Assessment of stormwater and snowmelt quality based on water management priorities and the consequent water quality parameters

(1)

Assessment of stormwater and snowmelt quality based on water management priorities

and the consequent water quality parameters

Helen Galfi

Department of Civil, Environmental and Natural Resources Engineering Division of Architecture and Water

ISSN 1402-1544 ISBN 978-91-7790-546-2 (print)

ISBN 978-91-7790-547-9 (pdf) Luleå University of Technology 2020

Helen Galfi Assessment of stormwater and snowmelt quality based on water management priorities and the consequent water quality parameter

Urban Water Engineering

(2)

quality based on water management priorities and the consequent water quality parameters

Helen Galfi

Luleå, 2020

Doctoral thesis

Division of Architecture and Water

Department of Civil, Environmental and Natural Resources Engineering SE-97187 Luleå University of Technology

Sweden

(3)

Luleå University of Technology 2020 www..ltu.se

(4)

ACKNOWLEDGEMENTS

The laboratory and field studies presented in this thesis were conducted in Luleå and Ös- tersund as activities of the research center of excellence Stormwater&Sewers (Dag&Nät) sponsored by the municipalities of Luleå, Skellefteå, Östersund, Boden and municipal water organisations of Umeå (Vakin) and Sundsvall (MittSverige Vatten & Avfall), and the Swed- ish Water and Wastewater Association (Svenskt Vatten). The snowmelt studies conducted in Luleå were supported by FORMAS, the Swedish Research Council for agricultural and spatial planning. The financial support from the Dag&Nät municipalities, Svenskt Vatten and FORMAS is gratefully acknowledged. Additional financial support for laboratory bac- teria analyses received from Åke och Greta Lissheds Stiftelse and Stiftelsen J. Gust. Richert foundation is also gratefully acknowledged. Finally, I would like to thank the Wallenbergss- tiftelsen – Jubileumslaget for supporting my attendance and presentation of research results at international conferences with peer-reviewed proceedings.

This work could not have been accomplished without the help of all staff at the Östersund municipality and Luleå University of Technology, that assisted me with sample collection and analyses. Special thanks go to my PhD student fellows in the Urban Water research group in Luleå and Jenny Haapala, Monica Sundelin and Lisbeth Köhler in Östersund, who per- sistently supported the field sampling and laboratory work during the entire data collection process. Thank you, Tony Johansson from MJK for the continuous help with the sampling devices in the field.

I am particularly thankful to my principal supervisor, Professor Maria Viklander, who offered me the chance to dive into an exciting study journey; thank you for your energy, knowledge sharing and for opening the opportunities for learning and managing my own projects!

Many thanks to Professor Jiri Marsalek, for his continuous advice, fruitful inputs and provision of assistance and supervision during the entire study process. It has been an honour to work under the supervision of two excellent researchers in the urban water field. I am also greatly indebted to my assistant supervisors, Godecke-Tobias Blecken and his successor Heléne Österlund, for providing advice and guidance during my studies. Special thanks go to my colleagues and co-authors Camilla Westerlund and Kerstin Nordqvist for their assistance, thoughtfulness and motivation during my work; you made this journey much more colourful with all your expertise.

I would like to thank Dmitrij for supporting me during this journey and engaging in chal- lenging discussions about science and life. Thank you, Sana Skandrani, for all your support with statistics, you opened up a new world to me and I hope to keep up our discussions through many years to come. Special thanks to my father, Bela Galfi, for your support with your technical expertise. Also, a big thanks to my colleagues in the City of Gothenburg for your continuous curiosity, cheerfulness and motivation to finish my thesis, I am glad to have you! I can´t thank enough my lovely friends in Luleå who cherished my life in Sweden from the very beginning, you all made this journey so much more special by sharing with me your spare time, joy and thoughts and being there and listening every time an experiment failed or every time I put too much pressure on myself.

Thank You!

Luleå, 2020

(5)

ABSTRACT

Stormwater and snowmelt pollution contributes to degradation of quality of the receiving waters. For assessing such impacts, it is effective to focus on specific causes of degradation, as done in this study of the quality of stormwater and snowmelt discharges into the receiving waters serving for supply of raw drinking water and water-based recreation. While the main priority were faecal indicator bacteria (FIBs), the understanding of their occurrence, and of other potential effects on the receiving waters, required addressing additional water quality parameters as well.

Exports of FIBs in stormwater and snowmelt discharged from four urban catchments yielded the following findings: (a) E.coli, with mean concentration of all stormwater data C_mean = 430 cfu (colony forming units)/100 mL, and enterococci (C_mean=1380 cfu/100 mL) were the best indicators of faecal pollution of stormwater, but total coliform (C_mean=3130 cfu/100 mL) and C. perfringens (C_mean=150 cfu/100 mL) were much less effective: the former indicator includes non-faecal bacteria and the latter one barely varied; (b) Among the different catchments, the central catchment with mixed land use produced the highest concentrations of FIBs; (c) FIB concentrations in snowmelt were significant only in the case of enterococci (400 cfu/100 mL); and, (d) Baseflows in two catchments were practically de- void of FIBs, with C_mean=10 cfu/100 mL for both E.coli and enterococci. Hence, there were no contributions of sanitary sewage to the storm sewer baseflows.

FIB concentrations varied with stormwater or snowmelt quality, described by associated parameters, which were identified by cluster analysis as: temperature, conductivity, TSS, flow rate, and TP. Such findings were used in statistical regressions indicating, that E. coli and enterococci could be statistically modelled in three of the four catchments, with deter- mination coefficients R² ranging from 38-66%. In spite of uncertainties, such modelling would be useful for future FIB monitoring, or for comparing remediation alternatives. Esti- mation of FIBs by microbial partitioning to settleable solids (represented by gully pot sediments) was infeasible, because these highly mineral sediments contained little FIBs.

Storm sewer outfall effluents were also analyzed for mineral (Al, Ca, Fe, K, Mg, Na) and anthropogenic indicator trace metal (TM) inorganics (Cd, Cr, Cu, Ni, Pb, Zn). The total mass of inorganics exported from the catchments by runoff or snowmelt was dominated by mineral inorganics, which were particularly high in baseflows. TM concentrations were compared to the tentative guidance limits suggested in Sweden as annual mean, or maximum event mean, total TM concentrations. Effluents from the catchments studied clearly exceeded the recommended values 5 times in the case of Zn.

Field studies drew attention to uncertainties in measured FIBs and solids. Automated sampling of greatly varying FIB concentrations was affected by sampling line water residuals, which can be minimized by short sampling lines and avoidance of sags in the sampling line. Stormwater and snowmelt solids were underestimated by the conventional TSS method requiring withdrawal of aliquots from total samples. This bias can be eliminated by using whole-sample methods; either the existing SSC (suspended sediment concentration) method, or the newly proposed (and easier to use) multiple filter procedure (MFP), filtering whole samples through progressively finer filters (pore sizes 25, 1.6 and 0.45 µm). The MFP produced data equivalent to those obtained with SSC, as confirmed by the Limits of Agreement (LoA) statistical procedure.

(6)

SAMMANFATTNING

Dagvatten från regn och snösmältning är viktiga komponenter i föroreningstransport till yt- vattentäkter som används för dricksvattenproduktion och badvatten. I denna studie har poten- tiella källor och transportvägar av mikrobiologiska föroreningar studerats i Östersund genom mätningar och utvärdering av fekala indikatorbakterier (FIB) i dagvatten som släpps ut i den närliggande Storsjön. Även andra föroreningar, så som metaller, ingick i studien för att bättre förstå källorna till och förhållanden mellan dessa parametrar och FIB.

Transport av FIB i dagvatten som släppts ut från fyra urbana avrinningsområden gav föl- jande resultat: (a) E. koli, med medelkoncentration av all data sammanlagt (C_mean = 430 cfu (kolonibildande enheter)/100 ml), och enterokocker (C_mean = 1380 cfu/100 ml), var de bästa fekala indikatorerna; (b) koliformer (C_mean = 3130 cfu/100 ml) och C. perfringens (C_mean = 150 cfu/100 ml) var mycket mindre effektiv som indikator; där koliformer inkluderar icke-fekala bakterier och C. perfringens uppvisade knappt någon variation mellan platser och provtag- ningstillfällen; (c) mellan de olika avrinningsområdena uppmättes de högsta halterna av FIB i ett centralt avrinningsområde med blandad markanvändning; (d) FIB-halter i snösmältning var endast signifikanta för enterokocker (400 = cfu/100 ml) och (e) i basflöden, vilket före- kom i två av avrinningsområdena, uppmättes nästan inga FIB. Följaktligen har dagvattnet i dessa områden inte påverkats av felkopplingar eller inläckage av spillvatten.

FIB-halter varierade i dagvatten från regn och snösmältning beroende av andra parametrar.

Hur dessa parametrar relaterade till FIB identifierades genom klusteranalys. Parametrarna var: temperatur, konduktivitet, suspenderade ämnen (TSS), flödeshastighet och fosfor. Dessa har vidare använts i regressionsanalys. E. koli och enterokocker kunde statistiskt model- leras i tre avrinningsområden med determinationskoefficienter, R², mellan 38-66%. Trots osäkerheter skulle sådan modellering vara användbar för framtida FIB-övervakning eller för att jämföra olika alternativ av åtgärder. Uppskattning av FIB genom provtagning av dagvat- tensediment i rännstensbrunnar lyckades inte, eftersom dessa mycket mineralhaltiga partiklar innehöll låga FIB-halter.

Dagvatten analyserades även för oorganiska ämnen såsom mineraler (Al, Ca, Fe, K, Mg, Na) och antropogena tungmetaller (Cd, Cr, Cu, Ni, Pb, Zn). Transport av oorganiska ämnen från avrinningsområden via dagvattenledningar dominerades av mineraler som uppvisade höga halter i basflöden. Tungmetallhalter jämfördes med riktvärden som föreslagits i Sverige. Dagvat- ten i respektive avrinningsområde överskred de rekommenderade värdena fem gånger för Zn.

Vid genomförda fältstudier påvisades osäkerheter i uppmätta FIB-halter och suspenderade ämnen. Automatiserad provtagning med kraftigt varierande FIB-halter påverkades av rester av vattnet i provtagningsslangarna, vilket kan minimeras genom kortare provtagningsslang utan böjar i provtagningslinjen. Vidare påvisades att mängden suspenderade ämnen i dagvatten underskattades med den konventionella TSS-metoden som innebär analys på delprov istället av helprov. Felkällor i uppskattning kan elimineras med hjälp av helprovsanalys;

antingen den befintliga SSC-metoden (suspenderad sedimentkoncentration) eller i denna studie föreslagen MFP (multipel filter procedure) som innebär filtrering av hela provet genom succesivt finare filter (porstorlekar 25, 1,6 och 0,45 µm). MFP genererade resultat som var ekvivalenta med de som erhölls med SSC, vilket bekräftas genom statistiska metoder.

(7)

(8)

LIST OF PAPERS

Paper I. Galfi, H., Österlund, H., Marsalek, J. and Viklander, M. (2016). Indicator bacteria and associated water quality constituents in stormwater and snowmelt from four urban catchments. Journal of Hydrology, 539: 125–140. DOI:10.1016/j.jhydrol.2016.05.006.

Paper II. Galfi, H., Österlund, H., Marsalek, J. and Viklander, M. (2017). Mineral and anthropogenic indicator inorganics in urban stormwater and snowmelt runoff: Sources and mobility patterns. Water, Air and Soil Pollution, 228:263. DOI: 10.1007/s11270-017-3438-x.

Paper III. Galfi, H., Haapala, J., Nordqvist, K., Westerlund, C., Blecken, G-T., Marsalek, J., Viklander, M. (2016). Inter-Event and Intra-Event Variations of Indicator Bacteria Concen- trations in the Storm Sewer System of the City of Östersund, Sweden. Journal of Environ- mental Engineering, 142(7): doi.org/10.1061/(ASCE)EE.1943-7870.0001067.

Paper IV. Nordqvist, K., Galfi, H., Marsalek, J., Westerlund, C., Viklander, M. (2014). Mea- suring solids concentrations in urban stormwater and snowmelt. Environmental Science:

Processes & Impacts, 16(9):2172-2183.

Paper V. Westerlund, C., Viklander, M., Nordqvist, K., Galfi, H. and Marsalek, J. (2011).

Particle Pathways during Urban Snowmelt and Mass Balance of Selected Pollutants. Proc.

12^th ICUD, Porto Alegre, Brazil, Sep.11-16, 2011.

Paper VI. Galfi, H., Nordqvist, K., Sundelin, M., Blecken, G-T., Marsalek, J. Viklander, M.

(2014). Comparison of Indicator Bacteria Concentrations Obtained by Automated and Man- ual Sampling of Urban Storm-Water Runoff. Water, Air and Soil Pollution, 225(9):2065.

Paper VII. Galfi, H., Österlund, H., Marsalek, J. and Viklander, M. (2018). Estimation of faecal indicator bacteria in stormwater by multiple regression modelling and microbial partitioning to solids. Proc. 11^th Int. Conf. on Urban Drainage Modelling (UDM 2018), Palermo, Italy, Sep. 23-26, 2018.

(11)

The author’s contribution to the scientific papers appended in the thesis is outlined below.

Paper no.

Development of idea

Research study design

Data collection

Data processing and analysis

Data interpretation

Publication process Manuscript preparation for submission

Responding to reviewers I Shared

responsibility

Shared

responsibility Responsible Responsible Shared

responsibility Responsible Shared responsibility II Shared

responsibility

Shared

responsibility Responsible Responsible Shared responsibility

Shared responsibility

Shared responsibility III Shared

responsibility

Shared

responsibility Responsible Responsible Shared responsibility

Shared responsibility IV No

Contribution No Contribution

No

Contribution Responsible Shared

responsibility Contributed Shared responsibility

V No

Contribution No Contribution

No

Contribution Contributed Contributed Contributed Contributed VI Shared

responsibility

Shared

responsibility Responsible Responsible Responsible Shared responsibility

Shared responsibility VII Shared

responsibility Responsible Responsible Responsible Shared responsibility

Shared responsibility

Responsible – developed, consulted (where needed) and implemented a plan for comple- tion of the task.

Shared responsibility – made essential contributions towards the task completion in collaboration with other members of the research team

Contributed – worked on some aspects of the task completion

No contribution – for valid reason, has not contributed to completing the task (e.g. joining the research project after the task completion)

(12)

1. INTRODUCTION

The risk of faecal contamination of recreational waters and the sources of drinking water by stormwater has been reported in a number of locations worldwide (Marsalek and Rochfort, 2004), including the City of Östersund in central Sweden (Widerström et al., 2014). Such risks are typically described by concentrations of faecal indicator bacteria (FIBs), which indicate the presence of pathogens. While the literature on FIBs in the temperate climate is relatively plentiful and generally deals with recreational waters in summer months (Ro- chelle-Newall et al., 2015; Ahmed et al., 2019), there is an obvious knowledge gap concerning FIBs in cool temperate climate, with lower temperatures and solar radiation levels, and differences in nutrient and organic matter availability. The cooler climate with low temperatures strongly affects FIB populations, and with respect to water supply, the interest in FIBs in source waters exists year round. To address this knowledge gap, FIBs in storm sewer effluents (i.e., stormwater, snowmelt, and baseflows) were studied. Besides FIBs, the study also addressed the associated water quality parameters, which acted as bacteria vectors, or agents affecting bacteria survival and reproduction, and as such, could be utilized in constructing statistical FIB models. Finally, heavy metal concentrations in storm sewer effluents were assessed against the available effluent quality guidelines, and potential bias in FIBs sampling and solids measurements was also addressed.

1.1. Aim and objectives

The overall aim of this thesis was to address the knowledge gap concerning FIBs in the cool temperate climate, with seasonal snow and low temperatures, by investigating the significance of urban storm drainage effluents (stormwater, snowmelt, and dry-weather baseflows) discharged into the receiving waters and by assessing the environmental significance of trace metals in such effluents. Specific objectives can be summarized as follows:

• Quantification and assessment of concentrations of the selected FIBs in stormwater, snowmelt and baseflow at drainage outfalls from four urban catchments with various land use draining into the receiving waters providing specific beneficial uses (Papers I, III, VI)

• Evaluation of indication performance of four common FIBs (Paper I)

• Identification of the associated water quality parameters, whose concentrations are related to those of FIBs, and examining the feasibility of using such parameters as explanatory variables in statistical modelling of FIB concentrations (Papers I , II, and VII)

• Examining the feasibility of using microbial partitioning to gully pot sediments for statistical modelling of E. coli and enterococci concentrations (Paper VII)

• Assessing the mineral and anthropogenic indicator inorganics in stormwater, snowmelt and baseflow with respect to sources, mobility patterns and potential effects on the receiving waters (Paper II, V), and

• Developing practical procedures for reducing bias in automated sampling of FIBs and the measurement of solids (total suspended solids, TSS) in stormwater and snowmelt samples (Papers IV and VI).

(13)

1.2. Thesis structure

The thesis is based on seven research papers, which are appended and numbered as Paper I-VII. Linkages among the paper topics and the physical environment, comprising the path- way between the catchment → storm sewer system → receiving waters, are shown in Fig. 1.

In essence, the main findings (Results) focus on the quality of urban drainage effluents discharged into the receiving waters (Papers I, II, III, VII), pollutant pathways during snowmelt (Paper V), and the remaining two papers (IV and VI) deal with bias in FIBs sampling and measurements of solids.

The thesis comprises seven chapters: In Chapter 1, Introduction, the knowledge gaps are identified with respect to FIBs and trace metals in stormwater and snowmelt in cool temperate climate, and the thesis aim and objectives are defined. Chapter 2, Background, presents the findings of a literature review indicating that while there is a fair number of references on FIBs in stormwater during the summer months in temperate climate, there is a lack of references on year-round occurrences of FIBs in cool temperate climate and on the related water quality parameters and their reliable measurement. Chapter 3, Methods, presents the research methods applied throughout the thesis project and encompasses the measurement of hydro-meteorological data (precipitation, air temperatures, sewer flow rates); the sampling of stormwater, snowmelt, storm sewer baseflows, and gully pot sediment; analyses of such samples for FIBs and physico-chemical parameters; and, statistical methods including basic statistical parameters, principal component analysis, cluster analysis, and the limits of agreement method. Chapter 4, Results, presents the main research findings including the characterization of stormwater, snowmelt and baseflows; transport patterns of runoff, FIBs, and TSS;

uncertainties in measurements of solids (total suspended solids, TSS) by the methods using partial aliquots withdrawn from whole samples; identification of parameters associated with FIB concentrations by PCA and cluster analysis; and, comparisons of trace metal concentrations with the available guideline limits for stormwater. Chapter 5, Discussion, provides the discussion of results in the context of published research, with attention paid to occurrence and transport of stormwater and snowmelt pollutants or contaminants (FIBs), estimation of bias in conventional FIBs and solids measurements, estimation of FIB concentrations by statistical modelling, and suggestions for improved stormwater sewer flow quality monitoring and impact assessment. Chapter 6, Conclusions, presents a list of the main research conclusions concerning FIBs occurrence in stormwater and snowmelt in urban catchments, improved FIB and solids measurement, identification of the parameters associated with FIBs, occurrences and transport of trace metals in stormwater and snowmelt, and their potential impacts on the receiving waters. Finally, Chapter 7 presents the references cited in the thesis.

(14)

Fig. 1. Schematic depiction of the topics studied

(15)

(16)

2. BACKGROUND

Advancing urbanization profoundly transforms the hydrological cycle of urban areas into the urban water cycle, comprising such major components as water supply, urban drainage, wastewater management, and the receiving water systems. These components are intercon- nected by numerous links (Marsalek et al., 2008) and therefore, require integrated sustainable management, which is also referred to as the total management of the urban water cycle.

In recent decades, numerous studies of such management focused on the nexus between water supply and sustainable stormwater management, with stormwater facilities serving as a source of subpotable water for urban waterscape, aquatic habitat provision, and the protection of downstream waters against pollution. Hence, even though the original interest in stormwater management was more or less limited to flood protection, the current objectives of stormwater management are much broader and include maintenance of water balance, prevention of geomorphic changes in watercourses, avoidance of increased flood risk potential, protection of water quality, and maintenance of appropriate biodiversity and opportunities for beneficial water uses (MOE, 2003). Such objectives reflect the strong interdependence between stormwater quantity and quality, and furthermore, in the current context, they need to be pursued within the frameworks of stormwater management sustainability, flexibility, and resiliency in a changing climate (Jabareen, 2013). Concerning the stormwater quality, which is formed and evolves as rain/stormwater moves through the urban landscape, it represents a fundamental issue in one half of the preceding objectives and can be generally defined as the chemical, microbiological and thermal energy characterization of stormwater and snowmelt.

A variety of natural events and anthropogenic activities, and conveyance of stormwater on both natural and man-made surfaces in urban landscapes, contribute to the development and degradation of stormwater quality, mostly by atmospheric deposition, wash-off of materials and pollutants accumulated on, or constituting, drainage surfaces, heat transfer from urban surfaces, and direct inputs from other urban sources of pollution, including the wastewater. Thus, stormwater pollutants originate from both anthropogenic and non-anthropogenic sources. The former ones include atmospheric deposition, traffic, industry, construction, infrastructure maintenance, household and garden waste, and releases from drainage surfaces (including building materials), the latter ones include catchment soils, groundwater, and wildlife (Edge et al., 2018). While in most situations the anthropogenic sources dominate stormwater pollution, non-anthropogenic (natural) sources may significantly contribute to drainage effluent pollution through releases of faecal microorganisms found in soils, manure and on the vegetation (Gagliardi and Kams, 2002), as well as in bird droppings on drainage surfaces and urban beaches (Edge et al., 2018). Another example of natural sources of pollutants may be soils, particularly the peat soils in the boreal regions, which are known to leach out heavy metals including Pb, Ni, Cu and Cr (Cory et al., 2006; Lidman et al., 2014), though at levels exerting small effects on water quality.

(17)

The pollutants entrained by stormwater are transported through the drainage system, which may include best management practices (BMP) or low impact development (LID) measures for quantity and quality control. Some fractions of pollutant burdens are immobilized in BMPs and LIDs or catchment soils, but the rest is discharged into the receiving waters.

Polluted stormwater discharges may cause acute or chronic impacts, on both human health (e.g., in recreational waters, or through contamination of water supplies) and aquatic life (particularly the persistent pollutants contributing to chronic toxicity). Heuristic studies of stormwater and snowmelt identified large numbers (hundreds) of chemicals and microbiological organisms in stormwater, however, the majority of the reported parameter concentrations may be environmentally insignificant, and their studies are constrained by high costs of data collection and analyses. More appropriately, the process of planning the environmental protection of urban waters should start with identifying the local impacts and their probable causes, and then focus on remediation. Thus, the studies of stormwater quality should focus on local environmental priorities and the underlying specific parameters, or groups of parameters of concern identified in urban water management planning (Marsalek, 2013).

The remaining presentation in the background section focuses on stormwater and snowmelt quality concerns in the cool temperate climate, including: distinctive characteristics of stormwater/snowmelt runoff; pollutants of concern (FIBs, heavy metals, solids); FIBs in stormwater, snowmelt and baseflow; water quality parameters associated with FIBs; heavy metals in stormwater and snowmelt; solids in stormwater and snowmelt; and bias in sampling FIBs and analysis of TSS in stormwater and snowmelt; and, knowledge gaps.

2.1. Distinctive characteristics of stormwater runoff in cool temperate climate

In cool temperate climate, urban snow accumulates in seasonal snowpacks (Silanpää and Koivusalo, 2013), which are further reshaped in urban areas by ploughing and eventual removal to local, or central, or remote snow disposal sites (Reinosdotter et al., 2006). Ur- ban pollutants and solids from urban surfaces accumulate in snowpacks over winter months (Malmqvist, 1978; Kuoppamäki et al., 2014; Moghadas et al., 2015; Vijayan et al., 2019) and are released and transported during snowmelt runoff, representing a significant part of the annual stormwater runoff. Hence, during snowmelt runoff, relatively large amounts of solids and chemicals are released during short periods into the receiving environment (Westerlund et al., 2003), compared to the stormwater runoff caused by rainfall. Snowmelt solids follow a typical pattern of high build-up during the winter season (up to six months in northern regions) followed by a much shorter period of wash-off during snowmelt. The amount of solids and chemicals accumulated during winter is affected by winter road maintenance, including grit and salt applications (Reinosdotter et al., 2006) on trafficked surfaces. Furthermore, such solids contain pollutants from vehicle exhausts, brake lining and tires wear (Malmqvist, 1983). The solids released from snowpacks are mostly large particles that can accumulate and clog storm sewers, because of low flow velocities and frozen pipes occurring during

(18)

snowmelt periods (Oberts, 2003; Lau and Stenstrom, 2005; Karlsson et al., 2010). However, recent studies showed that grit applied on wintery roads as a traction agent may be ground by vehicle tires into fine particles (VanDuijn et al., 2008). In the same context, trace metals and other pollutants adsorbed to solids and accumulated in the snowpack (Moghadas et al., 2015) are suddenly released during snowmelt into the receiving environment, with a risk of toxic impacts (Schillinger and Gannon, 1985; Hvitved-Jacobsen and Yousef, 1991; Viklander, 1998; Marsalek et al., 2008). Thus, a good quantification of solids sources, fluxes and sinks is a key factor for reliable design and maintenance of stormwater management systems as well as for assessing the impact of stormwater and snowmelt runoff on the receiving waters.

2.2. Municipal environmental priorities in the cool temperate climate

Consideration of municipal environmental priorities was guided by an example of a mid-size municipality in central Sweden, which is served by a separate storm sewer system, draining into the receiving waters used for recreation and water supply. In the absence of specific industrial pollution sources, the priorities include the protection of the recreational waters and drinking water sources, and aquatic biota. The pollutants of concern include FIBs, heavy metals originating mostly from traffic, and a carrier of both pollutants – solids.

2.2.1. FIBs

Discharges of urban effluents from storm sewers, including stormwater, snowmelt and baseflows, may contribute to faecal contamination of receiving waters and, thereby, cause serious impacts on human health, where such waters are used for recreation or drinking water supply.

Consequently, the concerns about the microbiological pollution of stormwater discharges were already reported in the early studies of stormwater quality (US EPA, 1983) and moti- vated numerous research studies of such pollution reviewed e.g. in Marsalek and Rochfort (2004), Rochelle-Newall et al. (2015), and Ahmed et al. (2019).

Even though the primary concerns are caused by pathogenic microorganisms, including bacteria, viruses, parasites and protozoa, their species and sources are too numerous and costly to investigate and, consequently, the assessment of the risk of faecal contamination of open waters is conventionally done by using faecal indicator bacteria (FIBs) (Ortega et al., 2009; Metcalf and Eddy, 2013). Typically, multiple FIB species are used in this assessment, because different types of FIBs possess various environmental survival characteristics (McFeters et al., 1972; McFeters et al., 1974) and can be used as surrogates of pathogens in stormwater monitoring to estimate the risk of pathogens transport into the receiving waters with stormwater runoff (Gerba, 2000). The most common FIB types used for assessing the microbiological quality of drinking water and of recreational waters include total coliforms, E. coli, enterococci and C. perfringens (EC, 1998; EC, 2000). In sanitary surveys of urban catchments, the assessment typically starts at storm sewer outfalls and proceeds in the upstream direction, if individual sources of FIBs need to be identified (Panasiuk et al., 2015).

(19)

Common sources of faecal microorganisms in stormwater include: (a) animal faeces and droppings (domestic pets, urban wildlife and particularly birds (Edge et al., 2018)), (b) wash- off of catchment surfaces including green surfaces; (c) solid waste collection; (d) lack of san- itation (Marsalek and Rochfort, 2004); urban farming and gardening (Desai and Rifai, 2010;

Tiefenthaler et al., 2011); and, urban infrastructure deficiencies, such as cross-connections between storm and sanitary sewers, accumulation of sediments in gully pots and sewers, and growth of bacteria in nutrient-rich standing waters in storm sewer systems (Olyphant et al., 2003; Jeng et al., 2005a; Coulliette and Noble, 2008; Rowny and Stewart, 2012; Panasiuk et al., 2015). Among the above sources, the most serious ones are cross-connections of storm sewers with sanitary sewers and misconnections of household wastewater to storm sewers (Ellis and Butler, 2015; Panasiuk et al., 2015), which may cause highly microbiologically polluted dry weather flows (also referred to as ‘baseflows’) in storm drains. Such flows require a closer scrutiny to ascertain, whether they contain sanitary sewage, which would cause public health risks.

In a spatially integrated approach to identifying stormwater pollution sources, the sources may be considered as associated with certain catchment characteristics, such as land use, the degree of development (Selvakumar and Borst, 2006), and anthropogenic activities, as well as the local climate, including air temperatures, humidity, solar radiation and the rainfall regime (Brezonik and Stadelmann, 2002; Ghafouri and Swain, 2005; Tiefenthaler et al., 2011; Rochelle-Newall et al., 2015). Stormwater from low-imperviousness (green) catchments may carry elevated concentrations of total coliform bacteria originating from the ambient environment (soils and vegetation), pets and wildlife faeces/droppings, and recreational activities (Desai and Rifai, 2010; Tiefenthaler et al., 2011). The survival of such bacteria may be supported by the strong presence of solids and nutrients in stormwater (Jeng et al., 2005b; Chudoba et al., 2013). On the other hand, stormwater runoff from more developed catchments yields higher bacteria concentrations attributed to the presence of debris (solid waste), human activities, and animal faeces (Desai and Rifai, 2010; Hathaway et al., 2010).

Thus, sources of the microbiological pollution in urban areas are ubiquitous and dispersed throughout the catchment.

The strength of microbiological pollution of stormwater discharges is affected by the characteristics of the causative storm events, which may activate the microorganism sources in wet-weather by wash-off and scouring of surface deposits, erosion of soils, and resuspension of solids in gully pots and storm sewers (Marsalek and Rochfort, 2004). Consequently, the historical contamination episodes are often associated with specific actual storms (Auld et al., 2004). Such contamination incidents were reported in a number of countries, including Sweden, where microbiological contamination episodes contributed to the impairment of drinking water supplies in several municipalities in 2010 and 2011 (Hrudey and Hrudey, 2014; Widerström et al., 2014). The search for causes of such contamination events motivat- ed studies of microbiological quality of urban stormwater discharged into the water bodies of concern. These studies can also serve for assessing the state of the urban sewer infrastructure, with respect to the prevention of entry of wastewater into separate storm sewer systems (Panasiuk et al., 2015).

(20)

Climatic conditions, including air temperatures, humidity and solar radiation, and differences in nutrient and organic matter availability, also affect FIB populations and inactivation (Noble et al., 2004; Rochelle-Newall, 2015), and change the dynamics of FIB populations. In warmer climates, the FIBs dynamics is different and higher temperatures result in higher FIB populations (Ahmed et al., 2019). This was documented for urban catchments in the subtrop- ical climate by McCarthy et al. (2012), who reported that the sources, growth, and die-off of FIBs were more important determinants of FIB levels than wash-off and transport processes during wet weather, since the most influential factors for event mean concentrations (EMC) of E. coli were antecedent catchment conditions and nutrient levels in stormwater. Converse- ly, one would expect reduced FIB populations in cool temperate climate, as indicated by seasonal FIB data.

Seasonal variations of specific FIBs in stormwater and particularly their dependence on air and water temperatures were reported by numerous authors (e.g., US EPA, 1983; Selvaku- mar and Borst, 2006), who reported that mean seasonal concentrations of total coliforms, E.

coli, and enterococci in stormwater were the lowest in winter, in both the temperate and cool tempered climates. The only exception to these findings might be C. perfringens, which are known for their persistent survival in the environment (Medema et al., 1997). Thus, it follows from temperature considerations that FIB concentrations in snowmelt, if detected, should be much lower than those in stormwater produced by rain events.

Differences in FIB concentrations in stormwater and snowmelt were of particular interest in this thesis project, because the study area (the City of Östersund, Sweden) is located in the subarctic region above the 60° latitude (i.e., in the cool temperate climate), where about 40%

of the total annual precipitation occurs in the form of snow (Bergström, 1993) and seasonal snowpacks last 4 to 6 months during the cold periods of the year. Thus, snowmelt contributes high percentage of the annual runoff and with respect to quality, differs from stormwater generated by rainfall. In particular, snowmelt runoff may carry elevated amounts of solids released during short periods of time (Westerlund et al., 2003) and those solids contain such adsorbed contaminants as heavy metals, bacteria, and PAHs (polycyclic aromatic hydrocar- bons) from anthropogenic activities (Schillinger and Gannon, 1985; Hvitved-Jacobsen and Yousef, 1991; Viklander, 1998; Marsalek et al., 2008). However, large quantities of mineral-origin solids on the catchment surface, their scouring by snowmelt runoff, and low air/

runoff temperatures are likely to produce different relationships between FIBs, solids and nutrients than those typical for rainfall runoff during warmer conditions (Bogosian et al., 1996). Reduced bacteria survival at low temperatures contributes to total coliform, E. coli and enterococci levels significantly lower in snowmelt than in rainfall generated stormwater.

The delivery of FIBs from the sources to sewer inlets and eventually to sewer outfalls is facilitated by surface runoff and its transport in sewers, which is commonly driven by rainfall or snowmelt. Thus, the microbiological pollution in urban catchments depends not only on catchment characteristics, but also on the local climate with respect to the precipitation regime (mostly transport issues), air and water temperature, and solar radiation affecting FIB

(21)

populations and their inactivation and survival (Noble et al., 2004). Occurrence and transport of FIBs in storm sewers is a highly dynamic complex process characterized by greatly varying bacteria concentrations resulting from time-varying contributions of numerous sources, transport of bacteria partitioned to solids, and their survival dynamics including die-off or regrowth in the water phase (Gonzales et al., 1992; Marsalek and Rochfort, 2004). Trans- port and survival of FIBs may be further affected by specific stormwater quality parameters, which may serve as transport vectors (e.g., solids in general, or total suspended solids, TSS), synergistic parameters (e.g., solids and nutrients), or antagonistic factors, which were hy- pothesized here as the parameters shown by statistical analysis to contribute to reduced FIB counts. For brevity, the stormwater/snowmelt quality parameters potentially affecting FIB populations are called here as the “associated parameters”. Besides their use in interpretation of FIBs data, the associated parameters should be helpful in planning future data collection and developing statistical models of FIB concentrations.

2.2.2. Parameters associated with FIBs

For documenting the relationships between FIBs and the associated parameters, one can cite the examples of TSS and nutrients: FIBs transported with suspended solids survive longer than those in the free water phase, because of their increased protection against ambient conditions (including solar irradiation) and the supply of nutrients associated with certain types of solids (Davies et al., 1995; Desmarais et al., 2002; Jeng et al., 2005b; Hathaway et al., 2010; Gao et al., 2011; Chudoba et al., 2013). Several studies have shown that FIBs and TSS concentrations increase in the receiving waters during and after stormwater discharges, when wet-weather FIB concentrations are considerably higher than during dry periods (Gannon and Busse, 1989; Dutka and Marsalek, 1993; Jeng et al., 2005, Salmore et al., 2006; Coul- liette and Noble, 2008). Furthermore, significant correlations were found between indicator microorganisms and TSS loads during wet-weather flows (Olyphant et al., 2003; Jeng et al., 2005b; Coulliette and Noble, 2008). Hence, there is a great variation in indicator bacteria concentrations in stormwater reported in numerous studies, review papers and stormwater quality databases (e.g., NURP, 1983; Duncan, 1999; Marsalek and Rochfort, 2004; Ahmed et al., 2019) and many authors attempted to explain this variation by catchment characteristics, rainfall/runoff processes, stormwater quality, and climate characteristics (particularly air and water temperatures, and solar radiation) (Noble et al., 2004; Desai and Rifai, 2010; Hathaway et al., 2010; Tiefenthaler et al., 2011; McCarthy et al., 2012).

The associated parameters can be identified in two ways: by adopting recommendations from the literature (Selvakumar and Borst, 2006; Jeng et al., 2005b; Hathaway et al., 2011;

Chudoba et al., 2013), or by statistical analysis of collected data (Ortega et al., 2009). In the latter case, the associated parameters are considered as explanatory variables contributing to explanation of FIB concentrations in stormwater and snowmelt. In sanitary surveys of catchments, both approaches are combined, starting with the data from the literature and expanding such a database for the data from the survey. One of the first steps is to look for microbial partitioning of FIBs to solids, where both TSS and settleable solids were used in

(22)

such an analysis (Jeng et al., 2005b; Hathaway et al., 2011; Chudoba et al., 2013). This search is implemented by applying PCA (principal component analysis) and cluster analysis to the FIB and related data, assuming, that the clustered parameters have the same source of variation (Jolliffe, 2002). Other associated parameters reported in the literature included water temperature, nutrients, and electrical conductivity (identified as an antagonistic parameter) (Noble et al., 2004; Jeng et al., 2005b; Chudoba et al., 2013). Electrical conductivity mir- rors dissolved ion concentrations (including road salts, minerals and trace metals) and may contribute to die-off of some FIBs with lower resistance to salt (Peters et al., 1991). Other associated (explanatory) parameters (biochemical oxygen demand, BOD, and chemical oxygen demand, COD) were recommended in a study in a warmer climate of South Korea (Paule-Mercado et al., 2016). Thus, for interpretation of the observed data, it is useful to expand the surveys of FIBs in stormwater for monitoring some additional physico-chemical parameters, including water temperature, TSS, conductivity, nutrients, and possibly major inorganic ions, which contribute to water conductivity. Furthermore, the interaction between FIBs, solids, water temperature, nutrients, and dissolved ion content was rarely studied in stormwater and snowmelt, especially in regions with the cool temperate climate, where certain stormwater quality parameters might support the FIB survival, and would be of interest in interpreting and extrapolating the FIBs data.

2.2.3. Trace metals

Trace metals in urban stormwater and snowmelt originate from both anthropogenic and natural sources (Sillanpää and Koivusalo, 2013; Kuoppamäki et al., 2014). As sources of metals, the anthropogenic activities dominate and, depending on local conditions, their strength can be tentatively ranked as follows: (a) traffic (Kayhanian et al., 2012; Huber et al., 2016), (b) wash-off or corrosion of building and structure envelopes (Sörme et al., 2001; Fuchs, 2006;

Petrucci et al., 2014; Huber et al., 2016), (c) atmospheric deposition (Gunawardena et al., 2013), (d) industrial releases, and (e) impurities in road salts and grit applied on roads in winter maintenance (Westerlund et al., 2003). This ranking is supported by Fuchs data (2006), indicating that the top two sources (traffic and building envelope washoff) contribute about two third of the trace metal burden in urban stormwater. Natural sources are generally small and include crustal leaching (Joshi and Balasubramanian, 2010) and particularly the leaching of Pb, Ni, Cu and Cr from peat soils (Cory et al., 2006; Lidman et al., 2014).

In view of the dominance of anthropogenic sources, the associated metals are also referred to as “anthropogenic indicators”. Among such indicators, the metals associated with traffic are particularly of environmental concern and generally include the following group of elements: Cd, Cr, Cu, Ni, Pb and Zn (Andradottir and Vollertsen, 2014). Individual elements and their association with specific-traffic related sources were identified as follows (Huber et al., 2016): Cd - tires and brakes (rotors), Cr - automobile bodies, Cu - brakes and tires, Ni - automobile bodies, and Zn - frame/body, brakes and tires. As cars and trucks evolve, the use of various materials is also changing and the above sources may need to be periodically reassessed. While the metal concentrations in stormwater are generally reported as “whole

(23)

sample” concentrations, for toxicity assessments or design of control measures, the concentrations of dissolved metals, or truly dissolved metals (Vijayan et al., 2019), which are more bioavailable, are of special interest. Evidence from field and lab studies of highway runoff indicates that the traffic related metals may occur in highway runoff samples at toxic levels, depending on traffic intensity (Marsalek et al., 1999; Bartlett et al., 2012). Consequently, acute toxicity is most frequently detected in runoff from freeways with AADT exceeding 100,000 (Marsalek et al., 1999).

In cool temperate climate, runoff quality varies seasonally, with the highest pollutant concentrations found in winter runoff and snowmelt (Oberts et al., 2000). This is caused by two factors: (a) higher releases of pollutants in cities during winter months (Hokerby and Malmqvist 1977), and (b) accumulation of pollutants in urban snowpack and snow deposits over extended periods of several months (Moghadas et al., 2015; Vijayan et al., 2019).

Hence, total metal concentrations in urban snowmelt generally greatly exceed those in rain runoff, and were observed in snow along two roads with annual average daily traffic (AADT)

= 15,000 – 20,000 as high as C_Cd = 4.6, C_Cu= 621, C_Pb = 567, and C_Zn = 1390 µg/L (Mogh- adas et al., 2015). In an ideal case, the pollutants would be released from melting snow by the preferential elution of dissolved pollutants (e.g., chloride, major ions, dissolved metals), with solids and hydrophobic chemicals either released during the final melt with fine solids (Schöndorf and Herrmann, 1987), or remaining on site as a residue (Viklander, 1997). The preferential elution magnifies concentrations of dissolved chemicals released early during the snowmelt process, compared to the average concentrations in snow. However, as shown by Westerlund (2007), the ideal case of preferential elution is strongly modified in urban areas, where the snowpack is highly disturbed by snow handling and the melt is impacted by anthropogenic activities (snow ploughing and removal, road salt applications, atmospheric deposition reducing snow albedo). Furthermore, salt applications change metal partitioning by enhancing the dissolved phase (Reinosdotter and Viklander, 2007). Nevertheless, in cool temperate climate and urban areas with AADT in the range of low tens of thousands of vehi- cles per day, there is an environmental concern about the traffic related metals and their potential impacts on the receiving waters quality in the form of chronic rather than acute effects.

2.2.4. Environmental assessment of the observed FIB and trace metal concentrations: Effluent quality guidelines

For assessing compliance with drinking water and recreational waters microbiological qual- ity, the two widely used FIBs are E. coli and enterococci, which may be sometimes comple- mented by total coliforms and C. perfringens (EC, 1998; EC, 2006). The reason for using multiple indicators follows from the fact that individual indicators belonging to different taxa exhibit various environmental survival and ability to cope with ambient conditions.

Thus, different indicators may perform well in specific conditions and serve well in stormwater monitoring and estimation of the risk of pathogens discharge into the receiving waters (Gerba, 2000). Where such discharges do occur, the resulting contamination of receiving waters used for drinking water supply and/or recreation can pose a direct threat to human health (Marsalek and Rochfort, 2004), as reported in Sweden in recent years (Widerström et

(24)

al., 2014). In the absence of Swedish guidelines for the assessment of FIB levels in storm sewer effluents, the observed levels can be assessed only qualitatively by comparison to the literature values reported elsewhere for similar climates. A summary of data in Mar- salek and Rochfort (2004) suggested that E.coli concentrations in runoff from smaller urban catchments without sanitary sewer cross-connections varied from 10³ to 10⁴ E. coli/100 mL, with the lower values corresponding to small residential catchments and the higher values corresponding to larger combined land use areas during warm seasons of the year. Lower counts would be expected in cool seasons (US EPA, 1983). Furthermore, concentrations of FIBs in stormwater discharged into the receiving waters are reduced by mixing and dilution, as would be the case in the study location in the City of Östersund, where the discharge of polluted stormwater into the near-shore zone of Lake Storsjön would be subject to mixing and dilution caused by longshore current and general water circulation in the lake.

The trace metals recognized as anthropogenic indicators (Cd, Cr, Cu, Ni, Pb and Zn) are frequently addressed in monitoring the environmental quality of freshwaters, particularly when tracking the priority pollutants, which may be toxic to the aquatic life (US EPA, 1993;

Eriksson et al., 2007; Singh et al., 2013). Therefore, a good understanding of such pollutants and their monitoring in stormwater and snowmelt runoff discharged to freshwaters is needed for a sustainable urban watershed management. A number of guidance documents exist for assessing the concentrations of anthropogenic indicator metals in stormwater discharged into the receiving waters, including the EC environmental quality standards for priority pollutants (EC, 2013) and two effluent guidelines for trace metal concentrations in stormwater discharges into Swedish receiving waters proposed by Alm et al. (2010) in a report to the Swedish Water and Wastewater Association, and a similar guideline developed by the City of Gothenburg (City of Gothenburg, 2013). Among these documents, the first deals with ambient water quality data, and the last two represent stormwater effluent data and, therefore, are directly applicable to the studies monitoring stormwater quality at the storm sewer outfalls.

Both sets of effluent guidance data, by Alm et al. (2010) and the City of Gothenburg (2013), are listed in Table 1.

Table 1. Recommended stormwater effluent quality

Trace metal µg/L Alm et al. (2010)¹ City of Gothenburg (2013)²

Cd _Tot 0.45 0.40

Cu _Tot 30 10

Ni _Tot 20 40

Pb _Tot 10 14

Zn _Tot 90 30

1 Annual mean effluent concentration

2 Maximum runoff event mean concentration

(25)

The guidance data in Table 1 provide a tool for assessing stormwater quality with respect to trace metals. As annual average, or maximum event concentrations, they could not identify, and do not apply to, acute toxicity episodes, which are unlikely to occur in smaller cities of central and northern Sweden, with relatively low volumes of traffic. However, the guideline concentrations are useful for making informed decisions concerning the need to address discharges of polluted stormwater into the receiving waters.

2.3. Bias in estimation of TSS and FIBs by automated sampling

Solids are one of the most important parameters in the assessment of stormwater and snowmelt quality, because they have an impact on both the operation of drainage systems and the water quality in the receiving waters. Suspended solids (TSS) are measured in the laboratory in one-litre samples, within the size range restricted to 1.6 µm ≤ D ≤ 2 mm (APHA, 2005). The importance of accurate, precise and standardized measurements of solids in stormwater or snowmelt cannot be overestimated, particularly in connection with the transport of FIBs and trace metals. Although TSS are the most frequently studied stormwater quality parameter, their sampling and measurement may be biased for a number of reasons, particularly when studying stormwater runoff with disproportionally large amounts of relatively coarse solids, as typical for snowmelt (Westerlund, 2007). The sources of bias include both the automated sampling method and the withdrawal of aliquots for measuring solids in the laboratory. The former issue follows from the lack of guidance for designing stormwater sampling programs with automatic samplers (Harmel et al., 2003), including the acceptable lift height (Clark and Siu, 2008; Degroot and Gulliver, 2010), the placement and orientation of the sampling line intake in the sampled flow, and the intake line velocity (Marsalek, 1978). Sampler intake should be pointing upstream, other- wise it under-samples suspended solids, and ideally, the intake velocity should be equal to the incoming flow velocity (Marsalek, 1978). In the sampling line, the particle fall velocity reduces the lift of heavier sand-like particles (D> 0.1 mm) into the sampler, compared to the lighter particles, and hence the collected sample is biased towards finer particles, with underrepresentation of heavier particles (Marsalek, 1978; Guo, 2007; Roseen et al., 2011). The preceding authors attributed the TSS sampling method errors to the lift height, sampler intake location, and non-uniform distribution of solids (and attached pollutants) in the water column with density stratification. With the current automated samplers, these issues are unavoidable and where coarser stormwater particles are of great importance, other methods of sampling, e.g., manual sampling, may need to be implemented. Another source of bias, in the measurement of TSS, arises during the preparation of samples for laboratory analysis and is caused by the underrepresentation of heavier particles in the laboratory withdrawals of sub-sample aliquots (from the typical sample volume of 1 L) for the conventional TSS analysis (Clark and Siu, 2008). Similarly, as in the field, such withdrawals under-sample coarser solids in aliquots from whole samples, and lead to a negative bias in measured TSS concentrations.

(26)

Another method available for measuring solids in water-solids mixtures is known as the suspended sediment concentration (SSC) method (ASTM, 2007), in which all the sediment material is extracted from large volume samples regardless of the particle sizes. However, the sediment extraction is time consuming, particularly when the samples are voluminous.

The choice of sampling techniques for collecting FIB samples, i.e., manual versus automated techniques (Ferguson, 1994), may also cause bias in FIB samples, not reported in the earlier literature. Automated sample collection prevails in the current practice for operational reasons (i.e., low costs, field personnel safety), but contributes to increased uncertainties in the measured FIB concentrations, because of transfer of water residuals in the sampling equipment between consecutive samples and the resulting cross-contamination. Depending on FIB concentration variation, the sampled concentrations can be biased by the residual from the previous sample.

2.4. Knowledge gaps

The findings in the Background section can be briefly summarized as follows: (a) There is a knowledge gap on FIBs populations in stormwater and snowmelt, and the related choices of faecal bacteria indicators in cool temperate climates with seasonal snow and low air temperatures; such data are of particular interest where urban storm sewer effluents discharge into the receiving waters serving for water supply and recreation; (b) For FIB modelling and the planning of stormwater controls, the water quality parameters associated with FIBs are of primary interest and should be monitored as well (particularly temperature, solids and nutrients); (c) Another group of parameters of interest are metals as anthropogenic effects indicators, particularly those related to traffic; and, (d) in field studies of FIBs and trace metals, potential bias caused by automatic sampling of stormwater and snowmelt for FIBs, and by under-sampling of coarser solids in the field, as well as in the laboratory withdrawals of sample aliquots in the standard TSS method, should be addressed and mitigated.

(27)

(28)

3. METHODS

The chapter on methods starts with the selection of stormwater quality parameters of interest, followed by descriptions of four urban catchments selected as study areas, field monitoring and sampling sites in the study areas, laboratory apparatus for investigating FIBs and solids measurement methods under controlled conditions, analytical methods used for analyses of field and laboratory samples, and finally the statistical methods applied in data analysis and interpretation. Details follow.

3.1. Selected stormwater quality parameters

The studied stormwater quality parameters can be divided into four groups: (i) faecal indicator bacteria (FIBs), (ii) physico-chemical parameters identified in the literature as co-varying with FIBs (associated parameters), (iii) mineral (geogenic) inorganics dominating the baseflow chemistry, and (iv) trace metals serving as indicators of anthropogenic pollution.

For study of microbiological quality of stormwater, four types of FIBs were selected from those commonly used in guidelines and regulations for protection of recreational waters and the safety of drinking water: total coliforms, E. coli, enterococci, and C. perfringens (EC, 1998; EC, 2006). The use of multiple indicators, known for different survival dynamics in the environment (Anderson et al., 2005), served for obtaining a more robust assessment of the microbiological water quality. Among these four FIBs, total coliforms represent an older type of FIB and are ubiquitous in the natural environment. While they occur in human faeces, they may also originate from other sources, such as soils and animal manure. For that reason, they are no longer recommended for the risk assessment of recreational waters but are used in water supply as an indicator of contamination by outside sources. E. coli is specific to faecal matter from humans and other warm-blooded animals, and is used in risk assessment of recreational freshwaters. Enterococci are known for their survival in marine waters, and consequently are recommended for use in recreational marine waters. Finally, C. perfrin- gens were also proposed as FIBs, because they are fully of faecal origin and associated with human species (Stelma, 2018). The main advantage of this FIB is its high persistence in the environment (Stelma, 2018), which favours the use of C. perfringens as indicators for tracing more distant sources, or sources in the marine environment.

The common physico-chemical parameters or their surrogates selected for study were reported in the literature as the parameters affecting survival dynamics or growth of FIBs. Such parameters include temperature, solids (both settleable and TSS, enhancing the FIBs survival), flow rate (possibly a co-variable with resuspended TSS), nutrients (P and N, supporting bacteria regrowth), conductivity (reducing bacteria survival), and pH (Davies et al., 1995; Charaklis et al., 2005; Jeng et al., 2005a,b; Chudoba et al., 2013). These parameters were discussed earlier in the Background chapter (Parameters associated with the pollutants of concern).

The selected mineral inorganics were adopted from the literature (Frost et al., 2015) and studied to identify the sources of baseflow at two of the four sampling sites and comprised six minerals (Al, Ca, Fe, K, Mg and Na). All these parameters are abundant in the Earth’s crust and confirmed the local groundwater as the source of baseflow in catchments B and D (Lundegårdh, 1984).

(29)

The selection of trace metals indicating anthropogenic pollution followed the recommendations by Eriksson et al. (2007) and the availability of chemical analyses in standard analytical packages offered by commercial laboratories. This indicator group comprised six trace metals: Cd, Cr, Cu, Ni, Pb and Zn, among which Cu, Pb and Zn are ubiquitous metals found in urban stormwater and the remaining are closely associated with traffic (Cd, Cr, Ni) (Huber et al., 2016).

3.2. Sampling sites

Four urban catchments with drainage outfalls along the Östersund waterfront on Lake Stors- jön were selected for studying stormwater and snowmelt quality in typical urban catchments with various land use (Fig. 2).

Fig. 2. Location map of the four studied catchments A, B, C, D (yellow marked) in central Östersund, Sweden

(30)

In terms of land use, the studied catchments represented typical land uses in Östersund, with a varying degree of urban development. The basic characteristics of the study catchments are listed in Table 2.

Table 2. Characteristics of the Östersund study catchments serving for sampling urban drainage effluents: stormwater, snowmelt and baseflow Catchment Area

(ha) Impervious-

ness (%) Baseflow in

dry weather Land use description

A 19 21 No Green areas (park land and urban

forest), with one residential street

B 21 50 Yes Residential area with single family

homes on grassed lots

C 36 60 No Central area with institutional build-

ings (university and municipal build- ings), park land, roads and streets,

and residential housing

D 2.2 80 Yes A hospital complex with surrounding

parking lots and streets

The GIS-data available for the four studied catchments was improved with time, therefore the presented catchment sizes and percent imperviousness may slightly deviate from the values in the published papers. Typical landscapes in the study area are shown in Fig. 3 below, with the catchments identified in the photos.

(31)

Fig. 3. Typical landscapes in the study area

3.3. Rainfall and temperature measurement

Rainfall and air temperature were monitored during sampling events with a tipping bucket rain gauge (type MJK, manufactured by MJK Automation) and a temperature logger (type Tinytag Plus 2, manufactured by Gemini Data Loggers). Both instruments were installed in the city centre (Fig. 4), and recorded data in one-minute intervals.

Fig. 4. Tipping bucket rain gauge installation in the city centre

(32)

Rainfall depths of the fully sampled events varied from 2.8 to 11.2 mm, and the event antecedent dry periods were at least three days to allow accumulation of pollutants in the catchment. Samples were collected during the entire rain events with the rainfall depth of at least 2 mm. This depth was determined as the minimum rainfall needed to produce simultaneous runoff in all the four catchments. Basic characteristics of the sampled events are listed in Table 3 below.

Table 3. Characteristics of the sampled rain and snowmelt run-off events Event

No.

Date Source of runoff Event rainfall (mm)

Event duration (h)

1 14.09.2012 Rain 3.2 1.5

2 26.09.2012 Rain 2.8 8.0

3 04.10.2012 Rain 6.0 2.0

4 26-28.02. 2013 Snowmelt 0 43.0

5 26.03.2013 Snowmelt 0 4.5

6 09.04.2013 Snowmelt 0 6.5

7 10.04.2013 Snowmelt 0 7.0

8 16.04.2013 Snowmelt 0 9.0

9 09.05.2013 Rain 5.6 7.0

10 03.06.2013 Rain 9.6 7.0

11 13.06.2013 Rain 11.2 9.0

12 06.03.2014 Snowmelt 0 9.0

13 21.03.2014 Snowmelt 0 11.5

14* 17.08.2014 Rain 26 14.0

*Not monitored in catchments A, C and D due to operational reasons

Assessment of stormwater and snowmelt quality based on water management priorities and the consequent water quality parameters