MEASUREMENT OF NITROGEN DIOXIDE AND SULFUR DIOXIDE BY
SATELLITE AND PASSIVE MONITORS IN NORTHEASTERN BRITISH
COLUMBIA, CANADA

By

S M Nazrul Islam
M.S., Jahangirnagar University, 2006

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
December 2014

© S M Nazrul Islam, 2014

UMI Number: 1526512

All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

Di!ss0?t&iori Publishing
UMI 1526512
Published by ProQuest LLC 2015. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346

Abstract
The Peace River district o f Northeastern British Columbia (B.C.) Canada is a region of
natural gas production that has undergone rapid development since 2005. Both satellite data
products and Willems badge passive sampler measurements of nitrogen dioxide (NO 2) and
sulfur dioxide (SO2) were used to assess the air quality implications from gas development
activities. Both satellite data products between 2005 and 2013 and Willems badge passive
samplers during six two-week exposure periods between August and November, 2013 have
been considered in this study. All satellite data products and passive monitoring of these two
pollutants in Northeastern B.C. found higher values in Taylor, Fort St. John, and Dawson
Creek. This spatial distribution of higher values has resulted from the large gas development
activities in these areas. The temporal analysis o f satellite NO 2 data revealed higher values
near Dawson Creek after 2007 with annual increment of 1.7%. It was also found that Taylor
is half as polluted as one o f the Canada’s largest non-urban SO2 emission source areas
(Canadian oil sands area in Alberta).

TABLE OF CONTENTS
Abstract

ii

Table of Contents

iii

List of Tables

v

List of Figures

vi

Acknowledgement

ix

Chapter One

Study Rationale and Research Questions
1.1.
Background
1.2.
Rationale of the study
1.3.
Objectives and research questions
1.4.
Layout o f the thesis
1.5.
References

1
1
2
4
5
7

Chapter Two

Literature Review and Methodology
2 .1 .
Introduction
2 .2 .
Literature review
2 .2 . 1 .
Natural gas development in Northeastern B.C.
2 .2 . 1 . 1 . Air quality aspects of gas development activities
Satellite remote sensing o f trace gases in the
2 .2 .2 .
troposphere
Global Environmental Multi-scale - Modeling Air
2.2.3.
Quality and Chemistry (GEM-MACH) model
2.2.4.
Passive monitoring o f air quality
2.3.
Methodology
2.3.1.
Study area
Satellite data products
2.3.2.
2 .3.2.1. Satellite data filtering
2 .3.2.2. Pixel averaging approach
Passive monitoring network
2.3.3.
2.3.3.1. Passive sampler preparation and analysis
2 .3.3.2. Calculation o f ambient concentrations from passive
samplers
2 .3.3.3. Accuracy and precision o f passive samplers
2.4.
References

9
9
9
9

Chapter Three

Satellite observations of Nitrogen dioxide and Sulfur dioxide
over Northeastern B.C., Canada
Abstract
3.1.
Introduction
3.2.
Satellite data products o f air quality

iii

12

15
18
19
21
21
22

23
24
25
27
31
32
34

39
39
39
44

3.3.
3.4.
3.5.
3.6.
3.7.
Chapter Four

Chapter Five

Tropospheric NO 2 column densities over Northeastern
B.C.
Ambient NO 2 concentration
Tropospheric SO2 column densities over Northeastern
B.C.
Conclusions
References

Passive monitoring measurements of Nitrogen dioxide and Sulfur
dioxide concentrations over Northeastern B.C., Canada
Abstract
4.1.
Introduction
4.2.
Methods
4.2.1.
Study area and design
4.2.2.
Sampler preparation and analysis
4.2.3.
Accuracy and precision o f passive samplers
4.3.
Results and discussion
4.3.1.
Ambient concentration o f NO 2 in Northeastern B.C.
4.3.2.
Ambient concentration o f SO2 in Northeastern B.C.
4.3.3.
Comparison o f passive with active (continuous)
monitor at collocated sites

49
54
55
58
61

72
72
73
78
78
80
83
84
84
87
89

4.4.

Conclusions

91

4.5.

References

93

Conclusions and recommendations
5.1.
Introduction
5.2.
Summary o f results and conclusions
5.3.
Recommendations for future work

Appendix

iv

104
104
104
108
109

List of Tables

Table 2.1. Basic characteristics o f space-borne instruments.

18

Table 2.2. Cross-track raw anomalies.

18

Table 2.3. Passive sampler deployment site IDs, nearest towns, location (latitude,
longitude), elevation, sampling period, and site location description.

27

Table 4.2. Passive measurement o f NO 2 , and SO2 concentrations (ppb) in
Northeastern B.C., Canada. The CV is also provided in the parentheses.

98

Table 4.3. B.C. ambient air quality objectives.

99

v

List of Figures
Fig. 2.1

Fig. 2.2

Fig. 2.3

Fig. 2.4

Fig. 3.1

The study area, (a) Unconventional gas play trends in Northeastern B.C.
(modified after OGC, (2012a)). White squares refer to towns, (b) Overall
gas production in B.C. from both conventional and unconventional wells
(modified after OGC, (2012a)), (c) Conventional oil and gas wells in 2012
in Northeastern B.C. (source: NPRI database, available at
http://www.ec.gc.ca/inrp-npri/default.asp?lang=en&n=lD892B9F-l.
accessed on: 20 September 2014), however, an unconventional wells
distribution map is not currently available in the NPRI database, and (d)
Passive sampling sites in the study area. Red squares refer to the major
towns across the study area. The passive monitoring network with major
towns is also presented in Fig. 2.3.

14

The location of the Montney formation in B.C. and Alberta, (a) Montney
formation with major rock types in B.C. and Alberta (adapted from NEB,
(2013)), (b) Montney regional fields in B.C. with dry, rich (wet) and oil
distribution (adapted from OGC, 2012a).

15

Study area with the locations o f 24 passive monitoring stations (+ symbols).
The study domain in northeast B.C. consists of 24 sites; Plaza 400 (in the
Prince George air shed, not included in the figure) was selected in order to
validate the passive sampler results with continuous monitoring data of NO 2
and SO2 . Besides Plaza 400, Site 14 and Site 16 were also chosen to
collocate with continuous monitoring stations; each of these sites measures
only SO2 (except Plaza 400 which also measures NO 2) along with
meteorology and data are archived by the B.C. Ministry o f the Environment
(available at: http://www.bcairqualitv.ca/readings/index.html~). Each o f the
total of 25 stations are exposed to ambient conditions from 15 Aug. 2013 to
mid Nov. 2013 with samplers being exchanged every two weeks (six
measurements per site), major towns o f the study area are also indicated
with red circles. The B.C. - Alberta border is along 120°W longitude.

26

a) The ‘Willems badge’ passive sampler: 1) Velcro® for sampler
deployment, 2) sampler body with opening at one end, 3) absorbent pad, 4)
spacer ring, 5) Teflon® membrane filter, 6 ) spacer ring, and 7) cap,
(adapted from Zbieranowski and Aheme, 2012a); b) Passive samplers were
deployed in the field o f this study.

30

OMI annual mean tropospheric NO 2 VCDs across a large area (B.C. and
Alberta), averaged over 2005-13 shown on a 6.0 x 6.0 km 2 grid and
calculated using different averaging radius. Plots a) and b) are showing
tropospheric VCDs from DOMINO-NO 2 and SP-NO 2 data products,
respectively with a 24 km averaging radius. An averaging radius o f 60 km
is used for the two NO 2 data products in c) DOMINO-NO 2 and d) SCIANO 2 . Note pixels from the both edges o f the swath o f OMI data products

VI

were not removed for plots (a), (b), and (c). The locations of Victoria,
Vancouver, Prince George, Fort St. John, Kamloops, Grande Prairie,
Calgary, Edmonton and Fort McMurray are indicated by different symbols
from left to right in each plot. The white rectangle o f each plot indicates the
present study area (Northeast B.C.).
Fig. 3.2

Average (2005-2013) tropospheric VCDs: (a) DOMINO-NO 2, (b) ECDOMINO-NO 2 , (c) SP-NO2, and (d) EC-SP-NO 2. Marker notation o f each
plot from left to right: Chetwynd, Taylor, and Dawson Creek

Fig. 3.3 Temporal variation of EC-SP-NO 2 : (a) 2005-2007, and (b) 2008-2013.
Marker notation of each plot from left to right: Chetwynd, Taylor, and
Dawson Creek.
Fig. 3.4 Comparison of NO 2 data products (2005-2013): (a) DOMINO-NO 2, and (b)
SCLA-NO2 . Here averaging radius is 60 km in both cases. Marker notation
of each plot from left to right: Chetwynd, Taylor, and Dawson Creek.
Fig. 3.5 Average (2005-2013) EC-OMI spatial distributions o f NO 2 vmr: (a) ECvmr, (b) GEM-MACH-vmr, (c) number of signals from EC-OMI-vmr in
each circle of 18 km radius, and (d) the same as (c) but using GEM-MACHvmr.

Fig. 3.6 Average (2005-2013) tropospheric SO2 VCDs: (a) NASA-SO 2-VCDS, (b)
SNR ofN A SA -S0 2-VCDs. Marker notation o f each plot: Taylor Town Site
station (+), Taylor South Hill Station (x), Taylor (o), and McMahon gas
processing plant (A).
Fig. 4.1

Site and exposure specific concentrations: (a) NO 2; and (b) SO2 . NO 2
passive samplers were not analyzed for exposure 5. The horizontal dash line
(red) in both figures represents the LOD.

Fig. 4.2 Spatial distribution of concentrations: (a) NO 2, and (b) SO2 . The units of
both NO 2 and SO2 in the color bar are ppb.
Fig. 4.3 Site and exposure specific box plots of both species. Values of all periods of
exposures were considered for site specific box plot (a. NO 2 , b. SO2 ) while
all sites values during each exposure period provide exposure specific box
plot (c. NO 2 , d. SO2). The line across the box represents the median,
whereas the bottom and top of the box show the locations o f the first and
third quartiles (Qi and Q 3). The whiskers are the lines that extend from the
bottom and top o f the box to the lowest and highest observations inside the
region defined by Qi-1.5(Q 3 - Qi) and Q 3 + 1.5(Q3 - Qi). The hinges in ‘c’
and ‘d’ show the 95% confidence interval of the median. Individual points
with values outside these limits (outliers) are plotted with ‘+’ signs.

Fig. 4.4

Fig. 5.1

Fig. A1

Comparison of passive with active (continuous) monitor at collocated sites.
a) passive NO 2 vs active NO 2 (continuous data from Plaza 400 station in
Prince George), b) passive SO2 vs active SO 2 (three active stations over a
total six period o f exposures have been considered, see Table 2.3). The
equation of the linear regression line (linear fit) o f each plot is also
included. 1:1 line is also provided in both figures for visual aid.

103

Spatial distribution o f satellite and passive observations of NO 2 and SO2 in
Northeastern B.C. a) Average passive NO 2 observations (Fig. 4.2a) and
long-term average EC-SP-NO 2 VCDs (Fig. 3.2d); b) average passive SO 2
observations (Fig. 4.2b) and long-term average SO2 VCDs (Fig. 3.6a). Note
that all passive SO2 locations are not included here since satellite SO 2
analysis was taken in a relatively small area. Also note that relative size of
circle in each plot refers the variation of concentrations (not using the same
scale of both plots).

107

Average (2005-2013) tropospheric SCD: (a) DOMINO-NO 2, (b) ECDOMINO-NO 2, (c) SP-NO 2, and (d) EC-SP-NO 2 . Marker notation o f each
plot from left to right: Chetwynd, Taylor, and Dawson Creek.

109

v iii

Acknowledgement

This thesis would not be possible without the help of a number of people. First and foremost
I sincerely thank Peter Jackson, my advisor, for sharing his enthusiasm for tropospheric air
pollution with me and for encouraging my independence in the field. I appreciated his quick
mind and ready laugh. I am indebted to Chris McLinden for introducing me to space-based
air quality data handling and for providing a delightfully personal component to my graduate
program. My appreciation also goes to Roger Wheate for being a committee member of my
graduate program.
Thanks to Raju Aryal and Talaat Bakri for helping me in the field work. I am also grateful to
many friends with whom I enjoyed a number of fun and stimulating intellectual exchanges
over the last two years. My parents, wife and daughter’s wonderful unconditional emotional

support throughout this graduate study lead to me work sincerely.
This graduate study was funded by BC Oil and Gas Commission (OGC), along with partial
supports from UNBC as Graduate Entrance Research Award and Peter’s NSERC Discovery
Grant. Jahangimagar University in Bangladesh is also acknowledged for providing me with
sufficient study leave to pursue the graduate study in UNBC.
Finally, I dedicate this thesis to my beloved parents who have given me a wonderfully
supportive childhood and exemplary work ethic.

IX

1. Study Rationale and Research Questions
1.1. Background
In 2012, the total production of natural gas in British Columbia (B.C.) was 40,482*106 m 3
(oil was 1 ,2 2 2 ><1 0 6 m3) with total estimated reserves (proven plus probable recoverable) of
1,138x 109 m 3 (oil was 19,108x 10 6 m3) which is a 146% increase over the natural gas
reserves estimated in 2006 (OGC, 2012). The exploitation of these vast reserves of natural
gas is a significant economic driver and revenue generator in the province, and as such, the
B.C. provincial government is planning on expanding this industry by promoting
development of liquefied natural gas (LNG) for export (MEM, 2012). This trend of
increasing reserve estimates is largely due to the successful development o f unconventional
gas extraction including the application of horizontal drilling and hydraulic fracturing
technology, which commenced extensively in 2007/2008. Most of this development has been
in the Montney formation and the Horn River Basin of Northeastern B.C. (Fig. 2.1 and 2.2)
and currently accounts for 60% o f B.C.’s total natural gas production (OGC, 2012). It is also
expected that natural gas from unconventional sources will continue to increase while
conventional pools will be depleted in the next few years (OGC, 2012). The Montney
formation is an unconventional Lower Triassic aged formation that includes dry, liquid rich
gas and oil in over-pressured siltstones that stretches over 200 km from the B.C. - Alberta
border near Dawson Creek to the B.C. foothills o f the Rocky Mountains. Recently, the US
EPA estimated that uncontrolled venting from guarding against over-pressuring wells during
well completions involving hydraulic fracturing can vent approximately 230 times more
natural gas to the atmosphere than wells without hydraulic fracturing (conventional well
drilling) resulting in significant air pollution (US EPA, 2011). People who lived within half a

1

mile of the unconventional wells had a greater risk o f developing non-cancer health effects
from short-term exposure to the high emissions of hydrocarbons than those living further
away (McKenzie et al., 2012).
1.2. Rationale of the Study
Concerns have arisen recently in Northeastern B.C. about increasing air pollution as a result
of accelerating natural gas production (Fraser Basin Council, 2013; Krzyzanowski, 2012;
MoE, 2014) without the simultaneous implementation o f available technological advances to
control emissions (Krzyzanowski, 2009). According to the 2010 emissions data reported to
Canada’s

National

Pollutant

Release

Inventory

(NPRI,

http://www.ec.gc.ca/pdb/websol/quervsite/auerv e.cfml. sulfur dioxide (SO 2) and nitrogen
dioxide (NO2) are the dominant species among all gaseous air pollutants in Northeastern B.C.
and are emitted from various stages o f oil and gas activities, mostly during the production
phase such as processing (e.g., flares, engines, and compressors), distribution (e.g., leaks of
pipelines and flanges), and from storage tanks (e.g., vaporization) (Krzyzanowski, 2012).
Once these two gaseous species enter the atmosphere they undergo chemical and/or physical
reactions which will subsequently contribute to the acidic deposition in terrestrial ecosystem
through both dry- and wet-deposition (Cox, 2003). These chemical compounds may impact
on human health through a number of environmental pathways including air, water, and soil
or through multiple pathways (Krzyzanowski, 2012). Although these two species have a
deleterious impact on the environment, only four permanent SO2 monitors have been
installed in the previous 15 years and there has been no monitoring o f N 0 2 (MoE, 2014),
which is considered insufficient by the Fraser Basin Council (2013) in the context of recent
unconventional gas development in the Montney formation. The location of these four SO2

2

monitors include: Pine River Hasler and Pine River Gas Plant near Chetwynd, Taylor
Townsite and Taylor South Hill near Taylor. Furthermore, communities in Northeastern B.C.
are dissatisfied by what they consider as insufficient information, from both the government
and the oil and gas sector, and the communities are also concerned with the perception o f a
lack of transparency with respect to specific oil and gas activities (Fraser Basin Council,
2013). Therefore, rigorous investigation and regulations are required for baseline assessments
o f air quality and adequate communication with the public should be conducted prior to oil
and gas resource development activities.
Recently it has been suggested from the satellite observations of NO 2 and SO2 over the
Canadian oil sands in Alberta (McLinden et al., 2012 and 2014) and space-based ambient
ground-level concentrations of N 0 2 (inferred from VCDs and model-simulated vertical
profiles) in North America (Lamsal et al., 2008) that satellite observations would be an
alternative and complementary to surface based measurements in Northeastern B.C. where
in-situ monitoring stations are limited or sparsely distributed. Therefore, the overall purpose
o f this study is to assess air quality between 2005 and 2013 using satellite remote sensing
observations of NO 2 and SO2 over portions of Northeastern B.C. that are undergoing a rapid
expansion in natural gas production. A fieldwork campaign with NO 2 and SO2 passive
samplers was also carried out in order to indirectly validate the remotely-sensed air quality
data. It is important to note that direct comparisons between VCDs and in-situ measurements
are difficult (McLinden et al. 2012) due to pixel-based (horizontally averaged) VCDs which
are vertically integrated column values assigned in the pixel center as the VCD of one pixel
(the smallest area of one pixel among all current satellite data products is 312 km ) while insitu measurements are point observations at ground level. To date, there has been no

3

systematic study of the satellite and passive monitoring measurement o f NO 2 and SO2 in
Northeastern B.C.
1.3. Objectives and Research Questions
The primary objective o f this research is to use satellite-based tropospheric observations to
analyze the air quality in Northeastern B.C. with particular focus on NO 2 and SO2 . The
spatial and temporal distribution of satellite observations over Northeastern B.C. will also
provide information on changes in air quality associated with oil and gas development
activities. The three months field study in 2013, using passive monitors and available in-situ
monitoring station information, will allow for an indirect validation o f the satellite
observations. In addition, this study will provide the necessary scope to do a comparative
analysis o f Northeastern B.C. with other known polluted areas such as Canadian oil sands in
Alberta. Specific questions addressed in this thesis include:
1) What are the levels o f these pollutants in this region?
2) A re the trends in satellite air quality observations between 2005 and 2013
related to the increased trends in oil and gas development activities in this
region?
3) H ow do these pollutants vary spatially and temporally?
4) A re satellite observations o f NO 2 and SO 2 from different data products
consistent?

This analysis uses several NO 2 satellite data products (VCDs) including the OMI (Ozone
Monitoring Instrument), SCIAMACHY (SCanning Imaging Absorption spectroMeter for
Atmospheric CHartographY) and the GOME-2 (Global Ozone Monitoring Experiment-2).
The OMI data products produced from different algorithms (e.g., NASA standard product,

4

DOMINO [Dutch OMI NO 2] data products, and EC [Environment Canada] data products)
have been tested. Only the NASA OMI data products have been used for SO 2 analysis
between 2005 and 2013. In addition to VCDs, air quality of Northeastern B.C. was analyzed
using space-based ambient concentration o f NO 2 in conjunction with a model-simulated
(GEM-MACH: The Global Environmental Multi Scale Modeling- Air Quality and
Chemistry) surface concentrations. Besides the pixel measurements, point observations were
also analyzed using Willems badge passive measurements of NO 2 and SO2 (August November, 2013) along with SO2 data from several in-situ monitoring stations.
1.4. Layout of the Thesis
This thesis involves two main components: 1) satellite observations of NO 2 and SO2 over
Northeastern B.C., Canada (Chapter three); and 2) passive monitoring o f NO 2 and SO2 over
Northeastern B.C., Canada (Chapter four). The satellite and model investigation using data
between 2005 and 2013 aims to address all the research questions o f this current study, while
the short term passive monitoring analysis also investigates this area separately with
particular focus on research question 1 and partially the question 3 (spatial distribution of
NO 2 and SO2). Each of the two main components has been written in manuscript format with
the intention o f submitting each chapter to a suitable journal for publication. Each manuscript
contains an abstract, introduction, and conclusion along with presenting data and methods
involved in each component and providing results with discussions for each section
separately. A reference list for each component is also provided at the end of each
manuscript.
Chapter two provides a background literature review regarding the satellite observations and
passive monitoring for the air quality study. A brief overview of oil and gas activities in

5

Northeastern B.C., with a particular focus on the unconventional Montney formation, is also
provided. In addition, a detailed discussion of all the methods associated with the analysis of
satellite data and passive monitoring have also been included in chapter two. Chapter five
provides a summary of all of results and a conclusion that can partially answer all o f the
research questions. Some recommendations for future work are also included at the end of
this chapter. The author of the present thesis is the principal investigator and the first author
of the two main chapters (three and four) as well as all other parts o f this thesis.

6

1.5. References
Cox, R.M., 2003. The use of passive sampling to monitor forest exposure to O 3, NO 2 and
SO2 : a review and some case studies. Environmental Pollution 126, 301-311.
Fraser Basin Council, 2013. Identifying Health Concerns Relating to Oil & Gas Development
in Northeastern B.C.-Human Health Risk Assessment Phase 1 Report. Available at:
http://www.health.gov.bc.ca/librarv/publications/vear/2 0 1 2 /Identifving-health
concems-HHRA-Phasel-Report.pdf (accessed on: 15 May 2014).
Krzyzanowski, J., 2009. The Importance of Policy in Emissions Inventory Accuracy-A
Lesson from British Columbia, Canada. J. Air & Waste Manage. Assoc. 59, 430-439.
Krzyzanowski, J., 2012. Environmental pathways o f potential impacts to human health form
oil and gas development in northeast British Columbia, Canada. Environ. Rev. 20,
122-134.
Lamsal, L.N., Martin, R.V., van Donkelaar, A., Steinbacher, M., Celarier, E.A., Bucsela, E.,
Dunlea, E.J., Pinto, J.P., 2008. Ground-level nitrogen dioxide concentrations inferred
from the satellite-borne Ozone Monitoring Instrument. J. Geophys. Res. 113,
D 16308, doi: 10.1029/2007JD009235.
McKenzie, L.M., Witter, R.Z., Newman, L.S., Adgate, J.L., 2012. Human health risk
assessment of air emissions from development of unconventional natural gas
resources. Science o f the Total Environment 424, 79-87.
McLinden, C.A., Fioletov, V., Boersma, K.F., Krotkov, N., Sioris, C.E., Veefkind, J.P.,
Yang, K., 2012. Air Quality over the Canadian oil sands: A first assessment using
satellite observations. Geophysical Research Letters 39, L04804.
McLinden, C.A., Fioletov, V., Boersma, K.F., Kharol, S.K., Krotkov, N., Lamsal, L., Makar,
P.A., Martin, R.V., Veefkind, J.P., Yang, K., 2014. Improved satellite retrievals of
N 0 2 and S 0 2 over the Canadian oil sands and comparisons with surface
measurements. Atmos. Chem. Phys. 14, 3637-3656, doi:10.5194/acp-14-3637-2014.
MEM (BC Ministry of Energy and Mines), 2012. British Columbia’s Natural Gas Strategy:
Fueling B.C.’s Economy for the Next Decade and Beyond. Available at:
http://www.gov.bc.ca/ener/popt/down/natural gas strategy.pdf (accessed on: 15 May
2014).
MoE (BC Ministry of Environment), 2014. Report on Initial Network Design-NE B.C. Air
Quality Network: BC Ministry o f Environment, Report to the SCEK Fund, BC Oil
and
Gas
Commission,
January
31,
2014.
Available
at:
http://www.bcairqualitv.ca/readings/northeast/pdfs/ne air monitor project report ne
twork design.pdf (accessed on: 15 May 2014).

7

OGC (BC Oil and Gas Commission), 2012. Hydrocarbon and By-Product Reserves in British
Columbia:
2012-BC
Oil
and
Gas
Commission.
Available
at:
httt>s://www.bcogc.ca/node/l 1111/download (accessed on: 15 May, 2014).
US EPA, 2011. Oil and Natural Gas Sector: New Source Performance Standards and
National Emission Standards for Hazardous Air Pollutants Reviews; 76 Federal
Register
52738,
August
23,
2011.
Available
at:
http://www.epa.gov/ttn/atw/oilgas/fr23aul 1.pdf (accessed on: 01 September 2014).

2. Literature Review and Methodology
2.1. Introduction
This chapter consists of two main sections: 1) a literature review; and 2) an overview of the
methodology. The literature review section introduces the topic o f natural gas development
in Northeastern B.C. with a particular focus on gas reserves in the Montney formation. The
air quality aspects o f these extensive gas development activities are briefly discussed
followed by a short theoretical background of satellite remote sensing o f tropospheric trace
gases and passive monitoring of ambient air quality. The methods section describes how the
NO 2 and SO2 data were collected and processed prior to analysis for the study area.
2.2. Literature review
2.2.1. Natural gas development in Northeastern B.C.
The BC Oil and Gas Commission (OGC), established in 1998, is the B.C. provincial single­
window regulatory agency with responsibilities for fair regulation of oil and gas development
activities (e.g., geophysical exploration, well drilling, pipelines, and gas processing) in B.C.
The services the OGC provide include reviewing applications for industry activity and
equitable participation in production, confirming industry complies with provincial
legislation and all regulatory requirements, coordinating with partner agencies, consulting
with First Nations, public safety, and protecting the environment (OGC, 2012a). Although
this is the single-window regulatory agency, it has become relatively less independent since
2002 as the provincial government appointed the Deputy Minister o f Energy and Mines to
the position of chair and director of the commission. This gives the Deputy Minister the
power o f casting a vote in case o f a tie and this casting vote is cause for concern because
environmental approvals will be decided by the chair of the commission whose Ministry’s

9

(Ministry of Energy and Mines) first objective is to increase investment in energy and
mineral resource development in B.C. (WCEL, 2003). Prior to 2002, environmental
approvals were issued by the Ministry o f Environment and then by a neutral commission
(OGC). The development of natural gas activities was fairly consistent from 1992 to 2003
followed by a significant expansion in development of natural gas activities (OGC, 2012a).
During the first couple o f years of the expansion period the production was dominated by
conventional wells (located in the B.C. part o f large Western Canadian Sedimentary Basin
which extends from southwestern Manitoba, southern Saskatchewan, Alberta, Northeastern
B.C. and the southwest comer o f the Northwest Territories) with only 20% o f B.C.'s total
production coming from the unconventional (horizontal drilling and multi-stage hydraulic
fracturing) Jean Marie and Deep Basin (DB) Cadomin drilling (Fig. 2.1b) (OGC, 2012a).
Contributions from conventional sources declined at 8 % per year with minimal drilling
occurring in 2012. In contrast, the development of the unconventional Montney and Horn
River regional plays has been increasing dramatically since its commencement in 2007/2008
(OGC, 2012a).
The Montney formation is an unconventional Lower Triassic aged formation that includes
dry, liquid rich gas in over-pressured siltstones (alternatively known as tight gas) that
stretches northwest 200 km from the B.C. - Alberta border near Dawson Creek to the B.C.
foothills of the Rocky Mountains (Fig. 2.2b). The unconventional mid-Devonian aged Horn
River Basin is a shale play with dry gas over-pressured shale (alternatively known as shale
gas) of the Muskwa, Otter Park and Evie Formations near Fort Nelson in Northeastern B.C.
(Fig. 2.1a). The areal extent o f the Montney formation in B.C. is approximately 29,850 km 2
(Fig. 2.2b) (OGC, 2012a) while the total area, including Alberta, is 130,000 km 2 (Fig. 2.2a)

10

(NEB, 2013). Although the total area in B.C. is small, the Montney formation’s most
marketable unconventional gas resources are located in this province. This is due to the
natural gas having a lower content o f natural gas liquid (NGL) and oil (natural gas with lower
content of NGL is generally known as dry gas) which makes it economically viable in the
largely siltstone dominant geologic formation with poor porosity and ultralow permeability.
This is in contrast to the more porous sandstone deposited to the east (Fig. 2.2a and b) (NEB,
2013). Low porosity and ultralow permeability within a formation (this combined feature
alternatively known as tight formation) also reduces the effectiveness of the conventional
well log (OGC, 2012b). In contrast, the unconventional horizontal drilling with multistage
hydraulic fracturing directs pressurized fluids, typically containing any combination of water,
proponent, and other chemicals, to penetrate tight rock formations in order to release the oil
and gas which requires high rate o f pumping and extended flow back to withdraw fracture
fluids and solids.
The Montney formation in B.C. is subdivided into two main regional fields: the Heritage
Field (south) and the Northern Montney Field (north) (Fig. 2.2b). The Heritage Field covers a
large area (561,120 ha) with favorable conditions of reservoir parameters (e.g., porosity,
permeability and pressure) for gas withdrawal, while the withdrawal o f gas in the Northern
Field is complicated due to the presence o f a disturbed belt in the Northern Rockies (OGC,
2012a). Consequently, much of the new gas development activities are occurring in the south
field which is approximately bounded from 55°N to 57°N and the Alberta border to 122°W
encompassing the towns o f Fort St. John, Taylor, Dawson Creek and Chetwynd (MoE,
2014). Unconventional gas production accounts for 60% of B.C.’s total production in 2012
(1.41 trillion cubic feet [tcf]) and the Montney formation contributed 40% o f the

11

unconventional gas produced from its 1,270 wells which represent 33% of the province’s
total remaining recoverable raw gas reserves at the end of 2012 (40.2 tcf) (OGC, 2012a).
Total production o f natural gas in 2012 is approximately 45% higher than the production in
2005 (OGC, 2012). The total estimated prospective raw natural gas resource in the Montney
formation is 1,965 tcf and the estimated marketable reserve is 271 tcf in B.C. while the
prospective raw and estimated marketable Montney reserve in Alberta is 2,309 and 178 tcf,
respectively. This total estimated marketable 271 tcf in B.C. is equivalent to 87 years of
Canada’s 2012 consumption (total Canadian natural gas demand in 2012 was 3.1 tcf) (NEB,
2013). The exploitation of these vast reserves o f natural gas is a significant economic driver
and revenue generator in B.C., and the province is planning to expand this industry by
promoting development o f liquefied natural gas (LNG) for export. It is estimated that 3 tcf
per year will be produced for meeting three LNG facility development goals by 2020 (MEM,
2012 ).

2.2.I.I. Air quality aspects of gas development activities
Raw natural gas in the B.C. Montney formation is comprised of dry and wet (rich) gas,
therefore, it is necessary to process the raw gas for the removal o f impurities other than
methane (such as: paraffinic hydrocarbons, nitrogen, carbon dioxide, hydrogen sulfide-H 2S,
helium) prior to supplying the pipeline as a marketable gas with specific grade for use in
domestic, commercial or industrial purposes (OGC, 2012a). Two major gas processing plants
(the Dawson processing plant in 16 km west o f Dawson Creek and the McMahon gas plant in
Taylor) are located in the Heritage Field (Fig. 2.2b) to process the raw natural gas generated
in this field. The C 0 2 and H2S from the raw gas processed in the Dawson processing plant is
blended and transported for further processing through a pipeline to the McMahon gas plant

12

(http://www. spectraenergy. com/Operations/N ew-Proi ects-and-Our-Process/N ew-Proi ects-in
Canada/Dawson-Processing-Plant/. accessed on: 19 September 2014). Previous research has
shown that SO2 and NO 2 are the dominant species among all gaseous air pollutants in
Northeastern B.C. and these pollutants are emitted from the various stages o f oil and gas
production including the processing (e.g., flares, engines, and compressors), distribution
(e.g., leaks of pipelines and flanges), and storage (e.g., vaporization from storage tanks)
(Krzyzanowski, 2012). Besides this, unconventional natural gas development consists of two
main phases: well development and production, and individual well development involves
three stages: pad preparation, well drilling and well completion (US EPA, 2010). Recently,
the US EPA estimated that well completions involving hydraulic fracturing in uncontrolled
manner can vent natural gas to the atmosphere resulting in air pollution. This is done to guard
against the over-pressuring of the well which results in approximately 230 times more natural
gas being vented compared to wells without hydraulic fracturing (conventional well drilling)
(US EPA, 2011). However, oil and gas industries are not required to report to the emissions
during the well development phase (e.g., in pilot phase or in exploration or drilling phase) to
the

National

Pollution

Release

Inventory

(NPRI)

(http://ec.gc.ca/inrp-

npri/default.asp?lang=En&n=02C767B3-C9FD-4DD7-8072. accessed on: 19 September
2014).

13

V riw

\

SWT.
124

Jcaa Marie

■» Moatacy

— g g H orn Rqr«*-~

2 CMTeatKmat

> * > TetaJ.NEBC
a.f)
__

BC Peace River Regioa

Fig. 2.1. The study area, (a) Unconventional gas play trends in Northeastern B.C. (modified after OGC,
(2012a)). White squares refer to towns, (b) Overall gas production in B.C. from both conventional and
unconventional wells (modified after OGC, (2012a)), (c) Conventional oil and gas wells in 2012 in
Northeastern
B.C.
(source:
NPRI
database,
available
at
http://www.ec.ec.ca/inrpnpri/default.asp?lang=en&n=lD892B9F-l. accessed on: 20 September 2014), however, an unconventional
wells distribution map is not currently available in the NPRI database, and (d) Passive sampling sites in the
study area. Red squares refer to the major towns across the study area. The passive monitoring network with
major towns is also presented in Fig. 2.3.

14

Alberta

1F

, British
. Columbia

/

\
Unconventional Montney
Regional Fields
Dry Ga&'Wet Ga&/Oii Piatnbtrtion
^^c.gu'? 1 ^.FraL, j «JXZ1L,

Fig. 2.2. The location o f the Montney formation in B.C. and Alberta, (a) Montney formation with major rock
types in B.C. and Alberta (adapted from NEB, (2013)), (b) Montney regional fields in B.C. with dry, rich (wet)
and oil distribution (adapted from OGC, 2012a).

2.2.2. Satellite remote sensing of trace gases in the troposphere
Lower tropospheric pollutant observations from satellite began with the GOME-1 instrument
(Global Ozone Monitoring Experiment, 1996-2003) aboard the ERS-2 satellite (Burrows et
al.,

1999) and has continued with SCLAMACHY (SCanning Imaging Absorption

spectroMeter for Atmospheric CHartographY, 2002-2012) aboard the ENVISAT satellite
(Bovensmann et al., 1999). Currently these measurements are made by the OMI (Ozone
Monitoring Instrument, 2004-present) on the Aura satellite (Levelt et al., 2006) and the
operational GOME-2 (2007-present) on the MetOP platform (Martin, 2008). The basic
characteristics of these instruments are listed in Table 2.1. The satellites have been designed
15

to yield information on the lower tropospheric trace gas constituents by flying in near-polar,
Sun-synchronous, low earth orbits; a typical orbit altitude is 705 km (Martin 2008). These
trace gas constituents (e.g., NO 2 and SO2), are processed as vertical column densities (VCDs)
from an integrated column amount by measuring solar back-scattered UV-visible radiation
commonly in nadir (down looking) geometry by these instruments. The determination of
tropospheric VCDs from the calibrated spectra involves three main steps (Boersma et al.,
2007): 1) a spectral fit to determine the slant columns densities (SCD), 2) elimination of the
stratospheric contribution to the SCD; and 3) conversion of tropospheric SCD to tropospheric
VCDs via an air mass factor (AMF), where the AMF accounts for the sensitivity o f the
instrument to the absorber (target species, such as: NO 2 and SO2) (McLinden et al., 2014).
The AMF describes the enhancement in absorption when light traverses a slant path through
a layer and, therefore, it represents a basis o f trace gas retrievals in the spectrum o f UVvisible wavelengths. It is important to note that the AMF is the ratio of the SCD to the VCD
and it is computed using a radiative transfer model because of the complex path o f the
scattered and reflected sunlight. The radiative transfer model simulates nadir radiances while
accounting for all relevant physical factors such as: multiple scattering by molecules and
aerosols, absorption by trace gases and reflection from the surface. The accuracy o f the AMF
calculation is largely governed by the accurate information of the absorber vertical profile
including cloud and aerosol information, surface reflectivity and surface pressure (McLinden
et al., 2014). The NASA OMI SO2 product uses a spatially and temporally invariant AMF
and it is 0.36.
These three sensors (SCIAMACHY, OMI, and GOME-2) differ in their horizontal resolution
of each observation (pixel) and among these, OMI is finer (Table 2.1). However, unlike the

16

others, OMI’s across-track resolution varies noticeably due to its 2-D detector which can
measure 60 across-track pixels simultaneously. Consequently, those pixels near the nadir are
roughly 30 km wide (across track width) while pixels near the edge are larger than 100 km,
however, the along track width o f each OMI pixel is invariant (13km). Some pixels o f OMI
data products are affected by the “row anomaly” (RA) error which is an anomaly that affects
the quality of the radiance data within a row o f detectors (primary data that is collected by
the instrument)

at all wavelengths

for a particular viewing

direction o f OMI

(http://www.knmi.nl/omi/research/product/rowanomalv-background.php). The RA changes
over time affecting radiance data and consequently impacting on the VCDs (OMI User’s
Guide, 2012). The first RA occurred in June 2007, however, the origin of the RA remains
unclear. The OMI team identified four probable reasons for the RA: 1) part o f the incoming
earthlight is blocked; 2) inhomogeneous illumination that causes a shifting of wavelength; 3)
stray sun light is reflected into the field-of-view; and 4) stray earthshine is reflected into the
field-of-view (Boersma et al. 201 la). Moreover, Boersma et al. (201 la) reported that various
RA correction algorithms have been developed since the first occurrence but to date no
correction algorithm works with reasonable agreement that effectively removes the
anomalies. Therefore, the OMI team recommends that researchers not use the affected cross­
track scenes; Table 2.2 lists the affected scenes to date.
Generally, all the satellite data products for both N 0 2 and S 0 2 offer the possibility to
improve our understandings of lower tropospheric trace gas concentrations (Lamsal et al.,
2008 and 2013; Lee et al., 2011), as well as helping with identification of emission sources
(Fioletov et al., 2011 and 2013; Lu and Streets, 2012; Martin et al., 2003), and atmospheric

17

chemistry, through testing and improvements to emission inventories, using top-down
modeling techniques (Boersma et al., 2008).
Table 2.1. Basic characteristics of space-borne instruments (Adapted from Lu and Streets, 2012).
Instrument
Satellite
Agency
Launch time
Period of operation
Type
Altitude
Time for one orbit
Inclination
Repeat time
Ascending or
Descending
Local equator overpass time
Spatial resolution
Global coverage time

GOME
ERS-2
ESA
Apr 21, 1995
1996-2003
Sun-synchronous
782 km
100.5 mins
98.52"
35 days
Descending

SCIAMACHY
Envisat
ESA
Mar 1,2002
2002-April 30,2012
Sun-synchronous
782 km
100.5 mins
98.52"
35 days
Descending

OMI
EOS-Aura
NASA
Jul 15,2004
2004-now
Sun-synchronous
705 km
98.8 mins
98.2"
16 days
Ascending

GOME-2
Metop-A
EUMETSAT/ESA
Oct 19,2006
2007-now
Sun-synchronous
840 km
101.7 mins
98.8"
5 days
Descending

10:30
40x320 km2
3 days

10:00
30x60 km2
6 days

13:30
13x24 km2
1 day

9:30
40x80 km2
1.5 days

Table 2.2. Cross-track raw anomalies (adapted from Boersma et al. 201 la).
Anomaly
Anomaly 1
Anomaly 2
Anomaly 3
Anomaly 4
Anomaly 5

Date since its occurrence
Since June 25th, 2007
Since May 1101, 2008
Since January 24th, 2009
Since July 5*, 2011
Since August 9, 2011

Affected cross-track positions (0-based)
53-54
37-44
27-44
42-45
41-45

2.2.3. Global Environmental Multi-scale - Modelling Air Quality and Chemistry
(GEM-MACH) model
This section is synthesized from Anselmo et al. (2010). GEM-MACH, established by
Environment Canada (EC) in November 2009, is the Canadian regional comprehensive air
quality on-line model. This model is used operationally for forecasting the concentration of
O 3, NO 2, and PM 2.5 as well as a full description of atmospheric chemistry (including gasphase, aqueous-phase, aerosol processes, and heterogeneous chemistry) and meteorological
processes. GEM-MACH is currently integrated twice per day to provide 48-hour forecast on
a 348x465 rotated latitude-longitude grid over North America with 15-km horizontal grid
spacing (also known as GEM-MACH 15) in 58 vertical levels extend from the surface to 0.1
hPa. This model uses the emissions inventory o f the US EPA and EC to assess the source of

18

tropospheric ozone precursors. However, time-dependent meteorological lateral boundary
conditions (LBCs) are taken from GEM 15 which is the EC operational weather forecast
model with a variable-resolution in global scale but found uniform 15-km grid spacing
centered over North America. The chemistry fields o f GEM-MACH are initialized by cycling
the 12-h forecast of the previous model run. The time step for the integration o f the chemical
processes by the GEM-MACH is 15.0 minutes and meteorology is integrated with a 7.5
minutes time step, therefore, chemical processes have been integrated every second time step
(i.e., 15 minutes) for the sake of reducing the model run time. No significant chemical
degradation has been found in the chemistry forecast with this modification. Recently,
McLinden et al. (2014) used the GEM-MACH output as a source o f profile information for
updating the AMF of OMI NO 2 and SO2 data products over the Canadian oil sands area and
they also reported that 15-km grid size of GEM-MACH was well suited for the OMI small
pixels (across track width ~30km), therefore, no additional smoothing was applied.
2.2.4. Passive monitoring of air quality
Passive diffusive samplers have been widely used across North America and Europe for the
assessment o f atmospheric NO 2 and SO2 concentrations (e.g., Bytnerowicz et al., 2010;
Campos et al., 2010; Vardoulakis et al., 2009; Zbieranowski and Aheme, 2012a). Passive
diffusion samplers can be broadly classified into tube, radial, badge and cartridge-type
samplers (Krupa and Legge, 2000) and these samplers can have substantially different uptake
rates, which make them more or less suitable to certain applications (Yu et al., 2008). For
example, tube-type samplers have generally lower uptake rates due to longer axial diffusion
path and smaller cross-sectional diffusion area, which makes them suitable for assessing
relatively long-term (e.g., monthly mean) ambient air quality levels. In contrast, the badge-

19

type and radial samplers have typically higher uptake rates due to the shorter diffusion paths
and larger diffusion areas and are suitable for assessing the relatively short-term (e.g., daily
mean) personal/occupational exposure to air pollution (Vardoulakis et al., 2009).
Furthermore, badge-type samplers are advantageous over tube-type since they have an
entrance filter to create a diffusion area free from turbulence to avoid wind effects (Van
Reeuwijk et al., 1998). The basic principle behind all the passive sampler measurements is
Fick’s law of diffusion of gases from the atmosphere into a sampler o f defined dimensions
onto an absorbing medium. The sampler’s theoretical uptake rate is a function o f the length,
L (m), and the cross-sectional area, A (m2), o f the stationary air layer within the sampler, and
can be calculated provided that the diffusion coefficient, D (mV1), of the gas o f interest is
known. In particular, tube and badge type passive samplers are used extensively to measure
atmospheric NO 2 (Tang et al., 2001). The advantages o f diffusive passive samplers include
that they are inexpensive, easy to deploy and use in the field for long term assessments, do
not require power, produce accurate results in both indoor and outdoor environments, and are
reliable for monitoring ecosystem exposure to gaseous pollution (Cox, 2003; Hafkenscheid et
al., 2009; Kot-Wasik et al., 2007; Namiesnik et al., 2005; Seethapathy et al., 2008;
Zbieranowski and Aheme, 2012a and 2012b). However, passive samplers have some
problems associated with the variability in environmental factors (e.g., temperature, relative
humidity, wind, and rain), unsuitability for short term monitoring, and need to be validated
with collocated continuous active monitors (Cox, 2003; Kot-Wasik et al., 2007; Krupa and
Legge, 2000; Runeckles and Bowen, 1999; Seethapathy et al., 2008).

20

2.3. Methodology
2.3.1. Study area
Northeast B.C. is a region o f plains, bordered by the Yukon and Northwest Territories to the
north, the Rocky Mountains to the southwest and the province o f Alberta to the east. It is the
largest of B.C.’s ecological regions, representing 21.8% of the land area of the province
(20,494,470 ha), but the least populated, with 1.6 % of the total population (69,000 people).
O f those residing there, approximately 13% self-identify to be o f aboriginal (i.e., First
Nations) descent. The Northeast region has one o f the most active economies in B.C. which
is mainly driven by oil and gas exploration and production. Due to this, the population o f the
northeast

is

expected

to

rise

to

almost

(http://www.welcomebc.ca/Live/about-bc/regions/northeast.aspx).

80,000

by

2030

There

are

extreme

differences in temperature between the warmest and coldest months of the year, in some
areas, for example in Fort St. John, the average daily temperature can range from -21 °C in
January to +14 °C in July. Major communities include Fort St. John, Fort Nelson, Taylor,
Dawson Creek, Chetwynd and Hudson’s Hope. Due to the increase in natural gas
development activities and the public concern about air quality in Northeastern B.C. this area
analyzed in the present study covers the Heritage Field of the B.C. unconventional Montney
formation (Fig. 2.Id, 2.2b, and also Fig. 2.3). This area is approximately bounded from 55°N
to 57°N and the Alberta border to 122°W, which roughly encompasses Fort St. John, Taylor,
Dawson Creek and Chetwynd. Therefore, this present study area covers not only the major
gas development area but also most of the major communities of Northeastern B.C.

21

2.3.2. Satellite data products
In this analysis, NO 2 data from three instruments (SCIAMACHY, OMI and GOME-2) from
2005-2013 were considered for analysis. While multiple OMI NO 2 VCD data products
currently exist, the two global primary products from the NASA standard product (SP)
(Bucsela

et

al.,

2013;

http ://disc. sci.gsfc. nasa. gov/Aura/data-

holdings/OMI/omno2 v003.shtml) and the Dutch OMI NO 2 (DOMINO) (Boersma et al.,
2 0 1 1 b;

http://www.temis.nl/airpollution/no2 .html) processed in near real time have been

included in the analysis. The reason for two OMI NO 2 data products in this analysis is to
understand the differences in NO 2 VCDs for the study area that arise from different
algorithms used to eliminate the stratospheric SCD. The DOMINO product simulates the
stratospheric NO 2 by assimilating OMI SCDs in a chemical data assimilation system,
whereas the SP executes a complex high-pass filtering approach (Dirksen et al., 2011). In
addition, the algorithm o f the SP product depends on monthly mean profiles from the Global
Modeling Initiative (GMI) (Bucsela et al., 2013) while DOMINO uses daily output from the
Tracer Model 4 (TM4) chemical transport model (Boersma et al., 2007). However, both o f
these models make use o f emission inventories appropriate for the late 1990s (McLinden et
al., 2014), when oil and gas activities in Northeastern B.C. were significantly lower than the
current level. Accurate absorption profiles of trace gases (e.g., NO 2) substantially govern the
accuracy o f the AMF calculation, and the AMF plays a fundamental role in determining
tropospheric VCDs from SCDs. Therefore, McLinden et al. (2014) recently recalculated the
AMF (known as EC-AMF) for North America based on an updated and higher resolution
absorption profile from a regional-scale (15 km x 15 km resolution) air quality model (the
Global Environmental Multi-scale - Modeling Air Quality and Chemistry, GEM-MACH;

22

Anselmo et al., 2010) using emission inventories from the US EPA and EC data for 2006,
high spatial and temporal surface reflectivity data, and an improved treatment o f snow.
Finally, two new OMI NO 2 data products from the tropospheric SCDs of SP and DOMINO
divided by the EC-AMFs instead of the original AMFs have been created by the EC
(McLinden et al., 2014) and these new data products (VCDs), EC-SP-NO 2 and ECDOMINO-NO 2, were included in this analysis. Data from these two new products were
collected from EC (Chris McLinden, EC, personal communication, 2014). In addition, NO 2
data products of SCIAMACHY between 2005 and 2011 were combined with the GOME-2
data products o f 2012-2013 in this analysis (this combined data products named here as
SCIA-NO 2 for simplicity) and this combination approach (Ghude et al., 2011) has been
implemented for the comparison with the OMI data products between 2005 and 2013. The
same retrieval techniques were used for NO 2 data products for both SCIAMACHY and
GOME-2 instruments and the flight time of the satellites carrying these two instruments is
almost identical (Table 2.1). Although multiple NO 2 data products have been included in this
study, analysis of SO2 data (2005-2013) was only limited to the OMI instrument (OMS02
specific

to

the

planetary

boundary

layer

[PBL],

version

3,

http://disc.sci.gsfc.nasa.gov/Aura/data-holdings/OMI/omso2 v003.shtml. this data product is
named hereafter as NASA-S0 2 for simplicity). This product is retrieved based on the NASA
Band Residual Method (BRM) using four wavelengths between 310 and 315 nm to quantify
SO2 absorption (Krotkov et al., 2006).
2.3.2.I. Satellite data filtering
Orbit based level 2 data products, in which the location of an observation is assigned to the
co-ordinates of the pixel center, from all these instruments (SCIAMACHY, OMI, and

23

GOME-2) were used in this study. However, as OMI measures 60 cross-track positions
(pixels), which are variable in size depending on the track position, 1 0 pixels on each swath
edge were excluded from analysis to limit the across-track pixel width to -4 0 km. In
addition, some cross-track positions (Table 2.2), affected by RA since June 2007, were
dynamically removed based on the RA flags in all OMI data products for both species. Daily
OMI data products were also filtered to remove data with high cloud radiance fraction (>0.2)
and large solar zenith angle (SZA), >75° and >60° for NO 2 and SO2 , respectively. The
stringent SZA threshold of 75° for NO 2 results in about 75% of the data being allocated for
April-September while the remaining 25% is for October-March. However, a 60° SZA
threshold for SO2 keeps 99% data between April and September o f each year (McLinden et
al., 2014). Surface albedo was limited to <0.3 for both SP and DOMINO data products from
OMI. Although year round N 0 2 data products from all these instruments were used,
following the Fioletov et al. (2011) protocol, only summertime (May-August) NASA-SO 2
data were used, since data from this period have the best signal-to-noise ratio (SNR: the ratio
between the mean values and the standard errors of the means). Furthermore, the data were
also restricted to values between -10 and +5 DU (1 DU = 2.69 x 1026 molecule/km2) to avoid
spikes from transient volcanic plumes.
2.3.2.2. Pixel averaging approach
A pixel-averaging or oversampling approach was applied in this study to analyze the satellite
data products of both gas species (Fioletov et al., 2011 and 2013). For this approach, a
geographical grid is defined in the study area and the average o f all pixels centered within a
given radius from each grid point is calculated. For example, a grid o f 224 km x 224 km with
1.2 km step and 1215 km x 1215 km with 6.0 km step is established for Northeast B.C. and a

24

larger regional area (most of B.C. and Alberta), respectively. An average of all valid pixel
values of a sensor within a given time window and falling within the specified radius from
the center of a grid cell is assigned to that grid point, where the pixel center is used as the
location of the measurement. Therefore, this approach provides a detailed subpixel-resolution
spatial distribution o f mean values o f the pollutant (Fioletov et al., 2011 and 2013). It is
important to note that, the selection o f an averaging radius determines the degree of
smoothing; averaging with a large radius reduces the noise, but it also reduces the spatial
resolution (Fioletov et al., 2013).
2.3.3. Passive monitoring network
Two-week average ambient concentrations o f NO 2 and SO2 were measured at 24 sites across
Northeastern B.C. (Montney formation), Canada (Fig. 2.Id and Fig. 2.3) during the period
August 2013-November 2013. The Taylor Town Site and Pine River Hasler, (Fig. 2.3) as
well as an additional site in central B.C. (B.C. Environment Plaza 400 monitoring location in
Prince George, B.C., Canada) were included in the study as they were part o f an established
B.C. air quality monitoring network, in order to assess the passive sampler performance. All
other sites (Table 2.3) were selected to cover the region of oil and gas development activities
in Northeastern B.C. (Fig. 2.1c), which corresponds to the Peace River region o f B.C. (Fig.
2.3). All sites were chosen to be free o f obstacles impeding wind flow and, with a few
exceptions, most sites were also at least 3.0 kilometers from major industrial sources and
from urban areas in order to provide a better estimate o f the overall spatial pattern o f ambient
levels by avoiding the local impact of point source emissions (Table 2.3). A few sites were
also placed at private homes to assess the ambient concentrations directly relevant to human
exposure (Table 2.3).

25

BC Peace River Region

1700

1300

1100

122°W

1 2 1 'W

120*W

Fig. 2 3 . Study area with the locations of 24 passive monitoring stations (+ symbols). The study domain in
northeast B.C. consists o f 24 sites; Plaza 400 (in the Prince George air shed, not included in the figure) was
selected in order to validate the passive sampler results with continuous monitoring data o f N 0 2 and S 0 2.
Besides Plaza 400, Site 14 and Site 16 were also chosen to collocate with continuous monitoring stations; each
o f these sites measures only S 0 2 (except Plaza 400 which also measures N 0 2) along with meteorology and data
are
archived
by
the
B.C.
Ministry
of
the
Environment
(available
at:
http://www.bcairQualitv.ca/readings/index.html). Each o f the total o f 25 stations are exposed to ambient
conditions from 15 Aug. 2013 to mid Nov. 2013 with samplers being exchanged every two weeks (six
measurements per site), major towns of the study area are also indicated with red circles. The B.C. - Alberta
border is along 120°W longitude.

26

Table 2 3 . Passive sampler deployment site IDs, nearest towns, location (latitude, longitude), elevation,
sampling period, and site location description (see also Fig. 2.3).
Latitude

Longitude

Decimal
degrees
55.16375

Decimal
degrees
-120.20661

965

55.30956

-120.12789

878

55.44800

-120.02722

963

55.54847

-120.10897

757

55.57211

-120.60486

808

55.85114

-120.37417

769

55.70142

-120.90508

865

55.91244

-120.50594

724

F9

Dawson
Creek
Chetwynd

55.79628

-121.07164

758

F10

Taylor

56.03336

-120.63308

816

FI 1

Chetwynd

55.82294

-121.60242

769

F12

Taylor

56.12161

-120.61697

481

F13

56.28986

-120.10883

764

55.89094

-121.90642

735

56.22453

-120.71636

694

56.42047

-120.26722

769

56.03253

-121.90086

515

56.30503

-120.85406

692

F20

Fort
St.
John
Hudson
Hoope
Fort
St.
John
Fort
St.
John
Hudson
Hoope
Fort
St.
John
Chetwynd

55.69753

-121.61836

627

F21

Chetwynd

55.56164

-122.01369

841

Site 14*

Taylor

56.15028

-120.68647

513

Site 16*

Chetwynd

55.60544

-121.97347

635

Indus 1

Dawson
Creek
Dawson
Creek
Prince
George

55.95631

-120.15775

688

55.79222

-120.48383

809

53.91511

-122.74169

597

ID

FI
F2
F3
F4
F5
F6
F7
F8

F14
F15
F16
F17
F18

Indus 2
Plaza
400*

Nearest
town

Dawson
Creek
Dawson
Creek
Dawson
Creek
Dawson
Creek
Dawson
Creek
Dawson
Creek
Chetwynd

Elevation
m

Sampling
period
dd.mm.yyyy

Site description*

17.08.201310.11.2013
17.08.201310.11.2013
17.08.201310.11.2013
17.08.201310.11.2013
18.08.201311.11.2013
16.08.201310.11.2013
18.08.201311.11.2013
16.08.201310.11.2013
18.08.201311.11.2013
16.08.201310.11.2013
15.08.201309.11.2013
16.08.201310.11.2013
16.08.201310.11.2013
15.08.201309.11.2013
16.08.201310.11.2013
16.08.201310.11.2013
15.08.201309.11.2013
16,08.201310.11.2013
15.08.201309.11.2013
15.08.201309.11.2013
16.08.201310.11.2013
15.08.201309.11.2013
17.08.201311.11.2013
18.08.201311.11.2013
12.08.201306.11.2013

Open field, near forested area
Open field, near forested area
Open filed, 1 Gas plant in 200m
Open field, 3 Gas plants immediate after 5km
Open field, near Heritage Highway, 52
Open field, near agricultural crops
Open field, near forested area
Open field, near agricultural crops
Open field, near agricultural crops
Private home lawn (open field)
Forested area, near Road 29
Private home lawn (open field)
Open field, 3 Gas plants immediate after Skm
Forested area, near Road 29
Private home lawn (open field)
Open field, near forested area
Open field, near forested area
Private home lawn (open field)
Private home lawn (center of Chetwynd town)
Open field, near forested area, 1 gas plant
immediate after 3 km
Collocated in Taylor Town Site station, (Center
of Taylor town) 2 gas plants in 1.5 km
Collocated in Pine River Hasler station,
(forested area) 1 gas plant immediate after 3 km
Open field, near agricultural crops, 2 gas plants
immediate after 3 km
Open field, near agricultural crops, 2 gas plants
immediate after 3 km
Collocated in Plaza 400 station, In the
Downtown of Prince George, B.C.

Gas plants were identified according to the NPRI database (NPRI, N 0 2 and S 0 2 emissions for Canada, 2012,
available at: http://www.ec.gc.ca/Ddb/websol/auervsite/Querv e.cfrot.

2.3.3.1. Passive sampler preparation and analysis
The Willems badge diffusive passive samplers (Fig. 2.4a) were provided by the laboratory o f
Professor Julian Aheme, Trent University, Ontario, Canada. Once the passive samplers were
27

exposed at the field locations they were returned to Trent University by post (in resealable
zip-lock plastic bags packaged in cardboard) for laboratory analysis. Exposed samplers
before laboratory analysis and unexposed samplers before deployment were refrigerated at
4°C. Samplers were exposed in triplicate for each species, for a total o f six samplers per site.
A two-week sampling frequency at a height o f 1.8 m was used (except Plaza 400 and Site 16,
which were located on the roof-top of a 4 and a 1 storey building, respectively), which
corresponds to the height of the active samplers they were collocated with. Each sampler was
mounted using Velcro® under a 127 mm diameter plastic cap that acted as a precipitation
and bird shield (Fig. 2.4b). There were a total of six exposure periods from August 15, 2013
until November 11, 2013. All NO 2 samplers in the 5th exposure period were not considered
for analysis due to a problem with sampler preparation. Also, two thirds of all sites during
first two exposure periods were monitored for both species with duplicate rather than
triplicate passive samplers due to a shortage of samplers. During each exposure period, five
unexposed samplers were retained as laboratory blanks. Besides laboratory blanks, travel
blanks or lot blanks (for each return shipment from Peterborough, Ontario to Prince George,
B.C.) and field blanks (for both species of each exposure) were also sent to and from sample
sites periodically throughout the study and compared to laboratory blank samplers to ensure
that the sampler cap and resealable bags were effectively protecting the samplers from
contamination between preparation, shipment, exposure and analysis.
Analysis o f NO 2 and SO2 data followed the protocol developed by Zbieranowski and Aheme,
(2012a and 2012b). The Willems badge passive sampler (Fig. 2.4a) has a cylindrical body
(diameter 28 mm and length is 15 mm) in which a specially treated filter paper absorbs a
specific gas from the air. A spacer ring separates the absorbent pad (Whatman 1820-090)

28

from a Teflon® membrane (Whatman 10 411 116) and this creates a stagnant air layer in
which diffusion of the desired gas occurs from ambient air onto the absorbent pad. Both the
Teflon® membrane and absorbent pad have the same cross-sectional area (5.309 x 10' 4 m2).
An additional spacer ring holds the assembly together and a cap on the top creates an airtight
seal when not sampling. The sampler body, spacer rings and cap o f all the samplers were
made from polystyrene vials (Greiner 214 111). The samplers were constructed at Trent
University, Ontario, Canada to the exact specifications of the original Willems badge sampler
for NO 2 (Van Reeuwijk et al., 1998) and the SO2 sampler was a modified design from
Bytnerowicz et al. (2005) (Zbieranowski and Aheme, 2012a and 2012b). Zbieranowski and
Aheme (2012a and 2012b) also reported that Willems badge passive sampler has been tested
to perform well against co-located active sampling methods. The passive samplers for N 0 2
were assembled in a glove box connected to an air hose leading from a filtration pack (NO 2 :
PVC tube ID 1 cm, length 80 cm stuffed with glass wool soaked in triethanolamine (TEA)
solution). However, SO2 sampler components were assembled without a filtration system.
Before assembly, absorbent pads for NO 2 samplers were cleaned in a bath of Milli-Q water at
a temperature of 95° C and then dipped in TEA solution (2 mL o f TEA [CAS: 102-71-6]
stirred in 48 mL of acetone [CAS: 67-64-1]) and left to dry for 15 min on a ribbed glass plate
prior to assembly (Nylasorb filters for SO2 did not require cleaning and coating as these
filters directly absorb S 0 2 from air). Teflon membranes for both species (N 0 2 and S 0 2)
samplers were cleaned in two baths o f Milli-Q water and 95% ethanol (50:50, v:v) and a
third bath o f 95% ethanol; all other sampler components were washed in Milli-Q water.
Samplers were capped and sealed inside sealed plastic bags and placed in a cardboard box

29

prior to transfer to University of Northern British Columbia (UNBC), B.C., Canada, by post
for deployment in the field.

Fig. 2.4. a) The ‘Willems badge’ passive sampler: 1) Velcro® for sampler deployment, 2) sampler body with
opening at one end, 3) absorbent pad, 4) spacer ring, 5) Teflon® membrane filter, 6) spacer ring, and 7) cap,
(adapted from Zbieranowski and Aheme, 2012a); b) Passive samplers were deployed in the field o f this study.

After samplers were transferred to Trent University for analysis, each badge (sampler) was
disassembled in a sealed glove box. Then the absorbent pad was placed in a vial containing 5
mL of Milli-Q water, shaken and left for 30 min. A two-times dilution was performed for
analysis; 2.5 mL was removed and placed in a vial containing 1 mL o f Milli-Q water to
which 1.5 mL o f reagent was added (total volume 5 mL). The reagent was prepared by 4.0 g
of sulphanilamide [CAS: 63-74-1], 10.0 g of tartaric acid [CAS: 87-69-4] and 0.1 g of
ethylene-diaminetetra- acetate (EDTA) [CAS: 6381-92-6] dissolved in 800 mL o f Milli- Q
water to which 0.1 g of N-l-naphtylethylene-diammonium dichloride [CAS: 1465-25-4] and
10 mL o f acetone [CAS: 67-64-1] were added and diluted to a total volume of 1000 mL with
Milli-Q water. Samples were left in dark conditions for 2-3 h and analyzed using a UV-VIS
Spectrometer (Perkin Elmer Lambda XLS) at a wavelength of 540 nm. A standard calibration
curve was developed from a set o f six sodium nitrite [CAS: 7632-00-0] standards o f known
nitrite concentration (0.0, 26.6, 132.8, 265.6, 398.4 and 531.2 p g N 0 2 L '1).

30

Nylasorb filters without coatings were placed in the Willems badge samplers as Nylasorb
directly absorbs SO2 from the air. Each badge (sampler) was disassembled after exposure.
The filter pad (Nylasorb filter) was placed in a 50 mL falcon tube containing 6 mL of Milli-Q
water. The samples were then placed on a shaker table for 30 min after which 5.5 mL o f the
sample was taken in an IC (Ion Chromatography) vial. Finally, each sample was analyzed on
a Dionex ICS-1100 Ion Chromatograph in a carbonate eluent through an IonPac AS22
analytical column.
2.3.3.2. Calculation of ambient concentrations from passive samplers
The calculation of ambient concentrations from passive samplers also follows the protocol
developed by Zbieranowski and Aheme, (2012a and 2012b). The quantity of ambient NO 2
and SO2 sampled on the absorbent pads was determined by subtracting the laboratory blank
(n =5 per exposure period) from exposed field samples (n=3 per site per species) with the
application of calibration and dilution factors:
Q = (Sf - S b) x / x d f

(1)

where Q is the total amount of NO 2 and SO2 sampled (pg), Sf is the average o f the field
samples (IC peak concentration [pg L'1] for SO2 and UV-VIS absorbance forNCh), Sb is the
average of the laboratory b lan k s,/is the calibration factor (slope between peak concentration
[IC] or absorbance [UV-VIS] and standard concentration) and df is the dilution factor. The
ambient concentration (C, pg m'3) of atmospheric SO2 was estimated from Q:
C= slope x (Q/7)

(2)

Where slope is the conversion factor from amount sampled (SO 2) to a dose (SO 2) and t is the
length of the sampler exposure period (hours).
The ambient concentration (C, pg m '3) of atmospheric N 0 2 was estimated from Q:

31

C=(Q-Rt)/ (A x t)

(3)

where R, is the total resistance of the Willems badge, A is the absorbent pad surface area
(5.309 x 10-4 m2) and t is the length of the sampler exposure period (s).
The method limit o f detection (LOD) for NO 2 and SO2 passive samplers were 0.3 ppb (0.54
pg m'3) and 0.03 ppb (0.07 pg m’3), respectively. These were estimated from three times the
standard deviation of laboratory blanks o f each corresponding species. No significant
differences were obtained in lab blanks over field or travel blanks for both species (p » 0.05
in Mest of all cases). The lab blank mean value was subtracted during calculation of ambient
concentrations o f each species.
2.3.3.3. Accuracy and precision of Willems badge passive samplers
The accuracy o f Willems badge passive diffusion samplers was assessed using the percentage
relative error (%) (also known as relative bias):
Accuracy = [(Cp - Ca) / Ca] x 100

(4)

Where Cp is the air pollutant concentration measured with a Willems badge passive sampler,
Ca is the concentration measured with the reference active method (i.e., automatic
chemiluminescence and UV Fluorescence method applied for NO 2 and SO2 measurement,
respectively

in

the

B.C.

continuous

air

quality

http://www.bcairqualitv.ca/assessment/monitoring-instruments.html.

network,
accessed

on:

url:
11

September 2014) and averaged over the same time period at exactly the same location. It
should be noted that the percentage relative error is used here as a simplified indicator of
accuracy and this method is also applied in other studies (e.g., Campos et al., 2010;
Vardoulakis et al. 2009). In addition, to estimate accuracy, the least-squares regression
equation coefficients of the passive diffusion measurements (dependent variable) against

32

automatic active measurements (independent variable) were also used as an indicator o f
linearity between these two methods. Finally, statistical significance (using a paired /-test to
compare the means of two measurements) was carried out to determine whether the
differences were significant between passive and active observations.
The precision of the Willems badge passive diffusion sampling measurements was evaluated
from the triplicate (also from duplicate, see section 2.3.3.1) sets of NO 2 and SO2 during each
deployment at all sites. The relative standard deviation (RSD), which is a statistical measure
o f repeatability (also called coefficient of variation - CV), was calculated for each pollutant
by dividing the standard deviation o f the triplicate samples by the mean concentrations over
the same time period and then multiplying by 100. This approach has also used to assess the
suitability o f passive diffusion samplers elsewhere (e.g., Vardoulakis et al. 2009;
Zbieranowski and Aheme, 2012a and 2012b).

33

2.4. References
Anselmo, D., Moran, M.D., Menard,S., Bouchet, V., Makar, P., Gong, W., Kallaur, A.,
Beaulieu, P.-A., Landry, H., Stroud, C., Huang, P., Gong, S., Talbot, D., 2010. A new
Canadian air quality forecast model: GEM-MACH15. Proc. 12th AMS Conf. On
Atmos. Chem, Jan 17-21, Atlanta, Ga, American Meteorological Society, Boston,
MA., 6 pp., http://ams.confex.com/ams/pdfpapers/165388.pdf (access on: 04
September 2014).
Boersma, K.F., Eskes, H.J., Veefkind, J.P., Brinksma, E.J., van der A.R.J., Sneep, M., van
den Oord, G.H.J., Levelt, P.F., Stammes, P., Gleason, J.F., Bucsela, E.J., 2007. Nearreal time retrieval of tropospheric NO 2 from OMI. Atmos. Chem. Phys. 7, 2103—
2118.
Boersma, K.F., Jacob, D.J., Becsela, E.J., Perring, A.E., Dirksen, R., van der A, R.J.,
Yantosca, R.M., Park, R.J., Wenig, M.O., Bertram, T.H., Cohen, R.C., 2008.
Validation of OMI tropospheric NO 2 observations during INTEX-B and application
to constrain NOx emissions over the eastern United States and Mexico. Atmospheric
Environment 42,4480-4497. doi:10.1016/j.atmosenv.2008.02.004.
Boersma, K.F., Braak, R., van der A, R.J., 201 la. Dutch OMI NO 2 (DOMINO) data product
v2.0:
HE5
data
file
user
manual.
Available
at:
http://www.temis.nl/docs/OMI NQ2 HE5 2.0 2011.pdf
(accessed
on:
04
September, 2014).
Boersma, K.F., Eskes, H.J., Dirksen, R.J., van der A, R.J., Veefkind, J.P., Stammes, P.,
Huijnen, V., Kleipool, Q. L., Sneep, M., Claas, J., Leitao,J., Richter, A., Zhou, Y.,
Brunner, D., 2011b. An improved tropospheric NO 2 column retrieval algorithm for
the Ozone Monitoring Instrument. Atmos. Meas.
Tech. 4, 1905-1928,
doi: 10.5194/amt-4-1905-2011.
Bovensmann, H., Burrows, J.P., Buchwitz, M., Frerick, J., Noel, S., Rozanov, V.V., Chance,
K.V., Goede, A.P.H., 1999. SCIAMACHY: Mission Objectives and Measurement
Modes. J.Atmos.Sci. 56,
127-150.
doi: http://dx.doi.org/10.1175/1520
0469(1999)056<0127:SMOAMM>2.0.CQ:2.

Bucsela, E.J., Krotkov, N.A., Celarier, E.A., Lamsal, L.N., Swartz, W.H., Bhartia, P. K.,
Boersma, K.F., Veefkind, J.P., Gleason, J.F., Pickering, K.E., 2013. A new
stratospheric and tropospheric NO 2 retrieval algorithm for nadir-viewing satellite
instruments: applications to OMI. Atmos. Meas.
Tech. 6 , 2607-2626.
doi: 10.5194/amt-6-2607-2013.
Burrows, J.P., Weber, M., Buchwitz, M., Rozanov, V., Ladstatter-Weifienmayer, A., Richter,
A., DeBeek, R., Hoogen, R., Bramstedt, K., Eichmann, K., Eisinger, M., Pemer, D.,
1999. The Global Ozone Monitoring Experiment (GOME): Mission concept and first

34

scientific
results.
J.
Atmos.
Sci.
0469( 1999)056<0151 :TGOMEG>2.0.CO;2.

56,

151-175.

doi: 10.1175/1520-

Bytnerowicz, A., Sanz, M.J., Arbaugh, M.J., Padgett, P.E., Jones, D.P., Davila, A., 2005.
Passive sampler for monitoring ambient nitric acid (HNO 3) and nitrous acid (HNO 2)
concentrations. Atmospheric Environment 39, 2655-2660.
Bytnerowicz, A., Fraczek, W., Schilling, S., Alexander, D., 2010. Spatial and temporal
distribution of ambient nitric acid and ammonia in the Athabasca Oil Sands Region,
Alberta. Journal of Limnology 69, 11-21.
Campos,V.P., Cruz, L.P.S., Godoi, R.H.M., Godoi, A.F.L., Tavares, T.M., 2010.
Development and Validation of Passive Samplers for Atmospheric Monitoring of
SO2, NO 2, O3 and H2 S in Tropical Areas. Microchem. J. 96, 132-138.
Cox, R.M,. 2003. The use of passive sampling to monitor forest exposure to O 3 , NO 2 and
SO2 : a review and some case studies. Environmental Pollution 126, 301-311.
Fioletov, V.E., McLinden, C.A., Krotkov, N., Moran, M.D., Yang, K., 2011. Estimation of
S 0 2 emissions using OMI retrievals. Geophys. Res. Lett. 38, L21811.
doi: 10.1029/2011GL049402.
Fioletov, V.E., McLinden, C.A., Krotkov, N., Yang, k., Loyola, D.G., Valks, P., Theys, N.,
Van Roozendael, M., Nowlan, C.R., Chance, K., Liu, X., Lee, C., Martin, R.V., 2013.
Application o f OMI, SCIAMACHY, and GOME-2 satellite SO2 retrievals for
detection of large emission sources. J. Geophys. Res. Atmos. 118, 11, 399-11,418.
doi: 10.1002/jgrd.50826.
Ghude, S.D., Kulkami, P.S., Kulkami, S.H., Fadnavis, S., van Der A, R.J., 2011. Temporal
variation of urban NOx concentration in India during the past decade as observed
from space. International Journal o f Remote Sensing 32:3, 849-861. DOI:
10.1080/01431161.2010.517797.
Hafkenscheid, T., Fromage-Mariette, A., Goelen, E., Hangartner, M., Pfeffer, U., Plaisance,
H., de Santis, F., Saunders, K., Swaans, W., Tang, Y.S., Targa, J., van Hoek, C.,
Gerboles, M., 2009. Review of the Application of Diffusive Samplers in the European
Union for the Monitoring o f Nitrogen Dioxide in Ambient Air. JRC Scientific and
Technical Reports. European Commission, Joint Research Centre, Institute for
Environment and Sustainability, p. 79.
Kot-Wasik, A., Zabiegala, B., Urbanowicz, M., Dominiak, E., Wasik, A., Namiesnik, J.,
2007. Advances in Passive Sampling in Environmental Studies. Anal. Chim. Acta
602, 141-163.
Krotkov, N.A., Cam, S.A., Krueger, A.J., Bhartia, P.K., Yang, K., 2006. Band residual
difference algorithm for retrieval of S 0 2 from the Aura Ozone Monitoring Instrument
35

(OMI).
IEEE
Trans.
Geosci.
doi: 10.1109/TGRS.2005.861932.

Remote

Sens.

44,

1259-1266.

Krupa, S.V., Legge, A.H., 2000. Passive sampling of ambient, gaseous air pollutants: an
assessment from an ecological perspective. Environmental Pollution 107, 31-45.
Krzyzanowski, J., 2012. Environmental pathways of potential impacts to human health form
oil and gas development in northeast British Columbia, Canada. Environ. Rev. 20,
122-134.
Lamsal, L.N., Martin, R.V., van Donkelaar, A., Steinbacher, M., Celarier, E.A., Bucsela, E.,
Dunlea, E.J., Pinto, J.P., 2008. Ground-level nitrogen dioxide concentrations inferred
from the satellite-borne Ozone Monitoring Instrument. J. Geophys. Res. 113,
D 16308. doi: 10.1029/2007JD009235.
Lamsal, L.N., Martin, R.V., Parrish, D.D., Krotkov, N.A., 2013. Scaling Relationship for
NO 2 Pollution and Urban Population Size: A Satellite Perspective. Environ. Sci.
Technol. 47, 7855-7861. dx.doi.org/10.1021/es400744g.
Lee, C., Martin, R.V., van Donkelaar, A., Lee, H., Dickerson, R.R., Hains, J.C., Krotkov, N.,
Richter, A., Vinnikov, K., Schwab, J.J., 2011. S 0 2 emissions and lifetimes: Estimates
from inverse modeling using in situ and global, space-based (SCIAMACHY and
OMI) observations. J. Geophys. Res. 116, D06304. doi:10,1029/2010JD014758.
Levelt, P.F., van den Oord, G.H.J., Dobber, M.R., Mslkki, A., Visser, H., Vries, J., Stammes,
P., Lundell, J.O.V., Saari, H., 2006. The Ozone monitoring instrument. IEEE Trans.
Geosci. Remote Sens. 44, 1093-1101. doi:10.1109/TGRS.2006.872333.
Lu, Z., Streets, D.G., 2012. Increase in NOx Emissions from Indian Thermal Power Plants
during 1996-2010: Unit-Based Inventories and Multisatellite Observations. Environ.
Sci. Technol. 46, 7463-7470.
Martin, R.V., Jacob, D.J., Chance, k., Kurosu, T.P., Palmer, P.I., Evans, M.J., 2003. Global
inventory of nitrogen oxide emissions constrained by space-based observations
of
N 0 2 columns. J. Geophys. Res. 108(D17), 4537.
doi: 10.1029/2003JD003453.
Martin, R.V., 2008. Satellite remote sensing of surface air quality. Atmospheric Environment
42, 7823-7843.
McLinden, C.A., Fioletov, V., Boersma, K.F., Krotkov, N., Sioris, C.E., Veefkind, J.P.,
Yang, K., 2012. Air Quality over the Canadian oil sands: A first assessment using
satellite observations. Geophysical Research Letters 39, L04804.
McLinden, C.A., Fioletov, V., Boersma, K.F., Kharol, S.K., Krotkov, N., Lamsal, L., Makar,
P.A., Martin, R.V., Veefkind, J.P., Yang, K., 2014. Improved satellite retrievals of

36

NO 2 and SO2 over the Canadian oil sands and comparisons with surface
measurements. Atmos. Chem. Phys. 14, 3637-3656, doi:10.5194/acp-14-3637-2014.
MEM (BC Ministry of Energy and Mines), 2012. British Columbia’s Natural Gas Strategy:
Fueling B.C.’s Economy for the Next Decade and Beyond. Available at:
http://www.gov.bc.ca/ener/popt/down/natural gas strategv.pd f (accessed on: 15 May
2014).
MoE (BC Ministry of Environment), 2014. Report on Initial Network Design-NE B.C. Air
Quality Network: BC Ministry of Environment, Report to the SCEK Fund, BC Oil
and
Gas
Commission,
January
31,
2014.
Available
at:
http://www.bcairqualitv.ca/readings/northeast/pdfs/ne air monitor project report ne
twork design.pdf (accessed on: 15 May 2014).
Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M., Wasik, A., 2005. Passive Sampling
and/or Extraction Techniques in Environmental Analysis: A Review. Anal.
Bioanal.Chem. 381, 279-301.
NEB (National Energy Board), 2013. Energy Briefing Note: The Ultimate Potential for
Unconventional Petroleum from the Montney Formation o f British Columbia and
Alberta. Published by National Energy Board, British Columbia Ministry o f Natural
Gas Development, Alberta Energy Regulator, British Columbia Oil and Gas
Commission in November 2013, ISSN 1917-506X. Available at: http://www.nebone.gc.ca/clfnsi/mrgvnfintn/nrgyrprt/ntrlgs/ltmtptntlmntnvfrmtn2013/ltmtptntlmntnvf
rmtn2013-eng.pdf (accessed on: 19 September 2014).
OGC (BC Oil and Gas Commission), 2012a. Hydrocarbon and By-Product Reserves in
British Columbia; 2012 - BC Oil and Gas Commission. Available at
https://www.bcogc.ca/node/11111/download (accessed on: 15 May 2014).
OGC (BC Oil and Gas Commission), 2012b. Montney Formation Play Atlas NEBC, October
2012,
BC
Oil
and
Gas
Commission.
Available
at
http://www.bcogc.ca/node/8131/download (accessed on: 26 November 2013).
OMI User’s Guide, 2012. Ozone Monitoring Instrument (OMI): Data User’s Guide. OMI
Team,
NASA.
Available
at:
http://di sc. sci. gsfc.nasa. gov/Aura/ dataholdings/additional/documentation/README.OMI DUG.pdf (accessed on: 10
November 2013).
Runeckles, V.C., Bowen, P.A., 1999. The use of calibrated passive monitors to assess crop
loss due to ozone in rural locations. In: Agrawal, S.B., Agrawal, M. (Eds.),
Environmental Pollution and Plant Responses. Lewis Publishers, Boca Raton, FL
( 2000).
Seethapathy, S., Gorecki, T., Li, X., 2008. Passive Sampling in Environmental Analysis. J.
Chromatogr. A. 1184, 234-253.
37

Tang, Y.S., Cape, J.N., Sutton, M.A., 2001. Development and Types o f Passive Samplers for
Monitoring Atmospheric NO2 and NH3 Concentrations. The Scientific World 1,513529.
US EPA, 2010. Greenhouse Gas Emissions Reporting from the Petroleum and Natural Gas
Industry Background Technical Support Documtent. Washington DC: Climate
Change Division, US Environmental Protection Agency. Available at:
http://www.epa.gov/ghgreporting/documents/pdf/2010/Subpart-W TSD.pdf
(accessed on: 01 September 2014).
US EPA, 2011. Oil and Natural Gas Sector: New Source Performance Standards and
National Emission Standards for Hazardous Air Pollutants Reviews; 76 Federal
Register
52738,
August
23,
2011.
Available
at:
http://www.epa.gov/ttn/atw/oilgas/fr23aul 1.pdf (accessed on: 01 September 2014).
Van Reeuwijk, H., Fischer, P.H., Harssema, H., Briggs, D.J., Smallbone, K., Lebert, E.,
1998. Field comparison o f two NO 2 passive samplers to assess spatial variation.
Environmental Monitoring and Assessment 50, 37-51.
Vardoulakis, S., Lumbrera, J., Solazzo, E., 2009. Comparative evaluation of nitrogen oxides
and ozone passive diffusion tubes for exposure studies. Atmospheric Environment 43,
2509-2517.
WCEL (West Coast Environmental Law), 2003. Pump It Out: The Environmental Coasts of
BC’s
Upstream
Oil
and
Gas
Industry.
Available
at:
http://wceI.org/sites/default/files/publications/Pump%20It%20Qut%20%20The%20E
nvironmental%20Costs%20of%20BC%27s%20Upstream%200il%20and%20Gas%2
01ndustrv.pdf (accessed on: 18 September 2014).
Yu, C.H., Morandi, M.T., Weisel, C.P., 2008. Passive dosimeters for nitrogen dioxide in
personal/indoor air sampling: a review. Journal of Exposure Science. Environmental
Epidemiology 18, 441-451.
Zbieranowski, A.L., Aheme, J., 2012a. Ambient concentrations of atmospheric ammonia,
nitrogen dioxide and nitric acid across a rural-urban-agricultural transect in southern
Ontario, Canada. Atmospheric Environment 62, 481-491.
Zbieranowski, A.L., Aheme, J., 2012b. Spatial and temporal concentration of ambient
atmospheric ammonia in southern Ontario, Canada. Atmospheric Environment 62,
441-450.

38

3. Satellite observations of Nitrogen dioxide and Sulfur dioxide over Northeastern B.C.,
Canada
Abstract
The Peace River district o f Northeastern British Columbia (B.C.) Canada is a region of
natural gas production that has undergone rapid development since 2005. In order to assess
air quality implications of this a satellite air quality study in Northeastern B.C., using
multiple data products for Nitrogen dioxide (NO2) and sulfur dioxide (SO2), is presented. The
spatial distributions among all data products illustrate consistently high values in both
pollutants over the Montney formation which has experienced an increase in unconventional
natural gas activities. The Montney formation includes the communities o f Taylor, Fort St.
John and Dawson Creek. The magnitude of these high values for both NO 2 and SO2 ,
quantified in terms of tropospheric vertical column densities (VCDs) as well as surface
concentrations (except for SO2), are approximately one fifth and half, respectively, o f the
values found in the Canadian oil sands area, an area known as one of the largest non-urban
pollution sources in Canada. Furthermore, the temporal analysis of NO 2 data products
revealed higher values near Dawson Creek after 2007 with annual increment of 1.7% which
is coincident with the commencement of unconventional gas development activities.
However, a lack of available year round data limited the temporal analysis for S 0 2.
Keywords: Northeastern B.C., Nitrogen dioxide (NO 2), Sulfur dioxide (SO 2), Taylor, GEMMACH, Natural gas, Volume mixing ratio (vmr), Montney formation.
3.1. Introduction
In 2012, the total production of natural gas in British Columbia (B.C.) was 40,482* 106 m 3
(oil was 1 ,2 2 2 * 1 0 6 m3) with total estimated reserves (proven plus probable recoverable) of

39

1,138* 109 m 3 (oil was 19,108* 106 m3). This is a 146% increase over the natural gas reserves
estimated in 2006 (OGC, 2012). The trend o f increasing reserve estimates is largely due to
the successful development of unconventional gas extraction including the application of
horizontal drilling and hydraulic fracturing technology in the Montney formation and the
Horn River Basin of Northeastern B.C. (OGC, 2012). The exploitation o f these vast reserves
of natural gas is a significant economic driver and revenue generator in the province, and as
such the B.C. provincial government is planning on expanding this industry by promoting
development of liquefied natural gas (LNG) for export. It is estimated that 84,951 * 106 m 3 per
year will be produced in order to meet goals o f developing three LNG facilities by 2020
(MEM, 2012). B.C. is currently the second largest natural gas producer in Canada, and
Canada is the 5th largest in the world (CAPP, 2013).
Natural gas is a nonrenewable fossil fuel that develops naturally over millions o f years from
the carbon and hydrogen molecules of ancient organic matter trapped within geological
formations. The two major geological formations in Northeastern B.C. are the Montney
formation and the Horn River Basin. The Montney formation is an unconventional Lower
Triassic aged formation that includes dry, liquid rich gas and oil in over-pressured siltstones
that stretches toward northwest 200 km from the B.C.-Alberta border near Dawson Creek to
the B.C. foothills o f the Rocky Mountains. The unconventional mid-Devonian aged Horn
River Basin is a shale play with dry gas over-pressured of the Muskwa, Otter Park and Evie
Formations near Fort Nelson in Northeastern B.C. Currently, these two formations that
utilize unconventional drilling which commenced extensively in 2007/2008, account for 60%
of BC’s total production. Conventional drilling methods were predominantly applied before
2007 (OGC, 2012). It is expected that natural gas from unconventional sources will continue

40

to increase while conventional pools will be depleted in the next few years (OGC, 2012).
Unconventional natural gas development consists of two main phases: well development and
production. Individual well development involves three stages: pad preparation, well drilling
and well completion (US EPA, 2010). Recently, the US EPA estimated that well completions
involving hydraulic fracturing in an uncontrolled manner can vent natural gas to the
atmosphere resulting in air pollution. This is done to guard against the over-pressuring o f the
well which results in approximately 230 times more natural gas being vented compared to
wells without hydraulic fracturing (conventional well drilling) (US EPA, 2011). Therefore,
people who lived within half a mile of the unconventional wells had a greater risk o f
developing non-cancer health effects from short-term exposure to the high emissions of
hydrocarbons than those living further away (McKenzie et al., 2012). The well development
phase is not the only source o f air pollution; the production phase involving flaring (either
purposefully or accidentally), processing, compressing, pipeline distribution, storage, etc.
also may lead to air pollution. The leading air pollutants from natural gas development
activities are hydrogen sulfide (PUS), SO2 from sulfur rich (or sour) gas, methane (CH4), non­
methane hydrocarbons typically volatile organic compounds (VOCs), and nitrogen oxides
(NOx) (Lattanzio, 2013).
Concerns have arisen recently in Northeastern B.C. about increasing air pollution resulting
from accelerating natural gas production (Fraser Basin Council, 2013; Krzyzanowski, 2012;
MoE, 2014) without simultaneous implementation o f available technological advances to
control emissions (Krzyzanowski, 2009). According to the 2010 emissions data reported to
Canada’s

National

Pollutant

Release

Inventory

(NPRI,

http://www.ec.gc.ca/Ddb/websol/quervsite/querv e.cfm). SO2 and NO 2 are the dominant

41

species among all gaseous air pollutants in Northeastern B.C. which are emitted from various
stages o f oil and gas activities but mostly from the production phase such as processing
(flares, engines, and compressors), distribution (leaks o f pipelines and flanges), and also from
storage tanks as vaporization (Krzyzanowski, 2012). However, oil and gas industries are not
required to report to the NPRI the emissions during the well development phase (e.g., in pilot
phase

or

in

exploration

or

drilling

phase)

npri/default.asp?lang=En&n=02C767B3-C9FD-4DD7-8072).

(http://ec.gc.ca/inrp-

Therefore,

it

has

been

suggested that both of these gaseous species are under-reported (Krzyzanowski, 2009) and
under-monitored (Krzyzanowski, 2011). Though significant emissions during production
have been reported by the NPRI, there are only four permanent SO2 monitors installed over
the previous 15 years (Pine River Hasler and Pine River Gas Plant near Chetwynd, Taylor
Townsite and Taylor South Hill near Taylor) and no monitoring o f NO 2 except for occasional
short term monitoring in some places in 2010 and 2011 using the B.C. MoE mobile air
monitoring laboratory. However, SO2 is also monitored at three additional industry operated
sites, with data they are not publicly accessible (MoE, 2014). Due to this growing concern
and high public demand, B.C. MoE recently has commissioned three new stations (Doig
River, Farmington Community Hall and Tomslake) which were deployed in December 2013
and January 2014 to monitor SO2 and total reduced sulfur (TRS) along with meteorology, as
a pilot study (MoE, 2014).
NO 2 and SO2 are gaseous species that undergo chemical and/or physical reactions in the
atmosphere and contribute to acidic deposition in terrestrial ecosystems as dry-deposited
gases or in dissolved form in precipitation, fog, and cloud (Cox, 2003) and also may impact
on human health through a number o f environmental pathways (Krzyzanowski, 2012). In

42

aerosol form, they can also impact visibility while NO 2 as a precursor to the formation of
photochemical oxidants can also lead to direct impact on human health (Cox, 2003). Previous
studies regarding air quality issues in northeastern B.C. and Alberta reported that elevated
levels o f SO2 in this territory lead to direct injury to natural vegetation (Legge et al., 1998).
Recently, as a result of the growing concern on potential negative impacts of gas extraction
and processing, the B.C. provincial government has initiated a three-phase study on the
health impacts of oil and gas activities in B.C.’s northeast region. Respondents during the
Phase-1 preliminary study complained o f personal health problems - such as asthma,
bronchitis, cancer, stress and sleep deprivation associated with oil and gas activities (Fraser
Basin Council, 2013). Besides the impact-oriented issues, some respondents were dissatisfied
by what they saw currently as insufficient information available to them from both the
government and the oil and gas sector, and a lack o f transparency with respect to specific oil
and gas activities. Consequently, it was suggested from the health impact study in Northeast
B.C. that rigorous investigation and regulations are required for baseline assessments o f air
quality and also adequate communication with the public should be conducted prior to oil
and gas resource development activities. This baseline information is not available from the
current ambient monitoring network due to the limited number o f monitoring sites, so new
monitoring locations should be assessed. Decisions on where to install new stations should be
based on background information such as which pollutants related to oil and gas
development activities are changing over time in the ambient air.
Recently it has been suggested from the satellite observations o f NO 2 and SO2 over the
Canadian oil sands in Alberta (McLinden et al., 2012 and 2014) and space-based ambient
concentration of NO 2 in North America (Lamsal et al., 2008) that satellite observations

43

would be an alternative and complementary to surface-based measurements. Therefore, the
overall purpose of this study is to assess air quality using satellite remote sensing
observations o f NO 2 and SO2 over portions of Northeastern B.C. that are undergoing an
expansion in natural gas production. To the best of our knowledge, there are no published
articles available that describe the air quality of Northeastern B.C. using satellite
observations. This chapter briefly introduces different satellite NO 2 and one S 0 2 data
products along with the importance o f AMFs, particularly for this study area, in section 2.
This section also describes the basic satellite data filtering criteria prior to obtain valid
scenes. In section 3, spatial distributions of NO 2 VCDs of the study area in the context of
known pollution sources over most of B.C. and Alberta have been analyzed by applying a
pixel-averaging approach to different data products followed by space-based and modelsimulated surface level NO 2 spatial distribution. SO2 VCDs were analyzed over a small
domain. This study also compares air quality in Northeastern B.C. with air quality in the
Canadian oil sands area, which is known as one o f the largest non-urban pollution sources in
Canada (McLinden et al., 2012).
3.2. Satellite data products of air quality
Lower tropospheric pollutant observations from satellites began with the GOME-1
instrument (Global Ozone Monitoring Experiment, 1996-2003) aboard the ERS-2 satellite
(Burrows et al., 1999) and has continued with SCIAMACHY (SCanning Imaging Absorption
spectroMeter for Atmospheric CHartographY, 2002-2012) aboard the ENVISAT satellite
(Bovensmann et al., 1999), and currently by the OMI (Ozone Monitoring Instrument, 2004present) in the Aura satellite (Levelt et al., 2006) and the operational GOME-2 (2007present) in the MetOP platform (Martin, 2008). The basic characteristics o f these instruments

44

are listed in Table 2.1. The satellites have been designed to yield information on the lower
tropospheric trace gas constituents by flying in near-polar, Sun-synchronous, low earth
orbits; a typical orbit altitude is 705 km (Martin 2008). These trace gas constituents (e.g.,
NO 2 and SO2), are processed to yield vertical column density (VCD) from the integrated
column amount by measuring UV-visible solar radiation in nadir (down looking) geometry.
The tropospheric VCDs from the calibrated spectra involves three main steps (Boersma et al.,
2007): 1) a spectral fit to determine the slant columns densities (SCD), 2) elimination of the
stratospheric contribution to the SCD; and 3) conversion of tropospheric SCD to tropospheric
VCDs via the air mass factor (AMF) where the AMF accounts for the sensitivity o f the
instrument to the absorber (target species, such as: NO 2 and SO 2) (McLinden et al., 2014).
The AMF describes the enhancement in absorption when light traverses a slant path through
a layer and therefore it represents a basis of trace gas retrievals in the spectrum o f UV-visible
wavelength. It is noted that the AMF is the ratio o f the SCD to the VCD and it is computed
using a radiative transfer model because o f the complex path of the scattered and reflected
sunlight. The radiative transfer model simulates nadir radiances and accounts for all relevant
physical factors such as: multiple scattering by molecules and aerosols, absorption by trace
gases, and reflection from the surface. The accuracy of the AMF calculation is largely
governed by the accuracy o f information on the absorber vertical profile, cloud and aerosol
information, surface reflectivity and surface pressure (McLinden et al., 2014).
In this analysis, NO 2 data from three instruments (SCIAMACHY, OMI and GOME-2)
(2005-2013) were used for analysis. While multiple OMI NO 2 VCD data products currently
exist, the two global, primary products from the NASA standard product (SP) (Bucsela et al.,
2013;

http://disc.sci.gsfc.nasa.gov/Aura/data-holdings/OMI/omno2 v003.shtml)

45

and

the

Dutch

OMI

NO 2

(DOMINO)

(Boersma

et

al.,

2011;

http://www.temis.nl/airpollution/no2 .html') processed in near real time have been included in
the analysis. The reason for using two OMI NO 2 data products in this analysis is to visualize
how NO 2 VCDs particularly for the study area vary with data products. These two OMI NO 2
data products are derived from different separation techniques of stratospheric SCDs
(Bucsela et al., 2013). The DOMINO product simulates the stratospheric NO 2 by assimilating
OMI SCDs in a chemical data assimilation system whereas the SP executes a complex highpass filtering approach (Dirksen et al., 2011). In addition, the SP algorithm depends on
monthly mean profiles from the Global Modeling Initiative (GMI) (Bucsela et al., 2013)
while the DOMINO product uses daily output from the TM4 (Tracer Model 4) chemical
transport model (Boersma et al., 2007). However, both o f these models make use o f emission
inventories appropriate for the late 1990s (McLinden et al., 2014), when oil and gas
development activities in Northeastern B.C. were significantly lower than the current level.
An accurate absorption profile of trace gases (e.g., NO 2) substantially determines the
accuracy o f the AMF calculation whereas the AMF plays a fundamental role in the
calculation o f tropospheric VCDs from SCDs. Therefore, McLinden et al. (2014) recently
recalculated the AMF (named as EC-AMF, EC stands for Environment Canada) for North
America based on updated and higher resolution absorber profile from a regional-scale (15
km x 15 km resolution) air quality model (the Global Environmental Multi-scale - Modeling
Air Quality and Chemistry, GEM-MACH; for more details refer to Anselmo et al., 2010)
using emission inventories from the US EPA and EC data for the year 2006, high spatial and
temporal

resolution

surface

reflectivity

from

the

Moderate

Resolution

Imaging

Spectroradiometer (MODIS) satellite instrument, and an improved treatment of snow.

46

Finally, two new OMI NO 2 data products were calculated from each o f the tropospheric
SCDs o f the SP and DOMINO by dividing them by the EC-AMFs instead o f the original
AMFs (McLinden et al., 2014). These new VCDs, named as EC-SP-N0 2 and EC-DOMINONO 2, were also included in this analysis (2005-2013). Besides these OMI NO 2 data products,
NO 2 data from SCIAMACHY between 2005 and 2011 were combined with the GOME-2
data for 2012-2013 in this analysis (the combined dataset is named as SCIA-N02). The
dataset created by this combination approach (Ghude et al., 2011) has been implemented for
the comparison with OMI data products between 2005 and 2013. Note that the same retrieval
techniques were used for NO 2 data products of both SCIAMACHY and GOME-2
instruments and flight time of the satellites carrying these two instruments is almost identical
(Table 2.1). Although multiple NO 2 data products have been included in this study, analysis
of SO2 data (2005-2013) were only limited to the OMI instrument (OMS02 specific to the
planetary

boundary

layer

[PBL],

version

3,

http://disc.sci.gsfc.nasa.gov/Aura/data-

holdings/OMI/omso2 v003.shtml, this data product is named here as NASA-SO 2 for
simplicity). This product is retrieved based on the NASA Band Residual Method (BRM)
using four wavelengths between 310 and 315 nm to quantify SO 2 absorption (Krotkov et al.,
2006). Generally, all these satellite data products o f both NO 2 and SO2 offer the possibility to
improve our understanding of lower tropospheric trace gas concentrations (Lamsal et al.,
2008 and 2013; Lee et al., 2011), as well as helping with identification o f emission sources
(Fioletov et al., 2011 and 2013; Lu and Streets, 2012; Martin et al., 2003), and atmospheric
chemistry, through testing and improvements to emission inventories, using top-down
modeling techniques (Boersma et al., 2008).

47

Orbit based level 2 data products, in which the location o f an observation is assigned to the
co-ordinates o f the pixel center, from all these instruments were used in this study. OMI
measures 60 cross-track positions (pixels) during each along track scan. The across track
width differs in size depending on the track position, however, along track width is constant
(13 km). Therefore, 10 pixels on each swath edge were excluded from analysis to limit the
across-track pixel width to -4 0 km. In addition, some cross-track positions are affected by an
error called the “row anomaly” (RA) since June 2007 (also see Table 2.2) due to the partial
external

blockage

of

the

radiance

port

on

the

OMI

instrument

(http://www.knmi.nl/omi/research/product/rowanomalv-background.php), were dynamically
removed based on the RA flags in all OMI data products o f both species. Daily OMI data
products were also filtered to remove data with high cloud radiance fraction (>0 .2 ) and large
solar zenith angle (SZA) (>75° for NO 2 and >60° for SO2). Note that, the stringent SZA
threshold o f 75° for NO 2 means that about 75% data is from April-September and the
remaining 25% is from October-March. However, using a 60° SZA threshold for SO 2 keeps
99% of the data between April and September of each year (McLinden et al., 2014). Surface
albedo was limited to below 0.3 for both SP and DOMINO data products o f OMI. Year
round NO 2 data products from all these instruments were used but, following Fioletov et al.
(2011), only summertime (May - August) NASA-SO 2 data were used in order to maximize
the signal-to-noise ratio (SNR: the ratio between the mean values and the standard errors of
the means) and also were restricted to values between -10 and +5 DU (1 DU = 2.69 * 1026
molecule/km2) to avoid spikes from transient volcanic plumes.

48

3.3. Tropospheric NO 2 column densities over Northeastern B.C.
The nine year average (2005-2013) of tropospheric NO 2 VCDs over major parts o f B.C. and
Alberta, Canada, from different data products have been visualized in Fig 3.1 in an effort to
provide the context for the satellite observed pollution levels across the Peace River region of
Northeast B.C. Places with known large pollutant sources, including several urban areas, are
included in this geographical area. Major pollutant sources in some o f these urban areas
(such as Vancouver, and Victoria) are primarily related to vehicle emissions while regions of
large emissions particularly north of Fort McMurray but also Calgary and Edmonton also
have significant emissions due to oil and gas activities. Fort St. John (* symbol in Fig. 3.1),
the main town located within the Montney formation o f Northeastern B.C., has very small
VCDs compared with larger emission regions (Fig. 3.1). The pixel averaging technique
described by Fioletov et al. (2011 and 2013) was employed in this analysis to obtain a
statistically significant signal. Initially a 6 x 6 km 2 grid was defined for this larger area and
all screened observations of SP and DOMINO NO 2 data products within 2005-2013 and
falling within a radius o f 24 km from the grid-cell center were averaged. However, a larger
radius is required for analyzing GOME-2 or SCIAMACHY NO 2 products by pixel averaging
technique due to these two data products having a coarser resolution (Table 2.1) (Fioletov et
al., 2013). To remain consistent, a 60 km radius was used during the intercomparison
between DOMINO-NO 2 and SCIA-NO2 products. This long term averaging of VCDs from a
large number o f observations provides some advantages such as the center and shape o f the
enhancement of a given source location can be determined. All the plots in Fig. 3.1 illustrate
consistently high values over the major known pollution sources in this larger region with
some differences (DOMINO vs SCIA, and DOMINO vs SP) in the values mostly due to: i)

49

overpass times - OMI observes early afternoon while the others observe mid-morning; and ii)
the separation o f stratospheric contribution to the total SCDs. These findings show strong
agreement with results from McLinden et al. (2012 and 2014) whose principal focus was to
investigate the satellite observations of NO 2 over the Canadian oil sands, located near Fort
McMurray, Alberta. Overall, Northeastern B.C. does not show any significant values o f NO 2
VCDs in the context of the large area examined which suggests further investigation for
greater detail considering a smaller domain.
The small domain, roughly 224 km x 224 km, spanning 54.8°N to 56.8°N latitude and
123.5°W to 120.0°W longitude was chosen as this area covers most o f the Montney
formation o f Northeastern B.C. where substantial conventional and unconventional natural
gas activities have been taking place. Owing to its superior resolution (Table 2.1), OMI
allows for greater details in this small domain. All OMI NO 2 data products were analyzed
using a 1.2 x 1.2 km 2 grid and averaging over an 18 km radius. Plots o f different OMI data
products are shown in Fig. 3.2. The elongated ellipses in the north-south direction in the
eastern side (near B.C.-Alberta border) of the small domain consistently were identified in all
the plots of Fig. 3.2. Note that a similar pattern of distribution has been obtained with the
tropospheric SCD (tropospheric VCDs of OMI NO 2 data products x corresponding AMFs)
(see appendix, Fig. A l) that suggests the signals with the VCDs have not being created by
the AMFs themselves. The highest value o f VCD was found near Taylor (~15 km south of
Fort St. John) where Station 1-Taylor (natural gas extraction station) is located and this is the
significant source of NO 2 according to the NPRI database (https://ec.gc.ca/inrpnpri/default.asp?lang=en&n= 1D892B9F-Q. Besides the high values near Taylor, elevated
levels were also captured in the northward direction toward Fort St. John, and also to the

50

south near Dawson Creek. These areas are close to regions of gas development, with
unconventional drilling in the Montney formation, and processing activities based on the BC
Oil and Gas Commission (OGC, 2012) and NPRI database. Although the places with
elevated VCDs have been captured consistently by all the OMI NO 2 data products,
significant differences in the values o f VCDs were identified. The SP NO 2 (Fig. 3.2c) is
larger than the DOMINO (Fig. 3.2a) by a factor of 2.6 over the community o f Taylor and this
is due to the different stratospheric VCD separation techniques involved in the algorithms o f
these two data products as well as the lower frequency o f negative tropospheric VCDs in the
SP data products (Bucsela et al., 2013). On the other hand, EC -D 0M IN 0-N 0 2 (Fig. 3.2b)
data products differ from DOMINO products (Fig. 3.2a) by a factor o f -1.9, and this is due to
the improved EC-AMFs. McLinden et al. (2014) reported that the improved EC-AMFs are
predominantly driven by the accurate input information o f the 2006 emission inventory based
on an updated profile shape and accounted for over 90% o f the average decrease in the
original AMF. The same is true when comparing the SP (Fig. 3.2c) and EC-SP-NO 2 (Fig.
3.2d) over Taylor B.C.: the latter one is larger by a factor of 1.3. A similar degree o f
difference was also reported by McLinden et al. (2014) over the Canadian oil sands region.
The reason for the difference between EC-DOMINO-NO 2 (Fig. 3.2b) and EC-SP-NO 2 VCDs
o f the small domain can be explained by the different algorithm used for stratospheric
separation. It should be recognized that this region overall has much lower VCDs compared
to other known polluted places such as the oil sands region north o f Fort McMurray, Alberta
(Fig. 3.1). From the NPRI data set it is found that maximum individual NOx emission over
the oil sands region is roughly 15,000 tonnes per year (t/y) and the long-term (2005-11)
averaged EC-SP-NO 2 tropospheric VCDs- reported 4.0 x 1 0 15 molecules/cm 2 (McLinden et

51

al., 2014). However, the maximum NOx emission from the largest source o f Northeast B.C.,
located in Taylor, is approximately 1,700 t/y which is approximately nine times less than in
the oil sands region. It is therefore expected that the OMI VCDs over Northeast B.C. will
also be lower than in the oil sands region by at least a factor of 9.0. The present study finds
8.0 x 1014 molecules/cm 2 as the maximum tropospheric VCD near the Taylor from EC-SPNO 2 products (Fig. 3.2d) which is approximately five times less than in the oil sands region.
This comparison puts the air quality near Taylor into an appropriate context.
In addition, the superior algorithm used for the SP NO 2 data products (Bucsela et al., 2013)
along with corrected EC-AMFs (McLinden et al., 2014) suggest that the EC-SP-NO 2 data
products will be used for temporal analysis. In this temporal analysis, NO 2 VCDs between
2005 and 2007 have been averaged, using the pixel averaging technique, and were compared
with the average values between 2008 and 2013. The averaging time interval between the
earlier and later periods was doubled as some o f the cross-track positions were affected by
RA since 2007 (Lu et al., 2013) (see also section 3.2). The dynamical RA flagging removes
the affected pixels so that the number o f valid observations in each pixel averaging radius (18
km) is decreasing with time, for example, valid observations in 2005-2007, 2008-2010 and
2011-2013 were found to be 394, 219, and 207, respectively. However, it was 422 in the
doubling time interval (2008-2013) phase which is nearly consistent with the observation
numbers between 2005 and 2007. Comparing the 2005-2007 (Fig. 3.3a) with the 2008-2013
(Fig. 3.3b) mean NO 2 VCDs, an increase in NO 2 with time is identified close to the
community of Dawson Creek and this increment likely reflects the unconventional drilling
commencement since after 2007 in the Montney formation (OGC, 2012). From these two
figures (Fig. 3.3a and 3.3b), the maximum VCD of the NO 2 enhancement over the study area

52

is seen to be increasing at rates of 1.7%/y. This is calculated from the maximum values of
these two time intervals, given that few months (December to February) o f each year retain
limited observations due to high pass filtering of satellite data (see also section 3.2). It is
important to note that 60% of the total 2 0 1 2 ’s production comes from unconventional
sources in B.C. and the Montney formation contributes 40% of the total 2012 unconventional
production, however, southern Montney formation (or Heritage Montney field) which is
located in the study area produces 60% o f the total Montney production (see also section
2.2.1) (OGC, 2012).
The NO 2 data products of different sensors have also been investigated for this small domain.
The tropospheric VCDs from SCIA-NO 2 (Fig. 3.4b) data products are nearly two times
higher than the values of DOMINO data products (Fig. 3.4a). It is likely that photochemical
loss of NO 2 during the afternoon has resulted in a lower concentration o f NO 2 in the PBL
during the OMI flight time, since the photochemical activity is not as strong in the morning
(SCIAMACHY and GOME-2 flight time, also refer to Table 2.1). Besides the photochemical
loss, diurnal variation in NO 2 columns also may be driven by changing emissions throughout
the day (Boersma et al., 2008), however, investigating this possibility is beyond the scope of
this study. The SCIA-NO 2 data products with coarse resolution (see also Table 2.1) cannot
depict the maximum values as distinctly over the large source (Fig. 3.4b) particularly in a
region having a cluster of sources (NPRI data base). This is due to the requirement of a larger
radius in the pixel averaging technique for analyzing data products having a coarse resolution
(Fioletov et al., 2013). The eastern boundary in the area of maximum value, previously
identified in Fig. 3.2, also implies the presence o f cluster of sources to the east o f the B.C.Alberta border near Dawson Creek and Fort St. John where the southern Montney formation

53

starts (for details refer to section 2.2.1). Thus, the maximum value over Taylor originally
identified by all OMI data products (Fig. 3.2) has been shifted towards the east rather than
west with 60 km averaging radius in OMI data (Fig. 3.4a).
3.4. Ambient NO 2 concentration
Although no in-situ ground-based instruments are currently available in the study domain it
is necessary to represent the air quality of the study area in an easily understandable format
rather than the less familiar vertically integrated quantities (VCDs). A thorough spatial
coverage of ground-level NO 2 measurements is needed for exposure assessments (Lamsal et
al., 2008). Currently, no stations are operating in Northeastern B.C. for this purpose. Lamsal
et al. (2008) inferred surface-level NO 2 concentrations or surface volume mixing ratio-vmr
(S) from OMI tropospheric VCDs (Q) by applying the ratio of surface-level NO 2
concentrations (S0) to vertical column densities (£?c) calculated by a global chemical
transport model, usually by the GOES-Chem:

s = (S g/Q g) x a

(l)

Environment Canada researchers identified the general consistency o f GEM-MACH with
GOES-Chem (McLinden et al. 2014) during simulation of NO 2 and SO2 in the PBL.
Therefore, GEM-MACH could be an alternative for the well-established GOES-Chem
model. Equation (1) was applied to the EC-VCDs using the same monthly mean profiles used
in the calculation of the EC-AMFs (for detail refer to McLinden et al., 2014) and the ratio is
obtained from the GEM-MACH (so ‘G’ in equation (1) stands for GEM-MACH). The output
from equation (1) is referred to as EC-vmr for simplicity. For this analysis EC-vmr (based on
EC-DOMINO-NO 2) and GEM-MACH-vmr were collected from EC (Chris McLinden,
Environment Canada, personal communication, 2014). The 2005-2013 average surface NO 2

54

EC-vmr and GEM-MACH-vmr maps are illustrated in Fig. 3.5, and were calculated using the
same pixel-averaging parameters as the VCDs in Fig. 3.2. The spatial distributions from ECvmr (Fig. 3.5a) and GEM-MACH-vmr (Fig. 3.5b) generally mimic o f spatial distribution of
VCDs (Fig. 3.2), with a maximum NO 2 EC-vmr of nearly 0.4 ppbv (parts per billion by
volume) and a maximum GEM-MACH-vmr of 0.6 ppbv. McLinden et al. (2014) reported the
maximum long term average NO 2 EC-vmr of 2.3 ppbv over oil sands area to be 5.75 times
higher than Northeastern B.C., which is consistent with the VCDs (also see section 3.3). The
GEM-MACH map (Fig. 3.5b) shows similar spatial patterns to the EC-vmr map, GEMMACH NO 2 where values are typically 50% larger than those of EC-vmr (Fig. 3.5a) through
the area of higher concentration. The primary reason for this might be that it is associated
with the large number of EC-DOMINO-NO 2 VCDs being removed due to high pass filtering,
especially after 2007 due to the RA issue (Table 2.2). This can be seen in the 50% drop in
valid signals of each pixel averaging circle (18 km radius) near to the area o f high
concentration in the EC-vmr map (Fig. 3.5c) compared to GEM-MACH-vmr map (Fig.
3.5d). However, NO 2 generally has a longer lifetime during winter with more shallow mixing
depths than in summer due to the photochemical reactions in the PBL (Lamsal et al., 2008).
Lamsal et al. (2008) also illustrated that the winter mean OMI tropospheric VCDs over the
United States and southern Canada is 32% lower than the corresponding value from GOESChem which, in turn, may limit the value o f ‘S’ in the equation (1).
3.5. Tropospheric SO 2 column densities over Northeastern B.C.
NASA-SO 2 tropospheric VCDs are presented in this section with a focus on the major
emission sources in Northeastern B.C. This is done because previous studies report that
elevated SO2 column densities which have sufficiently high SNRs representing statistically

55

significant values will only be achieved within about 50 km o f major emission sources using
the NASA-SO 2 data product (Fioletov et al., 2011). However, emission sources may affect
only 1-2 pixels (Fioletov et al., 2011) and the NASA-SO 2 VCDs have high noise levels
(Fioletov et al., 2011 and 2013; McLinden et al., 2012 and 2014). Therefore, NASA-SO 2 data
products with valid observations (see section 3.2) between May and August o f each year
throughout 2005-2013 were taken for analysis. Fig. 3.2-3.5 clearly indicates the consistently
high values o f NO 2 near Taylor throughout the study period, where Station 1-Taylor (natural
gas extraction industry) is a known large source o f NOx emissions. However, this industry
emits very little SO2, while the McMahon Gas Plant (gas processing industry) located very
close to the Station 1-Taylor is one o f the largest sources o f SO2 emission (emission rate is
~3,500 t/y) in Northeastern B.C. (NPRI data base). Consequently, a 64 km x 64 km domain
with Taylor roughly in the center was selected and a pixel averaging approach was also used
taking 24 km as a pixel averaging radius in an effort to reduce noise (Fioletov et al., 2013). It
should also be noted that local bias correction (or background level), as suggested in previous
studies (Fioletov et al. 2011 and 2013), can be accomplished by calculating the average o f all
NASA-SO 2 VCDs from pixels centered between 250 km and 300 km from the source
location and this average value (or local bias) was subtracted from all measurement near the
source. Due to the large distribution o f oil and gas industries in the Northeastern B.C. (see
also Fig. 2.1), the NASA-SO 2 VCDs from pixels centered between 416 km and 466 km
(north east of Taylor) with no source have been averaged monthly and subtracted from the
monthly mean o f each grid (1.2 x 1.2 km 2 grid) near the source. The high value (-0.18 DU)
of the long-term (2005-2013) average VCDs (Fig. 3.6a), after the local bias correction, is
seen to cover a large area with relatively weaker SNR (Fig. 3.6b) compared with that of the

56

NO 2 analysis (the SNR in all analysis of NO 2 data products remains >15 in all directions
except west of the source, McMahon gas processing plant). The relatively large area of
elevated SO2 is partially a result o f the pixel averaging approach, but may also indicate that
SO2 emitted from stacks (in the case o f McMahon gas processing plant) is spread over a large
area (McLinden et al., 2014). The longer lifetime o f SO2 than NO 2 (Lee et al., 2011) in the
lower atmosphere also may influence this high value across a large area. In general, this high
value is consistent with the recent air quality studies in Northeastern B.C. that reported large
sources in Taylor, and S 0 2 concentrations would likely be highest in Taylor (Krzyzanowski,
2011, MoE, 2014) and to the east o f Taylor (MoE, 2014).
The available ground based in-situ SO2 monitoring station located within few kilometers of
the source (McMahon gas processing plant) along with S 0 2 signals allows for a comparative
analysis with that of the oil sands area of Alberta. The high values o f SO 2 (2005-2013) over
the Taylor area (or around the McMahon gas processing plant) in terms o f NASA-SO 2VCDS,
and in-situ station (monitoring station name: Taylor Town Site, data is publicly available in
the B.C. Ministry o f Environment website, available at: http://envistaweb.env.gov.bc.ca/.
accessed on: 06 September 2014) are 0.18 DU (Fig. 3.6a), and ~2 ppb, respectively. These
are smaller than a factor of ~ 1.9 and 2.6, respectively in the corresponding parameters in the
oil sands area (McLinden et al., 2014). The ratio o f in-situ station to NASA-SO 2-VCDS in
both sites is highly consistent, Oil sands area: 14.8 (or 10.4 with EC-AMFs based VCDs)
(McLinden et al., 2014), and Taylor: ~11, which implies Taylor in Northeastern B.C. is
approximately two times less polluted in terms of SO2 . However, the NASA-SO 2 VCDs in
the oil sands area (due to annual SO2 emission rate from Syncrude Canada Mildred Lake
Plant Site ~ 80 kt/y) is as large as that from any other individual emissions source in Canada,

57

including the large base-metal smelting operations in Manitoba (Thompson Operations) and
Ontario (Sudbury nickel smelter complex) (McLinden et al., 2012). Assuming the air quality
data mentioned above is reliable, the large number of small sources in and around Taylor
(NPRI data base) may cumulatively account for the higher value than the expected level.
Another probable explanation is related to the geomorphological features o f Taylor (in a
valley) along with its industrial activities that may impede the efficient horizontal dispersion
of pollutants and lead to a strong thermal inversion, which would limit vertical mixing
mostly in the fall and winter (Ainslie and Jackson, 2009). Furthermore, the limited vertical
mixing is confirmed from the long term (2005-2013) average ambient concentration (~0.9
ppb) of the Taylor South Hill monitoring station (data is publicly available in the B.C.
Ministry of Environment website, available at: http://envistaweb.env.gov.bc.ca/. accessed on:
06 September 2014) which is located at 6 8 8 m above the mean sea level (MSL) and five km
south of the McMahon gas plant, whereas this gas plant and the Taylor Town Site station are
only one km apart in same elevation level (~ 470 m MSL) (Fig. 3.6). It should also be noted
that the oil sands region is relatively flat and also located away from a large number of
individual small sources. Since previous studies revealed that NASA-SO 2 data products can
detect statistically significant signals only from the sources emitting SO2 above 70 kt/y
(Fioletov et al., 2011), therefore, further study is suggested to validate the present satellite
SO2 observation in the study area.
3.6. Conclusions
This study presents satellite data of NO 2 (OMI and combination o f SCIAMACHY and
GOME-2) and SO2 (OMI) over the Northeastern B.C. and demonstrates a long term (20052013) spatial distribution using a pixel averaging approach with different data products from

58

these instruments (for SO2 only NASA SP data product). All the data products o f NO 2 VCDs
indicate that the maximum value is confined to the vicinity of Taylor, Fort St. John, and
Dawson Creek area where intensive natural gas development activities are taking place. In
this vicinity the maximum long term average NO 2 VCDs have been identified consistently in
Taylor owing to its position as a major source o f emissions throughout the study period.
However, temporal analysis revealed higher values near Dawson Creek after 2007 with
annual increment of 1.7% which implies that the commencement of unconventional gas
development activities also lead to high NO 2 VCDs. Maximum reported values vary
substantially with data products largely due to AMF, tropospheric VCDs retrieval algorithm,
and possibly instrument observation time (satellite flight time). Notably, the NO 2 VCDs from
SP are 2.6 times higher than those o f DOMINO and with the new EC-AMFs the difference
remains nearly the same, but values of both data products increased with the EC-AMFs. In
addition to the VCDs, this area was also spatially examined with the surface level NO 2
concentrations taking data from both satellite and model simulations. These two different
data sets reflect a similar spatial distribution o f NO 2 concentrations compared to the NO 2
VCDs, however, GEM-MACH simulated ambient concentrations are two times larger than
the satellite retrieval. Besides the multiple analysis of NO 2 in a relatively large area, SO2 was
analyzed only within few kilometers around the source (McMahon gas processing plant) also
located in Taylor using NASA-SP products and an in-situ monitoring station data and it
demonstrated that Taylor is half as polluted as Canada’s largest SO2 emission sources area,
the Canadian oil sands area, due to its morphological features along with large number of
individual sources in close proximity. All these pieces of evidence in this previously less
studied area suggest further investigation is required to directly validate the present satellite

59

investigation. This can be done using the high resolution air quality chemistry model (GEMMACH or WRF-Chem) coupled with accurate emission estimation rather than relying on
comparative analysis or in-situ stations without involving dispersion models (e.g.,
CALPUFF).
Acknowledgement
The BC Oil and Gas Commission is acknowledged for providing funding for this study
through a gift to UNBC. We acknowledge the free use o f tropospheric NO 2 column data from
the GOME-2, SCIAMACHY, OMI sensors from www.temis.nl. We also acknowledge the
NASA Goddard Earth Sciences Data and Information Services Center (GES DISC;
http://daac.gsfc.nasa.gov/) for free access o f the OMI NO 2 and SO2 data products. We also
acknowledge the B.C. Ministry of Environment for providing access to air quality data from
their

continuous

monitoring

network

(http;//envistaweb.env.gov.bc.ca/).

Finally,

Environment Canada is also highly acknowledged for providing the new satellite data
products with new EC-AMFs as well as GEM-MACH simulated surface concentrations of
N 0 2.

60

3.7. References
Ainslie, B., Jackson, P.L., 2009. The use o f an atmospheric dispersion model to determine
influence regions in the Prince George, B.C. airshed from the burning o f open wood
waste piles. Journal of Environmental Management 90, 2393-2401.
Anselmo, D., Moran, M.D., Menard,S., Bouchet, V., Makar, P., Gong, W., Kallaur, A.,
Beaulieu, P.-A., Landry, H., Stroud, C., Huang, P., Gong, S., Talbot, D., 2010. A new
Canadian air quality forecast model: GEM-MACH 15. Proc. 12th AMS Conf. On
Atmos. Chem, Jan 17-21, Atlanta, Ga, American Meteorological Society, Boston,
MA., 6 pp., http://ams.confex.com/ams/pdfpapers/165388.pdf (access on: 04
September 2014).
Boersma, K.F., Eskes, H.J., Veefkind, J.P., Brinksma, E.J., van der A.R.J., Sneep, M., van
den Oord, G.H.J., Levelt, P.F., Stammes, P., Gleason, J.F., Bucsela, E.J., 2007. Nearreal time retrieval o f tropospheric NO 2 from OMI. Atmos. Chem. Phys. 7, 2103—
2118.
Boersma, K.F., Jacob, D.J., Becsela, E.J., Perring, A.E., Dirksen, R., van der A, R.J.,
Yantosca, R.M., Park, R.J., Wenig, M.O., Bertram, T.H., Cohen, R.C., 2008.
Validation of OMI tropospheric NO 2 observations during INTEX-B and application
to constrain NOx emissions over the eastern United States and Mexico.
Atmospheric Environment 42, 4480—4497. doi:10.1016/j.atmosenv.2008.02.004.
Boersma, K.F., Eskes, H.J., Dirksen, R.J., van der A, R.J., Veefkind, J.P., Stammes, P.,
Huijnen, V., Kleipool, Q. L., Sneep, M., Claas, J., Leitao, J., Richter, A., Zhou, Y.,
Brunner, D., 2011. An improved tropospheric NO 2 column retrieval algorithm for the
Ozone Monitoring Instrument. Atmos. Meas. Tech. 4, 1905-1928, doi:10.5194/amt4-1905-2011.
Bovensm ann, H., Burrows, J.P., B uchw itz, M ., Frerick, J., N o el, S., R ozanov, V.V., Chance,
K.V., G oede, A .P.H ., 1999. SCIAM ACH Y: M ission O bjectives and M easurem ent
M odes. J.A tm os.Sci. 56,
127-150.
doi: http://dx.doi.org/10.1175/1520
0469(1999)056<0127 :SMOAM M>2.Q.CO:2.

Bucsela, E.J., Krotkov, N.A., Celarier, E.A., Lamsal, L.N., Swartz, W.H., Bhartia, P. K.,
Boersma, K.F., Veefkind, J.P., Gleason, J.F., Pickering, K.E., 2013. A new
stratospheric and tropospheric NO 2 retrieval algorithm for nadir-viewing satellite
instruments: applications to OMI. Atmos. Meas. Tech. 6 , 2607-2626.
doi: 10.5194/amt-6-2607-2013.
Burrows, J.P., Weber, M., Buchwitz, M., Rozanov, V., Ladstatter-WeiBenmayer, A., Richter,
A., DeBeek, R., Hoogen, R., Bramstedt, K., Eichmann, K., Eisinger, M., Pemer, D.,
1999. The Global Ozone Monitoring Experiment (GOME): Mission concept and first
scientific
results.
J.
Atmos.
Sci.
56,
151-175.
doi: 10.1175/15200469( 1999)056<0151 :TGOMEG>2.0.CO;2.
61

CAPP (Canadian Association of Petroleum Producers), 2013. The Facts on British Columbia
Natural
Gas
and
Crude
Oil.
Available
at:
http://www.capp.ca/getdoc.aspx?DocId=234418&DT=NTV (accessed on: 16 May
2014).
Cox, R.M,. 2003. The use of passive sampling to monitor forest exposure to O 3, NO 2 and
SO2 : a review and some case studies. Environmental Pollution 126, 301-311.
Dirksen, R. J., Boersma, K. F., Eskes, H. J., Ionov, D. V., Bucsela, E. J., Levelt, P. F.,
Kelder, H. M., 2011. Evaluation of stratospheric NO 2 retrieved from the Ozone
Monitoring Instrument: intercomparison, diurnal cycle and trending. J. Geophys. Res.
116, D08305, doi: 10.1029/2010JD014943.
Fioletov, V.E., McLinden, C.A., Krotkov, N., Moran, M.D., Yang, K., 2011. Estimation of
S 0 2 emissions using OMI retrievals. Geophys. Res. Lett. 38, L21811.
doi: 10.1029/2011GL049402.
Fioletov, V.E., McLinden, C.A., Krotkov, N., Yang, k., Loyola, D.G., Valks, P., Theys, N.,
Van Roozendael, M., Nowlan, C.R., Chance, K., Liu, X., Lee, C., Martin, R.V., 2013.
Application o f OMI, SCIAMACHY, and GOME-2 satellite SO 2 retrievals for
detection o f large emission sources. J. Geophys. Res. Atmos. 118, 11, 399-11,418.
doi: 10.1002/jgrd.50826.
Fraser Basin Council, 2013. Identifying Health Concerns Relating to Oil & Gas Development
in Northeastern B.C.-Human Health Risk Assessment Phase 1 Report. Available at:
http://www.health.gov.bc.ca/librarv/publications/vear/2 0 12 /Identifyinghealthconcems
-HHRA-Phasel-Report.pdf (accessed: 15 May 2014).
Ghude, S.D., Kulkami, P.S., Kulkami, S.H., Fadnavis, S., van Der A, R.J., 2011. Temporal
variation of urban NOx concentration in India during the past decade as observed
from space. International Journal o f Remote Sensing 32:3, 849-861. DOI:
10.1080/01431161.2010.517797.
Krotkov, N. A., Cam, S. A., Krueger, A. J., Bhartia, P. K., Yang, K., 2006. Band residual
difference algorithm for retrieval of SO2 from the Aura Ozone Monitoring Instrument
(OMI).
IEEE
Trans.
Geosci.
Remote
Sens.
44,
1259-1266.
doi:10.1109/TGRS.2005.861932.
Krzyzanowski, J., 2009. The Importance of Policy in Emissions Inventory Accuracy-A
Lesson from British Columbia, Canada. J. Air & Waste Manage. Assoc. 59, 430-439.
Krzyzanowski, J., 2011. Approaching cumulative effects through air pollution modelling.
Water Air Soil Pollut. 214: 1-4, 253-273.

62

Krzyzanowski, J., 2012. Environmental pathways o f potential impacts to human health form
oil and gas development in northeast British Columbia, Canada. Environ. Rev. 20,
122-134.
Lamsal, L.N., Martin, R.V., van Donkelaar, A., Steinbacher, M., Celarier, E.A., Bucsela, E.,
Dunlea, E.J., Pinto, J.P., 2008. Ground-level nitrogen dioxide concentrations inferred
from the satellite-bome Ozone Monitoring Instrument. J. Geophys. Res. 113,
D 163 08. doi: 10.1029/2007JD009235.
Lamsal, L.N., Martin, R.V., Parrish, D.D., Krotkov, N.A., 2013. Scaling Relationship for
NO 2 Pollution and Urban Population Size: A Satellite Perspective. Environ. Sci.
Technol. 47, 7855-7861. dx.doi.org/10.1021/es400744g.
Lee, C., Martin, R.V., van Donkelaar, A., Lee, H., Dickerson, R.R., Hains, J.C., Krotkov, N.,
Richter, A., Vinnikov, K., Schwab, J.J., 2011. SO2 emissions and lifetimes: Estimates
from inverse modeling using in situ and global, space-based (SCIAMACHY and
OMI) observations. J. Geophys. Res. 116, D06304. doi:10,1029/2010JD014758.
Lattanzio, R.K., 2013. Air Quality Issues in Natural Gas Systems. Congressional Research
Service (CRS) Reports prepared for Members and Committees o f Congress
(www.crs.gov). Available at: http://www.civil.northwestem.edu/docs/Tight-ShaleGas-2013/Air-Oualitv-Issues-Natural-Gas-Ratner-2013 .pdf) (accessed on: 16 May
2014).
Legge, A.H., Jager, H.-J., Krupa, S., 1998. Sulfur dioxide. In R. B. Flager (Ed.), Recognition
of air pollution injury to vegetation - A pictorial atlas. Pittsburgh, PA: Air and Water
Management Association.
Levelt, P.F., van den Oord, G.H.J., Dobber, M.R., Mslkki, A., Visser, H., Vries, J., Stammes,
P., Lundell, J.O.V., Saari, H., 2006. The Ozone monitoring instrument. IEEE
Transactions on Geoscience and Remote Sensing, doi: 10.1109/TGRS.2006.872333.
Lu, Z., Streets, D.G., de Foy, B., Krotkov, N.A., 2013. Ozone Monitoring Instrument
Observations o f Interannual Increases in SO 2 Emissions from Indian Coal-Fired
Power Plants during 2005-2012. Environ. Sci. Technol.47, 13993-14000.
Martin, R.V., Jacob, D.J., Chance, k., Kurosu, T.P., Palmer, P.I., Evans, M.J., 2003. Global
inventory o f nitrogen oxide emissions constrained by space-based observations
of
N 0 2 columns. J. Geophys. Res. 108(D17), 4537.
doi:10.1029/2003JD003453.
Martin, R.V., 2008. Satellite remote sensing o f surface air quality. Atmospheric Environment
42, 7823-7843.
McKenzie, L.M., Witter, R.Z., Newman, L.S., Adgate, J.L., 2012. Human health risk
assessment o f air emissions from development o f unconventional natural gas
resources. Science of the Total Environment 424, 79-87.
63

McLinden, C.A., Fioletov, V., Boersma, K.F., Krotkov, N., Sioris, C.E., Veefkind, J.P.,
Yang, K., 2012. Air Quality over the Canadian oil sands: A first assessment using
satellite observations. Geophysical Research Letters 39, L04804.
McLinden, C.A., Fioletov, V., Boersma, K.F., Kharol, S.K., Krotkov, N., Lamsal, L., Makar,
P.A., Martin, R.V., Veefkind, J.P., Yang, K., 2014. Improved satellite retrievals of
NO 2 and SO2 over the Canadian oil sands and comparisons with surface
measurements. Atmos. Chem. Phys. 14, 3637-3656, doi: 10.5194/acp-14-3637-2014.
MEM (BC Ministry of Energy and Mines), 2012. British Columbia’s Natural Gas Strategy:
Fueling B.C.’s Economy for the Next Decade and Beyond. Available at:
http://www.gov.bc.ca/ener/Dopt/down/natural gas strategy.pdf (accessed: 15 May
2014).
MoE (BC Ministry of Environment), 2014. Report on Initial Network Design-NE B.C. Air
Quality Network: BC Ministry o f Environment, Report to the SCEK Fund, BC Oil
and
Gas
Commission,
January
31,
2014.
Available
at:
http://www.bcairqualitv.ca/readings/northeast/pdfs/ne air monitor project report ne
twork design.pdf (accessed on: 15 May 2014).
OGC (BC Oil and Gas Commission), 2012. Hydrocarbon and By-Product Reserves in British
Columbia; 2012 - BC Oil and Gas Commission. Available at:
https://www.bcogc.ca/node/l 1111/download (accessed on: 15 May 2014).
US EPA, 2010. Greenhouse Gas Emissions Reporting from the Petroleum and Natural Gas
Industry Background Technical Support Documtent. Washington DC: Climate
Change Division, US Environmental Protection Agency. Available at:
http://www.epa.gov/ghgreporting/docmnents/pdf/2010/Subpart-W TSD.pdf
(accessed on: 01 September 2014).
US EPA, 2011. Oil and Natural Gas Sector: New Source Performance Standards and
National Emission Standards for Hazardous Air Pollutants Reviews; 76 Federal
Register
52738,
August
23,
2011.
Available
at:
http://www.epa.gov/ttn/atw/oilgas/fr23aul 1.pdf (accessed on: 01 September 2014).

64

List of Figures
Fig. 3.1. OMI annual mean tropospheric NO 2 VCDs across a large area (B.C. and Alberta),
averaged over 2005-13 shown on a 6.0 x 6.0 km 2 grid and calculated using different
averaging radius. Plots a) and b) are showing tropospheric VCDs from DOMINO-NO 2 and
SP-NO2 data products, respectively with a 24 km averaging radius. An averaging radius of 60
km is used for the two NO 2 data products in c) DOMINO-NO 2 and d) SCIA-NO 2 . Note
pixels from the both edges of the swath o f OMI data products were not removed for plots (a),
(b), and (c). The locations of Victoria, Vancouver, Prince George, Fort St. John, Kamloops,
Grande Prairie, Calgary, Edmonton and Fort McMurray are indicated by different symbols
from left to right in each plot. The white rectangle o f each plot indicates the present study
area (Northeast B.C.).
Fig. 3.2. Average (2005-2013) tropospheric VCDs: (a) DOMINO-NO 2, (b) EC-DOMINONO 2, (c) SP-NO2, and (d) EC-SP-NO 2 . Marker notation o f each plot from left to right:
Chetwynd, Taylor, and Dawson Creek.
Fig. 3.3. Temporal variation of EC-SP-NO 2 : (a) 2005-2007, and (b) 2008-2013. Marker
notation of each plot from left to right: Chetwynd, Taylor, and Dawson Creek.
Fig. 3.4. Comparison o f NO 2 data products (2005-2013): (a) DOMINO-NO 2 , and (b) SCLANO 2 . Here averaging radius is 60 km in both cases. Marker notation o f each plot from left to
right: Chetwynd, Taylor, and D aw son Creek.

Fig. 3.5. Average (2005-2013) EC-OMI spatial distributions o f NO 2 vmr: (a) EC-vmr, (b)
GEM-MACH-vmr, (c) number o f signals from EC-OMI-vmr in each circle o f 18 km radius,
and (d) the same as (c) but using GEM-MACH-vmr.
Fig. 3.6. Average (2005-2013) tropospheric S 0 2 VCDs: (a) NASA-S0 2-VCDs, (b) SNR o f
NASA-SO 2-VCDS. Marker notation of each plot: Taylor Town Site station (+), Taylor South
Hill Station (x), Taylor (o), and McMahon gas processing plant (A).

65

-1 2 5

-1 2 0

-1 1 5

-1 1 0 - 1 2 5

-1 2 0

-1 1 5

-1 1 0

Longitude
Fig. 3.1. OMI annual mean tropospheric N 0 2 VCDs across a large area (B.C. and Alberta), averaged over 200513 shown on a 6.0 * 6.0 km2 grid and calculated using different averaging radius. Plots a) and b) are showing
tropospheric VCDs from DOMINO-NO 2 and SP-N02 data products, respectively with a 24 km averaging
radius. An averaging radius of 60 km is used for the two N 0 2 data products in c) DOM INO-N02 and d) SCLAN 0 2. Note pixels from the both edges o f the swath of OMI data products were not removed for plots (a), (b),
and (c). The locations of Victoria, Vancouver, Prince George, Fort St. John, Kamloops, Grande Prairie, Calgary,
Edmonton and Fort McMurray are indicated by different symbols from left to right in each plot. The white
rectangle o f each plot indicates the present study area (Northeast B.C.).

66

H
«
O
e
E
_3
’o

U
.a
t:u

>
-1 2 3

-1 2 2

-1 2 1

-1 2 0 -1 2 3

-1 2 2

-121

-120

Longitude

Fig. 3.2. Average (2005-2013) tropospheric VCDs: (a) D 0 M IN 0 -N 0 2, (b) EC-DOMINO-N02, (c) SP-N 02,
and (d) EC-SP-NQ2. Marker notation of each plot from left to right: Chetwynd, Taylor, and Dawson Creek.

67

14
%

s

10

56

#*L
-120

-123

i

Longitude

Fig. 3.3. Temporal variation o f EC-SP-N02: (a) 2005-2007, and (b) 2008-2013. Marker notation of each plot
from left to right: Chetwynd, Taylor, and Dawson Creek.

68

14
X 10

-123

-122

-121

-120 -123

-122

-121

-120

Longitude

Fig. 3.4. Comparison o f N 0 2 data products (2005-2013): (a) DOMINO-NO 2 , and (b) SCIA-N02. Here
averaging radius is 60 km in both cases. Marker notation of each plot from left to right: Chetwynd, Taylor, and
Dawson Creek.

69

-123

-122

-121

-120 -123

-122

-121

-120

Longitude

Fig. 3.5. Average (2005-2013) EC-OMI spatial distributions o f N 0 2 vmr: (a) EC-vmr, (b) GEM-MACH-vmr,
(c) number of signals from EC-OMI-vmr in each circle of 18 km radius, and (d) the same as (c) but using GEMMACH-vmr.

70

SNR

120,6

-126.4
L o n ^ l« 4 t

Fig, 3.6. Average (2005-2013) tropospheric S 0 2 VCDs: (a) NASA-S02-VCD s, (b) SNR o f N ASA-S02-VCD s.
Marker notation o f each plot: Taylor Town Site station (+), Taylor South Hill Station (x), Taylor (o), and
McMahon gas processing plant (A).

71

4. Passive monitoring measurements of Nitrogen Dioxide and Sulfur Dioxide
concentrations over Northeastern B.C., Canada
Abstract
The Peace River district of Northeastern British Columbia (B.C.) Canada is a region of
natural gas production that has undergone rapid expansion since 2005. In order to assess air
quality implications of this, Willems badge passive samplers were deployed for six two-week
exposure periods between August 15 and November 11, 2013 at 24 sites across the region to
assess the ambient concentration o f nitrogen dioxide (NO2) and sulfur dioxide (SO 2). Site 14,
located in Taylor, recorded the highest concentrations o f both species (NO 2 : 9.1 ppb, SO2 :
1.91 ppb) relative to other sites during the whole study period (except the 1 st exposure period
of both species) which is consistent with its location near major sources. However, high
values from passive measurement of N 0 2 at site F20 (deployed near the center o f Chetwynd)
indicate an origin of N 0 2 from an urban setting as well. Observations o f both species from
other sites reflect relatively high concentrations near Fort St. John, Taylor and Dawson Creek
where unconventional oil and gas development activities are quite high. Although a few sites
in Northeastern B.C. recorded elevated concentrations of N 0 2 and S 0 2 during this
investigation, the concentrations remain well below B.C. (Provincial) ambient air quality
objectives. The average precision in terms o f relative standard deviation (RSD) for the
triplicate Willems badge passive samplers were 16.7% for N 0 2 and 18.2% for S 0 2. The
analysis o f results showed good agreement between ambient concentrations from N 0 2
passive samplers and ambient concentrations from co-located chemiluminescence analyzers.
Although, the average accuracies (percentage relative error) of three collocated passive S 0 2

72

sites falls within the recommended values, the accuracy o f each collocated site differ
significantly, thus further investigation is suggested.
Keywords: Unconventional drilling, Oil and gas, Northeastern B.C., Willems badge, air
quality, Nitrogen dioxide, Sulfur dioxide
4.1. Introduction
In 2012, the total production of natural gas in British Columbia (B.C.) was 40,482><106 m 3
(oil was 1 ,2 2 2 x 10 6 m3) with total estimated reserves (proven plus probable recoverable) of
1,138* 109 m 3 (oil was 19,108x 106 m3). This is a 146% increase over the natural gas reserves
estimated in 2006 (OGC, 2012a). The trend of increasing reserve estimates is largely due to
the successful development of unconventional gas extraction including the application of
horizontal drilling and hydraulic fracturing technology in the Montney formation and the
Horn River Basin of Northeastern B.C. (OGC, 2012a). The exploitation o f these vast reserves
of natural gas is a significant economic driver and revenue generator in the province, and as
such the B.C. provincial government is planning on expanding this industry by promoting
development of liquefied natural gas (LNG) for export. It is estimated that 84,951*106 m 3 per
year will be produced in order to meet goals o f developing three LNG facilities by 2020
(MEM, 2012). B.C. is currently the second largest natural gas producer in Canada, and
Canada is the 5th largest in the world (CAPP, 2013).
Natural gas is a nonrenewable fossil fuel that develops naturally over millions o f years from
the carbon and hydrogen molecules of ancient organic matter trapped within geological
formations. The two major geological formations in Northeastern B.C. are the Montney
formation and the Horn River Basin. The Montney formation is an unconventional Lower
Triassic aged formation that includes dry, liquid rich gas and oil in over-pressured siltstones

73

that stretches toward northwest 200 km from the B.C.-Alberta border near Dawson Creek to
the B.C. foothills of the Rocky Mountains. The unconventional mid-Devonian aged Horn
River Basin is a shale play with dry gas over-pressured of the Muskwa, Otter Park and Evie
Formations near Fort Nelson in Northeastern B.C. Currently, these two formations that
utilize unconventional drilling which commenced extensively in 2007/2008, account for 60%
of BC’s total production. Conventional drilling methods were predominantly applied before
2007 (OGC, 2012a). It is expected that natural gas from unconventional sources will
continue to increase while conventional pools will be depleted in the next few years (OGC,
2012a). Unconventional natural gas development consists o f two main phases: well
development and production. Individual well development involves three stages: pad
preparation, well drilling and well completion (US EPA, 2010). Recently, the US EPA
estimated that well completions involving hydraulic fracturing in an uncontrolled manner can
vent natural gas to the atmosphere resulting in air pollution. This is done to guard against the
over-pressuring of the well which results in approximately 230 times more natural gas being
vented compared to wells without hydraulic fracturing (conventional well drilling) (US EPA,
2011). Therefore, people who lived within half a mile of the unconventional wells had a
greater risk o f developing non-cancer health effects from short-term exposure to the high
emissions o f hydrocarbons than those living further away (McKenzie et al., 2012). The well
development phase is not the only source of air pollution; the production phase involving
flaring (either purposefully or accidentally), processing, compressing, pipeline distribution,
storage, etc. also may lead to air pollution. The leading air pollutants from natural gas
development activities are hydrogen sulfide (H2 S), SO2 from sulfur rich (or sour) gas,

74

methane (CH4), non-methane hydrocarbons typically volatile organic compounds (VOCs),
and nitrogen oxides (NOx) (Lattanzio, 2013).
Concerns have arisen recently in Northeastern B.C. about increasing air pollution resulting
from accelerating natural gas production (Fraser Basin Council, 2013; Krzyzanowski, 2012;
MoE, 2014) without simultaneous implementation of available technological advances to
control emissions (Krzyzanowski, 2009). According to the 2010 emissions data reported to
Canada’s

National

Pollutant

Release

Inventory

(NPRI,

http://www.ec.gc.ca/pdb/websol/quervsite/querv e.cfm), SO2 and NO 2 are the dominant
species among all gaseous air pollutants in Northeastern B.C. which are emitted from various
stages o f oil and gas activities but mostly from the production phase such as processing
(flares, engines, and compressors), distribution (leaks of pipelines and flanges), and also from
storage tanks as vaporization (Krzyzanowski, 2012). However, oil and gas industries are not
required to report to the NPRI the emissions during the well development phase (e.g., in pilot
phase

or

in

exploration

or

drilling

phase)

npri/default.asp?lang=En&n=02C767B3-C9FD-4DD7-8072).

(http://ec.gc.ca/inrp-

Therefore,

it

has

been

suggested that both of these gaseous species are under-reported (Krzyzanowski, 2009) and
under-monitored (Krzyzanowski, 2011). Though significant emissions during production
have been reported by the NPRI, there are only four permanent S 0 2 monitors installed over
the previous 15 years (Pine River Hasler and Pine River Gas Plant near Chetwynd, Taylor
Townsite and Taylor South Hill near Taylor) and no monitoring o f NO 2 except for occasional
short term monitoring in some places in 2010 and 2011 using the B.C. MoE mobile air
monitoring laboratory. However, SO2 is also monitored at three additional industry operated
sites, with data that are not publicly accessible (MoE, 2014). Due to this growing concern

75

and high public demand, B.C. MoE recently has commissioned three new stations (Doig
River, Farmington Community Hall and Tomslake) which were deployed in December 2013
and January 2014 to monitor SO2 and total reduced sulfur (TRS) along with meteorology, as
a pilot study (MoE, 2014).
NO 2 and SO2 are gaseous species that undergo chemical and/or physical reactions in the
atmosphere and contribute to acidic deposition in terrestrial ecosystems as dry-deposited
gases or in dissolved form in precipitation, fog, and cloud (Cox, 2003) and also may impact
on human health through a number of environmental pathways (Krzyzanowski, 2012). In
aerosol form, they can also impact visibility while NO 2 as a precursor to the formation of
photochemical oxidants can also lead to direct impact on human health (Cox, 2003). Previous
studies regarding air quality issues in northeastern B.C. and Alberta reported that elevated
levels of SO2 in this territory lead to direct injury to natural vegetation (Legge et al., 1998).
Recently, as a result of the growing concern on potential negative impacts o f gas extraction
and processing, the B.C. provincial government has initiated a three-phase study on the
health impacts of oil and gas activities in B.C.’s northeast region. Respondents during the
Phase-1 preliminary study complained o f personal health problems - such as asthma,
bronchitis, cancer, stress and sleep deprivation associated with oil and gas activities (Fraser
Basin Council, 2013). Besides the impact-oriented issues, some respondents were dissatisfied
by what they saw currently as insufficient information available to them from both the
government and the oil and gas sector, and a lack o f transparency with respect to specific oil
and gas activities. Consequently, it was suggested from the health impact study in Northeast
B.C. that rigorous investigation and regulations are required for baseline assessments o f air
quality and also adequate communication with the public should be conducted prior to oil

76

and gas resource development activities. This baseline information is not available from the
current ambient monitoring network due to the limited number o f monitoring sites, so new
monitoring locations should be assessed. Decisions on where to install new stations should be
based on background information such as which pollutants related to oil and gas
development activities are changing over time in the ambient air.
Therefore, the overall purpose of this study is to assess the spatial pattern o f ambient
concentrations of NO 2 and SO2 over portions o f Northeastern B.C. that are undergoing an
expansion in natural gas production. Ambient concentrations o f these species were measured
using passive diffusive samplers, which have been widely used across America and Europe
for the assessment o f atmospheric NO 2 and SO2 concentrations (e.g. Bytnerowicz et al.,
2010; Campos et al., 2010; Cape et al., 2004; Cox, 2003; Hafkenscheid et al., 2009;
Hagenbjork-Gustafsson et al., 1999; Hsu, 2013; Kirchner et al., 2005; Legge et al., 1996;
Tang et al., 1997; Tang et al., 1999; Tang et al., 2001; Van Reeuwijk et al., 1998;
Vardoulakis et al., 2009; Zbieranowski and Aheme, 2012a) and for filling gaps in monitoring
networks (Zbieranowski and Aheme, 2012a). To the best of our knowledge, there are no
published articles available that describe the ambient air concentrations o f Northeastern B.C.
using passive diffuse samplers. However, a network of industry-run passive monitors exists
in this region which is not publicly available (MoE, 2014). The advantages o f diffusive
passive samplers include that they are: inexpensive, easy to deploy in the field for long term
assessment, easy to operate, do not require electricity, are able to produce accurate results in
indoor and outdoor environments, and are reliable for monitoring ecosystem exposure to
gaseous pollution (Cox, 2003; Hafkenscheid et al., 2009; Kot-Wasik et al., 2007; Namiesnik
et al., 2005; Seethapathy et al., 2008; Zbieranowski and Aheme, 2012a and 2012b). Despite

77

having many advantages, passive samplers also have some problems associated with
environmental factors (temperature, relative humidity, wind, and rain), unsuitability for short
term monitoring, and need to be validated with collocated continuous active monitors (Cox,
2003; Kot-Wasik et al., 2007; Krupa and Legge, 2000; Runeckles and Bowen, 1999;
Seethapathy et al., 2008).
The Methods section of this paper includes information on the study area and the study
design followed by passive sampler preparation and analysis. The method to calculate
accuracy and precision of passive diffusion sampler is also included in the Methods section.
The ambient concentrations o f NO 2 and SO2 are described in the Results and Discussion
section. Comparison of the data obtained from passive samplers with continuous monitoring
is provided at the end o f the Results and Discussion section, with the conclusions
summarized at the end.
4.2. Methods
4.2.1. Study area and design
Northeast B.C. is a region of plains, bordered by the Yukon and Northwest Territories to the
north, the Rocky Mountains to the southwest and the province of Alberta to the east. It is the
largest of B.C.’s regions, representing 21.8% o f the land area of the province (20,494,470
ha), but the least populated, with 1.6 % o f the population (69,068 people). O f those residing
there, approximately 13% consider themselves to be of aboriginal (i.e., First Nations)
descent. The Northeast region has one o f the most active economies in B.C. and this active
economy is mainly driven by oil and gas exploration and production. Due to this, the
population

of the

northeast

is

expected

to

rise

to

almost

(http://www.welcomebc.ca/Live/about-bc/regions/northeast.aspx).

78

80,000

There

are

by

2030

extreme

differences in temperature between the warmest and coldest months o f the year, in some
areas for example in Fort St. John, average daily temperature can range from -21 °C in
January to +14 °C in July. Major communities are located in Fort St. John, Fort Nelson,
Taylor, Dawson Creek, Chetwynd and Hudson’s Hope where oil and gas development
activities are happening. In the present study, passive air quality sensors were deployed
around each of these communities (Fig. 2.3), except Fort Nelson, which is geographically
separated from the others.
Two-week average ambient concentrations of NO 2 and SO2 were measured at 24 sites across
Northeastern B.C., Canada (Fig. 2.3, Table 2.3) during the period August 2013-November
2013. Two sites were selected in existing air quality monitoring stations (Taylor Town Site
and Pine River Hasler, Fig. 2.3; and an additional site in central B.C. collocated at the B.C.
Environment Plaza 400 monitoring location in Prince George, B.C., Canada) as they were
part o f an established B.C. air quality monitoring networks in order to assess the passive
sampler performance. Besides these, all other sites (Table 2.3) were selected to cover the
region o f oil and gas development activities in the Montney formation of Northeastern B.C.,
corresponding to the Peace River region of B.C., based on National Pollution Release
Inventory data base (NPRI, Nitrogen oxide and sulfur oxide emissions for Canada, 2011,
available at: http://www.ee.gc.ca/inrp-npri/default.asp?lang=en&n=::lD892B9F-D. All sites
were chosen to be free of obstacles impeding wind flow, and, with a few exceptions (Table
2.3), most sites were also at least 3.0 kilometers from major industrial sources and from
urban areas in order to provide a better estimate of the overall spatial pattern o f ambient
levels by avoiding the local impact of point source emissions. A few sites (Table 2.3) were

79

also placed at private homes to assess the ambient concentrations directly relevant to human
exposure.
4.2.2. Sampler preparation and analysis
The Willems badge diffusive passive samplers (Fig. 2.4a) were provided by the laboratory of
Professor Julian Aheme, Trent University, Ontario, Canada. Once the passive samplers were
exposed at the field locations they were returned to Trent University by post (in resealable
zip-lock plastic bags packaged in cardboard) for laboratory analysis. Exposed samplers
before laboratory analysis and unexposed samplers before deployment were refrigerated at
4°C. Samplers were exposed in triplicate for each species, for a total o f six samplers per site.
A two-week sampling frequency at a height o f 1.8 m was used (except Plaza 400 and Site 16,
which were located on the roof-top of a 4 and a 1 storey building, respectively), which
corresponds to the height of the active samplers they were collocated with. Each sampler was
mounted using Velcro® under a 127 mm diameter plastic cap that acted as a precipitation
and bird shield (Fig. 2.4b). There were a total of six exposure periods from August 15, 2013
until November 11, 2013. All NO 2 samplers in the 5th exposure period were not considered
for analysis due to a problem with sampler preparation. Also, two thirds of all sites during
first two exposure periods were monitored for both species with duplicate rather than
triplicate passive samplers due to a shortage of samplers. During each exposure period, five
unexposed samplers were retained as laboratory blanks. Besides laboratory blanks, travel
blanks or lot blanks (for each return shipment from Peterborough, Ontario to Prince George,
B.C.) and field blanks (for both species o f each exposure) were also sent to and from sample
sites periodically throughout the study and compared to laboratory blank samplers to ensure
that the sampler cap and resealable bags were effectively protecting the samplers from

80

contamination between preparation, shipment, exposure and analysis. Travel or lot blanks
were not carried to and from the field.
Passive diffusion samplers can be broadly classified into tube, radial, badge and cartridgetype samplers (Krupa and Legge, 2000) and these samplers can have substantially different
uptake rates, which make them more or less suitable to certain applications (Yu et al., 2008).
For instance, tube-type samplers have generally lower uptake rates due to longer axial
diffusion path and smaller cross-sectional diffusion area, which make them suitable for
assessing relatively long-term (e.g., monthly mean) ambient air quality levels. In contrast, the
badge-type and radial samplers which have typically higher uptake rates due to shorter
diffusion paths and larger diffusion areas, are suitable for assessing relatively short-term
(e.g., daily mean) personal/occupational exposure to air pollution (Vardoulakis et al., 2009).
Furthermore, badge-type samplers are advantageous over tube-type since they have an
entrance filter to create a diffusion area free from turbulence to avoid wind effects (Van
Reeuwijk et al., 1998). The basic principle behind all passive sampler measurements is the
principle of diffusion of gases from the atmosphere into a sampler of defined dimensions
onto an absorbing medium, according to Fick’s law. The sampler’s theoretical uptake rate is
a function of the length, L (m), and the cross-sectional area, A (m2), of the stationary air layer
within the sampler, and can be calculated provided that the diffusion coefficient, D (m2s' 1), of
the gas of interest is known. In particular, tube and badge type passive samplers are used
extensively to measure atmospheric NO 2 (Tang et al., 2001). In this particular study, ambient
concentrations of gaseous NO 2 and SO2 were measured with the Willems badge passive
sampler which has been tested to perform well against co-located active sampling methods
(Zbieranowski and Aheme, 2012a; 2012b). The Willems badge passive sampler has a

81

cylindrical body (diameter 28 mm and length is 15 mm) in which a specially treated filter
paper absorbs a specific gas from the air (see also Fig. 2.4). Triethanolamine (TEA) coated
filter paper was placed inside the Willems badge passive sampler to absorb NO 2 from
ambient air while Nylasorb filters without chemical coating were used in the badge type
passive samplers as Nylasorb directly absorbs SO2 from the air. The methods o f Willems
badge passive sampler preparation, analysis and calculation are described in detail in
Zbieranowski and Aheme, (2012a and 2012b) (also see sections: 2.3.3.1 and 2.3.3.2). The
samplers for this study were constructed at Trent University, Ontario, Canada to the exact
specifications of the original Willems badge sampler for NO 2 (Van Reeuwijk et al., 1998)
and the SO2 sampler was a modified design from Bytnerowicz et al. (2005) (Zbieranowski
and Aheme, 2012a, 2012b). Samplers were capped and sealed inside zip-lock plastic bags
and placed in a cardboard box prior to transfer to University o f Northern British Columbia
(UNBC), B.C., Canada by post for deployment in the field. It is noted that Zbieranowski and
Aheme, (2012a and 2012b) did not measure S 0 2, however, design and analysis o f SO2 from
the Willems badge type sampler is similar to that of H N 0 3 sampler except for the generation
of a standard calibration curve during analysis with Ion Chromatography (Professor Julian
Aheme, Trent University, Canada, Personal Communication). The method limit o f detection
(LOD) o f N 0 2 and S 0 2 passive samplers were 0.3 ppb (0.54 pg m '3) and 0.03 ppb (0.07 pg
m'3), respectively and these were estimated from three times the standard deviation of
laboratory blanks of each corresponding species. It is also noted that no significant
differences were obtained in lab blanks over field or travel blanks for both species (p » 0.05
in t-test of all cases). The lab blank mean value was subtracted during calculation o f ambient
concentrations of each species.

82

4.2.3. Accuracy and precision of Willems badge passive samplers
The accuracy of Willems badge passive diffusion samplers was assessed using the percentage
relative error (%) (also known as relative bias):
Accuracy = [(Cp-C a) / C J * 100

(1)

Where Cp is the air pollutant concentration measured with a Willems badge passive sampler
while Ca is the concentration measured with the reference active method (i.e. automatic
chemiluminescence and UV Fluorescence method applied for NO 2 and SO2 measurement,
respectively

in

the

B.C.

continuous

air

quality

http://www .bcairaualitv. ca/assessment/monitoring-instruments.html.

network,
accessed

on:

url:
11

September 2014) and averaged over the same time period at exactly the same location. It
should be noted that the percentage relative error is used here as a simplified indicator of
accuracy and this method is also applied in other studies (Campos et al., 2010; Vardoulakis et
al. 2009). In addition to accuracy estimation, the least-squares regression equation
coefficients o f the passive diffusion measurements (dependent variable) against automatic
active measurements (independent variable) were also used as an indicator o f linearity
between these two methods. Finally, statistical significance (using a paired t-test to compare
the means of two measurements) was carried out to determine whether the differences were
significant between passive and active observations.
The precision of the Willems badge passive diffusion sampling measurements was evaluated
from the triplicate (also from duplicate, see section 4.2.2) sets of NO 2 and SO 2 during each
deployment at all sites. The relative standard deviation (RSD), which is a statistical measure
of repeatability (also called coefficient of variation - CV), was calculated for each pollutant
by dividing the standard deviation of the triplicate samples by the mean concentrations over

83

the same time period and then multiplying by 100. Initially, RSD was calculated for each
exposure period separately and then an average CV o f each site over the whole sampling
period was also provided (Table 4.1). This approach is also used to assess the suitability of
passive diffusion samplers elsewhere (Vardoulakis et al. 2009; Zbieranowski and Aheme,
2012a and 2012b). Furthermore, CV was also obtained from calculated concentrations over
the whole exposure period for each site o f both species in order to evaluate the spatial and
temporal variability across the study domain (Table 4.1).
4.3. Results and discussion
Ambient concentrations of NO 2 and SO2 across the Peace River region o f Northeastern B.C.
are presented in this section with a focus on sites o f higher concentrations. There are 6
periods o f two-week exposures (the 5th exposure period for NO 2 was not analyzed) between
mid-August and mid-November, 2013 at 24 sites across the region o f gas extraction in the
Montney formation o f Northeastern B.C.
4.3.1. Ambient concentration of N 0 2 in Northeastern B.C.
Concentrations of NO 2 across the study area had high spatial variability during the study
period (Fig. 4.1a, Table 4.1). Four locations (Site 14, F10, F20 and F15) (Fig. 4.1) had
comparatively higher concentrations than the other twenty sites. The maximum concentration
of NO 2 among all locations was recorded at Site 14 during the 2nd exposure period at 9.1 ppb.
This site also recorded the highest values among all sites across the whole period (except the
1st exposures when F10 recorded the highest value) with values ranging from nearly 4.0 to
9.1 ppb. The F20 and F10 sites also documented high concentrations o f NO 2 which were
slightly lower than at Site 14; concentrations in these two places varied from 3.5 to 7.4 ppb
and 2.7 to 8.1 ppb, respectively. NO 2 concentration at F I5 was also high (2.2 to 4.9 ppb) but
not as high as Site 14, F10 and F20 (Fig. 4.1a). In contrast, NO 2 concentrations at the other
84

20 sites (Fig. 4.1a) had lower values ranging from below the LOD to roughly 4 ppb with the
following different pattern of variation across the whole period of exposures. Note that the
four high concentrations sites (Site 14, F10, F20 and F15) show some temporal variation with
a decrease during and after the 3rd exposure period (September 14, 2013 to September 28,
2013), while the highest values were captured by passive samplers during the 2 nd exposure
period (August 31, 2013 to September 14, 2013), which might be associated with the changes
of source activities. However, this three months study is not long enough to justify the
temporal variation; therefore, further study at least for one year is recommended. In general,
NO 2 concentrations are higher when daytime sunlight is lower because there is decreased
photo dissociation of ozone and hydroxyl radical formation leading to reduced conversion o f
NO 2 to HNO 3 (Hertel, 2011). Since the sampling period is in a period of decreasing sunlight
from August through November, one might expect increasing NO 2 concentrations during this
time, all else being equal. But, there was a sharp decrease in NO 2 concentrations after the
temperature decreased from roughly 15°C to below freezing (-0.8°C) throughout the study
period (local air temperature data were collected from an adjacent weather station [url:
http://weather.gc.ca/citv/pages/bc-78 metric e.htmll which were then corrected to estimate
site specific air temperature using a lapse rate adjustment). This decrease on observed
concentration may indicate reduced sampling efficiency o f TEA (Triethanolamine) based
passive samplers under colder temperatures and reduced humidity (Vardoulakis et al., 2009).
TEA has been extensively used as a sorbent in passive samplers for assessing ambient NO 2
concentration since the mid 1970s with some studies suggesting that the sampling efficiency
of TEA may be reduced at low humidity and temperature (Hafkenscheid et al., 2009).
Zbieranowski and Aheme (2012a) during their Willems badge passive sampler-based study

85

of NO 2 in Southern Ontario, also noticed a sharp decrease in NO 2 concentrations across
southern Ontario, Canada after temperatures fell below freezing. The application of
correction factors for temperature and relative humidity following the method developed by
Plaisance (2004) did not influence the results of the current study. However, again, this non­
consistency might also be related to the fluctuation o f emissions from sources which do
fluctuate monthly (OGC, 2012b), as well as from changes in atmospheric dispersion.
Three of the sites (F10, F I5 and F20) with relatively elevated concentrations were located at
private homes (see Table 2.3), while Site 14 is also located close to private homes in Taylor.
The levels at these sites as well as at all other sites remain well below the annual national
Maximum Desirable Level (MDL) for NO 2 o f 32 ppb (Table 4.2).
The average NO 2 of all exposures at each site is shown spatially in Fig. 4.2a. This indicates
the highest average values are located near Taylor, Chetwynd, Fort St. John and Dawson
Creek. These areas are close to regions of gas development with unconventional drilling in
the Montney formation and processing activities based on the BC Oil and Gas Commission
(OGC, 2012a) (also see 2.2.1 section for more details) and NPRI database (NPRI, Nitrogen
oxide

and

sulfur

dioxide

emissions

for

Canada,

2011,

available

at:

http://www.ec.gc.ca/pdb/websol/quervsite/querv e.cfm). however, site F20 (in Chetwynd)
with high values of NO 2 is not as close to Oil and Gas development activities. Since this site
is located near the center of Chetwynd, vehicle emissions may be responsible for the elevated
concentrations of ambient NO 2 . Geddes et al. (2009) in their study reported that NO 2
concentrations were highest at Toronto West and Downtown, Toronto reflecting the
importance of transportation as an emission source o f NOx.

86

4.3.2. Ambient concentration of SO 2 in Northeastern B.C.
Ambient levels of SO2 across the study area except the Site 14 did not have as much spatial
variability as N 0 2 during the study period (Fig. 4.1b, 4.2b, Table 4.1). In contrast, Site 14
(Fig. 4.1b), had the greatest variation and on average the highest concentrations among all 24
sites. Similar to N 0 2, the maximum SO2 concentration was recorded at this site (Site 14)
during the 2nd exposure period with a value of 1.91 ppb (its lowest concentration was during
the 6 th exposure at 0.47 ppb, Table 4.1). Sites F13 and F10 had the highest S 0 2 levels during
the 1st exposure period (approximately August 15, 2013 to August 30, 2013; [see Table 2.3]
as all sites were not deployed on the same day). Concentrations at these two sites varied from
0.43 to 1.47 ppb and 0.64 to 1.26 ppb, respectively. Although these three sites (Site 14, F10
and F I3) recorded elevated concentrations o f SO2 during this study, the concentration remain
well below Level A annual BC (Provincial) objectives of 10 ppb (Table 4.2).
SO2 concentrations at the remaining sites (Fig. 4.1b) had low values ranging from below the
LOD to roughly below 1 ppb with no clear pattern of variation across the whole period of
exposures. Averaged values at all sites are shown spatially in Figure 4.2b. Similar to N 0 2,
the locations of high S 0 2 levels are found near Fort St. John, Taylor and Dawson Creek,
which are also in close proximity to locations of unconventional gas development in the
Montney formation and processing according to the BC Oil and Gas Commission (OGC,
2012a) and the NPRI database.
Site 14 (Taylor) had the highest ambient concentrations of both N 0 2 and S 0 2 with less
variation among all exposures compared to most other sites: it had a lower average CV (N 02:
30.0% and S 0 2: 43.8%, Table 4.1), suggesting consistent emissions and dispersion
throughout the study period. According to the NPRI database (NPRI, Nitrogen oxide and
sulfur

oxide

emissions

for

Canada,
87

2011,

available

at:

http://www.ec.gc.ca/pdb/websol/querysite/query e.cfm), two significant sources of NO 2 and
SO2 (Station 1-Taylor, McMahon Gas Plant) in Northeastern B.C. are located in Taylor
which is the most likely cause of the relatively high NO 2 and SO2 levels at Site 14 during the
study period. A recent study (Hsu, 2013), conducted with passive samplers in the Athabasca
Oil Sands Region of Alberta, Canada, found the highest concentration of passively sampled
SO2 and NO 2 (monthly exposure) were at 11.5 km from the largest stationary emission
source with amounts of 2.2 ppb (SO 2) and 5.7 ppb (NO2). These findings are consistent with
those of the present study in terms of levels as well as proximity to sources (Fig. 4.2, Table
2.3). Furthermore, recent air quality studies in Northeastern B.C. also reported that due to
large sources located in Taylor, SO2 concentrations would likely be highest in Taylor
(Krzyzanowski, 2011, MoE, 2014) and to the east o f Taylor (MoE, 2014).
Box plots o f both species at the 24 stations (Fig. 4.3) were produced in order to see the
variation in levels at each station and between exposures. These box plots provide a visual
impression o f the location and shape of the underlying distributions. Box plots show a large
spread in concentrations at Site 14 for both species (Fig. 4.3a and b). NO 2 had the greatest
range in concentrations during the 2nd exposure period followed by the 3rd and 1st (Fig. 4.3c).
Similarly, SO2 concentrations at all sites during the 2nd exposure have the largest variation
(Fig. 4.3d).
In the present study, the average precision in terms o f RSD (CV) for the triplicate Willems
badge passive samplers excluding Plaza 400 were: 16.7% for NO 2 (s.d. = 7.7, range: 7.437.5%), and 18.2% for S 0 2 (s.d. = 7.5, range: 5.4-34.2%) (Table 4.1). Though individual
outlier badges are statistically difficult to identify with 2 or 3 replicates (Vardoulakis et al.,
2009), we eliminated a few very obvious outliers (5 for NO 2, n= 122; 12 for SO2, n=147)

88

from the dataset. The precision of the NO 2 diffusion badges in particular was very consistent
with the values reported by Bush et al. (2001) for Palmes-type duplicate tubes exposed for 4week periods in urban background sites in the UK, however, reported precision of NO 2 in
this study is also different from some previous studies (Campos et al., 2010 [badge type
samplers]; Vardoulakis et al., 2009 [tube type samplers]). Similar to N 0 2, precision o f SO2
measurements of this study (5.4-34.2%) is also comparable to the value (10-25%) reported
by Ayers et al. (1998) where precision was expressed as mean percentage difference between
duplicates of the specially designed passive diffusion samplers (very similar to the Willems
badge samplers) in Australia, As noted before, there is no published study using passive
samplers in Northeastern B.C. to compare with the passive data o f the present study.
4.3.3. Comparison of passive with active (continuous) monitors at collocated sites
In this study, one site was collocated at the B.C. Ministry o f Environment Plaza 400
continuous (active) NO 2 monitoring site in Prince George (Table 2.3 and 4.1), as there is no
continuous monitoring of NO 2 in Northeastern B.C. Three sites (Site 14, Site 16, Plaza 400;
Fig. 2.3, Table 2.3) were collocated with SO2 continuous monitoring. Hourly continuous
monitoring data corresponding to the exposure period were obtained from the B.C. Ministry
of

Environment

(B.C.

MoE)

air

quality

website

(available

at:

http://www.bcairqualitv.ca/readings/index.html. accessed on: 11 September 2014) and
averaged for comparison with the passive data. The accuracy of NO 2 measurements with
passive samplers following eq. (1) varied between 3.0% to 34.6%, with an average o f 14.2%
for the concentration ranges 6.4-11.9 ppb atmospheric NO 2 . The statistical analysis was also
performed using the paired /-test to compare the means of these two different measurements
and has been found no significant difference between the results o f the concentrations

89

obtained by both passive and chemiluminescence methods (p=0.0722 which is higher than
the chosen 5% significance level). The correlation between the concentrations obtained with
passive samplers for NO 2 and the continuous monitors (chemiluminescence method) is
significantly strong (Fig. 4.4a) (R2=0.1S, R=0.89, p=0.0439 which is less than the chosen 5%
significance level).
The accuracy of the SO 2 passive sampler measurements deployed in Site 14, Site 16, Plaza
400 was verified following eq. (1) against the measurements of the collocated continuous
SO 2 analyzers at three sites. The accuracy o f passive samplers for SO2 ranged between 0 to
60.7% in the concentration range of 0.47 to 1.91 ppb at Site 14, -11.1% to 30.8% (except 2nd
exposure period which reported 285.7% higher value than the active measurement) in the
concentration range of 0.15 to 0.54 ppb at Site 16, and -58.0% to 20.0% in the concentration
range o f 1.41 to 2.22 ppb at Plaza 400, representing an average accuracy of 7.6% (or 23.0%
if also considering the 2nd exposure period of Site 16). Fortunately, the average accuracies of
diffusion passive samplers for both species satisfy the limit recommended by the European
Union (±25%) (Campos et al., 2010). In contrast, the statistical analysis (paired /-test to
compare the means) shows that there is no significant difference between the concentrations
obtained by passive diffusion samplers and continuous monitors in Site 16 and Plaza 400
while significant difference is attained for Site 14 (p=0.0319 under 5% significance level).
But, the linear regression results for SO2 show that a significant strong correlation is obtained
between passive and continuous measurements at site 14 (/?=0.94, /?=0.0057) while a poor
correlation exists at the other two sites (Site 16: R=Q.\7 and Plaza 400: R=-0.42) (Fig. 4.4b).
The proper explanation for these inconsistencies is beyond the scope o f this study.
However, continuous and passive measurement techniques have different purposes and uses,

90

therefore it is suggested that the data from these two methods can be suitably compared for
general trends rather than compared on an absolute basis (Hsu, 2013).
4.4. Conclusions
This air quality study using the Willems badge passive diffusion samplers in Northeastern
B.C. indicated higher concentrations o f NO 2 and SO 2 in the regions of high gas development
activities. Among all sites, the highest concentrations of both these species were recorded
(NO2 : 9.1 ppb, SO2 : 1.91 ppb) in Taylor where passive samplers were placed in close
proximity to gas development and processing activities in this territory suggesting that these
activities are an issue for air quality. Besides the gas processing areas, a site near a small
urban centre (Chetwynd) also recorded higher than average NO 2 levels, likely due to
emissions from vehicles. Although some o f the sites in Northeastern B.C. sampled during
this study recorded elevated concentrations of NO 2 and SO2, the concentrations remain well
below B.C. (Provincial) ambient air quality objectives.
The overall precision of this passive diffusion sampler is consistent with the precision found
in other studies elsewhere. The accuracy of NO 2 passive samplers was verified by
comparison with the collocated continuous monitoring station in Prince George, B.C. which
showed no significant differences between the two methods, suggesting that the passive
sampling of NO 2 is sufficiently accurate. Similar to N 0 2, no significant differences in the
means were also seen between passive and continuous SO2 measurements at Plaza 400
(Prince George) and Site 16 (study area) while significant differences seen at Site 14. In
contrast, Site 14 passive SO2 measurements are strongly correlated (i?=0.94) with the
collocated continuous measurements.

However, these inconsistencies in comparisons

suggest further study, using a combination o f different passive samplers (e.g., tube type) for

91

long term exposure, such as at least o f one year, owing to validate the Willems badge passive
samplers particularly for this study area and also to infer the temporal variation of
concentrations.
Acknowledgements
The authors would like to express appreciation to the BC Oil and Gas Commission through a
gift to UNBC for providing funding for this study. We also acknowledge the BC Ministry of
Environment for providing access to air quality data from their continuous monitoring
network.

92

4.5. References
Ayers, G.P., Keywood, M.D., Gillett, R., Manins, P.C., Malfroy, H., Bardsleys, T., 1998.
Validation o f passive diffusion samplers for SO2 and NO 2 . Atmospheric Environment
32:20, 3587-3592.
Bush, T., Smith, S., Stevenson, K., Moorcroft, S., 2001. Validation o f nitrogen dioxide
diffusion tube methodology in the UK. Atmospheric Environment 35, 289-296.
Bytnerowicz, A., Sanz, M.J., Arbaugh, M.J., Padgett, P.E., Jones, D.P., Davila, A., 2005.
Passive sampler for monitoring ambient nitric acid (HNO 3) and nitrous acid (HNO 2)
concentrations. Atmospheric Environment 39, 2655-2660.
Bytnerowicz, A., Fraczek, W., Schilling, S., Alexander, D., 2010. Spatial and temporal
distribution of ambient nitric acid and ammonia in the Athabasca Oil Sands Region,
Alberta. Journal o f Limnology 69, 11-21.
Campos,V.P., Cruz, L.P.S., Godoi, R.H.M., Godoi, A.F.L., Tavares, T.M., 2010.
Development and Validation of Passive Samplers for Atmospheric Monitoring of
SO2 , NO 2 , O 3 and H2 S in Tropical Areas. Microchem. J. 96, 132-138.
Cape, J.N., Tang, Y.S., van Dijk, N., Love, L., Sutton, M.A., Palmer, S.C.F., 2004.
Concentrations o f ammonia and nitrogen dioxide at roadside verges, and their
contribution to nitrogen deposition. Environmental Pollution 132,469-478.
CAPP (Canadian Association of Petroleum Producers), 2013. The Facts on British Columbia
Natural
Gas
and
Crude
Oil.
Available
at:
http://www.capp.ca/getdoc.aspx?Doc!d=234418&DT=NTV (accessed on: 16 May
2014).
Cox, R.M,. 2003. The use of passive sampling to monitor forest exposure to O 3, NO 2 and
SO2 : a review and some case studies. Environmental Pollution 126, 301-311.
Fraser Basin Council, 2013. Identifying Health Concerns Relating to Oil & Gas Development
in Northeastern B.C.-Human Health Risk Assessment Phase 1 Report. Available at:
http://www.health.gov.bc.ca/librarv/publications/vear/2 0 1 2 /Identifvinghealthconcems
-HHRA-Phasel-Report.pdf (accessed on: 15 May 2014).
Geddes, J.A., Murphy, J.G., Wang, D.K., 2009. Long term changes in nitrogen oxides and
volatile organic compounds in Toronto and the challenges facing local ozone control.
Atmospheric Environment 43, 3407-3415.
Hafkenscheid, T., Fromage-Mariette, A., Goelen, E., Hangartner, M., Pfeffer, U., Plaisance,
H., de Santis, F., Saunders, K., Swaans, W., Tang, Y.S., Targa, J., van Hoek, C.,
Gerboles, M., 2009. Review of the Application o f Diffusive Samplers in the European
Union for the Monitoring o f Nitrogen Dioxide in Ambient Air. JRC Scientific and
93

Technical Reports. European Commission, Joint Research Centre, Institute for
Environment and Sustainability, p. 79.
Hagenbjork-Gustafsson, A., Lindahl, R., Levin, J., Karlsson, D., 1999. Validation of a
diffusive sampler for NO 2 . Journal of Environmental Monitoring 1, 349-352.
Hertel, O., 2011. Nitrogen processes in the atmosphere. In: Sutton, M., et al. (Eds.), The
European Nitrogen Assessment. Cambridge University Press With Sections_
Authors/European Union, pp. 177-207 (Chapter 9).
Hsu, Yu-M., 2013. Trends in Passively-Measured Ozone, Nitrogen Dioxide and Sulfur
Dioxide Concentrations in the Athabasca Oil Sands Region of Alberta, Canada.
Aerosol and Air Quality Research 13, 1448-1463.
Kirchner, M., Jakobi, G., Feicht, E., Bernhardt, M., Fischer, A., 2005. Elevated NH 3 and NO 2
air concentrations and nitrogen deposition rates in the vicinity o f a highway in
Southern Bavaria. Atmospheric Environment 39, 4531-4542.
Kot-Wasik, A., Zabiegala, B., Urbanowicz, M., Dominiak, E., Wasik, A., Namiesnik, J.,
2007. Advances in Passive Sampling in Environmental Studies. Anal. Chim. Acta
602, 141-163.
Krupa, S.V., Legge, A.H., 2000. Passive sampling o f ambient, gaseous air pollutants: an
assessment from an ecological perspective. Environmental Pollution 107, 31-45.
Krzyzanowski, J., 2009. The Importance o f Policy in Emissions Inventory Accuracy-A
Lesson from British Columbia, Canada. J. Air & Waste Manage. Assoc. 59,430-439.
Krzyzanowski, J., 2011. Approaching cumulative effects through air pollution modelling.
Water Air Soil Pollut. 214: 1-4, 253-273.
Krzyzanowski, J., 2012. Environmental pathways of potential impacts to human health form
oil and gas development in northeast British Columbia, Canada. Environ. Rev. 20,
122-134.
Lattanzio, R.K., 2013. Air Quality Issues in Natural Gas Systems. Congressional Research
Service (CRS) Reports prepared for Members and Committees of Congress
(www.crs.gov). Available at: http://www.civil.northwestem.edu/docs/Tight-ShaleGas-2013/Air-Oualitv-Issues-Natural-Gas-Ratner-2Q 13 .pdf) (accessed on: 16 May
2014).
Legge, A.H., Nosal, M., Krupa, S.V., 1996. Modeling the numerical relationships between
chronic ambient sulphur dioxide exposures and tree growth. Can. J. For. Res. 26, 689695.

94

Legge, A.H., Jager, H.-J., Krupa, S., 1998. Sulfur dioxide. In R. B. Flager (Ed.), Recognition
of air pollution injury to vegetation - A pictorial atlas. Pittsburgh, PA: Air and Water
Management Association.
McKenzie, L.M., Witter, R.Z., Newman, L.S., Adgate, J.L., 2012. Human health risk
assessment o f air emissions from development of unconventional natural gas
resources. Science of the Total Environment 424, 79-87.
MEM (BC Ministry o f Energy and Mines), 2012. British Columbia’s Natural Gas Strategy:
Fueling B.C.’s Economy for the Next Decade and Beyond. Available at:
http://www.gov.bc.ca/ener/popt/down/natural gas strategy.pdf (accessed on: 15 May
2014).
MoE (BC Ministry o f Environment), 2014. Report on Initial Network Design-NE B.C. Air
Quality Network: BC Ministry of Environment, Report to the SCEK Fund, BC Oil
and
Gas
Commission,
January
31,
2014.
Available
at:
http://www.bcairqualitv.ca/readings/northeast/pdfs/ne air monitor project report ne
twork design.pdf (accessed on: 15 May 2014).
Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M., Wasik, A., 2005. Passive Sampling
and/or Extraction Techniques in Environmental Analysis: A Review. Anal.
Bioanal.Chem. 381, 279-301.
OGC (BC Oil and Gas Commission), 2012a. Hydrocarbon and By-Product Reserves in
British Columbia; 2012 - BC Oil and Gas Commission. Available at:
https://www.bcogc.ca/node/l 1111/download (accessed on: 15 May 2014).
OGC (BC Oil and Gas Commission), 2012b. Montney Formation Play Atlas NEBC, October
2012,
BC
Oil
and
Gas
Commission.
Available
at:
http://www.bcogc.ca/node/8131/download (accessed on: 26 November 2013).
Plaisance, H., 2004. Response of a Palmes tube at various fluctuations of concentration in
ambient air. Atmospheric Environment 38, 6115-6120.
Runeckles, V.C., Bowen, P.A., 1999. The use of calibrated passive monitors to assess crop
loss due to ozone in rural locations. In: Agrawal, S.B., Agrawal, M. (Eds.),
Environmental Pollution and Plant Responses. Lewis Publishers, Boca Raton, FL
( 2000 ).
Seethapathy, S., Gorecki, T., Li, X., 2008. Passive Sampling in Environmental Analysis. J.
Chromatogr. A. 1184, 234-253.
Tang, H.M., Brassard, B., Brassard, R., Peake, E., 1997. A New Passive Sampling System
for Monitoring SO 2 in the Atmosphere. Field Anal. Chem. Technol. 1, 307-314.

95

Tang, H., Lau, T., Brassard, B., Cool, W., 1999. A New All-Season Passive Sampling
System for Monitoring N 02 in Air. Field Anal. Chem. Technol. 3, 338-345.
Tang, Y.S., Cape, J.N., Sutton, M.A., 2001. Development and Types o f Passive Samplers for
Monitoring Atmospheric NO 2 and NH 3 Concentrations. The Scientific World 1,513529.
US EPA, 2010. Greenhouse Gas Emissions Reporting from the Petroleum and Natural Gas
Industry Background Technical Support Document. Washington DC: Climate Change
Division,
US
Environmental
Protection
Agency.
Available
at:
http://www.epa.gov/ghgrenorting/documents/pdf/2010/Subnart-W TSD.pdf
(accessed on: 01 September 2014).
US EPA, 2011. Oil and Natural Gas Sector: New Source Performance Standards and
National Emission Standards for Hazardous Air Pollutants Reviews; 76 Federal
Register
52738,
August
23,
2011.
Available
at:
http://www.epa.gov/ttn/atw/oilgas/fr23aul 1.pd f (accessed on: 01 September 2014).
Van Reeuwijk, H., Fischer, P.H., Harssema, H., Briggs, D.J., Smallbone, K., Lebert, E.,
1998. Field comparison o f two NO 2 passive samplers to assess spatial variation.
Environmental Monitoring and Assessment 50, 37-51.
Vardoulakis, S., Lumbrera, J., Solazzo, E., 2009. Comparative evaluation o f nitrogen oxides
and ozone passive diffusion tubes for exposure studies. Atmospheric Environment 43,
2509-2517.
Yu, C.H., Morandi, M.T., Weisel, C.P., 2008. Passive dosimeters for nitrogen dioxide in
personal/indoor air sampling: a review. Journal o f Exposure Science. Environmental
Epidemiology 18,441-451.
Zbieranowski, A.L., Aheme, J., 2012a. Ambient concentrations o f atmospheric ammonia,
nitrogen dioxide and nitric acid across a rural-urban-agricultural transect in southern
Ontario, Canada. Atmospheric Environment 62, 481-491.
Zbieranowski, A.L., Aheme, J., 2012b. Spatial and temporal concentration of ambient
atmospheric ammonia in southern Ontario, Canada. Atmospheric Environment 62,
441-450.

96

List of Tables
Table 4.1. Passive measurement o f NO2 and SO2 concentrations (ppb) in Northeastern B.C.,
Canada. The CV is also provided in the parentheses.
Table 4.2 . B.C. ambient air quality objectives.

List of Figures
Fig. 4.1. Site and exposure specific concentrations: (a) N 0 2; and (b) S0 2. NO2 passive
samplers were not analyzed for exposure 5. The horizontal dash line (red) in both figures
represents the LOD.
Fig. 4.2. Spatial distribution of concentrations: (a) NO2, and (b) SO2. The units o f both NO2
and SO2 in the color bar are ppb.
Fig. 4.3. Site and exposure specific box plots of both species. Values of all periods of
exposures were considered for site specific box plot (a. NO2, b. SO2) while all sites values
during each exposure period provide exposure specific box plot (c. NO2, d. SO2). The line
across the box represents the median, whereas the bottom and top o f the box show the
locations o f the first and third quartiles (Qi and Q3). The whiskers are the lines that extend
from the bottom and top of the box to the lowest and highest observations inside the region
defined by Qi-1.5(Q3 - Qi) and Q3 + 1.5(Q3 - Qi). The hinges in ‘c’ and ‘d’ show the 95%
confidence interval of the median. Individual points with values outside these limits (outliers)
are plotted with ‘+’ signs.
Fig. 4.4. Comparison o f passive with active (continuous) monitor at collocated sites, a)
passive NO2 vs active NO2 (continuous data from Plaza 400 station in Prince George), b)
passive S0 2 vs active S0 2 (three active stations over a total six period of exposures have
been considered, see Table 2 .3). The equation o f the linear regression line (linear fit) o f each
plot is also included. 1:1 line is also provided in both figures for visual aid.

97

Table 4.1. Passive measurement o f N 0 2 and S 0 2 concentrations (ppb) in Northeastern B.C., Canada. The CV is also provided in the parentheses.
Site IDs

Exposures/ NO;

!

Fxposures/SQ2

r'(cv)

2nd (CV)

3'd (CV)

4lh (CV)

6,h (CV)

Av (CV)

CV

|

r'(cv)

2nd (CV)

3,d (CV)

4* (CV)

5th (CV)

6* (CV)

Av (CV)

CV

FI
F2

3.8(17 7)
0 8 (26 5)

3.9(13)
Of

2 1 (15.7)
0 7 (20.4)

2 0 (1 0 1)
0.5(31.9)

1 8(15 6)
07(13.8)

2.8(14.4)
0.6(21 3)

37.3
493

J N/A
J 023(1.1)

0 1 (10 3)
0 28 (7.4)

0 07 (46 5)
0 16(126)

0.06(51)
0.06 (9.3)

0 18(2 4)
0 29 (24 1)

0.16(19)
0 14(17.7)

0 13(25.8)
0 19(14 2)

45.4
46.2

F3
F4
F5
F6
F7
F8
F9
F10
FI 1
F12
F13
F14
F15
F16
F17
F18
F20
F21
Indus 1
Indus2
Site 14
Site 16
Av (CV)

2.2 (23.4)
3 1 (25 1)
3 0(11 6)
2.4 (6 2 )
2 4(51 4)
4 0(12 8)
2.0(23.1)
8 1 (12 7)
2 7(36 2)
2 6(9.1)
2.0(47 3)
N/A
3.5 (30 7)
1 6 (1 1 9 )
1.3(18.5)
3.5(1)
5.7(14.7)
1.5(0 5)
3 3(21 2)
3.4(16 5)
5.6(13 6)
2.2(17.6)
3.1 (195)

0.7(97.5)
11 (14.2)
3 6(17 2)
4.2(16.6)
N/A
2 5 (49 4)
1.2 (48.2)
5 8(0 1)
3 1 (5 7)
4 0 (20 9)
2.8(9 7)
3.8(55)
4 9(14 2)
0 8(19 2)
16 (4.4)
3 1 (12.2)
7.4 (30.7)
1 2(10 6)
2 4(17 9)
3.0 (20)
9 1 (15 9)
3.0 (9 6)
3.2 (20 3)

1 3(13.4)
1.4 (7)
2 8(10.7)
2.8(5)
0 3 (6.1)
2 8 (18 2)
0.7 (86 3)
5.5(10.3)
1.9(10)
2.5(14.3)
1.0(17.8)
0 7(12.3)
3.4 (20.4)
0.9(10.8)
1 8(14 7)
2.0(17.9)
6.6 (28.8)
O f (8 8)
2.1 (27 9)
1.8 (12)
6 4 (9.1)
1.9(46)
2 3 (16.8)

0,4 (5 1)
1.3 (20.7)
1 4(1 6 5)
N/A
0.8 (21.7)
16(13 4)
0,5(19 8)
2.7(8.9)
1.1(4 5)
1 5(18 8)
0.7 (23.3)
0.9(16 7)
2.7(19 2)
0 8(33.9)
1.2(18 7)
1.2 (5.4)
3.5(21.8)
0.4(13 4)
1 7 (1 9 3 )
1 5(19 1)
3.9 (7 2 )
13(7.7)
1 5(16.4)

0 9 (33.2)
1.0(25.2)
1.9(2 4)
2.2 (4.7)
0.4 (0.6)
2 4 (4.7)
0 6 (1 0 1)
3.0(5)
15 (6 2)
2 4 (20 3)
1.2(18.1)
0.8(1 9)
2 2 (23 1)
0.9 (6 7)
1.7 (7.6)
2.1 (144)
3.7 (4.1)
0.6(10 6)
2.0(12)
1.7(17 4)
5.9 (0.6)
1.6(15.3)
1.8(11 4)

1 1 (34 5)
1 6(18.4)
2 6(11.6)
2 9(8 1)
1.0(19 9)
2.7(19 7)
1.0 (37 5)
5.0 (7.4)
2 1 (12 5)
2 6(16 6)
1.6(23.2)
1.6(9 1)
3.4 (21.5)
1.0(16.5)
1 6(12 7)
2.5(10.1)
5.4 (20.0)
0 8(8.7)
2 3(19.6)
2 3(17)
6.2 (9.2)
2 0(10.9)
2 4(16 7)

62 3
53.4
33.7
29 7
95.0
32 1
59.3
44.4
38.2
33,1
53 9
960
304
32.5
18.3
37.6
31.8
69,6
25.6
37.0
30.0
32.5
44.3

j 0 32(11 4)
j 0 14(34.3)

| 0 26(7,8)
J 0 51 (28.8)
1.26(18 9)
J 0 14(132)
1 0 81(15 9)
1 47(37 7)
J N/A
J 0 43(26 7)
j 0.25(2.7)
0.07(0.4)
j 0 45(3 5)
j 055(21 1)
J 0.42(4.6)
j 0.38(2.7)
j 0.44(19 6)
j 1.08(18 4)
j 0,37(25.4)
j 0.47(14.5)

0.6(7 5)
0.25 (22)
0.25 (33.2)
0.43 (4.5)
0.42(1.1)
0.4 (53.7)
0 54 (42.8)
0 89(5 8)
0 18(2 2)
1 01 (25 6)
0.71 (70 3)
0.15(124)
0.82(1.8)
0.85(6.5)
0.07 (22.7)
0 83 (12)
0.41 (14 3)
0 54(18 6)
0 58(27.6)
0 64(13 8)
1 91 (10 8)
0.54(35 5)
0 5 6(193)

0.26(15 2)
0 0 9 (1 8 1)
0.04 (36)
0 09 (20.4)
0 18(4 4)
0.18(17 1)
0.17(14.6)
0 64(10 6)
0.22 (28.2)
0.44(16 3)
0.56(5.4)
0.1 (15 6)
0.59(11.2)
0 5(1.2)
0 09 (36 7)
0 42(26.1)
0.28 (15.3)
0.28 (20)
0.33 (22 9)
0.24(8 1)
1 66(19 3)
0 24 (9 6)
0 33(18.0)

0 15(13.1)
0.11 (63 2)
0.07 (42.6)
N/A
0 09(1 5)
0.21 (10.6)
0.35(18.7)
0 66(16 5)
0.12(64.5)
0.61 (20.8)
0.43 (9 6)
0.14(24 7)
0.51 (18 7)
0 23(20 2)
0.1 (9.2)
0 37 (35 6)
0 41 (44 6)
0 06(22 5)
0 29(10.3)
0 25(10 5)
0.98(10 6)
0.15(13 1)
0.28(23.5)

0.26(16.5)
0.19(34 8)
0.2 (22)
0.35(10.1)
0.42 (2.1)
0.16(34 5)
0.74 (9.7)
0.72(8 4)
0.08(8 8)
0.56 (23 7)
0.78(13)
0.04(59.3)
0.51 (3.2)
0.44(9.4)
0.09(7.5)
0.39(12.2)
0 .2 (7 8 )
0.1 (22)
0.42 (7.5)
0 58 (8.6)
1 01 (6.3)
0 16(28.4)
0.37(15.9)

0 36 (21.4)
0.15(23.5)
0 18(37.4)
0.66 (6.4)
026(18.1)
0 43 (6)
0 5 5 (1 )
0 83(13.9)
0 13(15.7)
0 53(12.2)
0 47 (20.3)
0.15(13.7)
0.49 (6)
047(15.8)
0.17(30 7)
0.54 (4.1)
0.35(13.3)
0 06 (32.4)
0.58(4.8)
0 51 (6.1)
0.47(14 4)
0 17(12.9)
0.37(15.3)

0.33(14.7)
0 16(32.3)
0.14(34.2)
0.37(10.3)
0 29 (5.4)
0.27 (24.3)
0 48(17.3)
0 83(11.0)
0 15(23.8)
0.66(19.7)
0,74(23.7)
0,12(25.1)
0.56(8 1)
0.46(10.6)
0.10(21 3)
0.50(18)
0.37(19.0)
0 24 (23 1)
0 43(14 6)
0 44 (9 4)
1.19(12 2)
0 27(19 9)
0.40(18.2)

46.9
37.3
50.4
556
47.0
42 1
40 8
277
33.7
32.0
52 1
40.7
24.7
49 1
37 7
34.5
32.9
83 9
28 9
37.8
43 8
57 1
428

Plaza 400

8 7 (1 7 2 )

^

9.9(158)

6.4(8 7)

11 9(2 6 9 )

9 5 (19.8)

216

[ 2 22(0.6)

2.56(76 1)

1.89(4 5)

1 41 (126)

1 5 7 (1 9 7 )

163(6.6)

1.88(239)

23.3

((114
/1.1)}

0.13(59)

J 0.32(17)
j 0 36(3)

Note: concentrations are below the LOD. N/A indicates missing passive samplers during field exposures. Av (CV) stands for average passive concentration and
average CV in the parenthesis. Columns Av (CV) under each section (N 0 2 and S 0 2) were calculated from all exposures o f each site, however, rows o f Av (CV)
were from all sites o f each exposure. In addition, column CV under each section was calculated from the observed concentrations o f each site across the whole
study period. All CV values are reported here in percentages (detail in section 4.2.3).

98

Table 4.2. B.C. ambient air quality objectives.

Pollutant
S02

Averaging Time

Air Quality Objective
(PPb)

Annual

10
20
30
60
100
140
170
340
340
140
250
32
53
106
160
213
532

24 hour

1 hour

3 hour
no2

Annual
24 hour
1 hour

Level
A or lower
B
C
A or lower
B or upper
C
A or lower
B or upper
C
Lower
Upper
MDL
MAL
MAL
MTL
MAL
MTL

Abbreviations: MDL=National Maximum Desirable Level; MAL=National Maximum Acceptable Level;
MTL=National Maximum Tolerable Level; A, B and C=Provincial Level A, B and C Pollution Control
Objectives (B.C.). “Lower” represents discharges as applying to sensitive environmental situations; and
“Upper” represent discharges as applying to where it can be shown that unacceptably deleterious changes will
not follow.
Source:
Provincial
Air Quality Objective Information
Sheet (available
at:
http://w w w.bcairaualitv.ca/reports/pdfs/aaotable.pdf. accessed on: 10 O ctober 2013).

99

0

Expl

Exp2
Exp3
Exposure

Exp4

Exp6

5
4.5
4

3

2.5
2

SO (Jig m

is

1.5
1

0.5
0

Expl

Exp2

Exp3
Exposure

Exp4

Exp5

Exp6^

Fig. 4.1. Site and exposure specific concentrations: (a) N 0 2; and (b) S 0 2. N 0 2 passive samplers were not
analyzed for exposure 5. The horizontal dash line (red) in both figures represents the LOD.

100

NCg (ppb)

5 0 2 (ppb)

Fig. 4.2. Spatial distribution of concentrations: (a) NO 2 , and (b) SO 2 . The units o f both NO 2 and SO 2 in the
color bar are ppb.
101

Fig. 4.3. Site and exposure specific box plots o f both species. Values o f all periods o f exposures were
considered for site specific box plot (a. N 0 2, b. S 0 2) while all sites values during each exposure period provide
exposure specific box plot (c. N 0 2, d. S 0 2). The line across the box represents the median, whereas the bottom
and top of the box show the locations of the first and third quartiles (Qi and Q3). The whiskers are the lines that
extend from the bottom and top o f the box to the lowest and highest observations inside the region defined by
Q i- 1.5(Q3 - Qi) and Q3 + 1.5(Q3 - Qi). The hinges in ‘c ’ and ‘d’ show the 95% confidence interval o f the
median. Individual points with values outside these limits (outliers) are plotted with *+’ signs.

102

+ M ia 4 0 0
y - 0.85i + 2.4; R-0.OT

NO; (ppbJ-ContinuoiM

3.5

x P liza 400; y - -0.27x + 2.6; R - -0.42
+ Site 14;
o S ilt 16;

y = 1.5x-0.19; R = 0.94
y - 0.35x + 0.2; R - 0.17

o.
1.5

0.5

221
1.5
SO (ppb)-Continuous
SO.

3.5

I

Fig. 4.4. Comparison o f passive with active (continuous) monitor at collocated sites, a) passive N 0 2 vs active
N 0 2 (continuous data from Plaza 400 station in Prince George), b) passive S 0 2 vs active S 0 2 (three active
stations over a total six period of exposures have been considered, see Table 2.3). The equation of the linear
regression line (linear fit) o f each plot is also included. 1:1 line is also provided in both figures for visual aid.

103

5. Conclusions and Recommendations
5.1. Introduction
Satellite observations o f NO 2 and SO2 between 2005 and 2013 have been considered to
investigate the air quality o f Northeastern B.C. which has undergone rapid natural gas
development since 2005. The satellite observations used in this study are vertically integrated
column densities that extend up to few kilometers from the surface to the troposphere. In this
study, space-based and model-simulated NO 2 ambient concentrations have also been
investigated. In addition, Willems badge passive samplers were also deployed for six twoweek exposure periods between August 15 and November 11, 2013 at 24 sites across this
region to assess the ambient concentration o f NO 2 and SO2 . The study area is bounded from
55°N to 57°N and the B.C.-Alberta border to 122°W encompassing the towns of Fort St.
John, Taylor, Dawson Creek and Chetwynd. The following section highlights the overall
findings from this study.
5.2. Summary of results and conclusions
The findings o f this study answer the research questions that are listed in chapter one.
Question #1 “what are the levels o f these pollutants in this region ?" is answered through the
satellite investigation as well as passive monitoring. All the data products o f NO 2 VCDs
indicate that the long-term average (2005-2013) o f elevated pollutant levels are located in the
vicinity of Taylor, Fort St. John, and Dawson Creek with maximum values of 8.0 * 10 14
molecules/cm 2 (EC-SP-NO 2) and the maximum value varies substantially with data products
largely due to AMF, tropospheric VCDs retrieval algorithm, and instruments observation
time (satellite flight time). Most notably, are the NO 2 VCDs from SP which are 2.6 times
higher than those of DOMINO and with the new EC-AMFs the difference remains nearly the

104

same, but values of both data products increased with the EC-AMFs. Besides the multiple
analysis of NO 2 in a relatively large area, SO2 was analyzed only within few kilometers
around the source (McMahon gas processing plant) also located in Taylor using NASA-SP
products and is reported to be 0.18 DU SO2 VCDs (May-August o f 2005-2013). This
relatively larger value implies Taylor is half as polluted as Canada’s one o f the largest nonurban SO2 emission source areas (i.e., Canadian oil sands area). Willems badge passive
monitoring o f NO 2 and SO2 also found larger values in the same area that were identified by
the satellite observations. The three months average passive monitoring NO 2 and SO2
maximum concentrations are reported in Taylor with 6.2 ppb and 1.19 ppb, respectively,
while model-simulated and space-based nine year average NO 2 surface concentrations are 0.6
ppb and 0.4 ppb, respectively. However, two in-situ monitoring stations, installed these
different altitudes in Taylor, reported nine years average SO 2 concentrations o f 2 ppb (in the
station located at the valley bottom) and 0.9 ppb (at a nearby location above the valley)
which indicates possible trapping o f pollutants due to less vertical mixing of SO 2 in Taylor.
Passive and in-situ monitoring of SO2 provides reasonable agreement, however, space-based
and model-simulated NO 2 concentrations largely varies from passive monitoring might be
due to substantial difference of area coverage by these two methods, such as: satellite-based
concentration refers to an area o f roughly 312 km2 (typical one small OMI pixel) while
passive sensor takes the point observation. In addition, long-term average, short life time of
NO 2 and also quick photolysis of NO 2 during the mid-day (inter-comparison o f OMI and
SCIA data products justify this quick photodissociation of N 0 2), particularly in summer may
also explain the reason o f differences between the concentrations obtained from satellite and
passive observations.

105

Question #2 “are the trends in satellite air quality observations between 2005 and 2013
related to the increased trends in oil and gas development activities in this region?” is
addressed by the two time intervals (2005-2007 and 2008-2013) EC-SP-NO 2 data products
analysis. Note that limited valid SO2 data products availability all the year round does not
allow this study to investigate SO2 trend analysis. Comparing the 2005-2007 with the 20082013 mean NO 2 VCDs, an increase in NO 2 with time (1.7%/y) is identified close to the
Dawson Creek and this increment likely reflects the unconventional drilling commencement
since after 2007 in the Montney formation. 60% of 2012 total production comes from
unconventional sources in B.C. and the Montney formation contributes 40% of the total 2012
unconventional production, however, the southern Montney formation (or Heritage Montney
field) which is located in the study area provides 60% o f the total Montney production.
Question #3 “how do these pollutants vary spatially and temporally?” is answered by both
satellite and passive observations. All satellite NO 2 data products, simulated model output of
NO 2 and passive monitoring of NO 2 and SO2 reflect almost the same pattern of spatial
distributions and these distributions show that Taylor, Fort St. John, Dawson Creek and
Chetwynd (not in order) are the most polluted areas of the study. Satellite SO2 data products
were not considered for large scale spatial analysis, however, these data products also found
consistently high values in Taylor. Satellite NO 2 data products were also analyzed temporally
and have revealed higher values near Dawson Creek since after 2007 due to the
commencement of extensive unconventional natural gas development.
Question #4 “are satellite observations o f NO 2 and SO 2 from different data products
consistent?” is addressed by analyzing several satellite data products and comparing the
spatial distribution of both satellite and passive monitoring observations. Passive monitoring

106

observations were also validated against collocated active (continuous) measurements. As
mentioned before, satellite data products are represented by pixels while passive sensor
reflects the in-situ observation, therefore, direct validation with passive monitoring is not
feasible. However, the spatial distributions of both pollutants using these two measurement
techniques have been found consistent distribution (Fig. 5.1). In addition, consistent spatial
distributions from all kinds of satellite data reflect the consistency of satellite data products.
Furthermore, consistent outcome in the previous investigation that is replicated in this study
from the comparative analysis of satellite data products for the larger area (most part o f B.C.
and Alberta) using similar method of previous study also provides a strong indication of
acceptability of present findings.

•ia

-ins

-1M -121.5 -HI

-12M -l»

. 1 2 0 .*

Lm | M

.120.6

-I20.4

-120.1

-120

Umghadt

Fig. 5.1. Spatial distribution o f satellite and passive observations of N 0 2 and S 0 2 in Northeastern B.C. a)
Average passive N 0 2 observations (Fig. 4.2a) and long-term average EC-SP-N02 VCDs (Fig. 3.2d); b) average
passive S 0 2 observations (Fig. 4.2b) and long-term average S 0 2 VCDs (Fig. 3.6a). Note that all passive S 0 2
locations are not included here since satellite S 0 2 analysis was taken in a relatively small area. Also note that
relative size of circle in each plot refers the variation of concentrations (not using the same scale o f both plots).

107

5.3. Recommendations for future work
A number of developments would improve the usefulness o f satellite observations of
tropospheric trace gases, including a suitable technique being developed for year round valid
NO 2 and SO2 satellite data availability at high latitudes. Future work should also integrate the
analysis of satellite observations with in-situ aircraft measurements to directly validate the
satellite observations of NO 2 and SO2 with a particular focus on the Northeastern B.C. and to
provide information about other tropospheric constituents over the natural gas development
area. The spatial and temporal coverage of the aircraft measurements are also important.
Global models and updated emission inventory information will be essential in this
integration. Air quality dispersion models in conjunction with the satellite data products
should be considered for air quality study especially in this complex terrain area. Long-term
passive monitoring of NO 2 and SO2 with multiple passive sensors should be carried out to
validate the passive sensors in respect o f meteorology, geography and also concentration of
particular pollutants.
Finally, it is suggested that a denser network o f air quality stations in Northeastern B.C. with
also including NO 2 monitoring facilities be established due to the recent extensive natural gas
development activities. Quality assurances should be completed once the real time raw data
has been taken. However, in general data older than three months from the current date
available in the B.C. air quality data archive website have gone through a quality assurance
review but the data are still subject to change. Oil and gas companies are also requested to
share real time calibrated monitoring information publicly.

108

Appendix
Tropospheric slant column densities (SCD) in Northeastern B.C.
The SCD was calculated by multiplying the tropospheric VCD o f four different OMI NO 2
data sets (DOMINO-NO 2 , EC -D 0M IN 0-N 02, SP-N02, and EC-SP-N02) over Northeastern
B.C. with the corresponding tropospheric AMF.

-123

-122

-121

-120 -123
Longitude

-122

-121

-120

Fig. A l. Average (2005-2013) tropospheric SCD: (a) DOMINO-N02, (b) EC-DOMINO-N02, (c) SP-N 02, and
(d) EC-SP-N02. Marker notation o f each plot from left to right: Chetwynd, Taylor, and Dawson Creek.

109