Enhancement of stormwaterltu.diva-portal.org/smash/get/diva2:1478309/FULLTEXT02.pdf · 2020. 11....
Transcript of Enhancement of stormwaterltu.diva-portal.org/smash/get/diva2:1478309/FULLTEXT02.pdf · 2020. 11....
-
LICENTIATE T H E S I S
ISSN 1402-1757ISBN 978-91-7790-697-1 (print)ISBN 978-91-7790-698-8 (pdf)
Luleå University of Technology 2020
Snežana Gavrić E
nhancement of storm
water quality in grass sw
ales
Department of Civil, Environmental and Natural Resources EngineeringDivision of Architecture and Water
Enhancement of stormwater
quality in grass swales:Removal and immobilisation of metals
Snežana Gavrić
Urban Water Engineering
131907-LTU-Gavric.indd Alla sidor131907-LTU-Gavric.indd Alla sidor 2020-11-13 10:092020-11-13 10:09
-
Enhancement of stormwater quality
in grass swales: Removal and immobilisation of metals
Snežana Gavrić
Luleå, 2020
Urban Water Engineering
Division of Architecture and Water
Department of Civil, Environmental and Natural Resources Engineering
Luleå University of Technology
-
Printed by Luleå University of Technology, Graphic Production 2020
ISSN 1402-1757
ISBN: 978-91-7790-697-1 (print)
ISBN: 978-91-7790-698-8 (pdf)
Luleå 2020
www.ltu.se
-
i
Preface This licentiate thesis presents a summary of my research work carried out in the Urban
Water Engineering research group in the Department of Civil, Environmental and
Natural Resources Engineering at Luleå University of Technology. The work was carried
out as a part of a research cluster Stormwater&Sewers, a collaboration between Swedish
municipalities of Luleå, Skellefteå, Östersund, Boden, municipal water organisations
Vakin, MittSverige Vatten & Avfall, VASYD, the Swedish Water & Wastewater
Association, and the Urban Water Engineering research group. The research was
financed by the Swedish Research Council Formas (Grant no. 2015-778) and
DRIZZLE-Center for Stormwater management, funded by the Swedish Governmental
Agency for Innovation System (Vinnova), project 2016-05176.
First and foremost, I would like to express my great gratitude to my supervisors Maria
Viklander and Günther Leonhardt for their supervision, valuable feedback and
encouragement throughout the years. My great gratitude also goes to Jiri Marsalek and
Heléne Österlund. Thank you Jiri, for your scientific guidance at some pivotal moments
in my studies, for sharing your great knowledge with me and for all the feedback and
valuable comments. Heléne, thank you for your advice and support and help with both
practical and scientific questions. I also would like to thank Anna-Maria Perttu for her
help during my first months as a PhD student and Alexandra, for the help with Swedish
translations.
The field study would not be possible without the help of the staff from Luleå
municipality who helped me with logistics and provided me with information about the
studied catchments. I want to thank Peter Rosander for helping me to make my soil
sampler and solve all other practical issues. I want to thank Stefan Marklund for help in
contacting people from kommun and Kerstin Nordqvist for her help in the laboratory. I
also want to thank all my colleagues in the Urban Water Engineering research group for
providing friendly work environment.
Finally, I would like to thank all the people that made my time in Luleå pleasant: my
tango group for welcoming me in their community and my friends from other research
groups for fun times away from work. I want to thank all my family and friends in Serbia,
and especially my parents, Boško and Milica, thank you for all the love and support. To
Ivan, for being such a wonderful father, which enabled me to fully concentrate at work.
Thank you for believing in me and helping me manage everything at work and home.
To Luka, my wonderful boy, I love you dearly, thank you for being you.
Snežana Gavrić Luleå, November 2020
-
ii
-
iii
Abstract Grass swales are common elements of green drainage infrastructure used in urban
catchments to provide stormwater quantity and quality control. Concerning stormwater
quantity, swales convey runoff and attenuate stormwater volumes and peaks by enhancing
hydrological abstractions and providing dynamic storage. Furthermore, grass swales are
effective in treating stormwater runoff from trafficked surfaces. Swales are typically
designed as long shallow channels with dense grass and permeable soils, and thereby create
favourable conditions along the turf-stormwater interface for processes enhancing
stormwater quality and reducing pollutant concentrations in swale effluents.
The thesis aim is to advance the knowledge of short-term performance of grass swales in
removal of total metals, with respect to such influential factors as concentrations of metals
and solids (TSS) in the inflow, swale geometry, and grass-soil characteristics. The
literature reviewed showed that solids were the most frequently investigated parameter
and the enhancement of stormwater quality by settling gained most attention in the earlier
research. On the other hand, studies of swale performance in removal of other stormwater
pollutants, such as total and dissolved metals, were limited, as was the understanding of
the physical-chemical-biological processes facilitating the removal of other-than-solids
pollutants.
Since swales are generally recognized as being effective in removing metals from
stormwater through infiltration into swale soils, and the associated metal immobilisation
in soils, long-term operation of swales may lead to accumulation of pollutants in, and
contamination of, swale soils. Such conditions need to be remedied by relatively costly
swale maintenance. A field study was conducted to characterize the soil chemistry of
three swales, which serve for stormwater drainage and winter storage of snow cleared
from adjacent trafficked areas. The swales studied served three catchments with different
land use in the City of Luleå. Swales provided drainage of commercial, downtown, and
residential catchments and drained roads with various traffic intensities. The study
findings showed that the soil in the oldest swale, next to the road with the highest traffic
intensity, contained the highest concentrations of most of the investigated metals. For
example, the mean lead (Pb) concentration at this swale was ~70 mg/kg DW, compared
to
-
iv
-
v
Sammanfattning Svackdiken är ett vanligt inslag i grön infrastruktur, som i urbana områden används för
att kontrollera dagvattenflöden gällande både kvantitet och kvalitet. Svackdiken reglerar
dagvattenkvantitet genom att transportera avrinning och dämpa både dagvattenvolymer
och toppflöden via ökade hydrologiska abstraktioner och tillhandahållande av dynamisk
lagring. Dessutom är svackdiken effektiva för att rena dagvattenavrinning från trafikerade
ytor. Svackdiken utformas vanligtvis som långa, grunda, kanaler med tätt gräs och
permeabel jord och skapar därigenom gynnsamma förhållanden längs med gränssnittet
gräs-dagvatten för processer som förbättrar dagvattenkvalitet och reducerar
föroreningskoncentrationer i utsläpp från svackdiken.
Avhandlingens mål är att öka kunskapen om kortsiktig prestanda hos svackdiken avseende
avlägsnandet av metaller, med hänsyn till påverkande faktorer såsom koncentrationer av
metaller och partiklar (TSS) i inflödet, svackdikets geometri samt egenskaper hos gräs och
jord. Litteraturstudien visade att partiklar var den mest frekvent studerade parametern och
att förbättring av dagvattenkvaliteten genom sedimentation fick mest uppmärksamhet i
tidigare forskning. Å andra sidan var studier av svackdikens prestanda med avseende på
avlägsnande av andra dagvattenföroreningar, såsom totala och lösta metaller, begränsade,
liksom förståelsen av de fysikaliska-kemiska-biologiska processerna som främjar
avlägsnandet av andra föroreningar än partiklar.
Eftersom svackdiken allmänt anses vara effektiva för att avlägsna metaller från dagvatten
genom infiltration och associerad metallimmobilisering i jord, kan långsiktig användning
av svackdiken leda till ackumulation av föroreningar i, och kontamination av, svackdikens
jordar. Sådana förhållanden behöver åtgärdas med relativt kostsamt underhåll. En
fältstudie genomfördes för att karaktärisera markkemin i tre svackdiken som används för
dagvattenavledning och vinterförvaring av snö undanröjd från intilliggande
trafikområden. De studerade svackdikena, tjänade tre avrinningsområden med olika
markanvändning i Luleå. Svackdikena avleder avrinning från ett handelsområde,
innerstad och bostadsområde och med vägar av olika trafikintensitet. Resultaten från
studien visade att jorden i det äldsta svackdiket, bredvid vägen med högst trafikintensitet,
innehöll högst koncentrationer av de flesta undersökta metallerna. Till exempel var
medelkoncentrationen av bly (Pb) ~70 mg/kg DW (torrvikt), jämfört med
-
vi
-
vii
Table of Contents Preface ......................................................................................................................... i
Abstract ...................................................................................................................... iii
Sammanfattning ............................................................................................................ v
List of papers ............................................................................................................... ix
1. Introduction ......................................................................................................... 1
1.1. Aim and Objectives ......................................................................................... 2
1.2. Thesis structure ............................................................................................... 3
2. Background .............................................................................................................. 5
2.1. Swale design and characteristics ....................................................................... 5
2.2. Swale hydrological performance ...................................................................... 8
2.3. Swale performance in treating stormwater runoff ............................................. 8
2.4. Effect of stormwater infiltration on soil media quality .................................... 10
2.5. Conceptual models of grass swales ................................................................. 13
2.6. Knowledge gaps and future research ................................................................ 14
3. Methods ............................................................................................................. 15
3.1. The literature review ..................................................................................... 15
3.2. Study sites ..................................................................................................... 15
3.3. Soil sampling ................................................................................................. 16
3.4. Infiltration measurements .............................................................................. 18
3.5. Laboratory analysis ........................................................................................ 18
3.5.1. Soil parameters .................................................................................... 18
3.5.2. Analysis of total metal concentrations .................................................. 19
3.5.3. Sequential extraction analysis ............................................................... 19
3.6. Grit material applied during the winter road maintenance ............................. 20
3.7. Data analysis .................................................................................................. 21
3.8. Computation of the metal burdens in swale soils ........................................... 21
3.8.1. Background concentrations of metals in swale soils .............................. 23
3.9. Modelling methods (StormTac Web) ............................................................ 24
4. Results ................................................................................................................... 27
4.1. Physico-chemical characteristics of soils in the studied swales ........................... 27
4.2. Metal removal during runoff conveyance in grass swales .................................. 30
4.3. Metal burdens in swale soils ............................................................................. 34
-
viii
4.4. Total metal concentrations in swale soils .......................................................... 35
4.5. Metal mobility ................................................................................................. 40
4.6. Winter road maintenance ................................................................................ 42
5. Discussion .............................................................................................................. 43
5.1. Factors affecting the estimated metal burdens in swale soils .............................. 43
5.1.1. Metal sources ............................................................................................ 43
5.1.2 Mean metal Zn, Cu and Pb burdens in soil layers ...................................... 44
5.2. Selected factors influencing the removal of metals from stormwater transported
in grass swales ......................................................................................................... 46
5.2.1. Hydraulic functioning of various swales sections ....................................... 46
5.2.2. Swale soil properties: pH, LOI, EC and sorption capacity (CEC) .............. 48
5.3. Swale maintenance .......................................................................................... 50
6. Conclusions ............................................................................................................ 53
7. References ............................................................................................................. 55
-
ix
List of papers
Paper I Gavrić, S., Leonhardt, G., Marsalek, J., Viklander, M. (2019).
Processes improving urban stormwater quality in grass swales and filter strips: A
review of research findings.
Science of the Total Environment, 669, 431-447.
Paper II Gavrić, S., Larm, T., Österlund, H., Marsalek, J., Wahlsten, A., Viklander, M. (2019).
Measurement and conceptual modelling of retention of metals (Cu, Pb, Zn) in
soils of three grass swales
Journal of Hydrology, 574, 1053-1064.
Paper III Gavrić, S., Leonhardt, G., Österlund, H., Marsalek, J., Viklander, M. Metal enrichment of soils in three urban drainage grass swales used for seasonal
snow storage
Submitted to Science of the Total Environment
Assessment of contribution to the above papers
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 Contributed Shared
responsibility
Responsible Shared
responsibility
Shared
responsibility
Shared
responsibility
Shared
Responsible
II Contributed Shared
responsibility
Responsible Shared
responsibility
Shared
responsibility
Shared
Responsible
Shared
responsibility
III Shared
responsibility
Shared
responsibility
Responsible Shared
responsibility
Shared
responsibility
Shared
Responsible
N/A
Responsible – developed, consulted (where needed) and implemented a plan for
completion of the task.
Shared responsibility – made essential contributions towards the task completion in
collaboration with other members in 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)
N/A – Not applicable
-
x
-
1
1. Introduction The goal of the EC Water Framework Directive (WFD) (Directive 2000/60/EC, 2000)
is to achieve good qualitative and quantitative status of all water bodies in member states,
and this goal is further supported by the Environmental Quality Standards Directive
(EQSD) (Directive 2008/105/EC, 2008) and the Groundwater Directive (GD)
(Directive 2006/118/EC, 2006). Meeting the WFD objectives is particularly challenging
in river basins with high concentration of urban areas, which are recognized for multiple
pollutant sources impacting on both surface waters and the groundwater. With respect
to urban drainage, which is addressed in this thesis, one of the promising pollution
prevention and remediation measures is incorporation of green infrastructure (GI)
elements into urban and suburban catchments, with the objective of providing local
control of polluted stormwater runoff generated mostly on impervious areas. Grass swales
represent common GI elements that can be advantageously used, instead of stormwater
pipes, to drain runoff from trafficked areas. In addition to runoff conveyance, swales also
attenuate stormwater flow volumes and peaks, and reduce pollutant concentrations
through the interaction of flow with grass-soil media (Schueler, 1987). In climates with
seasonal snow, swales also serve for storage of snow cleared from streets, roads and
sidewalks (Backstrom and Viklander, 2000).
Swales ability to remove pollutants during actual rainfall and snowmelt events and
irrigation experiments, when runoff is conveyed through the swale, can be referred to as
short-term performance. Usually such a performance is estimated by comparing the
quality of stormwater before it enters and after it leaves the swale. On the other hand,
the performance of swales, that have been operated for many years, in retaining
particulate pollutants can be estimated from their soil chemistry. Soil chemical quality is
the result of swale long-term operation, i.e., drainage of a series of many runoff/snowmelt
events, processes occurring between the events (e.g., evapotranspiration, plant uptake)
and even external actions (maintenance of swales, reconstruction). These effects are
accounted for in swale long-term performance in immobilizing pollutants in their soils.
Studies investigating short-term swale performance in pollutant removal focused mostly
on selected swale characteristics, such as the longitudinal slope, geometry (the length and
cross-section), and the grass species and density, in order to develop the “best” swale
designs for stormwater quality control. A general analysis of the database derived from 59
swale studies showed that swales were efficient in removing TSS and particulate metals
(Zn, Pb, Cd and Cu) from conveyed runoff flows (Fardel et al., 2019). The removal of
solids has been investigated extensively, which led to the development of computational
methods for the settling of discrete particles (Deletic, 2001; Deletic and Fletcher, 2006).
Long-term exposure to polluted stormwater runoff from roads and parking lots in turn
leads to an elevated content of traffic-related metals in roadside soils (Lind and Karro,
1995; Achleitner et al., 2007). Generally, the highest metal concentrations were observed
-
2
in the upper soil layers (typically 0-5 cm) near the point of runoff inflow into the swale
(Tedoldi et al., 2017a), from which the pollutant concentrations declined with distance
and increasing depth below the soil surface (Tedoldi et al., 2017b). Although the metal
enrichment in roadside soils was investigated in some earlier studies, there is a lack of data
for swales serving not just for runoff control in warm seasons, but also for snow storage
during the winter season.
In spite of a fair number of studies on the role of grass swales in stormwater management,
the research of swales is still ongoing, because of the needs to cover the variety of climatic
conditions, swale design characteristics and operating conditions in urban catchments.
Swales usually provide a link between impervious surfaces (e.g., highways, streets, parking
lots, roofs, etc.) generating stormwater runoff, which partly ingresses into swale soils and
partly is conveyed through the swale, and the separate sewer systems. In that sense, swales
can be viewed as transport links between the pollutant sources and the receiving
environments, including both groundwater aquifers and surface waters. Thus,
investigations of both short- and long-term environmental performance of swales are of
great importance for creating opportunities for implementing environmental protection
by, and establishing maintenance needs of, grass drainage swales.
1.1. Aim and Objectives The vast majority of literature references reported on the studies, in which grass swales
were exposed only to rain-generated runoff and the associated pollution. However, in a
climate with seasonal snow addressed in this thesis, swales are also exposed to snowmelt
from intermittent melts of fresh snow on trafficked pavements and the melts of polluted
snow stored in swales during the winter. Urban snowmelt is generally more polluted than
rain runoff, because of winter road maintenance involving salt and grit applications on
roads and seasonal activation of other pollution sources (Vijayan, 2020).
This thesis aims to investigate the short- and long-term operation of grass swales serving
for stormwater drainage and seasonal snow storage. The specific thesis objectives are as
follows:
1. To advance the knowledge of processes affecting total metal concentrations in
stormwater runoff passing through grass swales, with respect to influential factors,
including inflow pollutant concentrations, swale geometry, and characteristics of
grass-soil media (Paper I).
2. To estimate metal enrichment of, and metal burdens in, soils of three urban grass
swales serving for stormwater drainage and seasonal snow storage, on the basis of
soil chemistry data (Papers II and III).
Much of the discussion of swale environmental performance focuses on traffic-related
metals, because of their common occurrence in road runoff at toxic levels and potentially
acute effects on biota in the receiving waters.
-
3
1.2. Thesis structure The thesis includes three papers referred to as Papers I-III. Paper I is a review paper,
which synthesizes and critically reviews the findings of previous research on the processes
affecting pollutant transport with runoff in grass swales and on grass filter strips. Paper II
presents a method, which uses soil chemistry data and planning level modelling to
estimate the metal burdens in swale soils. Paper III is a field study examining vertical and
horizontal profiles of swale soils, in order to advance the understanding of metal
distribution and mobility in urban grass swales operated in the cold climate with seasonal
snow. A synthesis of these three papers is presented in Figure 1.
The thesis is divided into seven chapters. Chapter 1 introduces the topic of grass swales
and their importance in stormwater management, and finishes with the thesis aims and
objectives. Chapter 2 presents a background of swale quantity and quality performance
in stormwater control, with special focus on the consequences of long-term infiltration
of polluted stormwater and snowmelt into swale soils. The investigated field sites and the
methods used in the three papers are described in Chapter 3. In Chapter 4, the main
results of the licentiate thesis are presented and followed by the discussion of results in
Chapter 5. The main study conclusions are presented in Chapter 6. The list of references
cited is presented in Chapter 7. Finally, the thesis papers are appended at the end of the
thesis.
Figure 1: The relationship among the papers included in the thesis
-
4
-
5
2. Background This chapter describes the role of grass swales in green urban drainage infrastructure and
presents an overview of the state-of-the-art knowledge addressing: (i) swale hydrology
and design, (ii) swale performance in removing pollutants from conveyed stormwater
runoff, (iii) enrichment of swale soils with metals (or other pollutants) entering soils
through infiltration, and (iv) modelling stormwater quality processes in grass swales.
2.1. Swale design and characteristics Progressing urbanisation results in higher runoff peaks and volumes of stormwater of
impaired quality travelling faster from the catchment to the receiving waters. One
solution for attenuating these negative effects of urbanisation and restoring some measure
of balance between the “natural” and “urban” drainage is the incorporation of green
infrastructure (GI), such as grass swales, into the landscape of urban catchments. Swales
are vegetated shallow drainage channels commonly used along impervious surfaces, such
as highways, roads, and parking lots, in order to reduce stormwater runoff volumes and
peaks, and remove some stormwater pollutants during runoff conveyance through a dense
grass layer and filtration through swale soils. Swales typically fall into three main
categories: (i) standard swales, (ii) dry swales, and (iii) wet swales. Standard and dry swales
are similar in their appearance, however, the latter feature a filter bed of specially
processed soil (an engineered soil) and an under-drain pipe to enhance infiltration
(Woods Ballard et al., 2015). On the other hand, wet swales are designed to operate with
permanent standing water and wetland vegetation (Woods Ballard et al., 2015). Typically,
swales receive runoff through lateral inflows over side slopes, either freely without
obstructions or through regular brakes in the curbs, and/or longitudinal inflows from the
upstream sources, e.g., from a bridge or a roof. Moreover, swales can be constructed as a
part of a stormwater treatment train, in combination with e.g. permeable pavement
parking lots, grass filter strips, or bioretention cells, in order to provide sufficient
treatment for meeting environmental objectives (WSUD, 2006; Revitt et al., 2017).
Sometimes the swale designs include grass filter strips (GFS) and/or check-berms across
the swale bottom section. GFS are considered as pre-treatment devices of swales,
however, some studies compared swale performance with and without GFS and
concluded that the main pollutant removal still occurs along the bottom of the swale
channel (Stagge et al., 2012). Check-berms serve to enhance infiltration into the swale
bed by increasing the flow depth. Figure 2 shows typical operating conditions of standard
swales.
There are some general design recommendations for swales that may be further modified
to meet specific local conditions. Swales are usually constructed to treat stormwater runoff
from small catchments, up to 1-2 ha in size, in order that the generated flow depths and
velocities allow for stormwater quality processes to occur (WSUD, 2006). The
recommended swale cross-sections are trapezoidal or triangular, with a rounded bottom
section and a recommended bottom width of 0.5-2.0 m, to allow for easier maintenance
-
6
and prevention of concentrated flows (Woods Ballard et al., 2015). Longitudinal slopes
are recommended not to exceed 2%, and inclusion of check-berms is recommended for
slopes steeper than 4%, while side slopes should be less than 1:3 (33%) (USEPA, 1999).
To achieve stormwater quality control, swales should perform well in pollutant removal
for the majority of events occurring on an annual basis (Woods Ballard et al., 2015). For
example, a design event for water quality swales should generate flow depths below the
grass height, flow velocities < 0.3 m/s, and adequate times of travel of runoff along the
swale (Woods Ballard et al., 2015). Even for major design rain events, with return periods
50 – 100 years, flow velocities within swales should be
-
7
Table 1: Methods used for determining swale characteristics in the previous research
Swale characteristics Method Ambient environmental and operational
conditions: Swale surroundings; channel
erosion and the state of turf cover;
accumulations of debris (litter) and sediment;
maintenance access; functionality of the
drainage pipe, condition of the drop inlet,
etc.
Visual observations (during regular inspections)
Swale geometry
Manual measurement at repeated distances along the
swale (e.g., 5 m); RTK-GPS survey and building the
digital elevation model (DEM) (Rujner et al., 2018)
Grass characteristics
Grass type
Processing digital photographs of the grass area with
imaging software to estimate the dominant grass
species (Ming-Han et al., 2008)
Grass cover
[grass blade/cm2]: Counting the blades within 10 cm2
quadrants (Deletic and Fletcher, 2006);
[%]: Processing digital photographs of the grass area
with imaging software to estimate grass coverage
(Ming-Han et al., 2008; Pan and Shangguan, 2006);
Aerial photographs imported to AutoCAD to draw
polygons around areas with bare soils plus visual
estimation (Winston et al., 2012)
Grass blade width [cm] Measuring ≈ 50 random grass blades (Deletic and
Fletcher, 2006)
Above-ground biomass per unit area
(if herbs are more dominant than grasses)a
An open cylinder (diameter 0.12 m and height 0.1
m) is placed in the centre of two adjacent 0.25 m2
quadrants to harvest the contained above-ground
biomass, the collected material is oven-dried and
weighed (Mazer et al., 2001);
Three plants randomly taken from a 15 cm2 quadrant,
aerial parts of plants were cut and dried at 35 °C for
one week, after which the dry residue was weighed
(Leroy et al., 2017)
Soil physical characteristics
Infiltration capacity [cm/h] Double ring infiltrometer test (Deletic and Fletcher,
2006; Rujner et al., 2018; Young et al., 2018)
Soil texture
From wet sieving and hydrometer analysis of two 13
cm long soil cores to determine % clay, % silt and %
sand (García-Serrana et al., 2017);
From granulometry of a 30 cm soil layer to determine
% clay, % silt and % sand (Rujner et al., 2018)
Soil compaction
Cone penetrometer (soil was considered compacted
if the cone index exceeded 2,070 kPa in the upper
7.6 cm) (Winston et al., 2012) a Herbs may produce high above-ground biomass despite low plant density (Mazer et al., 2001)
-
8
2.2. Swale hydrological performance Draining trafficked areas requires fast removal of generated runoff from the impervious
surfaces, in order to maintain safe road conditions (e.g., avoidance of hydroplaning) and
passage of emergency vehicles. Since swales are typically located next to roads and
highways, their design needs to provide good hydrological performance, including
attenuation of stormwater runoff volumes and peaks, and safe runoff conveyance.
Research has shown that swales are very effective in controlling runoff from small rain
events by completely infiltrating the runoff into soils, and avoiding any outflow (Davis et
al., 2012, Purvis et al., 2018, Young et al., 2018). This is why some authors noted that
small storms did not generate enough runoff in swales to allow for stormwater sampling.
For moderate rain events, swales attenuate runoff volumes and peaks, while for large
events, their main function is flow conveyance (Davis et al., 2012).
Rushton (2001) reported a mean runoff volume reduction of 30% in a catchment with
swales, compared to a similar catchment without swales. Bäckström et al. (2002)
performed field experiments on seven swales (5-10 m long) by pumping water mixed
with sediments into the swale at the upstream end (inflow rate 0.5-1.5 L/s) and observed
inflow volume reductions of 33-66%. Lucke et al. (2014) studied four field swales (30-
35 m long) by feeding in water with pollutants (TSS, TP and TN) at the upstream end
(0.5-2.0 L/s) and observed mean runoff volume reduction of 52%, which depended on
the initial soil moisture content. Jiang et al. (2017) monitored two swale sections (5 m
long) and reported mean runoff volume reductions of 78-98% for five monitored actual
events. Young et al. (2018) studied two swales (210-230 m) draining highway runoff and
observed mean volume reduction of 87% for 65 rainfall events.
The swale grass layer further enhances the swale hydraulic function, compared to bare
soils. Dense grass increases surface roughness by slowing down the runoff and increasing
infiltrated runoff volumes (García-Serrana et al., 2017), which also limits the effect of
slope on infiltration rates (Morbidelli et al., 2016).
2.3. Swale performance in treating stormwater runoff Runoff generated on impervious surfaces of urban catchments contains a variety of
pollutants from numerous anthropogenic activities. Draining such a polluted runoff into
grass swales can provide local treatment in well-designed swales (e.g., with dense grass
turf, mild bottom slopes, good infiltration rates, etc.). Identification of swale
characteristics, which are beneficial for stormwater quality control, resulted from studies
addressing grass swale performance in enhancing stormwater quality. These studies can
be divided into three groups: (i) laboratory and field assessment of simulated inflows, (ii)
field assessment of actual rainfall events, and (iii) computer modelling studies.
The first group of studies often aimed to advance the understanding of small-scale
processes occurring in swales with respect to influential factors. Typical controlled
-
9
investigations supplied synthesized stormwater runoff at the swale upstream end (Deletic
1999; Bäckström 2002; Deletic and Fletcher 2006; Lucke et al., 2014), but less frequently
also over the swale side slopes (Fardel et al., 2020). Developed algorithms for fluxes of
pollutants of different types and characteristics, which resulted from such controlled
experiments, can be verified using the measured data from the second group of studies.
Field studies of swale short-term performance in pollutant removal during actual rainfall
events typically investigated swales next to highways (e.g. Barrett et al., 1998; Winston
et al., 2010) or urban roads (e.g. Bäckström et al., 2006), and less frequently parking lots
(Rushton, 2001). Moreover, studies of swale field performance for actual rainfall events
are very important for collecting high quality data for testing standard urban drainage
modelling packages (e.g. SWMM, Mike SHE) and for investigating swale performance
in larger-scale systems (i.e., incorporated into the drainage system).
Looking at larger-scale processes is important, because these processes affect the
generation and quality of runoff entering the swale. During dry periods,
evapotranspiration (ET) restores swale infiltration capacity (Deletic, 2000) and compared
to bare soils, evapotranspiration is enhanced by the grass cover (Hino et al., 1987). At the
same time, pollutants accumulate on the catchment surfaces (including the swale surface)
as a result of dry atmospheric deposition. Even in dry weather, the pollutants accumulated
on the contributing drainage surfaces may be transported into grass swales by vehicle
induced turbulence, wind, street sweeping and snow clearance from pavements. During
wet weather, the accumulated pollutants are washed into the swale via runoff and vehicle-
generated splash water (Werkenthin et al., 2014), and additional pollutants enter swales
through direct precipitation (i.e., wet atmospheric deposition). Leroy et al. (2016)
sampled infiltrated water from a swale section receiving only atmospheric deposition (wet
and dry) and compared it to a swale section receiving also road runoff. The authors
measured lower concentrations in the former section, but of the same order of magnitude,
concluding that atmospheric wet and dry deposition should not be neglected in the
conditions of their study (Leroy et al., 2016).
Moreover, specific catchment characteristics, such as, land use type (e.g., highway,
secondary road, residential area), percentage of imperviousness, slopes, etc., affect the
stormwater runoff pathway. Different pathways of stormwater before reaching the swale
will affect the stormwater quality (i.e., the pollutant inflow concentrations reaching the
swale). Studies have shown that pollutant removal in grass swales is affected by pollutant
inflow concentrations. For example, Stagge et al. (2012) suggested that swales can treat
total phosphorus (TP), if its concentration in the inflow exceeds 0.7 mg/L, while
Bäckström et al. (2006) suggested that inflow concentrations of TSS >40 mg/L are
needed to produce positive removals. Winston et al. (2011) observed that the largest
increases in total nitrogen (TN) and TP concentrations, after conveyance through GFS,
may be caused by low inflow concentrations. Also, Winston et al. (2010) investigated
two wet swales and two standard swales (with GFS) draining highways with asphalt
-
10
pavement with a porous friction course (PFC) overlay. The swale length was 30.5 m and
the GFS length was 8 m (Winston et al., 2010). TSS inflow concentrations, to the GFS,
were reduced by runoff passage over the PFC to concentrations in the range 10-31 mg/L
and negative removals of TSS were observed after conveyance over GFS (Winston et al.,
2010). The authors explained such a TSS export by the irreducible TSS concentrations
in the influent, ~10 mg/L (Winston et al., 2010). The irreducible concentrations are the
minimum (residual) outflow concentrations that cannot be further reduced by stormwater
management facilities (Schueler, 2000).
Abundance of solids in urban areas, their release throughout the urban catchments and
the role in transporting other pollutants (e.g., adsorbed metals) (Liu et al., 2015), all
contribute to the fact that solids are the most investigated quality parameter in studies of
grass swale inflows and outflows. In addition, total and dissolved metals, nutrients, traffic-
associated hydrocarbons and oxygen-demanding constituents are also often analysed,
compared to, e.g., chloride and faecal indicator bacteria, which are studied less frequently.
The pollutant type and characteristics are important to consider, because actual
stormwater quality processes causing pollutant removal in overland flow over grass
depend on these influential factors. For example, settling and infiltration processes, were
found important in removing solids (usually described as total suspended solids (TSS))
(Mendez et al., 1999; Barrett et al., 2004; Stagge et al., 2012), while plant uptake
contributed to retention of metals in the roots and the above ground biomass (Leroy et
al., 2017).
2.4. Effect of stormwater infiltration on soil media quality Filtration of stormwater through swale soils is an important process for enhancing
stormwater quality, as shown by sampling subsurface flow from a swale underdrain pipe
(Purvis et al., 2018; Fardel et al., 2020). Filtration of road runoff through swale soils
resulted in a significantly cleaner outflow in the underdrain pipe, compared to untreated
road runoff, with respect to TSS, total volatile suspended solids (VSS), enterococcus, E.
coli, and turbidity (Purvis et al., 2018). For example, concentration reductions in the
underdrain outflow were 88% (TSS) and 87% (VSS), while reductions in the overflow
were substantially lower, 10 and 21%, for TSS and VSS, respectively (Purvis et al., 2018).
Fardel et al., (2020) conducted controlled field experiments to examine Zn, pyrene,
phenanthrene and glyphosate removals through standard and filtration swales, and found
that chemical removal efficiencies were significantly higher in the subsurface outflow,
compared to the overflow. In the same study by Fardel et al. (2020), for all experiments,
mass removals of Zn, pyrene, phenanthrene and glyphosate were higher in the filtration
swale than in the standard swale. Moreover, Leroy et al. (2015) observed that filtration
through the dense root system of grass captured suspended solids (SS) with attached PAHs
and limited the transfer of PAHs into deeper soil layers.
-
11
The above discussed research studies acknowledged that, in a long term, filtering polluted
stormwater runoff from trafficked surfaces can pose some risk of contamination of swale
soils, with contaminants reaching deeper soil layers and even leaching into the
groundwater. Monitoring pollutant concentrations in the field can be used to
characterize pollutant concentrations, provide spatial or temporal summary of
environmental contamination, and demonstrate or enforce the compliance with standards
or guidelines (Gilbert, 1987), to name a few examples. Moreover, field and laboratory
studies can provide data for studying pollutant transport and quantifying the relationships
that control the levels and variability of pollutant concentrations in time and space
(Gilbert, 1987). Often, a composite sampling is performed in order to reduce the cost of
sample analyses, but at the risk of losing the individual sample information and dilution
of the samples (Mason, 1992).
The fate and transport of metals and PAHs in the soils of the infiltration-based Sustainable
Urban Drainage Systems (SUDS) has been reviewed by Tedoldi et al. (2016). Metals are
often reported in higher concentrations in the topsoil layer, and such concentrations
decrease with the soil depth (Tedoldi et al., 2016). The thickness of the “topsoil layer”
differs among the studies. In many studies, the topsoil layer was considered to be 5 cm
thick (Lind and Karro 1995; Norrström and Jacks 1998; Achleiter et al., 2007; Ingversten
et al., 2012; Rommel et al., 2019), but in other studies (Rushton 2001; Hjortenkrans et
al., 2006) a smaller depth was considered (0-3 cm). According to Mason (1992) airborne
pollutants and pollutants that are strongly bound to soil particles are found in the top
15 cm, while pollutants from long-term deposition are found in the layers deeper than
15 cm.
There are multiple traffic-related sources of metals, e.g. vehicle operation, tire and brake
wear, vehicle washing, and road abrasion that contribute to metal pollution in stormwater
runoff (Müller et al., 2020). In many studies, soils were sampled next to the roads with
various traffic intensities, in order to assess the contribution of traffic to the metal
pollution in roadside soils. Carrero et al. (2013) sampled soils next to: (i) an old secondary
road exposed to high traffic (>60 years of service), (ii) a newer highway road (20 years
old with 28,200 AADT), and (iii) two roundabouts (1 and 5 years old). PCA analysis
showed that samples from the old secondary road clearly differed from the remaining
samples by having higher concentrations of traffic related metals (Carrero et al., 2013).
For example, Pb concentrations of 630 mg/kg in the old road case far exceeded the
concentrations of Pb
-
12
sorption capacity can lead to migration of metals into the deeper layers (Tedoldi et al.,
2016). In another case, the high clay content (19.1%) provided high cation exchange
capacity enabling metal sorption (Leroy et al., 2016). Moreover, decrease in pH can result
in mobilisation of metals (Bäckström et al., 2004), but neutral pH 6-7 reduces the risk of
occurrence of metals in the dissolved fraction, as reviewed by Rieuwerts et al. (2015).
Bäckström et al. (2004) sampled the water draining through the soil at a depth of 50 cm
below the soil surface, at various distances from the road edge. The sampling was done
in the soils next to two roads in mid Sweden during one year and the analytical protocol
included pH, electrical conductivity (EC), inorganic carbon (IC), total organic carbon
(TOC), chloride, sulphate, and metals (Cd, Cu, Pb, Zn, Na, Ca, Mg, K, Fe and Al). The
authors found strong significant correlations between chloride and metals, and electrical
conductivity and metals (Bäckström et al., 2004).
Lastly, swales in cold climate regions with seasonal snow have an additional function, i.e.,
storage of snow cleared from roads and parking lots. This is a very useful swale function
for maintaining safe driving conditions during the winter, since snow can be quickly
cleared from roads into swales. Comparisons of different snow management scenarios
(i.e., transport of snow to, and storage in, central or local snow storage sites, with or
without the use of swales) showed that storage of snow in the swales had a favourable
impact on costs and long-term traffic-related pollution emissions (Reinosdotter et al.,
1998).
Lind and Karro (1995) sampled soils next to two roads (AADT = 11,400-34,000) in
southern Sweden, after the first eight years of operation. The authors observed that
drainage of stormwater contributed to a metal (Zn, Pb and Cu) enrichment of soils.
Norrström and Jacks (1998) sampled soil next to a 29-year old highway (40,000-50,000
AADT) in southern Sweden. The first 15 cm of soil were sampled by coring at 20
locations along two transect lines at 0.5 and 2.5 m distances from the pavement edge and
the cores were divided into 5 cm slices, which were composed for individual transects
(Norrström and Jacks, 1998). The authors measured the highest Pb concentration (542
mg/kg) in the top 5 cm at 0.5 m from the highway. Hjortenkrans et al. (2008) sampled
two swales draining about 20-year old highways (20,700-22,300 AADT) in the South of
Sweden and produced composite samples, comprising at least seven sub-samples, to
represent the metal concentrations at different depths and distances from the highway. At
one site the highest Pb concentration (200 mg/kg) was measured at the 0.4 m distance,
10 cm below the surface, with the upper layer concentrations being lower (Hjortenkrans
et al., 2008). The authors explained this by the annual depositions of sand used in winter
road maintenance and the phase-out of Pb from gasoline (Hjortenkrans et al., 2008). The
characterisation of soil pollution in cold climate swales is important, because of specific
operating conditions, including the drainage from roads serviced by applications of salt
and anti-skid materials, and the effects of melting of the stored snow during the winter.
-
13
2.5. Conceptual models of grass swales Computer modelling studies strive to cope with the complexity of urban drainage systems
and integration of a large number of drainage elements. Modelling studies of grass swales
can be divided in two groups, according to the nature of the models used: (i) Studies
based on research models, and (ii) Studies of applications of standard urban drainage
modelling packages (e.g., SWMM, Mike SHE, Music, and others).
The first group represents semi-empirical models for computing TSS removal from
stormwater flow over grass surfaces, e.g., the Kentucky Method (Tollner et al., 1976) and
the Aberdeen equation (Deletic, 2000), which resulted from controlled laboratory
experiments. This group of models still attracts a lot of research interest and efforts to
verify the original equations for other conditions than those, for which they were
developed. For example, the Aberdeen equation was recently tested to predict TSS
removal efficiencies in two swales and the modelling results for six rain events were
compared to the actual field data (Hunt et al., 2020). The modelled event removal
efficiency was a weighted average of the removal efficiencies of each particle size
(calculated using Aberdeen method) (Hunt et al., 2020). The maximum difference
between the modelled and actual removal efficiencies was 20% and the authors noted
that the smallest difference (1-6%) was observed for two events when the flow depth was
close to the nominal grass height. There is a research need for more data on other-than-
solids pollutant removals and transport in swales, in order to gain more knowledge on
influential factors and quantify processes other than settling in swales.
The models in the second group were originally developed for larger (catchment) scales
and are continually being refined, in order to simulate small stormwater control facilities,
such as grass swales, with sufficient accuracy. For example, Niaizi et al. (2017) reviewed
papers on SWMM applications and found only a small number of studies, out of 150
peer-reviewed papers, describing the use of SWMM for modelling stormwater pollutant
reductions by GI. One study (Jia et al., 2014) compared the following drainage scenarios:
(i) impervious areas, and (ii) impervious area reduced by incorporating GI features
(including a grass swale in a treatment train). However, the pollution reduction by the
swale and representation of the stormwater quality processes in the swale was not the
focus of the study. A number of recent studies focused on modelling runoff quantity
control by grass swales using standard urban drainage modelling packages, Mike SHE,
SWMM, and WinSLAMM (Flanagan et al., 2017; Xie et al., 2017; Rujner et al., 2018;
Young et al., 2018; Wadhwa and Kumar, 2020; Broekhuizen et al., 2020). This work is
relevant to the studies in this thesis, since reliable quantity simulations are needed to
model well the water quality. However, an even bigger obstacle in modelling pollutant
reductions in swales is the lack of understanding of physical, chemical and biological
processes taking place in grass swales.
-
14
2.6. Knowledge gaps and future research
There is a continual need to expand the existing knowledge of design and operation of
drainage swales into widely varying and previously unexplored conditions, and assess the
underlying limitations of the past research. This assessment was conducted at the start of
the thesis project and its findings are briefly summarized below. Recognizing the inherent
emphasis of the thesis project on producing new experimental data and physico-chemical
concepts, the modelling of swale operation was excluded from the above knowledge gap
analysis.
A brief overview of the state of knowledge of urban grass drainage swales in this chapter
indicates that the analysed studies were conducted mostly in the temperate climate,
without seasonal snow, with the exception of the pioneering work by Bäckström (2003).
To achieve a good control of experimental conditions, laboratory or field research studies
mostly considered well-defined but less-complex class of swale layouts:
(i) Land cover/use serviced by drainage swales: mostly adjacent to urban roads or
highways, very few studies addressed parking lots, or residential lands
(ii) Generation of runoff inflow – mostly by irrigation water, or by actual rain events;
rarely by snowmelt (from drained surfaces, or stored snow)
(iii) Runoff inflow into swales – in studies applying swale irrigation, the inflow entered
at the upstream end only, with a few exceptions of supplementing the longitudinal inflow
with lateral inflows as well; in studies of actual rainfall events, both longitudinal and lateral
inflows were considered; lateral inflow – typically from one side only
(iv) swale cross-sections – mostly trapezoidal or triangular; many studies addressed
treatment in the bottom section only, neglecting treatment/infiltration on side slopes
(v) swale surface – turf, natural or synthetic (the latter was used in lab studies); rarely bare
earth in lab studies comparing turf with bare earth
(vi) swale soils – investigated in some field studies, within some distance (0-5 m) from
the road pavement and typical depths (0-30 cm)
(vii) swale water quality process studied – a vast majority of studies focused on solids
settling as the most important quality enhancement process; relatively few studies pursued
stormwater filtration through swale turf and soils, or the resulting effects on soil chemistry
The licentiate phase of the planned PhD project should strive to reduce or close the
above knowledge gaps by focusing on stormwater quality enhancement in urban grass
swales providing drainage and snow storage for various land use (for comparative
purposes), conveying actual runoff and snowmelt entering the swale at the upstream end
as well as on one or both sides, and focusing on the treatment of stormwater by infiltration
into swale soils.
-
15
3. Methods
3.1. The literature review The aim of the critical review paper was to provide a systematic overview of the state-
of-the-art knowledge of processes that serve to remove pollutants from stormwater runoff
flowing over grass surfaces, with respect to influential factors. The primary sources of
information were laboratory, field and modelling studies of stormwater quality processes
occurring during stormwater runoff over standard and dry grass swales and grass filter
strips (GFS). Literature research focused on peer reviewed articles, academic theses,
conference proceeding papers, books, reports and design guidelines in Scopus, Web of
Science and Google Scholar databases. The references listed in the reviewed articles were
also examined. Searches included a variety of related keywords e.g. “grass swale”,
“vegetative swale”, “grass ditch”, “drainage swale”, “dry swale”, “grass filter strip”, etc.
3.2. Study sites In this thesis project, three grass swales serving for stormwater drainage and seasonal snow
storage were selected for study using such selection criteria as: (i) Well-functioning swales
with clearly delineated inflows (on one or both sides) and outflows, (b) Coverage of a
variety of sites with various soils, land use and traffic intensity, and (c) A general suitability
with respect to the site proximity, access and field crew safety. Three sites meeting these
conditions were selected in the City of Luleå, Sweden (Paper II and III). The climate at
the study location is a cool temperate climate, characterized by long winters, with the
snow season starting in October-November and snow remaining on the ground until
April. The mean annual temperature is 1.4 ºC. The studied sites represent different land
use types, i.e., a swale in a commercial catchment (swale L1), a swale next to the busiest
road in the city in the downtown area (swale L2), and a swale in the residential catchment
(swale L3). The first swale (L1) receives runoff from a parking lot (408 m2), a small part
of a building roof (5 m2), and a single-lane road (241 m2) with the average daily traffic
(ADT) of ~ 2,750. The second swale (L2) receives lateral stormwater runoff from a two-
lane road (728 m2) with the highest traffic intensity (ADT ~ 11,650) among the studied
locations. This swale receives road runoff only from one side, because the other side
features a continuous curb preventing any stormwater runoff discharge into the swale
(further called the no-runoff (NR) side). The third swale (L3) receives runoff from a
parking lot (287 m2), a roof (812 m2), a grassed area (726 m2), and a two-lane road (520
m2) with ADT ~ 2,500. The age of the three studied swales was estimated from the years
of construction of the roads next to the swales; thus, the swales years of operation at the
time of the soil sampling campaign was 57 years for swale L2, and 38 years for swales L1
and L3. There are uncertainties concerning the years of swale construction, which
depended on when the road was completed and possible swales modifications in the
following years, resulting from road reconstruction or other building activities in the
catchments. All the three swales are used for snow storage during winter road
-
16
maintenance, which includes the clearance of snow from the roads and parking lots
adjacent to the swales and applications of anti-skid materials (grit). Road salt (NaCl) is
applied only as an additive to grit material, to prevent its freezing and formation of clumps
in the grit material. Such a salt/grit mixture is applied only in early or late winter, when
temperatures are above -6° C and salt is effective in melting the ice layer formed on grit
particles. In early spring, after the winter season (end of April-beginning of May), the
residual grit is brushed off the roads and parking lots and collected for disposal. As an
example, Figure 2 shows swale L1 in the commercial catchment, before and after the
winter. It can be seen from Figure 2 that the stored snow may remain on the swale
ground even after the sweeping and removal of the residual grit from the roads and
parking lots.
Regular maintenance of the three studied swales includes: (i) regular mowing of grass in
the summer, and (ii) removal of gravel accumulations from the swales, which is done
once a year in early spring, after snow melted away and swales became dry, but before
the grass layer was established.
3.3. Soil sampling
In each swale studied, a 20 m long section was selected for soil sampling, which was done
in October 2017, using a stainless-steel core sampler with a 5 cm diameter and the length
of 30 cm. The section received only direct lateral runoff from the adjacent road and/or
parking lot, and the measured soil chemistry was used to examine if there were differences
in metal concentrations in soils draining different land covers. Because of soil
characteristics variation along the swale, samples were collected at three cross-sections 10
m apart to allow for statistical analysis. In order to investigate the metal concentrations
along the runoff flow path, at each cross-section, samples were collected at the distances
Figure 2: Swale L1 in the commercial catchment. The picture on the left side shows the swale before
the soil sampling campaign (September 2017) and the picture on the right shows the same swale after the winter (April 2018).
-
17
of 40 and 80 cm from the edge of the pavement and at the deepest point of the cross-
section, in the swale bottom section. Using a stainless-steel knife, each soil core was
divided into 5 cm slices representing individual samples, which were placed in a plastic
bag, refrigerated and kept in cold storage (up to 7 days) until analysed. At all three swales,
top three soil layer samples (0-5, 5-10 and 10-15 cm) and the deepest layer sample from
the swale bottom section were analysed, while for the swale sides there were some
exceptions:
(i) In swale L3, no soil samples could be collected from the swale side draining
the parking lot, because of the presence of gravel from the parking lot
construction.
(ii) Only the top layer (0-5 cm) samples from the side draining road (L3) and
parking lot (L1) were analysed, because some deeper soil layers, at a 40 cm
distance from the pavement edge, were highly compacted and did not allow
sample extraction.
(iii) In swale L2, only the top layer (0-5 cm) samples from the no-runoff side were
analysed.
Figure 3 shows an example of sample distribution at three cross-sections of swale L1;
the black coloured symbols identify, which samples were analysed. The same sampling
pattern was applied in swales L2 and L3, with minor exceptions listed above. In total,
96 individual soil samples were collected and analysed.
Figure 3: Distribution of 30 cm deep soil cores collected at the three studied swales (obtained from
Paper III). Black colour indicates, which samples were subject to the analysis at swale L1. All lengths are
in cm, unless indicated otherwise.
-
18
In order to investigate the swale topography and runoff contributing areas, location data
(x-y-z coordinates) were collected at numerous points along the swale using a real-time
kinematic-GPS device (model GeoMax Zenith35 Pro TAG) with the precision of 1.5
cm (for x and y) and 2 cm for z. The location data was used to build the TIN (Triangular
irregular networks) surface in AutoCAD Civil 3D software.
3.4. Infiltration measurements In order to investigate the swale infiltration capacity, field measurements were performed
in September 2018 using the Modified Phillip Dunne (MPD) infiltrometer (ASTM,
2018). Infiltration measurements were performed at undisturbed sites, which were
covered with turf, along the three sampled cross-sections (Figure 3), within ~ 30 cm of
the corresponding sampling points, and at two additional points at 120 and 200 cm
distances from the pavement edge. The saturated hydraulic conductivity was calculated
according to the method developed by Upstream Technology Co. following the ASTM
standard (ASTM, 2018). The best-fit values of saturated hydraulic conductivity (Kf,best_fit)
were calculated using the method developed by Weiss and Gulliver (2015):
Kf,bestfit = 0.32(Kf,arit) + 0.68 (Kf,geo) (1)
where,
Kf,arit and Kf,geo represent the arithmetic and geometric means of the saturated hydraulic conductivity, respectively.
The measured data was also compared to the literature data on infiltration capacities of
soils of various textures to inform about the soil texture of the studied sites.
3.5. Laboratory analysis
3.5.1. Soil parameters
Soil samples were prepared according to the standard ISO 11464 (2006) with minor
changes, and analysed for electrical conductivity (EC) and pH in the university
laboratory. The samples were air dried and dry sieved in the laboratory using a vibratory
sieve shaker (Retsch AS200) and a stainless sieve (mesh size of 2 mm). Soil lumps
remaining on the sieve were crushed using pestle and mixed with the < 2 mm fraction
(ISO 11464, 2006). The fraction > 2 mm, which generally included grass roots and
stones, was excluded from analyses. Measurements of EC were done according to the
standard ISO 11 265 (1994) using the CDM210 conductivity meter. Measurements of
pH (in a 1:5 suspension of soil in water) were done according to the standard ISO 10390
(2005), using the WTW pH 330 instrument. Chloride and loss on ignition (LOI) analyses
were done by an accredited commercial laboratory (ALS Scandinavia AB, Luleå).
Chloride analysis was done according to standard DIN EN ISO 12457–4 (2003). The
LOI analysis was determined at 1000°C and reported in % of sample dry weight (DW).
-
19
3.5.2. Analysis of total metal concentrations
Total concentrations of 13 metals have been examined in this thesis. The group consists
of common urban-related metals Zn, Cu, Pb, Cd, Cr, Cu, Ni and Co, all of which,
except Co, are considered stormwater priority pollutants (Eriksson et al., 2007). This
group was expanded to include W, Mn, Ti, V and Ba, which were all reported as traffic-
related elements, originating from such sources as e.g., asphalt, tire and brake wear, and
tire studs (Apeagyei et al., 2011; Mummullage et al., 2016; Huber et al., 2016). Zr was
also selected since it was validated as a tracer exhibiting a concentration deficit in sediment
accumulations in Sustainable Urban Drainage Systems (Tedoldi et al., 2018). This deficit
is caused by dilution of sediments containing Zr from anthropogenic sources by sediments
of mineral origin. Analyses of total metal concentrations were done by ALS Scandinavia
AB in Luleå. Metals Cd, Cu, Co, Ni, Pb and Zn were determined by digestion in a
heating block with nitric acid, while for the remaining metals (Cr, V, Ba, Mn, Ti, W
and Zr), 0.1 g of dried sample was fused with 0.4 g LiBO2 (lithium metaborate) and
subsequently dissolved in dilute nitric acid. The total metal concentrations were analysed
using Inductively Coupled Plasma Sector Field Mass Spectrometry (ICP-SFMS)
following SS EN ISO 17294-1, 2 and EPA-method 200.8. All metal concentrations were
reported in mg/kg DW except for Mn and Ti, which were reported as MnO and TiO2,
respectively, and converted to mg/kg DW. The laboratory performing the analysis
reported analytical uncertainties in concentrations of Cd, Co, Cu, Pb and Zn as 19-33%
of the reported values.
3.5.3. Sequential extraction analysis
Results of analyses of total metal concentrations were complemented by results from a
five-step sequential extraction analysis and the residue analysis. In this procedure,
extractants of increasing reactivity are sequentially applied, so that the successive fractions
exhibit lesser mobilities and lower risks of metal release due to changes in the ambient
environmental chemistry. Such changes may include the changes in pH and other factors
(Stone and Marsalek, 1996). A set of 11 soil samples from swale L2, which was noted for
the highest metal concentrations among the three swales studied, were selected for this
analysis. The selected samples included three top layer samples (0-5 cm) - two on the
road shoulder and one on the no-runoff side; and, eight samples from two soil cores from
the swale bottom section. Each core comprised samples from four layers of successively
increasing depths. The selected samples were analysed by ALS Scandinavia AB in Luleå
using a five-step sequential extraction analysis, and the residual analysis, following the
method adopted from Hall et al. (1996a, 1996b). The total metal concentrations were
analysed using ICP-SFMS following SS EN ISO 17294-1, 2 and EPA-method 200.8.
Analytical uncertainties in the reported concentrations of all metals in all steps were in
the range of 17-37% of the reported values, except for Zn in step 2, which had higher
uncertainty (range of the uncertainty for the 11 samples analysed was 48-62%). Samples
were ground prior to the first extraction step. Concentrations were reported in µg/L and
recalculated to mg/kg DW. The five extraction steps are listed below:
-
20
Step 1 (Fraction 1): Extraction of 1 g sample with 10 ml 1.0 M acetate buffer (pH 5)
by shaking for 6 h at room temperature to remove and measure adsorbed and
exchangeable metals and carbonates.
Step 2 (Fraction 2): Extraction of the solid residue from Step 1 with 50 ml 0.1 M
pyrophosphate solution (pH 9) by shaking for 1 h at room temperature to remove and
measure labile organic forms, which are the forms associated with reaction sites such
as those present in humic and fulvic substances (Hall et al. 1996b).
Step 3 (Fraction 3): Extraction of the solid residue from Step 2 with 10 ml 0.25 M
hydroxylamine hydrochloride for 4 h at 50°C to remove and measure amorphous
Fe/Mn oxides.
Step 4 (Fraction 4): Extraction of the solid residue from Step 3 with 15 ml 1 M
hydroxylamine hydrochloride in 25% acetic acid for 3 h at 90°C to remove and measure
crystalline Fe oxides.
Step 5 (Fraction 5): Removal and measurement of stable organic forms and
sulphides by adding 0.75 g potassium chlorate to the solid residue from Step 4 followed
by adding 15 ml 12 M hydrochloric acid for 30 min at room temperature and then 10
ml 4 M nitric acid for 20 min at 90°C.
Additionally, metal residuals were also determined. The residual content of Ba, V and
Cr, was determined according to ASTM D3682: 2013 and ASTM D4503: 2008 (fusion
with LiBO2). For obtaining the residual content of Cd, Ni, Pb, Zn, Cu, and Co, the
samples were digested with HNO3/HCl/HF according to SS EN 13656: 2003. The ICP-
SFMS analyses were carried out according to SS EN ISO 17294-2: 2016 and EPA-
method 200.8: 1994. The residual concentrations were reported in mg/kg DW.
Analytical uncertainties in the reported residual concentrations of metals were 14-34% of
the reported values.
3.6. Grit material applied during the winter road maintenance Three samples of stocked anti-skid grit materials were collected from the municipal
storage in April 2018:
- Material A (aggregate sizes 2-6 mm)
- Material B (aggregate sizes 4-8 mm)
- Material C (aggregate sizes 0-6mm) + salt
This material is applied on the roads, parking lots and bicycle paths throughout the winter
and, once applied, may be ground by vehicle tires.
A single sample of each material was analysed for total metal concentrations by the ALS
Scandinavia AB laboratory in Luleå. Prior to the analysis, the material was crushed and
the total metal content was analysed as described in section 3.5.2. Analytical uncertainties
-
21
in concentrations of Cd, Co, Cu, Pb and Zn were reported by the laboratory as 18-29%
of the reported values.
3.7. Data analysis The most data analysis was performed using Microsoft Excel. In Paper II, the proprietary
conceptual model StormTac Web was used, as explained in detail in section 3.10.
Statistical software MiniTab was used in Paper III to examine the normality of data (using
the Anderson-Darling normality test). Because some variables were not normally
distributed, the Spearman rho correlation coefficient (ρ) was calculated in MiniTab to
examine correlations among different metals and between metal concentrations and soil
properties. The correlation is considered strong if ρ≥0.60, and significant, if the p-value
is
-
22
The mean metal concentration (C) [mg/kg] and standard deviation (STDEV) were
calculated in each 5 cm layer using all the samples collected in that layer. In the case of
swale L2, which receives stormwater runoff only from one side, the no-runoff side
samples were excluded from calculations of metal burdens. Since no samples were
analysed in the layers 15-20 and 20-25 cm of swales L1 and L3, the mean metal
concentrations and STDEV for those layers were assumed equal to those in the layers 10-
15 and 25-30 cm, respectively.
During dry sieving of soil samples, each sample was split into two sub-samples with
particles 2 mm, and the corresponding sub-sample masses were recorded
for further use in calculating the mean total soil mass in the 5 cm slice (msample [g]) as well
as the fraction of the total sample material < 2 mm [%]. The volume of a 5 cm slice from
the soil corer equals the sample volume (Vsample [m3]), which can be calculated as the
volume of a cylinder with a 5 cm diameter and the height of 5 cm. Soil density (ρsoil
[kg/m3]) was then calculated from the total soil mass in the 5 cm slice and the sample
volume (equation 2):
ρ𝑠𝑜𝑖𝑙 [kg
𝑚3] =
𝑚𝑠𝑎𝑚𝑝𝑙𝑒[𝑔]
1000𝑉𝑠𝑎𝑚𝑝𝑙𝑒 [𝑚
3](2)
Each 5 cm swale soil layer volume (Vlayer) [m3] was calculated as a product of the swale
area [m2] and the layer thickness (5 cm). The area included an upstream swale section,
which was not sampled and, therefore, it was assumed that the soil chemistry data from
the sampled 20 m swale section was representative for the entire swale length.
The mass of soil in each 5 cm layer (Msoil,layer) [kg] was calculated as a product of the soil
density (ρsoil) and the layer volume (Vlayer). Since the metal concentrations were analysed
only in the sieved material (< 2 mm), the mass of soil < 2 mm in each 5 cm layer (M
-
23
3.8.1. Background concentrations of metals in swale soils
Because the metal content of swale soil samples is a result of both the soil background
(natural) content plus inputs from anthropogenic activities, the background soil metal
burden (Mback) [kg] was subtracted from that calculated from soil samples (as explained in
the previous section). The thought that the samples from the no-runoff side of swale L2
could be used to obtain the native soil chemistry was rejected, because only top soil
samples (0-5 cm) were analysed at that location, and furthermore, such samples could be
contaminated by polluted snow cleared from the road into the swale and also by
atmospheric deposition. Since all the three swales were built using native soils, the metal
concentrations in the deepest analysed layer (25-30 cm) were assumed to provide the best
estimates of the native soil metal content (Table 2).
Table 2: Mean metal concentrations and standard deviation (STDEV) in the deepest sampled layer
Cu [mg/kg DW] Pb [mg/kg DW] Zn [mg/kg DW]
L1 (25–30 cm) 5.7 ± 1.1 2.9 ± 0.8 16.6 ± 3.0
L2 (25–30 cm)1 19.3 ± 1.8 80.1 ± 11.1 61.4 ± 8.5
L3 (25–30 cm) 7.1 ± 0.7 4.4 ± 1.3 17.3 ± 4.5
1Because only two samples were collected from the deepest layer in swale L2, instead of the standard deviation, the
differences of the actual concentrations from the mean of the two samples available for the layer are shown in the
table.
Cumulative metal burden (Mtot) for the 30 cm thick soil layer was calculated as a sum of
metal masses in individual layers (Mmetal), reduced by the background metal burden in
each layer (Mback). Uncertainty of the cumulative metal burden (stot) was calculated by
considering the uncertainty of the mean metal concentrations in the layer (STDEV) using
the law of propagation of uncertainties (equation 4). In such calculations, the
uncertainties in estimating the mass of soil < 2 mm in each 5 cm layer (M< 2 mm) were not
accounted for.
𝑠𝑡𝑜𝑡 = √𝑠𝑚𝑒𝑡𝑎𝑙,12 + 𝑠𝑚𝑒𝑡𝑎𝑙,2
2 + … + 𝑠𝑚𝑒𝑡𝑎𝑙,62 + 𝑠𝑏𝑎𝑐𝑘,1
2 + 𝑠𝑏𝑎𝑐𝑘,22 + ⋯ + 𝑠𝑏𝑎𝑐𝑘,6
2 (4)
where,
smetal,i – uncertainty of the metal mass in the layer obtained by multiplying the standard
deviation of metal concentrations in the layer by the mass of soil material
-
24
3.9. Modelling methods (StormTac Web) A proprietary source-based model StormTac Web was selected to simulate the annual
metal loads of Cu, Pb and Zn discharged into the studied swales and the annual metal
mass retained in their soils. This model was selected for the following reasons: (i) it is
widely used in Sweden for planning and design of stormwater treatment facilities (STFs)
and their maintenance, and (ii) it is a parsimonious model requiring little input data. The
input data included: (i) annual precipitation (rain + snow), (ii) the land use type, including
the corresponding runoff contributing area and the volumetric runoff coefficient, and (iii)
the StormTac Web database standard pollutant concentrations provided for each land use
(Larm, 2000).
The model estimates the annual stormwater pollutant load entering and leaving the swale
[kg/year] (Lin and Lout, respectively), from which the annual load retained in the swale is
determined using site-specific functions. StormTac Web database contains pollutant
reduction efficiencies [%] derived from flow-proportional input and output data for
specific STFs (e.g., swales, biofilters, etc.). The database also includes site-specific data of
the STFs, including the ratio of the STF area to the reduced watershed area (i.e. the
watershed area multiplied by the runoff coefficient). The model calculates regression
equations (RE [%]) for calculating the STF reduction efficiencies. The regression
equation for swales includes the ratio of the swale area to the reduced drainage area,
empirical field data of the pollutant removal by swale compiled in the StormTac Web
database, and additional site specific characteristics, such as inflow concentrations (Larm
and Alm, 2016). The annual load leaving the swale after treatment (Lout [kg/year]) is
calculated using equation 5:
𝐿𝑜𝑢𝑡 = 𝐿𝑖𝑛 −𝑅𝐸
100∗ 𝐿𝑖𝑛 (5)
The annual pollutant mass retained in the swale soil is calculated from the pollutant runoff
input loads minus the loads leaving the swale at the downstream end, after treatment (Lin-
Lout). The StormTac Web model calculates only the annual loads added to the soil from
the polluted stormwater and groundwater, without consideration of the native metal
mass. The retained annual metal loads calculated by the model were then multiplied by
the swale age to obtain the total metal loads to be compared to the loads estimated from
soil samples.
StormTac Web uses adjustment factors (F = 0-10) for each land use category, except the
roads, to calculate land use specific standard concentrations. A factor of 5 indicates
standard conditions of the land use, while factors < 5 and > 5 indicate that the
concentrations should be adjusted towards the minimum or maximum values,
respectively, in the model database. For the roads, the standard concentration is calculated
based on the road traffic intensity (ADT). Available historical ADT data for the three
studied sites were obtained from the Luleå municipality, in order to examine the
-
25
sensitivity of the calculated retained metal loads to the ADT values. Since the modelled
retained loads were little sensitive to historical variations in ADTs at the study sites, the
mean ADT was used for calculating the retained metal loads. Moreover, the default
options for the land use factor (F = 5) was used for other land covers as well (e.g., the
parking lot, roof etc.). This scenario was called a “standard scenario”.
In order to estimate the modelling uncertainty in calculating the yearly retained metal
mass in swale soils, two additional scenarios were developed for estimating the minimum
and the maximum annual metal loads retained in the swale. Two parameters were
adjusted to obtain those two scenarios, i.e., the land use input concentration and the
swale pollutant reduction efficiency, because these two parameters are directly related to
the modelling of stormwater quality. For the scenario serving to calculate the minimum
retained loads, the minimum traffic intensity from the available historical data was used,
and the minimum factor (F=0) was set for land uses other than roads to obtain the
minimum input concentrations (Cmin) and the associated minimum yearly metal loads
discharged into the swale. Moreover, the swale pollutant reduction efficiency was set to
the minimum values estimated from the StormTac Web database (REmin). For the
scenario estimating the maximum retained loads, the maximum factor (F=10) was set for
each land use and the maximum ADT from the available historical data for the study
location was used, in order to maximise the yearly metal loads into the swale. Also, the
maximum swale pollutant reduction efficiency (REmax) from the model database was set
for this scenario. Finally, the metal burdens calculated from model outputs from these
three scenarios (standard, min and max) were compared to the cumulative metal burdens
from soil samples (reduced by the native soil burdens).
-
26
-
27
4. Results This chapter presents synthesis of results in the following order:
(a) The soil characteristics that influence metal immobilisation in swales (infiltration
capacity, loss on ignition (LOI), pH and electrical conductivity (EC)) are presented
in Section 4.1. for the three swales studied,
(b) Short-term metal removal efficiencies of grass swales during runoff conveyance,
reported in previous research, are synthesized in Section 4.2.
(c) Metal burdens retained in swale soils, as a consequence of a long-term exposure to
stormwater runoff, are presented in Section 4.3.
(d) Total metal concentrations in the soils of the three studied swales are presented in
Section 4.4., and
(e) Metal content of the traction material used in the study area in winter road
maintenance is presented in Section 4.5.
4.1. Physico-chemical characteristics of soils in the studied swales
The critical review of grass swales and filter strips identified two main categories of
removal processes of pollutants, other than solids, in overland flow over grass swales and
grass filter strips. Firstly, removal of the particulate fraction (i.e., pollutants attached to
solids) by filtration through the grass and settling. Secondly, removal of the dissolved
fraction from stormwater runoff through infiltration into swale soils, where adsorption,
chemical precipitation, microbial degradation and plant uptake take place.
Thus, stormwater infiltration into swale soils, which can be estimated by measuring the
hydraulic conductivity (Kf), is particularly important for assessing swale operation. Best-
fit estimates of hydraulic conductivities (eq. 1 in Methods, section 3.5) for the three
studied swales, which were built using native soils (Paper III), are shown in Table 3.
Comparison of data from Table 3 to the recommended permeability values for swale
design (Kf > 1.3 cm/h; USEPA, 1999) and dry swales with engineered soils (Kf >3.6
cm/h; Ingvertsen et al. 2012) shows that all three studied swales meet or exceed the
recommended infiltration rates. Particularly, the swale side draining the parking lot next
to swale L3 exhibits significantly higher infiltration rates compared to other swale
sections. This can be explained by the presence of gravel leftover from the parking lot
construction, which also interfered with soil sample collection mentioned in Section 3.5.
-
28
Kf,best fit [cm/h]
Swale L1 L2 L3
R40 20.7 4.4 1.4
R80 11.2 4.8 4.9
R120 8.6
8.4
R200
21.4 4
Bottom 6.8 7.9 9.9
PL80 4.8
72.4
PL40 9.3
143.4
Examination of infiltration rates can provide an insight into the contact time between the
percolating stormwater and the soil media. High hydraulic conductivities enable
infiltration of significant stormwater volumes into the soil, but also accelerate transport
through the soil matrix. The contact time affects adsorption of the dissolved metals;
longer contact times result in more effective adsorption of metals, until an equilibrium
state is reached (Yousef et al., 1985; Scholes et al., 2008). For example, Pb is mostly
associated with the particulate fraction, while Cu and Zn may also occur in appreciable
quantities in the dissolved fraction (Huber et al., 2016). In order to examine whether
there is a relationship between the measured infiltration rates and the metal
concentrations in the soil, Kf,best fit was related to the mean Zn, Pb and Cu concentrations
(in all the layers) at a certain distance from the pavement edge, as shown in Table 4. No
relationship was observed between the hydraulic conductivity and the mean metal
concentrations (Zn, Pb, and Cu). Achleitner et al. (2007) also did not find any specific
relationship between the hydraulic conductivity and mean metal concentrations of Zn,
Cu, and Pb. When comparing hydraulic conductivities Kf, best fit for the bottom section of
the three studied swales (6.8-9.9 cm/h) against those reported by Rawls et al. (1982) for
soils of various textures, the calculated Kf, best fit corresponded to the soils classified as a
loamy sand (6.11 cm/h). Other soil properties, which may affect immobilisation of metals
in the soil matrix, i.e., LOI, pH and EC, are presented in Tables 5 and 6.
Table 3: Best-fit estimates of the saturated hydraulic conductivity Kf [