value which characterized the • Figure 3.1 - Greyscale view of micropore filter through BioQuant system boundaries of the measured objects, in this case the sediment aggregates (Figure 3.2). The image was then translated by the image analysis system into 1-bit depth (black/white), with black areas being the particles as defined by the user- 36 set boundary value, and white being the empty area around the particles. The particles were then measured for a given number of properties including perimeter, maximum length, and area (Figure 3.3). . "' ., ~ · •. . -1 ..... • ,. ~ - . ·- .. .. • ~ \"' • .- ., l I • -• .;, . •' .. . .... • I . . .. .. ~ Figure 3.2 -Black and white image with user-defined particle/non-particle areas. u -<:> 0 ~ ' ~ ., • D 1: IJ I " . D. . 0 ... ~ ~ rftd 0 0 0 .. ll . • t:>"" 0 ~ 0 0' "' ~ 'U • <>o ... . ¢ Figure 3 -Particle edges defined and measured by BioQuant. () .. 37 This procedure was repeated in a grid pattern over the area of the filter, creating a database of floc-dimension data. Sample size was determined in part by the concentration of the aggregates on the filter, but a target of 104 particles per filter was considered adequate even when there were many more particles available. The microscope lenses used in the image capture were a 4x objective and a lOx ocular. This resulted in an image size of 512 by 512 pixels covering 1700 x 1700 pm of the filter. Therefore, each pixel in the image represented 3.32pm2 • Given that a single pixel cannot give an adequate representation of a particle outline, the minimum effective resolution of this methodology is 9- 12 pm (three or more pixels). To ensure that this lower limit was not influencing the results of analysis, several filters were re-measured with a lOx objective, lOx ocular. Pixel size with this setup was less than one pm, which was the effective resolution. 3.4.2.3.2 Fractal analysis The particle data from the BioQuant analysis were analyzed for fractal dimensionality at different times of the year. Values of D, D 1 and D2 were calculated for each sample. Each value is a different measure of fractal dimensionality. Dis a measure of the area-perimeter relationship in a set of shapes. Collections of natural objects tend to have an area-perimeter (A, P) relationship of: p oc A D/2 38 Euclidean objects have aD value of one (De Boer, 1997). Values of D greater than one indicate that as area increases, perimeter increases at a greater than expected rate. This means that these larger shapes have more edge complexity. D 1 and D2 are measurements of one- and two-dimensional fractal dimensions, respectively. They are the exponents in power functions of the form: and A oc [ D2 where A = area, P = perimeter and 1= maximum chord length. Values of D 1 greater than one indicate more complex particle outlines and values of D2 less than two indicate more intricate shapes (De Boer, 1997). 3.5 Results 3.5.1 Variability of Test Conditions While the emphasis of this project was to determine the effect of pulp mill effluent on fine sediment structure, a secondary result from the sample design was an evaluation of the natural variability of fine sediment structure over an annual period. It was recognized that the conditions within the river itself could play a role in the variation in sediment structure. For example, the volumetric ratio between the river water and effluent, the temperature in the river, and the flow energy all vary considerably over the period of a year. Sampling over the annual range of conditions allows a comparison of the magnitude of any flocculation effect caused by the effluent and the effects of natural riverine variation. 39 3.5.1.1 River discharge The Fraser River at Prince George has a mean base flow of 150 m 3·s-1 when icecovered, and a mean peak flow of 2500 m 3·s-1 (Vine, 1995). According to measurements by the Water Survey of Canada (WSC) at Shelley, B.C., during the 1997 period of field sampling, there was a measured range of 152 m 3 ·s-1 to 3210 m 3 ·s-1 (Figure 3.4). This peak flow represents an approximately one in 25 year event (G. Davidson, MoELP, pers. comm.). The spring peak occurred on or around June 6th, 1997, while the lowest flows were recorded in January and February when the Fraser River was ice-covered. This variation in discharge represents a 20 fold change. Along with the variation in discharge come associated effects of varying flow energy, turbulence, bed shear, and sediment load. The sample sites for this project were on the margin of the river, and not subject to the full energy of the flow. Nevertheless, the sample sites did experience dramatic changes in the rate and energy of flow. For example, during the winter season, a velocity measurement at 60% depth was 0.35±0.1 m·s-I, whereas on the shoulders of the spring peak, flow velocity at 60% depth was 0.80±0.1ms-1 • This translated into shear velocities of 0.035±0.01 m·s-1 (1.2 N · m-2) during low flow and 0.079±0.01 m·s-1 (6.3 N · m-2) during the shoulder seasons. It was not possible to measure flow velocity or produce a velocity profile during the peak discharge, due to the extremely high water level, or in the winter, due to the ice. Measuring the velocity profile was time-consuming, as long-period eddies were present in the moving water, such that each measurement had to have a maximum and minimum value, reflecting the variance. A typical example was the surface velocity measurement taken at the upstream site 40 August l 5 t: over a period of a minute a minimum velocity of 0.64 m·s-1 and a maximum velocity of 0.81 m·s-1 were measured. -a 3000 (.) Q) If) M E -- - -6 N 5 E 4 z I I.. 3 cu Q) 2 J: en Ice on River • 2000 1000 1 •• • 0 600 0 ...J t'J) E en en 1- 400 200 0 20 If) (.) r::: cu t'J) II.. 0 10 0~ 0 ~ - 30% volatile by weight), warm (-20°C warmer than the river) with high ionic concentration (see Appendix A for chemical composition of the effluent). 3.5.2.1 Conductivity The conductivity of the effluent was measured at 1800±100 p.S·cm-1 • The background conductivity of the river water was higher in the winter (220±20 p.S·cm-1) than in the summer (150±20 p.S·cm-1), but altogether an order of magnitude lower than the effluent. The introduction of the effluent below the control site was expected to raise the conductivity at the two downstream sites above the background. The strength of the conductivity signal is dependent upon the mixing of the plume (lowering the concentration of the effluent) and 44 the volume into which the plume is mixing. As can be noted by comparing Figures 3.4 and 3.6, the grey line denoting the relationship of effluent concentration over time varies as the inverse of river discharge. Relative conductivity was calculated by taking the measurement at the control upstream site as 1.0, and the measurements at the effect sites as proportional to that measurement. There was a strong signal at the far-field downstream site, where the measured conductivity varies in a pattern very similar to that of the effluent concentration, indicating that by the time the plume has reached the far-field downstream site, it is well mixed and bathes the site regularly. The near-field downstream site, on the other hand, shows fluctuations which are usually less strong than the far-field downstream site. This would indicate that the near-field site was marginal to the plume, and often poorly mixed. 0 II $ 1.14 1.12 ec 1.10 ~ 1.06 "iii -- Relative Conductivity 0.006 Effluent Concentration ....._Upstream -a- Near-field ........_Far-field 0.005 Q) - 1.08 ·::; c 1.02 Q) > :;:::; 1.00 Qi 0.98 0 () CCI a: ::= w 0.003 ~ ..... c .._. c 0 u 1.04 :::l "0 0.004 :::l 0 >. a --c Q) (.) c 0 0.002 0 0 () c Q) :::l :;:::: 0.001 LiJ 0.96 ~ ~~ ~~~ ~~ 0 ':>v<::' ':is-4. ~ ~ ~ ~ 0(} ~ Q 0'-' Date Figure 3.6- Record of relative conductivity upstream and downstream of the effluent plume Feb. - Dec. 1997 45 3.5.2.2 Bacterial concentrations The samples taken in February, 1998 for bacterial analysis showed a large increase in the measured bacterial concentrations at the far-field downstream site relative to the levels at the control site (Table 3.1). The effluent had a bacterial concentration of 1.5xl04 colonies·mL-1 compared to 2 colonies·mL-1 measured in the sample taken from the control site. The sample taken at the far-field downstream site had 3 x 102 colonies·mL-I, or two orders of magnitude higher than the control. Table 3.1 -Bacterial concentrations- February 1998 water sampling Site Upstream (Control) Far-field downstream Effluent Number of Colonies I mL High I 1 2.5 290 390 14800 17400 Low The second batch of samples, collected between May and July, 1998, also showed an increase in the numbers of bacteria collected at the effluentinfluenced sites. The effluent had a bacterial concentration of 5.6x104 colonies·mL-1 . While the sediment samples taken from the upstream control site had a median concentration of l.Ox104 colonies· g-1, the samples taken at the nearfield downstream site had a median concentration of 1.8 x 104 colonies· g-1 • The samples taken at the far-field downstream site had a median concentration of 1.8 x 105 colonies·g-1. 46 3.5.3 Variation in sediment characteristics over time The characteristics of the sediment load were expected to change over the course of 1997, due to the variations in the conditions in the Fraser River as described in section 3.6. Both the constituent particles and the structure of the floes would vary with the energy of flow, the concentration of bacteria and sediment, and the organic composition of the sediment. The characteristics of the sediment were evaluated using two separate measurements: the Absolute Particle Size Distribution (APSD) and Effective Particle Size Distribution (EPSD). The absolute distribution is a measurement of the inorganic constituent particles. All organic material is removed and the floes are completely disaggregated. The effective distribution attempts to measure the sediment as it exists in the water column. Aggregates and organic material are preserved. Measuring the EPSD requires much care to not break or create aggregates. 3.5.3.1 Absolute particle size distribution (APSD) Partial results of the analysis are presented in Figure 3.7 for five water samples taken during different flow conditions over the period of 1997. Each sediment spectrum signifies the distribution of the particle sizes (x-axis) within the sample by concentration (y-axis), with the smaller particles represented on the left of the graph. Increasing populations of each size category move the spectrum upwards in the graph. The graph is logarithmic on both axes. The sediment spectra in Figure 3.7 represent decreasing TSS from the top to the bottom of the graph. The October 17th storm peak which had a measured TSS concentration over 600 mg·L-1 is the uppermost spectrum, and the measurement 47 from March 7th during the initial ice melt is the lowest. Two characteristics of the sediment spectra are of interest for comparison. The first is the shape of the left limb of the spectrum, also referred to as the source slope of the curve because it highlights differences in the source of the sediment (Kranck and Milligan, 1983, Kranck et al, 1996). It has been calculated as the slope of the regression of the points between 1 and 10 p.m. -~ - Oct. 17 a= 2o9o m = 0.536 1 0 > c July4 a= 1480 m= 0.352 0 :;::::; ca ..... c +-' Q) (.) June 6 a= 3210 m =0.534 0.1 c 0 0 Dec.12 a=257 m=0.113 0.01 0.001 18.38 March 7 a=273 m = 0.138 ~ 0.1 ~ 10 ~ 100 Particle Size (J..Lm) Figure 3.7- Coulter spectra of representative times of the year. m = sediment source slope, Q =river discharge (m 3·s-1). Modal size is circled and labelled (Jl.m). Source slope is measured from the size classes between 1 and 10 Jl.m. 48 A steeper slope (e.g. Oct. 17th) indicates a smaller proportion of very fine particles being moved, while a flatter curve such as seen in the March ~ spectrum indicates that very fine particles make up a large portion of the fine sediment being moved. The second characteristic of interest is the position of the mode on the curves, which is determined as the size class exhibiting the highest concentration. A mode further to the right indicates that larger size classes are being transported, which is usually a sign of greater entrainment energy or bottom shear stress. The source slopes of the summer and winter spectra are dramatically different (Figures 3.7 and 3.8). 0.07 • • 0.06 0.05 E 6 Q) -c 0 • 0.04 • * • • • 0.03 • •• • ~ 0.02 • • 0.01 •• 0 • • • •• • • • • • • • • • • • River partly or fully covered with ice .K' -0.01 0 500 1000 1500 2000 2500 3000 3500 River Discharge (m3/sec) Figure 3.8 - Source slopes measured during different flow conditions. Mean values of ice covered data set= 0.022±0.02 while the ice free means= 0.047±0.02 49 The summer source slopes are straight and steeply inclined when compared to the winter curves which are flat, and actually curve upwards at the lower limit of the resolution of the Coulter Multisizer. In the summer curves, a pattern in the form of the source slopes emerges. As the TSS and flow energy increase, the source slope steepens. This pattern does continue to the winter curves which have very shallow source slopes which reflect the low energy environment and the ice-covered conditions of the Fraser River drainage. The modes of the spectra shown in Figure 3.7 do not show a definitive relationship with river discharge or TSS. Figure 3.9 displays the modes of all spectra measured, and what becomes apparent is that the modal size increases 35.0 River partly or • - 30.0 25.0 E • 20.0 ! .6 Q) "'0 0 :::2: fully covered with ice .K' .. • • • 15.0 • • 10.0 • • • • .•• • • • •• • 1000 1500 2000 • • • • 5.0 0.0 0 500 2500 3000 River Discharge (m 3/sec) Figure 3.9- Modal particle sizes measured during different flow conditions. Mean values of ice covered data set= 23±10 Jlm while the ice free means= 15±7 JLm. 3500 50 during the times when the Fraser River is either wholly or partially ice-covered (Figure 3.4). These sampling dates have low discharge levels, and would therefore be designated low-energy. The measured modes were compared between the ice-covered and ice-free groups, and the relationship was significant (t = 5.84, p < 0.001). Note that only particles as large as very fine sands are represented in these Coulter spectra. Coarser material was not collected in the samples at this height in the water column over the range of flows. 3.5.3.2 Effective particle size distribution (EPSD) The results of the BioQuant analysis can be presented very similarly to those of the Coulter analysis of APSD. Figure 3.10 shows the EPSD for the same dates as the results from Section 3.7.1. Five curves are again shown, with relative volumetric contribution by size class increasing on the y-axis and increasing particle size on the x-axis. The only difference in format between this graph and the APSD graph (Figure 3.7) from the previous section is that the values on the yaxis have not been presented as a concentration (volume/volume). This means that the increasing TSS which shifted the different curves away from each other in Figure 3.7 is not a factor, leaving the curves on top of one another. The results shown in Figure 3.10 have been selected to reflect the same conditions as those in Figure 3.7, which were partial results selected as representative of different flow regimes over 1997. Unlike the APSD graph, the steepness of the left limb of the spectra is not the source slope as in Kranck et al. (1996), as the lower resolution of the measurements precludes measurement of particles between 1 and 10 microns in 51 diameter. The position of the modal size class is of primary importance, a higher one indicating more frequent occurrence of larger particles. In this graph, an increase in modal size shifts the curve upwards as well as to the right. In the spectra for these aggregated particles, neither the modal size nor the slope of the smaller particles shows the same pattern as the APSD, yet one curve does show a dramatic shift: the March 7th curve. This shift represents a greater proportion by volume of large particles in the measured sample. This result contrasts with the APSD results which shows the March 7th sample to have the highest relative proportion of fine particles. -JIE-7·Mar -a-&Jun ---.-4-Jul --+-- 1 7- O::t ---*-1 2- Dec 1000 ~ 10 ~~ 100 ~ 10 0 0 Particle Diameter (lim) Figure 3.10- Effective particle size distributions at representative times of the year To demonstrate the shift in these EPSD curves which can occur under varying environmental conditions, Figure 3.11 shows a series of five samples 52 taken on the rising limb of the spring flood 1997. This graph is the same as the previous graph, showing the EPSD. The dates coincide with the winter low flow, initial melt, first water rising, rising limb and two weeks past the flow peak, respectively. Each graph in chronological order has a mode which is lower than the previous, until the June 27" measurement, which is higher than the previous curve. This shows that with increasing discharge and resultant shear velocity in the water column, there is a tendency towards a smaller effective particle size. ~ Feb 07 Mar 14 Apr 11 May 08 Jun 27 ~ 1 ~ 10 ~ 100 -+- 164 m3/s _._ 265 m3/s m3/s -<>- 1940 m3/s - - 2030 m3/s ----&- 657 ~ 1000 Particle Size (J.Lm) Figure 3.11 -Effective particle size distributions under changing environmental conditions To test this result statistically, the Kolmogorov-Smirnov z-value was used. The K-S z tests the similarity of two populations. If the descriptive statistics are significantly different, (ie. mode, median, mean, skewness, kurtosis), this test returns a positive result. Table 3.2 shows that the March 4th sample population 53 does not differ significantly from the February 7th base case. However, subsequent dates do differ significantly from the February 7th case, indicating that the EPSD is significantly different during periods of high water as compared to the low water environment. The graphs indicate that this difference is toward larger particles in low flows. Table 3.2 - Kolmogorov- Smirnov test results indicating a deviation in the EPSD from the base case (Feb 07) February 07 (base) March 14 Aprilll May08 June 27 K-S z value 1.065 1.690 2.269 2.037 p 0.206 0.007 < 0.001 < 0.001 Figure 3.12 illustrates the pattern in effective particle sizes observed over the course of 1997. Three curves represent temporal changes in the D50, D84 and D99 particle sizes determined from BioQuant measurements. The D50, or median effective particle size, shows little variation, with slightly, but not significantly, higher particle sizes in the winter (January- March). The D 84 and D99 represent the 84th and 99th percentile of particles: the larger particles measured. These show a much stronger pattern of larger particle sizes in the winter months. Stepwise multiple regression analysis was performed on the D84 observations shown in Figure 3.12 with the variables discharge, temperature, total suspended sediments and organic matter content. The distributions of the variables for the regression indicated that all but D84 were positively skewed so the analysis was performed with log-transformed values. The only variable 54 which was significantly associated with D84 was discharge (t = -5.17, p < 0.001). The regression of log discharge vs. D84 has an r 2 value of 0.372* (Figure 3.13). 90 E :::1. Q) N 80 70 60 U5 50 (.) 40 Q) '+=' ..... cu a_ 30 Dso ~ ~~~ «.q}> ~~ ~ ~~ ~~~ ~ ~ ~~~ e:;0'<' oG- ~ ~ (:i' Date Figure 3.12- D50, D84, and D99 of effective particle size distributions (Jan.- Dec., 1997) Note that the log-corrected discharge is significantly related to shear velocity (r2 = 0.26*, p = 0.015) (Figure 3.14). Yet measured shear velocity does not regress strongly with particle size (r2 = 0.02, p = 0.82). This may be due to the periods when the shear measurements were not taken: the peak flows and the icy conditions. These times correspond to the most dramatic shifts in D 84, but it was not possible to measure the velocity gradient at these times. This selective measurement schedule may have biased the relationship between particle size and measured shear. 55 In an investigation of the relationship between measured shear and discharge, removing the two outliers (marked with arrows in Figure 3.14), 28 26 24 E' ~ . . ~ 22 " (}• ~ ~ ~ G4 ---1 ~~ "C) 0 (}• 0 0 (}• 0 ~ (}• '!>" 0 (}• Date Figure 3.15- Bootstrapped EPSD data, boxplots of D84 measurements around October, 1997 river discharge maximum Table 3.5- Environmental conditions at the control site at each week depicted in Figure 3.15 and statistical significance of site on D84 Group Date Gl Sept. 12 Sept. 19 Sept. 26 Oct. 3 Oct. 10 Oct. 17 Oct. 24 Oct. 31 G2 G3 G4 3.5.4.3 Q m /sec 3 591 638 512 1020 745 2090 998 930 ShearV m/sec 0.03907 0.03991 0.04261 0.05871 0.04676 0.05302 0.05794 TSS mg/L 30.53 22.76 15.88 98.55 27.66 621.36 31.17 22.58 Temp ·c 12.4 9.2 12.0 8.1 4.3 5.3 4.2 4.2 OM % 3.7 4.1 4.6 3.5 4.8 3.2 4.6 5.5 Site diff. F,p 197,0.001 3.16, 0.05 3.77, 0.03 4.20, 0.02 30.1, 0.001 0.87, 0.35 16.8, 0.001 22.4, 0.001 Fractal analysis Measurements of D and D 1, showed no deviation larger than the standard error from the D and D 1 values of Euclidean objects. D2 values of less than the 63 Euclidean value of two were measured, indicating an elongation of the larger particles. Table 3.5 shows a subset of the calculated fractal values. Table 3.6- Sample fractal dimensions at different times of the year. r values represent strength of relationships (section 3.4.2.3.2) Date Site D rL D1 rL D2 rL Feb. 14 Apr. 11 May 23 May 23 May 23 Oct. 03 Oct. 17 Oct. 24 UP UP UP ON 1.076±0.07 1.104±0.07 1.100±0.07 1.104±0.08 1.112±0.07 1.100±0.08 1.086±0.08 1.072±0.07 0.908 0.919 0.909 0.891 0.912 0.877 0.875 0.893 1.004±0.06 1.018±0.06 1.021±0.06 1.023±0.06 1.025±0.06 1.015±0.06 1.011±0.06 0.994±0.06 0.946 0.945 0.939 0.936 0.946 0.927 0.931 0.933 1.694±0.17 1.687±0.16 1.685±0.16 1.652±0.16 1.681±0.16 1.619±0.16 1.632±0.16 1.652±0.16 0.856 0.861 0.851 0.835 0.861 0.814 0.816 0.827 BCR UP UP UP S1te labels: UP= upstream, DN = near-held downstream, BCR = far-field downstream 3.5.4.4 Resolution effect The particles measured on the micropore filters yielded a slightly smaller EPSD when measured at a higher level of magnification (lOx ocular, lOx objective). The resolution of the image improved from a pixel size of 10 pm to a pixel size of < 1 pm. This allowed greater sizing accuracy of small particles. However, it also reduced the number of larger particles sampled, shifting the distribution towards smaller particles due to the maintaining of a sample size of 105 particles. Figure 3.16 shows two pairs of EPSD spectra. Each pair represents the EPSD as measured from duplicate filters 97082201 and 97082202, which were both taken on the same date at the control site. The pairs illustrate the differences in the spectra when the data capture is done using the 4x objective lens or the lOx objective lens. The diagram also indicates the good reproduceability of the methods for these filter duplicates. 64 Re-analyzing site differences at the higher magnification did not reveal any stronger or different patterns. The results of the fractal analysis did not change at the higher magnification (Table 3.7). Repeated measurements taken from the same filters showed no measurable differences in the resulting spectra, and variation among duplicate filters was also very small, as noted above. Table 3.7- Fractal dimensions at the control site using lOx objective. Date Site D r:l D1 r:l Dz rL Oct. 03 Oct. 17 Oct. 24 UP UP UP 1.134±0.09 1.124±0.10 1.094±0.08 0.893 0.897 0.893 1.082±0.07 1.078±0.07 1.041±0.06 0.945 0.947 0.943 1.738±0.17 1.748±0.18 1.715±0.16 0.874 0.875 0.858 1000000 . - - - - - - - - - - - - - - Measured with 4x Objective "; u " 'C M" .!E a. Measured with 1Ox Objective ::::a. cn..,a> c E Q) :I ~~ :I I:T w ~ 10 ~ 100 ~ 1000 Particle Size (pm) Figure 3.16- Effective Particle Size Distributions for identical filters using different levels of magnification. 65 3.6 Discussion 3.6.1 Effluent plume influence There was substantial evidence that the downstream sites were being influenced by the effluent plume. Both the tests for conductivity and bacteria gave positive results. Temperature did not provide a measurable signal. 3.6.1.1 Conductivity and temperature Measurements at the near- and far-field sites on a weekly basis both indicated higher conductivity levels than found at the control site. This effect is attributed to the high measured conductivity of the effluent plume (Marks, 1996; Vine, 1996). However, the levels measured at the near-field site were less elevated and less regular than the levels measured further downstream. As seen in Figure 3.6, the conductivity levels at the far-field site were elevated from the background levels proportional to the calculated effluent concentration. As the discharge in the river increased, the conductivity signal at the far-field site weakened. This is indicative of the effluent being well-mixed but present at this site. At the near-field site, given similar mixing, a higher conductivity would be expected, since the effluent plume is more concentrated nearer to its release. Some very strong signals were measured, but not regularly. Given this information and the bacterial information, it was concluded that the effluent plume at the near-field site was poorly mixed and the site was influenced only by the more dilute margin of the plume. This does not invalidate the data gathered at the near-field site, but does place the emphasis of analysis on samples taken from the far-field site. 66 Since measurable increases in river temperature due to the heated effluent were not recorded at either site at any time, it is reasonable to conclude that the heating effect of the effluent is negligible. 3.6.1.2 Bacterial concentration Along with the consistent positive results in the tests of conductivity at the far-field site, the fact that the bacteria concentration measured was 20 times the background level in the summer and approximately 200 times the background in the winter was unsurprising. The levels measured at the near-field site were of acute interest. Twice the number of colonies were cultured from the sediment samples taken from this site as compared to the control. This is in spite of the observation that the near-field site was only marginal to the plume. As the effluent sampling techniques are comparable between February and May-June we can identify a five fold increase in bacterial activity in the effluent, likely reflecting temperature differences. A detailed analysis of the species and concentrations of bacteria found in the effluent was not part of the scope of this project, although it has been shown that bacteria can produce flocculants (Bar-Or and Shilo, 1987, Droppo et al., 1996, Droppo et al., 1997, Sudo et al., 1997). 3.6.1.3 Total suspended sediment concentration The measured TSS in this project showed the expected variation with the annual variation in the river discharge. Levels were greatly elevated during the rising limb of the spring freshet and the fall storm-event peak in early October. In previous work by Kranck et al. (1992), Church and Krishnappan (in press) and 67 Milligan (1998), the suspended sediment concentration is identified as one of the factors in particle flocculation. The wide range of sediment concentrations observed during 1997 (Figure 3.4) should be adequate to represent the annual ranges which could be reasonably expected. 3.6.2 Variation of APSD with environmental conditions The Coulter Multisizer analysis of the constituent inorganic particles suspended in the Fraser River showed that the suspended sediment during icecovered flow regimes has a different sediment source slope with a slightly larger modal particle size (24 J.lm ± 10J.1m) than ice-free flow regimes (15 J.lm ± 7 J.lm), which tend to be more energetic. 3.6.2.1 Source slope The source slopes of the Coulter spectra measured during ice-covered conditions were concave, with similar low concentrations of -1 J.lm particles and -10 J.lm particles, but a depletion of particles in the 2-6 J.lm range (Figure 3.7). In view of the fact that the low flows are able to entrain and transport particles < 10 J.lm, it is interesting that the distribution of particles smaller than 10 J.lm is not similar to that under higher flow conditions. Under icy conditions, the source of material for transport would be largely from the channel bottom, as there would be little or no runoff from the surrounding land. It is likely that these smaller particles are not available on the channel bottom as single grained particles. During high runoff periods, the source slopes were steeper(- 0.54). Terrestrial and bank inputs would be much more important as sediment sources, providing greater inputs of material over all size ranges. 68 3.6.2.2 Modal particle size Stoke's Law: (Eqn. 1) describes a simple principle: given two particles with different diameters and the same density, the smaller will settle more slowly. A particle with half the diameter of another will have roughly one eighth the mass, reducing the force exerted by gravity. Countering this is the reduction in the fluid resistance to settling from the decreased particle size. The end result is a relationship where settling rate increases with the square of particle diameter. Particles are kept in suspension by the energy of flow providing upwardly directed turbulent currents which counteract the effect of gravity. As more energy is added to the system, more intense turbulence is generated, allowing larger particles to remain in suspension. If the flow energy were solely responsible for the modal size of the APSD, we would expect that in the low-flow, under-ice environment, the modal particle size would be smaller than in open water, high-discharge flow. What was found instead was that on average, modal particle sizes were slightly larger in the low flow than in those from the peak flow of the year. Of interest here is how low the modal values measured in both high and low flow are, considering the energy available for entrainment and transport. The inorganic constituent grain spectra (Figure 3.7) represent the material moving in the water column near the bank at 0.6 of the 1.4 m depth. Very little mass is represented above 20 p.m or in the medium to large silt size range. This must reflect the sources of the material available (glaciolacustrine deposits) 69 (Kranck et al., 1996, Petticrew, 1996) more so than the ability to transport (Knighton, 1984). Recall that sands and gravels which are being transported near the bed of the river during the high flows are not represented in this mid water column sampling. 3.6.3 Variation of EPSD with environmental conditions It is apparent that there are differences between the particle size distributions measured using BioQuant and using the Coulter Multisizer, when the sediment samples have been subjected to low temperature ashing and dissociation by sanification (Figures 3.7, 3.8 and 3.9). The reduced modal particle size and greater numbers of smaller particles in the Coulter spectra indicate that the sediment suspended in the Fraser River has a degree of aggregation. The amount of flocculation in the Fraser River's sediment load is minimal by marine standards (Logan and Wilkinson, 1990, Kilps et al., 1993, Kranck et al. , 1993) and only marginally flocculated by freshwater comparisons (Droppo and Ongley, 1994, Droppo et al., 1996, De Boer, 1997, Droppo et al., 1998b). This study found that the D50 effective particle size was approximately 10 pm, while the D84 effective particle size was approximately 20 pm (Figure 3.12). The modal size of the constituent particles was not that much different than the D 84 aggregated particle size. The maximum floc size measured was only ~ 50 pm during the summer and fall months. This compares poorly with marine floes, which can reach sizes of 103 to 104 pm in quiescent conditions (Kilps et al., 1993). This lack of aggregation in the suspended load of the Fraser River may be partly related to its low organic matter content when compared to agriculturally-influenced or 70 biologically active streams such as the Nith River (maximum floc size 138 pm) (Droppo and Ongley, 1994) and Forfar and O'ne-ell Creeks, (645 pm and 406 pm, respectively) (Petticrew, 1996). It may also be due to the mineralogy of the sediment (Milligan and Hill, 1998). This relatively unflocculated state is supported by the results of the fractal analysis. The sampled particles showed only slight variations from the fractal values of Euclidean shapes. A departure from the characteristics of Euclidean shapes (circles, squares) is an indicator of object complexity, and in this case would have indicated aggregation. Fractal analysis has been used successfully (De Boer and Stone, 1998) to identify differences in sediment source material in agricultural streams, even when the floc structures were predominantly less than 100 pm. Despite the small sizes of floes in the sediment of the Fraser River, there was a significant increase in the effective sizes of particles measured in the winter, when compared to the summer. The only significant environmental factor associated with changes in Fraser River floc sizes was discharge, which we use as a surrogate for bottom shear stress. The interactions between discharge, temperature and suspended sediment noted in both the full data set and the September to October data set (Figure 3.15, Table 3.4) do not present a clear pattern. The general annual pattern of smaller D84 associated with increasing discharge that was statistically supported and noted in Figure 3.13 is not reflected in the fall data set. The median particle size values of the upstream site in Figure 3.15 illustrate that D84 tends to increase with a increasing discharge. This smaller data set is interesting in that these dates exhibiting high discharge and high D 84 also have the highest concentrations of suspended sediment 71 sampled. This also occurs in the fall of the year, when soluble organic delivery from the watershed would be highest, promoting bacterial activity in the water column. The combined effect of shear, bacterial activity and suspended sediment concentration is presumably influencing the flocculation process but the data set and the techniques used to collect the data impose limits on the evaluation of these interactions. 3.6.4 Variation in EPSD among sites It was confirmed that the water sampled at the downstream sites was in fact an effluent/water mix. Given the makeup of the effluent, a warm, organic and bacterially laden liquid, it was reasonable to expect that it would promote flocculation. The results indicate that this is the case, but not to a large degree. The Kolmogorov-Smirnov tests for site differences show that the particle populations at the far-field downstream site are significantly different from those at the control site in 38 of the 60 tests run. The incidence of positive results do not relate directly to the discharge in the Fraser River. This indicates that the ability of the effluent to behave as a flocculant is not strictly dependent on concentration, although there were many more positive results during the low flow period of November and December. In any event, the concentration of the effluent is inversely proportional to the discharge in the Fraser River, which is being used as a surrogate measure for shear. Any site effect correlated with concentration would also be correlated with shear. Bootstrapping the data to look more closely at individual weeks, showed that there was a tendency for the far-field downstream site to have larger D 84 and D99 values than both the near-field and control sites (Figure 3.15). The clearest 72 patterns of increasing particle size downstream were observed during weeks of falling flow, and were insignificant or reversed during weeks of rising or peak flow. Given that the effluent acts in most cases as a flocculant, its effect is small. When compared to the larger changes in the particle morphology due to the change in the river conditions over the course of a year, the effect of the effluent is overshadowed. The largest increase in D 84 attributed to the influence of the effluent was approximately one micron. The change attributed to the change of the seasons is approximately ten microns. When comparing the D99 values (some of the largest particles in the sample), the picture is similar, with the largest shift attributable to effluent being nine microns, when compared to the seasonal shift of 30 microns. The effect of the effluent appears to be an order of magnitude smaller than the natural variation. These results do not indicate a strong flocculation effect from the effluentsediment interaction, in agreement with the prior work by Evans (1996), but in conflict with the results of Krishnappan and Lawrence (in press). A possible confounding factor in the particle sizes seen at the far-field downstream site is the influence of bed roughness. As noted previously, the far-field downstream site is situated on riprap, pieces of rock averaging 40 em in diameter. This constitutes a dramatic increase in bed roughness, which results in increased flow energy dissipation in the form of turbulence (Knighton, 1984). Increased turbulence results in increased shear stress, decreasing the mean floc size (Krishnappan et al., 1994, Krishnappan and Lawrence, in press). Whether or not this is mitigating the flocculation effect of the effluent is unknown. 73 Krishnappan and Lawrence (in press) found that the sediment upstream of the effluent diffuser was completely dissociated (D50 = 12pm), and was flocculated only 300 m downstream of the diffuser with a median particle size of 20 pm. This conflicts with this study's findings that the upstream and downstream particle populations are similarly minimally flocculated. Krishnappan and Lawrence (in press) conducted their field measurements in the early fall of 1996, from a boat which was meant to remain in a constant position in the plume-300m downstream of the diffuser. From the results in Figure 3.12, early fall is the point of the year when there is the lowest level of natural flocculation (as measured at the control site). In this, the data from this study are in agreement with theirs. Also notable are the results in Figure 3.15, which showed in some cases significant site differences in D84 particle size. Therefore, it may be that a combination of factors are causing the conflict between their published results and those of this study: • Krishnappan and Lawrence's (in press) location of the downstream measuring site (300 m downstream in a boat, avoiding both confounds of margin to the plume (near-field downstream site), and riprap (far-field downstream site) • the conflicting methodologies, which have not been calibrated against each other, have generated data which are in agreement (little flocculation at upstream site, an increase downstream), but not in the scale of the effect Laser particle size analyzing techniques, although ideal in being non- destructive to floes and able to be used for measuring both the effective and absolute particle size distributions, are new, and have not as yet had their results 74 comparatively analyzed with established methodologies (Phillips and Walling, 1995). 75 Chapter 4 Fine sediment settling and storage under the influence of pulp mill effluent 4.0 Methodology This study was aimed at investigating the settling characteristics of the suspended sediment in the Fraser River and any alteration caused by the influence of the effluent from the Northwood Pulp Mill. In order to accomplish this, a measure of sediment settling and storage was necessary. Increasing flocculation can result in greater settling velocities (Hill and Nowell, 1995, Milligan and Hill, 1998) and thereby increase the rate at which sediment could theoretically settle out of the water column. It also would change the sediment class sizes which could settle out, allowing finer class sizes to settle as part of larger aggregates. To distinguish and measure the effects of altered sediment behaviour, the amount and characteristics of sediment settling to the river bed had to be measured and compared under the influence of the effluent plume and at the control site. To accomplish this, a set of sediment traps was devised, which would be affixed to and embedded in the river gravels to gather a representative sample of the sediment settling to the bed of the Fraser River. 4.1 Sample site locations The primary criteria for location of sample sites on the Fraser River were similar to those of the suspended sediment site selection and included the location relative to the diffuser and similarity of bed conditions. Using these criteria, two sites were designated, a control site upstream of the diffuser, and 76 another downstream of the diffuser. These sites were approximately five metres upstream of the "upstream" and "near-field downstream" sample locations in the concurrent study of the sediment in the water column (chapter 3). The sites for this part of the study were placed slightly upstream to minimize the effects from regular wading being done for the water sample collection. 4.1.1 Position relative to diffuser The effluent diffuser from the Northwood Pulp Mill has six nozzles, positioned in pairs 10 meters apart, starting approximately 25 meters from the river bank at low water. The upstream sample site was positioned approximately 50 meters upstream of the diffuser. This site was chosen as close as possible to the diffuser to minimize the possibility of having any confounding effects introduced in the distance between the sample sites. As can be seen in the air photo (Figure 1.2), the plume released from the diffuser remains a coherent, definable body until well downstream of the scope of this experiment. The plume broadens downstream, to the point where it influences the water right up to the river bank. As the sediment entrapment would be conducted near the bank, the downstream location was placed at the point where air photos and previous work showed this to occur. 4.1.2 Site characteristics While the position relative to the diffuser provided the rough positioning for the sample sites, it was the bed and flow characteristics of the sites which determined their final location. As noted in Chapter 3, the most restricted location was the near-field downstream site. It was located at the junction of the 77 effluent plume and the bank 300 m downstream of the diffuser. This site was characterized by a coarse gravel bed, with a D 84 of approximately two centimeters. The bed sloped gently towards the center of the channel, allowing wading out for a distance of several meters. The upstream control site was positioned so that it would physically resemble the near-field downstream site. It was similar in bed slope, flow conditions, and bed roughness. 4.2 Trap design This part of the study had two objectives: estimating the total mass of sediment received by the bed and a size characterization of that sediment. This resulted in two separate design goals. One goal was to build a trap which had minimal impact on the bed and the water flow over it, as altering either of these factors could modify the mass of sediment deposited, by creating conditions which would either discourage settling (flow acceleration, decreasing permeability) or encourage it (flow deceleration). The second design goal was to create a trap to collect sufficient fine sediment for the evaluation of the in situ (or effective) particle size structure of the fine sediments. To meet these goals, two trap designs were adopted. 4.2.1 Infiltration bags Infiltration bags were used to capture the settling sediment with a minimum of bed disturbance allowing a measure of the mass of stored fine sediment. Placed properly, an infiltration bag was only in evidence by a patch of gravel which had little or no sand-sized particles and some rope protruding from 78 the riverbed. The design for the infiltration bags was taken directly from Lisle and Eads (1991). The infiltration bag consisted of an 18 em metal ring affixed to a non-permeable bag. The bag was made of heavy-duty nylon fabric, sewn together with an inverted seam to increase water retention. The bag is folded into the metal ring and placed in the apex of a hole approximately 30 em deep with a 18 em wide base excavated from the riverbed (Figure 4.1). The hole was then filled with native gravel from the Fraser River's bed which has been washed of all clasts smaller than 6.35 mm (b-axis). Ropes were attached to the ring and . left protruding from the surface of the gravel. Immersion times varied from three to 30 days, with a median immersion time of five days. At the end of the immersion period, the metal ring was drawn up through the gravel by the emergent ropes. The bag unfolded as the ring was drawn up, capturing the sediment sample for analysis. The traps containing the fines and gravel were returned to the University of Northern British Columbia for analysis. t [...--- 8 A ~ t cables--( . : ~~ ~ ~ : ' I \ ' ' ' : I ~ gravel t :: '' '' chain hoist ' : . : : I ~ ~ '' '' ~ collapsed infiltration bag Figure 4.1- Infiltration bag installation (A) and recovery (B) from Lisle and Eads (1991 ) 79 4.2.1.1 Laboratory procedure- Infiltration bags Once in the lab, the loaded infiltration bags were emptied into 10L plastic buckets with minor irrigation of the bag to ensure the removal all of the fine sediments. As the original cleaned gravel was not to be part of the analysis, the complete sample was washed through a 6.35 mm mesh to remove it. At this point, the sample consisted of a mix of gravels and sands in the bottom of the bucket with a suspended mix of silt and clay above it. This mix was wet sieved through a series of increasingly fine sieves to separate the collected sediment into size classes. The sieve sizes chosen for this experiment had mesh sizes of 2000,500,250 and 150 microns. The water and sediment which passed through the finest sieve was re-collected and allowed to settle over a period of days. The supernatant was drawn off, allowing collection and drying of the sample. All the size classes were dried and weighed to 0.1 grams. 4.2.2 Sealable tubes To gather a sediment sample for fine sediment size characterization (< 63 pm), a sealable trap was envisioned. The literature indicated two such trap designs were already in use. One design, as in Larkin and Slaney (1996), consisted of a lidded plastic bucket filled with cleaned gravel (see section 4.2.1), and embedded in the riverbed flush with the surrounding gravel. The lid was then removed for the duration of the collection period, at the end of which the lid was replaced and sealed until ready for analysis. This design allowed for effective capture of vertically-moving sediment, but prevented the trapping of sediment moving through the gravel pore space (Larkin and Slaney, 1996). This 80 sampling design was subject to scour of the top layer of gravel in the bucket, altering the flow around the trap (Rex, 1997, pers. comm.). The second design, as in Phillips and Walling (in press), very similar to that of Larkin and Slaney (1996), consisted of a flat, rectangular bucket, similarly embedded in the riverbed, with a remotely-triggered lid. This allowed the trap to be put in place during low flow events and remain shut until a river event which the researcher wished to measure. In the fast-moving current of the Fraser River, horizontal sediment movement within and above the gravels was thought to be important. A new trap was designed in the form of an open-ended tube, to be positioned parallel to flow on top of the bed. Each end of the 7.5 em wide by 45 em long tube was enclosed by a 6.35 mm wire mesh, which held the washed gravel within the tube. Each end of the tube was threaded to a lid, which was removed once the tube was in place on the riverbed and returned prior to removal after the immersion period. The result is a sealable gravel matrix which allowed water to percolate through it with the river flow. The flow around and through the tubes was tested in a laboratory flume using a Doppler velocimeter. When the water velocity in the flume was 0.20 m·s-I, flow moving through through the gravel-filled tube was measured at between 0.10 and 0.14 m·s-1. Field measurements of the flow immediately above the bed in the Fraser River showed a range of 0.12 m·s-1 to 0.30 m·s-I, indicating that the flume measurements should be a reasonable representation of the conditions in the field. 81 4.2.2.1 Laboratory procedure- Sealable tubes The tubes were transported to the University of Northern British Columbia, where they were unsealed from one end and their contents poured through the 6.25 mm sieve into lOL plastic buckets. The resulting sediment and suspension was allowed to settle over a period of 2-5 days, after which the supernatant was drawn off. The sediment was then subsampled for organic analysis, inorganic grain size (Coulter) analysis, and settling rate analysis. 4.2.2.1.1 Organic analysis The methods used to determine organic I inorganic ratio followed American Public Health Association (1995) standard procedures. This involved drying at no·c and ashing at 55o·c of the materials to isolate the dried mass and dried non-volatile mass, respectively. 4.2.2.1.2 Inorganic grain size analysis A Coulter Multisizer was used to characterize the inorganic, absolute particle size distribution (APSD) of the sediments collected in the infiltration bags and tube traps, using a 200 pm aperture tube as the samples did not exhibit significant numbers of particles exceeding this size. These samples were treated with 30% hydrogen peroxide (v /v) under heat until all organic material was removed. The samples were then dried, powdered with a mortar and pestle, and weighed. The samples were then sealed and shipped to the Bedford Institute of Oceanography in Dartmouth, Nova Scotia, where the particle size analysis was conducted. The resultant inorganic sediment spectra were then returned to the University of Northern British Columbia. 82 4.2.2.1.3 Settling rate analysis To measure whether or not there were any structural differences between the fine sediment samples collected at the two field sites, representative subsamples of the collected sediment were released into a column of pre-filtered river water for optical settling rate analysis. The subsamples which were measured were taken from the tube traps, which were agitated once to remove them from the traps, but then allowed to settle with minimum disturbance. Differences in the sediment samples (organic matter, biological activity, aggregate structure) would be measured photographically in the settling tube. The water column was a 1.5 m plexiglass tube containing roughly 30 litres of water. A calibrated set of markings (mm) on the inside rear of the tube allowed for measurement of the sediment particles as they moved downward through the camera's field of view. A black and white digital camera, a charge-coupled device (CCD) with a resolution of 512 by 512 pixels, was connected to a Intel-based PC running Empix Imaging's Northern Exposure software. This made an automated image grabbing system, which recorded the current time (accurate to 10-2 sec.) on each image. A run of 45 images could be grabbed in just over a minute and a half. The resultant images had square pixel dimensions of 55 pm ± 10 pm. The images were then analyzed via a custom-developed settling rate-measuring application. Measures of particle size and settling velocity allowed for the derivation of particle Reynolds numbers: (Eqn. 2) 83 where Re is the Reynolds number, xa is the projected area diameter of the particle, Pwis the density of the fluid, u is the terminal settling velocity, and J1 is the dynamic viscosity of the fluid. Particle density was derived by first calculating the correction factor according to its Reynolds number: k5 = (Eqn. 3) ~ for particles in Stokes' region (Re < 0.2), ~ = 5.31 - ~ (Eqn. 4) for particles in Newton's regime (1000 < Re < 3 x 105), k"" [k5 - (0.43/k5 ) 112] • (1000- Re)/(1000- 0.2) + (0.43/k5 ) 112 (Eqn. 5) for particles in the transition range (0.2 < Re < 1000), where k5 is the Stokes correction t ~ is the Newton correction factor, k is the transition correction factor and ~ is the two-dimensional shape factor. Using these correction factors, the densities could be calculated for Stokes' region: (Eqn. 6) for Newton's regime: (Eqn. 7) and for the transition zone: (Eqn. 8) where P£is the excess density of the particle, Pwis the density of the fluid, and g is the acceleration due to gravity (Namer and Ganczarczyk, 1993). 84 4.3 Results 4.3.1 Variability of Test Conditions 4.3.1.1 River discharge The measurements for this part of the study were conducted in the latter part of 1997 (Oct.- Dec.) and the summer of 1998 (April- Aug.). The river discharge peaked twice during this time period, on the 18th of October, 1997 and the 28th of May, 1998 (Figure 4.2). Both of these peaks had flow rates of 2.2 x 103 m 3 ·s·1 • The winter low flow between 1997 and 1998 was 2.4 x 102 m 3 ·s·t, which was higher than that recorded the previous winter (1.6 x 102 m 3 ·s-1). The summer peak flow rate of 1998 was only half that of 1997, which was an approximately one in 25 year event (G. Davidson, MoELP, pers. comm.). 2500 , - - - - - - - - - - - - , - - - - - - -- 1997 1998 ---------- 2000 uGl .!!! 1 1500 ... Gl Cl Ill .t:. u Ill 0 1000 Date Figure 4.2- Fraser River discharge at Shelley, B.C., during period of measurements. 85 4.3.1.2 Effluent discharge The discharge of effluent from the Northwood Pulp Mill was relatively stable, with sudden, dramatic decreases due to temporary shutdowns or slowdowns. A summary of 1997's effluent discharge is presented in Figure 3.5. Assuming complete mixing of the effluent plume into the river flow, the effluent plume accounts for only 0.05% of discharge during peak flow, and 0.5% during low flow. As can be seen in Figures 3.5, this variation of an order of magnitude is not due to variation in effluent discharge, but in the volume of water which is receiving it. Due to the incomplete mixing of the plume within the study reach, these figures are artificially low, perhaps by an order of magnitude. Measurements of the suspended solids in the Northwood effluent indicated a suspended solid concentration of 42±5 mg·L1 of which >30% is organic matter. Taking a representative effluent discharge rate of 1.4 x 108 litres per day, the Northwood Pulp Mill adds 5.9 x 103 kg of suspended solids to the Fraser River each day. During high flow in the Fraser River, the TSS concentration can be as high as 600 mg·L·I, and during low flow, as low as 3 mg·L-1 . Therefore, the effluent plume has a lower TSS concentration than the surrounding water during peak flows in the Fraser River. 4.3.2 Effluent characterization The effluent released from the Northwood Pulp Mill is a diluted medium for disposal of the byproducts of the pulping process. These byproducts, termed "biosolids" by Northwood Pulp and Timber G. Nylund, Northwood Pulp and Timber, pers. comm.), are largely organic (>30% volatile), warm ~ C warmer 86 than the river) (Evans, 1996, Biickert, 1997, unpublished data) with high ionic concentration (Table 4.1), though much lower than that of sea water (108,000 Table 4.1- Average concentration of ions in Northwood effluent and resulting effective Na+concentration, from Evans (1996). Ionic Species Na+ Ca..:+ AP+ Fej+ Mean Concentration (mg/L) 490 195 6.79 7.37 4.3.2.1 Valence Factor 1 64 (2°) 729 (3°) 729 (3°) Molecular. Wt. Ratio 23/23 23/40 23/27 23/56 Effective Na+ cone. (mg/L) 490 7176 3077 2207 Conductivity The ionic concentration of the effluent manifested itself in a very high conductivity. Direct measurements of samples of effluent indicated a conductivity of 1800 p.S·cm·1• The background conductivity of the river water was higher in the winter (220±20 p.S·cm-1) versus the summer (150±20 p.S·cm-1) , but altogether an order of magnitude lower than the effluent. The introduction of the effluent below the control site was expected to raise the conductivity at the downstream sample site above the background. The increase in conductivity was dependent upon the mixing of the plume (lowering the concentration of the effluent) and the volume into which the plume was mixing. As can be seen in Figure 3.6, the near-field downstream sample site showed fluctuations which were usually less strong than the far-field downstream site. This indicated that the near-field site was only marginal to the plume, and often poorly mixed. 87 Further conductivity evidence that the downstream sample site was indeed influenced by the effluent plume is shown in Figures 4.3 and 4.4, which graph measured conductivity at the upstream and near-field downstream sites at one second intervals over a period of several hours using a YSI recording temperature/ conductivity probe. The probes were positioned in the same fashion as the gravel tubes, situated on the gravel bed parallel to flow . These tests were performed July 22nd, 1998 and October 19th, 1998 to measure during periods of similar flow conditions (904 m 3·s·1 and 714 m 3 ·s·1, respectively), but different temperature regimes (ll.s·c and 4.2·c, respectively). Both graphs show slightly elevated conductivity at the downstream site. Figure 4.3 presents data taken in July 1998, showing strong fluctuations in the conductivity at the downstream site, indicating eddies of higher effluent concentration at the site. The data from October 1998 shows consistent downstream elevation of conductivity with a less variable signal over the four hour period. This difference may represent a change of behaviour of the effluent plume due to the increased buoyancy of the heated effluent (Marks, 1996) when the density of the water in the river is at the maximum for H 20, 4·c. 88 0.122 Conductivity (Downstream) 0.120 ........ 0.118 ..... E 0 (/) -E 0.116 >. ·;::;: u 0.114 :J "C c 0 0 0.112 Conductivity (Upstream) 0.110 ~ 09:00 09:30 10:00 10:30 11 :00 11 :30 12:00 12:30 13:00 13:30 Time (HH :MM) Figure 4.3- Short-term record of conductivity taken July 22, 1998 from 9 am to 2 pm 0.094 Conductivity (Downstream) 0.092 ..... E 0 ch 0.090 -- 0.086 c 0 0 0.084 E 0.088 >. ·;::;: :;::; 0 :J "C Conductivity (Upstream) 0.082 ~ 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 Time (HH:MM) Figure 4.4- Short-term record of conductivity taken Oct. 19, 1998 from 10 am to 2 pm 89 4.3.2.2 Bacterial concentrations To establish that significant numbers of bacteria are associated with the effluent plume and present at the downstream sample site, samples of raw effluent, and sediment samples at the effect and control sites were collected during May, June and July of 1998. Over this time period, the temperature in the water rose from soc to 12°C, and the temperature of the effluent rose from 25°C to 32°C. Culturing the samples showed that the effluent had a bacterial concentration of 5.6x104 colonies·mL-1 (Figure 4.5). By comparison, the samples taken from the control site had a median concentration of l.Ox104 colonies·mL-1 • The sample taken at the downstream site had a median concentration of 1.8 x 104 colonies·mL-1, or twice the concentration of the background. This relationship was tested using an independent-samples t-test, and was found to be not significant (t = -1.32, p = 0.203). This is undoubtedly due to the small sample sizes (n = 10 and 9) and three outliers. 6CXXJO :::::1 .e Ill Cl) "E 0 0 ~ *17 40000 c: 013 016 0 ~ "E Cl) u c: 0 u 20000 lU ·;;:: Cl) ulU Ql 0 N• Upstream Downstream Raw Effluent Site Figure 4.5- Boxplot of bacterial concentrations in river samples and in raw effluent 90 4.3.3 Trapped sediment The traps were successful in trapping large amounts of sediment of all size classes. Due to the size of the clasts in the gravel matrices which constituted the trapping mechanism, only clasts smaller than 6.35mm could be analyzed; all others were considered part of the original matrix. 4.3.3.1 Quantity Both the tube trap design (section 4.2.2) and the infiltration bags (section 4.2.1) were very successful at trapping large amounts of sediment in a short amount of time. The exact amount caught by the tube traps is unknown, as their contents were never rigorously flushed and collected. The mass of samples collected was consistently sufficient for the four grams required for organic and Coulter analysis with a large remainder for long-term storage. Depending on the TSS and the time left in the water, some traps yielded well over 300 grams of sediments finer than 6.35 mm. The infiltration bags, which were designed to quantify the amounts of sediment captured, were also very successful (Figure 4.6). As much as 60-70 grams per day were captured. Surprisingly, there does not appear to be a relationship between the measured TSS and the amount captured per day (r2 = 0.04, p = 0.16). Neither does this appear to be a determined by the length of time submersed, as all of the bags that captured more than 50 g per day were in the water more than seven days, and all the bags that captured less than 15 g per day were in the water four days or less. 91 70 :§ ~ 0 ..... Q) a. ..c C> :::J cu (.) IJ) IJ) cu :: • 60 • • •• • 50 • • .. 40 •• • • • •... • 30 • i • • 20 • • 0 •r t 10 0 *• 100 200 300 400 500 600 Total Suspended Sediment (mg/L) Figure 4.6- Mass of sediment caught by the infiltration bags per day vs. the total suspended sediments 4.3.3.2 Breakdown by size class An evaluation of the sediment size class categories caught in the infiltration bags, shows an interesting point: during all flow conditions a large mass of fine sediment is deposited to the riverbed (Figures 4.7 and 4.8). During periods of falling flow Guly 9 -15, July 30- Aug. 11, 1998), over 30 grams of sediment smaller than 150 microns were deposited per day in an area of 255 cm2 • Figure 4.7 shows the breakdown of sediment caught at the control site, and Figure 4.8 shows the breakdown at the site influenced by the effluent plume. Each bag removed from the river makes one set of five columns in the graph. A taller column represents a greater rate of sediment accumulation in that size class. There does not appear to be any pattern of greater quantities of fine sediment accumulating under the influence of the effluent plume, as might have been expected with increased flocculation. The proportions of fine to coarse 92 material at the control site is mimicked downstream. The percentage of the mass represented by the finest category is not significantly different by site (t = -0.73, p = 0.476) and neither is the absolute mass of the finest category (t = -1.79, p = 0.106). As well the total mass accumulated in all size categories is not significantly different between sites (t = -1.32, p = 0.210). 40 > 250f.im I > 200011m < I 50 !1m 'I!IIII'!il./" 35 / > 5001'm ...... > 1501'm 30 § 1: 25 "g I c. 20 .c al ::I ..:a:"'.."' u 15 10 - 5 - 0 - l Lr Lf & .J" J ..r{ s ..........r ID 950625 950709 950715 950715 950720 950720 950730 950730 950506 950506 95051 I 950511 ~ .c 95051 5 950515 Date (YYMMDD) Figure 4.7 - Breakdown of sediment trapped in infiltration bags at control site. 93 40 > 250!im > 200011m 35 I ~ / > 50011m < 150!1m ' 150!im > 30 :§ i: ... "CC 25 Q) a. E m ::s 20 "' 15 ....."' .. ::E t:::-- 10 - - r--- 5 - - r--- 0 laJ ....E .I. .,fl .li Lf _d [ ...J JJ .. _r fil_r ...[ ..d 980628 980709 980709 980715 9807 15 980720 980720 980730 980730 980806 980806 980811 9808 11 980818 9808 18 Date (YYMMDD) Figure 4.8 -Breakdown of sediment trapped in infiltration bags at effluent-influenced site. 4.3.3.3 Absolute particle size distribution Investigation of the size composition of material < 150 p.m analyzed using the Coulter Multisizer also indicates that there is little difference between the upstream and downstream sites. The sediment from the tube-type traps and infiltration bags showed almost identical absolute particle size distributions (APSD) at the upstream and downstream sites (Figures 4.9 and 4.10). Figure 4.9 represents a subsample of the sediment taken from each tubetype trap in late 1997. Each curve signifies the distribution of the particle sizes within the sample, with the smaller particles making up the leftmost portion. 94 Increasing populations of each size category move the curve upwards in the graph. The graph is logarithmic on both axes. When differentiating APSD curves, there are three points to observe: the mode of the distribution, the inflection point, and the source slope. The mode of the curve is the size class which is most common in the sample, and for noncohesive sediments the modal size reflects both the energy of entrainment and the conditions for deposition. The size class of the mode can reflect the erosional environment of surface sediments as particles finer than the mode may have been winnowed away. Alternately, the mode can reflect the depositional conditions, where particles of given sizes are preferentially settled or trapped by the gravels. As these experiments began with gravels washed of all sediments < 6.35 mm the resultant sediment spectra are presumed to better reflect the depositional environments. In Figure 4.9, the modes of all curves are nearly identical, not varying with date or sample site. This is expected, as the sample times were all under very similar flow conditions in the Fraser River. The inflection point is the particle size at which the right-hand limb of the curve drops off with too few particles measured to graph. This is tied closely to the mode as a measure of the energy of the depositional environment. There is no difference in the inflection points of the curves in Figure 4.9. The source slope of the curve is the left limb of the APSD curve. Both the shape (curvature) and angle of the source slope can vary with the conditions under which the sediment was deposited. A flatter curve, with a greater concentration of fine particles is an indicator of flocculated sediment (Kranck et al., 1996). The curves in Figure 4.9 have nearly identical source slopes. As with 95 mode, there is no differentiation by site or date. This is an indication that there may not be enhanced flocculation at the downstream site. The APSD curves in Figure 4.10, representing the sediment samples from the infiltration bags are very similar to the curves in Figure 4.9. The curves are very similar to each other with nearly-identical modes, inflection points and source slopes. This implies that there are no differences in the deposition environment associated with enhanced flocculation at the effluent-influenced sample site. The modes of the samples in the tube type traps are smaller by ~ 10 pm. .......... Nov.lB UP ···•·····Nov.lB UP ··-•-Nov. lB UP ....,.... Nov.lB DN ·····<>-·····Nov. 18 DN ,. ... Nov. 25 UP .....,.- Nov.25 DN -•- Nov.28 UP --. Nov.28 UP --. Nov.28 UP -g..... Nov.28 ON ---.. Nov.28 DN -f>- Nov.28 DN -+-D ec.OZ UP -+-Dec.OZ UP -+-Dec.OZ UP - A - Dec.OZ DN ~ - A - Dec.OZ DN 0.1 ~ 1 ~ 10 ~ 28 ~ 100 Equivalent Spherical Diameter (J.Lm) Figure 4.9- Absolute particle size distribution of sediment yielded during winter sampling by tube-type traps. Mode identified as 28 Jl.m .. Site is marked UP (upstream) or DN (downstream) 96 10 - • - Nov. 18 UP ····• ····--Nov.1 8 UP --9-- Nov. 18 DN - - Nov.25 UP -.t- Nov.25 UP -n- Nov.25 DN -t>- Nov.25 ON - ~ 0 - - Nov.28 UP ..c 0) 'Q) ~ - - Nov.28 UP - - Nov.28 UP -s--- Nov.28 DN ··B····· Nov.28 DN -&···Nov.28 DN -.t-Dec.02 UP --.-De c.02 UP --6--Dec.OZ DN --6--Dec.OZ DN 0.1 0.1 1 10 37 100 Equivalent Spherical Diameter (1-1m) Figure 4.10- Absolute particle size distribution of sediment yielded during winter sampling by infiltration bags. Mode identified as 37 JI-m. Site is marked UP (upstream) or DN (downstream) 4.3.3.4 Organic matter content Volatile solid analysis of the sediment yielded by the tube-type traps showed that the sediment caught is very low in organic material. As shown in Figures 4.11 and 4.12, in almost all cases the organic content in the samples was lower than that in the suspended sediment in the water column. The mean organic content of the 1997 sediment samples was 4.18%, with a standard deviation of 6.08. The mean content of the suspended sediment was 4.66% with a standard deviation of 1.22. The mean organic content of the 1998 sediment samples was 2.52%, with a standard deviation of 0.63. The mean content of the suspended sediment was 6.18% with a standard deviation of 2.48. 97 ~ + + Organics in trapped sediment o Organics in water column 30+----------------------------------------------------- ~ ·2 10+----------------------------------------------------Ill (ll C) 0 D D u~ D D ~ D D D D ~~~~ D :! ~ D t • D 0+---------------r---------------r---------------r----. "November 1 "october 1 September 1 "oecember 1 Figure 4.11- Organic content of1997 sediment samples by date, with organic content of suspended sediment for reference. 15 ,....... D I0 '#. '-" u ·c: .... C) 0 D D nn 5 D 8 ••• •• + ouooolb May 1 D •• . • •• • D D Drifl Apnl1 ~ D D D Ill D D D D Ill + Organics in trapped sediment o Organics in water column D ..r1 ~ D D ~ t June 1 D D D D D oo D D ~ July 1 ~ ~ August 1 ~ Figure 4.12- Organic content of1998 sediment samples by date, with organic content of suspended sediment for reference. Between the sample sites, there was no significant difference in the organic content of the trapped sediment (t = 0.92, p = 0.364) (Figure 4.13). There is a large amount of overlap between the boxplots for each site, and the dark, black line indicating the median value is similar as well. 98 6 998 997 5 5 I 4 4 3 3 2 2 I ~ I I I I 1 .!d 1 c: ~ m 0) 0 0 13 70 58 Near-field Upstream Near-field 12 N= Upstream Figure 4.13- Organic content of samples from tube-type traps by site and year 4.3.3.5 Settling velocity and density Settling rate analysis of the tube-type trap samples showed no significant ~ between the sample sites in the 1997 data (t = -0.02, p = 0.987) and the 1998 data (t = -0.91, p = 0.363). Some summary statistics on the collected data appear in Table 4.2. Table 4.2- Selected results of settling analysis. Year 1997 Particles measured (n) Largest particle 269 1454pm Highest density 1686 kg·m.j Lowest density 1002 kg·m.j I Details rate = 6344 prn/ s dens. = 1007 kg/m3 size = 137 p.m rate = 5930 prn/ s size = 329 p.m rate = 109 prn/ s 1998 853 1025pm 3092 kg·m.j 1007kg·m.j I Details rate = 3377 prn/ s dens. = 1007 kg/m3 size= 61 p.m rate = 3567 prn/ s size= 1025 p.m rate = 3377 prn/ s 99 0217 HXXXl om HXXXl ·= ·= ·= BCXXl 5CXXl "'~ I ·= 2CXXl .. t;; "' I I I I I I I 0 i 2CXXl .. ';:;' & Upstream 1998 12CXXl 1997 12CXXl 0 Near-field 8!1' 8:! 8U IIIII "'' "' "' Near-field ~ ~ Upstream Figure 4.14- Settling velocity distributions of sediment samples taken in 1997 and 1998 Further calculation of the density of particles from the measured settling rates also showed no significant differences at an a-level of 0.05 between the sample sites in either the 1997 data (t = 1.80, p = 0.072) or the 1998 data (t = 0.61, p = 0.542). I BOO .., *111 1500 ~ :;; 0 .. 0" "" ~ g.,= ., 911 "'' 1400 I I ~ I I Near-field 1998 2400 2200 .... 2000 ;; 1800 w ;;, 1600 ~ ~ .. 1200 0 J()(XJ "'" "'" •m *"' ;m om Ill I I Upstream Near-field 1400 ~ " 1CXXl Upstream 2';00 1997 Figure 4.15- Density distributions of sediment samples taken in 1997 and 1998 There was what appears to be exponential relationship between the calculated density of the observed particles and the size of their longest axis (Figures 4.16, 4.17). This shows that the larger the particle is, the less dense it is, indicating flocculation and aggregation processes at work in either the water column or while the sediment is stored in the gravels. 100 ~ 3000 -- 2500 c 2000 ....-.. ('I) E 0') ~ >- +-' "(j) Q) 0 <><> <> 1500 <> 1000 200 0 400 600 800 1000 1200 ~ ~ Size (Jlm) Figure 4.16- Density of particles versus diameter; 1997 results 3500 <> 3000 ....-.. -('I) E 0') ~ <> <> 2500 >- +-' "(j) c: Q) 0 2000 1500 0 ~~~~~~~~~~ 200 400 600 ~ 800 Size (Jlm) Figure 4.17- Density of particles versus diameter; 1998 results 1000 1200 101 4.4 Discussion 4.4.1 Effluent plume influence There was substantial evidence that the downstream site was being influenced by the effluent plume. Both the tests for conductivity and bacteria reflect the presence of the effluent at the downstream site. Temperature did not provide a measurable signal. 4.4.1.1 Conductivity and temperature Weekly measurements of conductivity at the downstream site indicated higher levels than found at the control site. This effect is attributed to the higher ionic content and conductivity of the effluent plume compared to the fresh Fraser River water. Compared to sea water, it has low ionic concentration (12,950 mg·L1 effective concentration of Na+in effluent vs. 108,000 mg·L-1 effective Na+in sea water (Evans, 1996)). However, the conductivity levels measured at the downstream trap location were less elevated and less regular than levels measured further downstream. Some very high conductivity values were measured, but not consistently (Figure 3.6). This fact led to the concern that the downstream site was being missed by the effluent plume entirely. Conductivity data loggers placed in the Fraser River at the near-field and control sites showed that there was an increase in the conductivity measured downstream although in some flows the signal was intermittent (Figures 4.3 and 4.4). Given this information and the bacterial data, it was concluded that the effluent plume at the near-field site was poorly mixed and the site was influenced by the more dilute margin of the plume. As the sediment traps were 102 placed in the river for periods between 3 and 30 days, the concern about the intermittent influence of the plume at the downstream site (period= 0.5 hours) was mitigated. Since measurable increases in river temperature due to the heated effluent were not recorded at the downstream site, it is reasonable to conclude that the heating effect of the effluent is negligible. However, there was one apparent temperature effect caused by the effluent plume. During January and February of 1997, when the Fraser River was covered by ice up to half a metre thick at the upstream site, the portion influenced by the plume was completely ice-free. Evidently the amount of heat provided by the plume was enough to discourage the formation of ice, but not enough to measure between sites. This effect did not influence the results from this part of the study, as during ice-covered conditions it was not practical to place traps at the control site. 4.4.1.2 Bacterial concentration The presence of elevated bacteria levels in the effluent plume was a very strong marker of effluent influence on the downstream site. The levels measured at the near-field site were of acute interest as the conductivity indicator of plume presence had thus far been weak. From the sediment samples taken, on average twice the number of colonies were cultured from the site influenced by the plume as compared to the control. This is in agreement with the observation that the downstream site was only marginal to the plume, as higher bacterial counts were measured further downstream in the suspended sediment portion of this study. 103 4.4.2 Trap effediveness The two trap designs both worked well, collecting sufficient sediment for analysis. The tube-type trap was designed for qualitative analysis of the fine sediments which would be trapped in its gravel matrix as the river water percolated through it. What was anticipated was a very small sediment sample in a large volume of water in the tube, which would be removed and analyzed in a fashion very similar to the water samples taken in the suspended sediment component of this study. The effective particle size distribution was to be measured using BioQuant analysis, investigating the sizes of in situ particles stored in the gravels. However, the amount of sediment which was trapped in the tubes meant any visual analysis was impractical due to the high sediment concentration in the samples. To accommodate this change a settling analysis of the particles was undertaken to characterize the size, shape, density and settling speed of the sediments stored in the traps. The infiltration bags were designed for quantitative analysis. The processing of the samples once they had left the river would be vigorous to remove all the sediment from the bags and categorize them by size. Any aggregation of particles would be altered or destroyed. It was also anticipated that there would be some loss of silt- and clay-sized particles during the removal of the bag traps from the riverbed and return to the lab. As it turned out, the mass of material lost in handling did not significantly alter the size spectra as seen in Figures 4.9 and 4.10. The consistency of the results suggests that there were no significant losses. 104 4.4.3 Particle quantity and coarse sizing None of the particle quantity or size analyses showed a significant difference between the samples taken under the influence of the effluent plume and those taken at the control site. There were, however, interesting observations to be made about the amount of sediment trapped in varying size classes. In all conditions, large amounts of sediment (Larkin and Slaney, 1996) were collected in both types of traps. The amount caught per day did not correlate well with the TSS concentration during the traps' immersion in the Fraser River (Figure 4.6). This is most likely due to a combination of three reasons: 1) During high flow, typified by high shear velocities and high TSS, relatively small amounts of sediment <150 pm were trapped (see June and July results, Figures 4.7 and 4.8), which makes up the bulk of the sediment samples in lower flow conditions. 2) The one set of samples taken during the highest flow conditions(> 2000 m 3 ·s-I, 600 mg·L-1 TSS) were in place for 30 days, and were saturated by fine sediment when withdrawn. 3) The greatest rates of deposition were during periods of falling water, especially in the shallows near the bank of the river where the traps were placed. These periods did not show the highest TSS concentration (highest usually on rising limb (Knighton, 1984)), but there was a large amount of deposition. The relationship between trapping efficiency and the time left in water did not show the "saturation effect" as in Larkin and Slaney (1996). The traps which 105 were left in the water the shortest amount of time were typified by the lowest amounts of sediment caught per day. This is confounded by the flow conditions during the period of immersion, including TSS concentration and whether or not the flow was rising or falling. It was physically easier to place and withdraw traps during times of low, steady flow, and this is reflected in the times of the year which were sampled: mid-summer and late fall. The importance of fine sediment deposition in these results is most likely a result of this, and using these sample methods more frequently during high flow (were it possible) may have shown greater importance of the salta ted, coarser clasts in the infiltration bags. 4.4.4 Sediment characteristics 4.4.4.1 APSD The results of the comparison of constituent inorganic particle size analysis by Coulter analysis were negative. There were no observable differences between the particle size spectra within a site over time or between sites. The samples were all collected under similar flow conditions. If the pulp mill effluent were enhancing flocculation, an increase in the amount of very fine sediment fractions (< 10 p.m) would be expected at the downstream site. There was no observable site difference. The particle size spectra from the trap samples differed from those of particles collected from the water column (Figure 3.7). Those samples were taken at 60% depth in 1.4 m of water, and indicate smaller modal particle sizes and flatter source slopes. This is anticipated, due to stratification of the suspended load, with finer particles more prevalent further from the bed. 106 Of note is the similarity of the spectral shape between the two trap types. Their source slopes and inflection points are almost identical, with a slightly larger modal particle size in the infiltration bags. This is remarkable, given that the traps were radically different in design and intent. The tube traps were sealed while still in place on the river bed, but it was expected that some of the fine sediments would be lost from the infiltration bags during extraction from the bed. However, from the results in Figures 4.3 and 4.4, it appears as though the bag design is sound for fine sediment analysis. The tube traps do not appear to be as effective at collecting particles sandsized and bigger. The tube samples were not sieved to remove particles> 150 p.m as were the bag samples, yet did not show large numbers of particles greater than 63 p.m. 4.4.4.2 Organic matter content The organic matter content of the trapped samples was significantly lower than that of the suspended sediment captured concurrently by means of grab samples. The explanation for this is relatively simple. The percent organic material by weight was performed on the tube trap samples, which contained proportionately larger clasts (as large as coarse sands). Each sand grain contains no volatile solids whatsoever, being completely mineral, biasing the sample organic matter measurement. What was more difficult to explain was the lack of extra organic material at the downstream site, which should have been influenced by the organic effluent plume. This can be taken as either an indicator of a) the site being marginal to the plume or b) little or no flocculation effect caused by the plume, 107 resulting in no significant increase in particle settling rates. Given the conductivity and bacterial results and the results of the suspended sediment portion of this study, (b) seems more likely. 4.4.4.3 Settling rate and density The settling analysis provided the best representation of the in situ sediment characteristics once it had been trapped in the gravel matrix. The particles observed were much different from those observed using BioQuant in the suspended sediment portion of this study. Many, much larger, particles were noted. The maximum size exceeded 1 x 103 microns. The largest particles observed from the water column were 100 - 300 microns. There was some concern as to whether or not these particles were being created during the storage period between collection and the settling analysis. Some of the samples had been refrigerated wet for up to 30 days prior to settling. To evaluate this, four sets of samples were tested for storage effects over time. One subsample from each was settled immediately upon collection, and then the remainder was settled in subsequent weeks, to a maximum of 28 days postcollection. There was no change observed in the size structure of particles being observed, so it was concluded that there was little or no effect of storage. There was no evidence of a difference in the settling behaviour of the particles captured at the effluent-influenced site. The settling behaviour would be determined by the amount of aggregation in the sample. The amount of aggregation would be influenced by the conditions in the gravels, primarily organic content (previously shown to not differ between sites) and the bacterial activity (Petticrew, 1996, 1998). The doubled population density of bacteria at the 108 downstream site does not appear to be sufficient to cause greater amounts of aggregation in the trapped sediment. Petticrew and Droppo (in review) observed two distinct types of particles in the sediment resuspended from biologically active gravel bed rivers. One set of particles, which were characterized as small and dense (most likely single grains or tightly-packed aggregates) settled very rapidly for their size. The second set of particles were typically much larger and settled slowly. These are hypothesized to be floes, with a large proportion of organic material and water within their volume. These two types of particles were observed in the Fraser River sediments as evidenced by the tails of distributions in Figures 4.16 and 4.17. However, the bulk of the particles were between 100 and 300 p.m. The larger particles in this study ( > 300 pm) must be flocculated, as they barely exceed the density of water. The few small (< 100 pm), dense particles could be single grains or compact aggregates. Quartz grains have a density of 2650 kg·m-3 while we measured densities between 1500 and 3000 kg·m·3 for particles of this size. The 1997 r 2 value of the log/log relationship between particle size and density was 0.59, p = 0.04, and the 1998 was 0.47, p = 0.03. The photography was the single, largest source of error in this method. The problems encountered are summarized here: 1) Resolution: The resolution of the images captured was 512x512 pixels. The resultant pixel size was 50±10 microns, with different values among different experiments depending on the precise setup of the camera equipment. This meant that the minimum size of particle which could be easily identified between frames was ~ 180 microns. 109 2) Focus: The focal depth of the lens was approximately one em, which was only a small fraction of the 10 em depth of the settling tube. Particles travelling outside this zone were not measured. 3) Frame rate: The hardware used could only capture one frame every two seconds. Given the three centimeter vertical portion of the settling tube that was within the field of view, any particle travelling faster than two em · s-1 did not appear on three sequential images, making particle identification difficult. The fastest-moving particles (single grains and compact aggregates) were difficult to identify. This resulted in an underrepresentation of faster-moving compact particles, which is evidenced by the lack of a vertical tail to the distribution in the smaller particle sample in from 1997 (1997 n = 269, 1998 n = 853) (Figure 4.16). 110 Chapter 5: Conclusions In Chapter 3, it is shown that the suspended sediment moving in the Fraser River system at Prince George is flocculated. Comparing the median and maximum floc sizes, the amount of flocculation in these sediments is less than seen in other river systems (Droppo and Ongley, 1994, Petticrew, 1996, DeBoer, 1997, Droppo et al., 1998a). This is likely due to the sources of the sediments moving in the Fraser River system, which do not have a large proportion of organic material. During the winter months, when the discharge in the Fraser River ~ 150 m 3 ·s·1 and the temperature is ~O OC the floc sizes measured were significantly larger (D84 ~ 10 pm) than those seen in the summer, high flow months. This difference is attributed to the low energy conditions associated with the low discharge, which would allow larger floes to exist in the water column without being broken by shear stress. The effect of the pulp mill effluent on the morphology of the suspended sediments in the Fraser River was not dramatic. There was a tendency towards larger particle sizes at the measurement sites under the effect of the effluent plume, but the difference in sizes was very small. Bootstrapping of the very large sample sizes resulted in statistically significant differences among the control and two effect sites. The difference in D84 seen among sites was ~ 1 micron, whereas the difference which was attributable to the difference between summer and winter flows was ~ 10 microns. Defining the relationship between changing aggregate size and changing shear was not possible with the shear measurement technique used. The measured shear was shown to correlate with log discharge, and D84 particle size 111 was shown to correlate with log discharge, but shear did not correlate well with D 84 • Despite this, the river discharge was taken as a surrogate for flow energy and shear within the water column. This was sound, as it was not possible to measure the velocity gradients during the times of the year when the most dramatic changes in flow and particle size occurred, but discharge measurements were available for the entire year. The results from the sediment traps suggest that in the near-field (300 m downstream) of the diffuser, the addition of pulp mill effluent to the water column does not create a large enough flocculation effect to cause measurable increases in deposition rates. Large amounts of sediment were collected in the traps at both the effect and control locations which, when fractionated and sized using a Coulter Multisizer, showed no difference in composition. This indicates that very similar mechanisms are responsible for the deposition at both sites; i.e. no effluent-augmented settling. Further experiments showed no significant differences in the organic content, settling rates and density of the sediment collected at the effect site. As a whole, the addition of pulp mill effluent to the Fraser River above Prince George does not appear to have a large effect on the near-field morphology or settling behaviour of its suspended sediment. Flocculation is a natural process in the Fraser River sediments, which varies due to the changing flow conditions. The variations observed during the annual cycle of the river were an order of magnitude larger than those observed due to the addition of the pulp mill effluent. In the near-field of the effluent outfall, no enhanced sediment accumulation is evident. 112 Even without enhanced sediment accumulation caused by the effluent plume, large deposits of fine sediment were noted at both the control and effect sites. A survey of stored contaminants was not within the scope of this project. As fine sediments are known to adsorb some organic and inorganic contaminants (McBride 1994), there is potential for contaminants introduced within the effluent to be stored within the gravels of the riverbed. Recommendations for further work Although this is the most comprehensive study to date of the flocculation effect at the Northwood Pulp Mill site, some questions remain: 1) is the difference in flocculation observed by Krishnappan and Lawrence (in press) as compared to those in Chapter 3 due to the differences in methodology? 2) is there a far-field sediment accumulation effect as in Krishnappan et al. (1994)? Due to the time and financial constraints placed on this study, it was not feasible to use laser diffraction or laser reflection equipment to measure the in situ particle size distributions. Krishnappan and Lawrence's (in press) work with the Malvern particle size analyser indicated different flocculation regimes than were observed in this study, which used the more rudimentary bottle-sample technique, which can modify the particle size structure (Phillips and Walling, 1995a, Milligan, 1997). Krishnappan and Lawrence (in press) record median particle sizes of ~ 12 pm upstream of the diffuser and ~ 22 pm downstream, where this study measured median particle sizes of ~ 10 pm both upstream and 113 downstream. Direct comparisons of these two measurement techniques would aid in clarifying this issue. Perhaps of more value would be an emulation of the techniques of Krishnappan et al. (1994), where two or more cross-sections of the Fraser River would be measured for depth-integrated total suspended sediment concentration, and a sediment flux calculated. Differences in the sediment flux upstream and downstream of the Northwood Pulp Mill could then be related to the influence of the pulp mill outfall. The Fraser River at Prince George is large and fast flowing, therefore to accomplish this method accurately very detailed, repetitive measures of suspended sediment would be required to account for natural temporal and spatial variability. There is an opportunity for this research, as there is a reach of -3 kilometers downstream of the pulp mill where there are no further anthropogenic influences. After this point, there are two smaller pulp mills, and then the city of Prince George, which would be major confounding factors (effluent sources). This would measure whether or not there is a far-field increase in fine sediment deposition, complementing the work shown here on the near-field deposition rates. Alternatively, a continuation of the sediment trap work shown here could be used to extend the results to the far-field reach of the Fraser River. It was impractical to place traps at the far-field sample site of this study, due to the riprap placed to protect the British Columbia Railway. Further downstream (1.5 to 2 km), the channel of the Fraser River turns away from the cutbank and returns to a straight reach banked by gentle gravel slopes. At this point, the effluent plume is well mixed, and it is possible to place both infiltration bags and tube-type traps in the bed. Access is limited to this portion of the river, but not 114 impossible. Any deposition increase caused by the effluent would be measurable at this distance from the diffuser. As with the cross-section proposal, downstream flexibility is limited by the two additional pulp mills immediately upstream of the city of Prince George. A valuable supplement to further trap work would be a study of the path the effluent plume follows under different flow regimes, including its path through the bends in the Fraser River downstream of the study area for this project. Conductivity cross-sections of the channel would be the most effective means of doing this, as conductivity is the most prominent signal of the effluent's presence in the Fraser River. This would help to define the best locations for situating future traps where effluent presence is assured in concentrations as high as possible. 115 Refereqces American Public Health Association, 1995. Standard Methods for the Examination of Water and Wastewater, APHA, Washington, D.C. Arakel, A., Hanna, R., McConchie, D., 1995. "Sediment quality in Australian river basins: a compartmental approach to assessment of a regional sediment-variability factor," Effects of Scale on Interpretation and Management of Sediment and Water Quality, IAHS publ. no 226, (ed. by W.R. Osterkamp), 246-252. Asselman, N.E., Middelkoop, H. 1995. "Floodplain sedimentation: quantities, patterns and processes," Earth Surface Processes and Landforms, 20,481499. Bar-Or, Y., Shilo, M., 1987. "Characterization of macromolecular flocculants produced by Phormidium sp. strain J-1 and by Anabaenopsis cirlularis PCC 6720," Applied and Environmental Microbiology, 53, 2226-2230. Bura, R., Cheung, M., Liao, B., Finlayson, J., Lee, B.C., Droppo, I.G., 1998. "Composition of extracellular polymeric substances in the activated sludge floc matrix," Water Science and Technology, 37, 325-333. Church, M., Krishnappan, B.G., in press. "Sediment sources, transport and potential modelling approaches in the Fraser River from Hansard to Mission," in: Health of the Fraser River Aquatic Ecosystem: A Synthesis of Research Conducted Under the Fraser River Action Plan, (ed. by C. Gray and T. Tuominen) Environment Canada, Vancouver. DOE FRAP 1998-11. De Boer, D.H., 1997. "An evaluation of fractal dimensions to quantify changes in the morphology of fluvial suspended sediment particles during baseflow conditions," Hydrological Processes, 11,_415-426. De Boer, D.H. and Stone, P.M., 1998. "Spatial and temporal variability in fractal dimensions of suspended solids in two southem Ontario rivers," Modelling Soil Erosion, Sediment Transport and Closely Related Hydrological Processes, IAHS Publ. 249 (ed. by W. Summer, E. Klaghofer and W. Zhang), 445-453. Droppo, I.G., Flannigan, D.T., Leppard, G.G., Jaskot, C., Liss, S.N., 1996. "Floc stabilization for multiple microscopic techniques," Applied and Environmental Microbiology, 62,3508-3515. Droppo, LG., Jeffries, D., Jaskot, C., Backus, S., 1998a. "The prevalence of freshwater flocculation in cold regions: a case study from the Mackenzie river delta, Northwest Territories, Canada," Arctic, 51, 155-164. Droppo, LG., Walling, D.E., Ongley, E.D., 1998b. "Suspended sediment structure: implications for sediment and contaminant transport modelling," Modelling Soil Erosion, Sediment Transport and Closely 116 Related Hydrological Processes, IAHS Publ. 249 (ed. by W. Summer, E. Klaghofer and W. Zhang), 437-444. Droppo, I.G., Leppard, G.G., Flannigan, D.T., Liss, S.N., 1997. "The freshwater floc: a functional relationship of water and organic and inorganic floc constituents affecting suspended sediment properties," Water, Air and Soil Pollution, 99, 43-54. Droppo, I.G., Ongley, E.D., 1994. "Flocculation of suspended sediment in rivers of southeastern Canada," Water Research, 28, 1799-1809. Dunne, T. and Leopold, L.B., 1978. Water in Environmental Planning, W.H. Freeman and Company, New York, 818 pp. Evans, W.J., 1996. Flocculation of Fraser River Sediments Due to Pulp Mill Effluents. MSc thesis, University of British Columbia, Canada. Everett, D.H., 1994. Basic Principles of Colloid Science. Royal Society of Chemisty, Cambridge. Finlayson, J.C., Liao, B., Droppo, I.G., Leppard, G.G., Liss, S.N., 1998. "The relationship between the structure of activated sludge floes and the sorption of hydrophobic pollutants," Water Science and Technology, 37, 353-357. Gonzalez, J.M., Whitman, W.B., Hodson, R.E., Moran, M.A., 1996. "Identifying numerically abundant culturable bacteria from complex communities: an example froma lignin enrichment culture," Applied and Environmental Microbiology, 62, 4433-4440. Harley, J.P, Prescott, L.M., 1996. Laboratory Exercises in Microbiology, 3'd Edition. William C. Brown, Dubique, lA, 484 pp. Hill, P.S. Nowell, A.R.M., 1995. "Comparison of two models of aggregation in continental shelf boundary layers," Journal of Geophysical Research, 100, 22749-22763. Kilps, J.R., Logan, B.E., Alldredge, A.L., 1993. "Fractal dimensions of marine snow determined from image analysis of in situ photographs," Deep Sea Research,41, 1159-1169. Knighton, D., 1984. Fluvial Forms and Processes, John Wiley and Sons, Ltd., New York. Kranck, K., 1979. "Dynamics and distribution of suspended particulate matter in the St. Lawrence Estuary," Naturaliste Canadien, 106, 163-173. Kranck, K., 1980. "Variability of particulate matter in a small coastal inlet," Canadian Journal of Fisheries and Aquatic Science, 37, 1209-1215. 117 Kranck, K., Milligan, T.G., 1983. "Grain-size distributions of inorganic suspended river sediment," SCOPE/UNEP Sonderband, 55,525-534. Kranck, K., Petticrew, E., Milligan, T.G., Droppo, I.G., 1993. "In situ particle size distributions resulting from flocculation of suspended sediment," Coastal and Estuarine Study Series, 42, 60-74. Kranck, K., Smith, P.C., Milligan, T.G., 1996. "Grain-size characteristics of finegrained unflocculated sediments 1: 'One-round' distributions," Sedimentology, 43, 589-596. Krishnappan, B.G., Lawrence, G., in press. "Interactions of pulp mill discharges with the Fraser River and its sediments," in: Health of the Fraser River Aquatic Ecosystem: A Synthesis of Research Conducted Under the Fraser River Action Plan, (ed. by C. Gray and T. Tuominen) Environment Canada, Vancouver. DOE FRAP 1998-11. Krishnappan, B.G., Stephens, R., Kraft, J.A., Moore, B.H., 1994. "Size distribution and transport of suspended sediment particles in the Athabasca River near Hinton," NWRI Contribution, 32 pp. Larkin, G.A., Slaney, P.A., 1996. "Calibration of a habitat sedimentation indicator for use in measuring the effectiveness of watershed restoration treatments," Watershed Restoration Management Report, British Columbia Ministry of Environment, Lands and Parks, 5, 14 pp. Lau, Y.L., 1990a. "Temperature effect on settling velocity and deposition of cohesive sediments," NWRI Contribution, 99, 12 pp. Lau Y.L., 199Gb. "Physical Processes of Pollutant Transport in the Aquatic Environment." NWRI Contribution 159, 15 pp. Lick, W., Huang, H ., 1993. "Flocculation and physical properties of floes," Coastal and Estuarine Studies: Nearshore and Estuarine Cohesive Sediment Transport, 42, 21-39 Lick, W., Huang, H., Jepsen, R., 1993. "Flocculation of fine-grained sediments due to differential settling," Journal of Geophysical Research, 98, 1027910288. Lisle, T.E., Eads, R.E., 1991. "Methods to measure sedimentation of spawning gravel," USDA Forest Service Research Note, PSW-411, 7 pp. Liss, S.N, Droppo, I.G. Flannigan, D.T., Leppard, G.G., 1996. "Floc architecture in wastewater and natural riverine systems," Environmental Science and Technology, 30, 680-686. Logan, B.E., Wilkinson, D.B., 1990. "Fractal geometry of marine snow and other biological aggregates," Limnology and Oceanography, 35, 130-136. 118 Marks, B.J., 1996. Initial Dilution of a Horizontal Jet in a Strong Current. MSc thesis, University of British Columbia, Canada. McBride, M.B., 1994. Environmental Chemistry of Soils. Oxford University Press, New York. Milligan, T.G., 1996. "In situ particle (floc) size measurements with the Benthos 373 plankton silhouette camera," Journal of Sea Research, 36.5, 93-100. Milligan, T.G., Hill, P.S., 1998. "A laboratory assessment of the relative importance of turbulence, particle composition, and concentration in limiting maximal floc size and settling behaviour," Journal of Sea Research, 39, 227-241. Mudroch, A., Duncan, G.A., 1986. "Distribution of metals in different size fractions of sediment from the Niagara River," Journal of Great Lakes Research, 12, 117-126. Namer, J., Ganczarczyk, J.J., 1993. "Settling properties of digested sludge particle aggregates," Water Research, 27, 1285-1294. Ongley, E.D., Krishnappan, B.G., Droppo, I.G., Rao, S.S., Maguire, R.J., 1992. "Cohesive sediment transport: emerging issues for toxic chemical management," Hydrobiologia, 236, 177, 187. Partheniades, E., 1993. "Turbulence, flocculation and cohesive sediment dynamics," Coastal and Estuarine Studies: Nearshore and Estuarine Cohesive Sediment Transport, 42, 40-59. Petticrew, E.L., 1996. "Sediment aggregation and transport in northern interior British Columbia streams," Erosion and Sediment Yield: Global and Regional Perspectives, IAHS Publ. 236 (ed. by D.E. Walling and B.W. Webb), 313-319. Petticrew, E.L., 1998. "Influence of aggregation on storage of fine-grained sediments in salmon-bearing streams," in: Forest-fish conference: land management practices affecting aquatic ecosystems. (tech. coords.: M.K. Brewin and D.M.A. Monita), Natural Resources Canada, Edmonton. NOR-X-356. Petticrew, E.L., Biickert, S.L., 1998. "Characterization of sediment transport and storage in the upstream portion of the Fraser River (British Columbia, Canada)," Modelling Soil Erosion, Sediment Transport and Closely Related Hydrological Processes, IAHS Publ. 249 (ed. by W. Summer, E. Klaghofer and W. Zhang), 383-391. Petticrew, E.L., Droppo, I.G., in review. "The morphology and settling characteristics of fine-grained sediment from a selection of Canadian rivers," 21 pp. 119 Phillips, J.M., Walling, D.E., 1995a. "An assessment of the effects of sample collection, storage, and resuspension on the representativeness of measurements of the effective particle size distribution of fluvial suspended sediment," Water Research, 29, 2498-2508. Phillips, J.M., Walling, D.E., 1995b. "Measurement in situ of the effective particlesize characteristics of fluvial sediment by means of a field-portable laser backscatter probe: some preliminary results," Marine and Freshwater Research, 46, 349-357. Phillips, J.M., Walling, D.E., in press. "The particle size characteristics of finegrained channel deposits in the River Exe basin, Devon, U.K.," 38 pp. Rao, S.S., Weng, J.H., Krishnappan, B.G., 1988. "Particle Associated Contaminant Transfer in the Yamaska River," NWRl Contribution 104,20 pp. Sudo, H., Yamada, A., Kokatsu, K., Nakamura, N., Matsunaga, T., 1997. "Development of a phosphate-removal system using a marine photosynthetic bacterium Chromium Sp.," Applied Microbiology and Biotechnology, 47, 78-82. Tambo, N., Watanabe, Y., 1979a. "Physical characteristics of floes-!: the floc density function and the aluminum floc," Water Research, 13,409-419. Tambo, N., Watanabe, Y., 1979b. "Physical characteristics of flocs-II: strength of floc," Water Research, 13, 421-427. Van Loosdrecht, M.C.M, Lyklema, J., Norde, W., Schraa, G., Zehnder, A.J.B., 1987a. "Electrophoretic mobility and hydrophobicity as a measure to predict the initial steps of bacterial adhesion," Applied and Environmental Microbiology, 53, 1898-1901. Van Loosdrecht, M.C.M, Norde, W., Schraa, G., Zehnder, A.J.B., 1987b. "The role of bacterial cell wall hydrophobicity in adhesion," Applied and Environmental Microbiology, 53, 1893-1897. Vine, J., 1996. Random Walk Model Applicable to Rivers. MSc thesis, University of British Columbia, Canada. 120 Appendix A Chemical Analysis of Northwood Pulp Mill Effluent provided on request by Northwood Inc. 121 RESULTS OF ANALYSIS - WATER File No. 2625C NPT Comp. Parameter Physical Tests pH Total Suspended Solids Volatile Suspended Solids 7.65 59 45 Dissolved Anions Chloride Fluoride 229 0.04 Nutrients Ammonia Nitrogen Total Kjeldahl Nitrogen Nitrate Nitrogen Nitrite Nitrogen Total Phosphorous Total Metals Aluminum Anitmony Arsenic Cadmium Chromium N N N N p 0.760 2 . 77 <0.005 <0.001 1. 41 1. 75 T-Al T-Sb T-As T-Cd T- Cr 0.0003 0.0015 <0.0002 <0.015 Cobalt Copper Lead Mercury Molybdenum T- Co T- Cu T-Pb T-Hg T-Mo <0 . 015 0 . 013 <0.001 <0 . 0001 <0 . 005 Nickel Selenium Titanium Vanadium Zinc T- Ni T- Se T- Ti T- V T-Zn <0 . 020 <0 . 0005 <0.030 <0.005 0.046 D-Na 470 Dissolved Metals Sodium < = Less than the detection limit indicated . Reults are expressed as milligrams per litre except for pH . Comp. = Composite .