 GUIDANCE FOR NO2 DISPERSION MODELLING
                       IN BRITISH COLUMBIA




          British Columbia Ministry of Environment & Climate Change Strategy
                           Environmental Protection Division
                            Environmental Standards Branch
                                        Clean Air




                             Victoria, British Columbia, Canada




Version: July 2022
                                                      Obtain a Copy

Copies of this document are available from the Ministry website.




                                                   Contact Information

Comments or questions on the Guidance can be sent to: bcdispersion.model@gov.bc.ca

Users are encouraged to check the Ministry website for any Guidance updates.




                                                          Clean Air
                                             Environmental Standards Branch
                                            Environmental Protection Division
                          British Columbia Ministry of Environment and Climate Change Strategy
                                               PO Box 9341, Stn Prov Govt
                                            Victoria, British Columbia, Canada
                                                         V8W 9M1




                                                    Document History

April 2021:               DRAFT released

November 2021:            October 2021 version released

August 2022:              July 2022 version released (minor revisions)




                                            IMPORTANT NOTE
Modelling practitioners are expected to read and follow the BRITISH COLUMBIA AIR QUALITY
DISPERSION MODELLING GUIDELINE in addition to following the NOX and NO2-specific guidance in this
document.




Version: July 2022
                                 Guidance on NO2 Dispersion Modelling in British Columbia



                                                                 TABLE OF CONTENTS

      INTRODUCTION .............................................................................................................................................. 1
           PURPOSE .......................................................................................................................................................1
           NOX TO NO2 CHEMISTRY .................................................................................................................................1
           INTRODUCTION TO CURRENT BC ENV MODELLING GUIDELINES...............................................................................1
                    Assessment Levels ........................................................................................................................2
                    1.3.1.1 Level 1 – Screening Models ............................................................................................2
                    1.3.1.2 Level 2 – Air Dispersion Models .....................................................................................2
                    1.3.1.3 Level 3 – Advanced Air Disperion Models ......................................................................3
           NO2 MODELLING TECHNIQUES ..........................................................................................................................4

      SUMMARY OF CHANGES FROM THE LATEST GUIDELINE ................................................................................ 5

      NO2 MODELLING METHODS ........................................................................................................................... 7
           AIR DISPERSION MODELS .................................................................................................................................7
                    AERMOD and AERSCREEN ............................................................................................................7
                    3.1.1.1 AERMOD ........................................................................................................................7
                    3.1.1.2 AERSCREEN ....................................................................................................................7
                    CALPUFF .......................................................................................................................................8
                    3.1.2.1 CALPUFF .........................................................................................................................8
           NO2 MODEL TECHNIQUES GUIDANCE .................................................................................................................8
                    Recommended NO To NO2 Conversion Techniques.....................................................................8
                    3.2.1.1 Tier 1: Total Conversion .................................................................................................9
                    3.2.1.2 Tier 2: Ambient Ratio Method........................................................................................9
                    3.2.1.3 Tier 3: Ozone Limiting Method and Plume Volume Molar Ratio Method ....................11
                    Alternative NO2 Modelling Techniques ......................................................................................14
           MODEL POST-PROCESSING .............................................................................................................................14
                    Conversion of NO To NO2 Using Post-Processing Tools .............................................................14
                    Baseline NO2 Concentrations .....................................................................................................15
                    3.3.2.1 Monte Carlo method for baseline ................................................................................17

      MATERIALS REQUIRED FOR A MODEL PROJECT APPLICATION ...................................................................... 19
           FILES INCLUDED IN SUBMITTAL.........................................................................................................................19
           DISPERSION MODELLING PLAN ........................................................................................................................19
           QUALITY ASSURANCE AND MODEL PERFORMANCE ..............................................................................................20

APPENDIX A: NO2 AND NOX DATA FOR ARM2 IMPLEMENTATION........................................................................ 22

APPENDIX B: IN-STACK RATIOS OF NO2 AND NOX ................................................................................................ 28

APPENDIX C: OZONE DATA .................................................................................................................................. 32
    OZONE DATA AVAILABLE IN BRITISH COLUMBIA .............................................................................................................32
    OZONE DATA IN NORTHEAST BRITISH COLUMBIA............................................................................................................34
    OZONE DATASETS FOR USE IN NO2 MODELS .................................................................................................................36

APPENDIX D: SUBSTITUTION METHODS FOR MISSING OZONE DATA ................................................................... 42




Version: July 2022                                                                                                                                              Page i
                           Guidance on NO2 Dispersion Modelling in British Columbia



ACKNOWLEDGEMENTS
The Ministry of Environment and Climate Change Strategy commissioned Trinity Consultants to assist in developing this
guidance. The guidance also benefitted from the contributions of modelling and atmospheric-science experts who are
experienced in the application of dispersion models for regulatory processes in British Columbia. The expert community
who contributed includes modelling practitioners from consultants (Hemmera, Golder, Stantec, WSP) and Metro
Vancouver. Industrial associations, including Canadian Association of Petroleum Producers and Canadian Fuels Association,
also provided valuable comments and contributions.




Version: July 2022                                                                                                Page ii
                            Guidance on NO2 Dispersion Modelling in British Columbia



     INTRODUCTION

       PURPOSE

The British Columbia (BC) Ministry of Environment and Climate Change Strategy (ENV or the Ministry) has prepared this
document to provide guidance on the modelling of ambient nitrogen dioxide (NO2) concentrations from nitrogen oxides
(NOX) emissions to supplement and replace aspects of the existing modelling guidance. For applications in the Metro
Vancouver region, applicants should consult with the Metro Vancouver Regional District, which regulates air quality and
makes air dispersion model decisions in the region.1

The primary goal in providing guidance specific to NO2 modelling is to provide clarity for applicants on the assumptions and
methods appropriate for NO2 models in the province, which increases accuracy of modelling and leads to a streamlined
model review process and consistency across projects.

       NO X TO NO 2 CHEMISTRY

The term NOX generally refers to the mono-nitrogen oxides: nitric oxide (NO) and NO2. Other less common nitrogen species
include nitrogen trioxide (NO3), nitrous oxide (N2O), dinitrogen dioxide (N2O2), dinitrogen tetroxide (N2O4), and dinitrogen
pentoxide (N2O5).2 NO and NO2 account for the vast majority of directly emitted nitrogen species. Some NOX emission estimates
also include nitrous acid (HONO) such as the U.S. EPA MOVES model for estimating vehicle emissions. It is possible for the NOX
within a stack or tailpipe to form nitrous acid or nitric acid (HNO3) prior to being emitted to ambient air.

NOX concentrations reach equilibrium in ambient air, during daylight, and can be represented as a ratio of NO to NO2. The value
of this ratio is dependent on the intensity of sunshine, concentration of ground-level ozone (O3), and concentrations of other
reactive species.3 The presence of excess molecular oxygen (O2) can also result in NO2 formation, as the O2 reacts with NO.
Emissions of NOX from combustion occur when the nitrogen in combustion air converts to NO and NO2 (thermal NOX) and when
available nitrogen in the fuel oxidizes (fuel NOX); the majority of the NOX in combustion exhaust is in the form of NO.4 The
conversion of NO to NO2 in the atmosphere must be considered in all air dispersion model NO2 predictions. This document
outlines the various assumptions and methods that can be used to model NO2 concentrations from a given project or facility as
accurately as possible.

       INTRODUCTION TO CURRENT BC ENV MODELLING GUIDELINES

This document serves to supplement the BC Air Quality Dispersion Modelling Guideline (the Guideline). The Guideline outlines
the levels of assessment available for various project scenarios, including descriptions of accepted methods for developing each
model. Section 8.2 of the Guideline discusses NO to NO2 conversion and associated modelling guidance, and Section 8.1.4
describes how to determine NO2 baseline levels using monitoring data. Section 2 of this document outlines the changes in this
NO2 modelling guidance relative to the latest Guideline.




1
  http://www.metrovancouver.org/
2
  All oxides of nitrogen species are referred to as NOZ and non-NOX species referred to as NOY; i.e., NOX + NOY = NOZ
3
  https://www3.epa.gov/ttncatc1/dir1/fnoxdoc.pdf
4
  Many techniques and models apply an assumption that 90% of NOX in a typical combustion process is NO and 10% is NO2.
See Appendix B for more details.


Version: July 2022                                                                                                      Page 1
                            Guidance on NO2 Dispersion Modelling in British Columbia



         ASSESSMENT LEVELS

Air dispersion model analyses can be categorized by the associated complexity of the project’s model requirements. Different
model assessment levels may be appropriate depending on the site-specific conditions and project emissions considerations and
the regulatory context related to a given model scenario. As described in Section 2 of the Guideline, Level 1 Assessments
generally apply the AERSCREEN model, while Level 2 or Level 3 Assessments generally apply the AERMOD or CALPUFF model.
The following sections provide the details as specified in Section 1.5 of the Guideline describing what conditions warrant a Level
1, Level 2, or Level 3 Assessment.

The latest Guideline states that advanced NO2 modelling techniques are only acceptable for Level 3 Assessments. This updated
guidance allows advanced NO2 modelling techniques for any assessment level.


1.3.1.1 LEVEL 1 – SCREENING MODELS

A Level 1 Assessment is appropriate when decisions can be made based only on an estimate of the possible worst-case air
quality, independent of where or when they occur. Screening Assessments are appropriate for situations such as:

        “Go, no-go” evaluations (a critical acceptance/rejection criteria are exceeded or not exceeded: typically for low-risk
         sources).

        Permit/approval decisions for low-risk sources.

        Preliminary identification of air quality issues associated with proposed new sources or modifications to existing
         sources.

        Planning purposes (internal resources required to conduct assessment, need to consider other studies to support
         the decision-making process, need to contact other agencies).

        Identification of the need for more detailed modelling using Level 2 or 3 Assessment approaches (if exceedances of
         short-term objectives are predicted).

        Confirmation of refined model results that appear unusually high or low.


1.3.1.2 LEVEL 2 – AIR DISPERSION MODELS

A Level 2 Assessment provides a more realistic and detailed determination of air quality dispersion than what is provided in a
Level 1 Assessment. This level requires a refined model that uses a time series of hourly meteorological data and the geophysical
conditions representative of the site. A Level 2 Assessment is appropriate when:

        A Screening (Level 1) Assessment indicates the potential for an exceedance of ambient objectives.

        There is a need to produce a maximum concentration for different time averages and distributions of the
         concentrations and/or depositions in time and space that reflect the actual meteorological conditions.

        The contaminant can be reasonably modelled by a straight-line, steady-state, Gaussian plume model with no
         chemical transformation. Although more complicated processes may be occurring (i.e., curved plume trajectory), a
         more complicated model that explicitly treats these processes is not necessary depending on the purposes of the
         modelling and the zone of interest. For example, if the area of interest is within 100 m, then curvilinear trajectories
         and chemical transformations are likely not critical.




Version: July 2022                                                                                                       Page 2
                            Guidance on NO2 Dispersion Modelling in British Columbia



       The emissions are from small sources (e.g., a small compressor station stacks) where the greatest concentrations
        are in the order of 100 m downwind.

       The source is considered to be low risk.

       The purpose is for a standard/generic permit or amendment process (such as an emissions reduction).

       To define conditions under which well test flaring can occur.

       Supporting other air management related investigations such as:

              o   identify potential contributing sources;

              o   identify worst-case meteorological conditions;

              o   identify areas of air quality concern;

              o   analyze historical air quality trends; and

              o   design monitoring networks (locations, contaminants, sampling period, frequency).


1.3.1.3 LEVEL 3 – ADVANCED AIR DISPERION MODELS

Level 3 Assessments require refined models and corresponding input data, resources and model operator expertise to properly
account for these factors. These models require detailed meteorological, geophysical and source input that may include:

       representative meteorological data (e.g., wind speed, wind direction, temperature, turbulence, and mixing height)
        at a number of sites in the domain of interest.

       Detailed emission inventories for point, line, area and volume sources in an airshed that could vary in time and
        where there is a mix of urban, industrial and natural sources.

       The speciation, emissions and time variation of different contaminants.

In general, a Level 3 Assessment is recommended in situations where:

       A Level 1 or 2 Assessment indicates predicted exceedances of ambient objectives.

       The purpose of the assessment requires detailed time and space variation of the concentrations.

       It is important to account for multiple source types, chemical transformations and effects associated with complex
        topography such as causality, calms, curvilinear plume trajectories, spatial variations in turbulent mixing.

       A source is considered to be high risk.

More specifically, Level 3 Assessments are recommended in situations where there is a need to:

       Evaluate air quality consequences under a permitting or Environmental Assessment process for large industrial
        developments that have considerable social, economic and environmental consequences.

       Assess contaminants resulting from non-linear processes (e.g., deposition, ground-level ozone, particulate
        formation, visibility).

       Evaluate consequences of air quality management approaches that involve multi-source, multi-sector
        contributions from permitted and non-permitted sources in an airshed.




Version: July 2022                                                                                                   Page 3
                            Guidance on NO2 Dispersion Modelling in British Columbia



        Provide information to support environmental, human and economic effects studies.

        Examine specific receptors that may be sensitive or of special interest such as individual residences, sensitive
         ecosystem areas.

        Assess contaminants in meteorologically complex situations (“complex flow” such as mountain valley flows,
         reversals, sea breeze, and fumigation).

        Assist in understanding of the underlying source and meteorological causes of episodes.

        NO 2 MODELLING TECHNIQUES

In addition to determining the appropriate model used to estimate NO2 by the complexity of the assessment (e.g., AERSCREEN
for a Level 1 Assessment to the CALPUFF modelling system for a refined Level 3 Assessment), the techniques for predicting NO2
can vary by complexity. At the conceptual centre of any NO2 model analysis are the assumptions regarding the conversion of NO
to NO2 in the atmosphere. The following lists different conversion treatments recommended by various agencies:

        100% of emitted NOX converts to NO2

        Ambient Ratio Method (ARM)

        Ozone Limiting Method (OLM) and Plume Volume Molar Ratio Method (PVMRM)

The OLM and PVMRM conversion techniques are considered most complex and require the most additional model inputs, but it
is worth noting that a more complex methodology does not necessarily yield the most accurate or lowest NO2 modelled
concentration. Therefore, modellers should consider the contributing scientific factors behind a given NO2 modelling technique
and choose a method that is appropriate for the situation. In addition to those conversion treatments listed above, there are
alternative NO2 modelling techniques, such as RIVAD/ARM3, RIVAD/ISORROPIA chemical transformation schemes within
CALPUFF. Section 3.2 below provides guidance on how to apply each technique.




Version: July 2022                                                                                                     Page 4
                           Guidance on NO2 Dispersion Modelling in British Columbia



     SUMMARY OF CHANGES FROM THE LATEST GUIDELINE

The latest Guideline included guidance on approaches for selecting NO2 baseline levels and methods for NO to NO2
conversion. The following list summarizes differences (if any) between the new NO 2 modelling guidance presented in this
document and each Guideline section related to NO2 baseline and NO to NO2 conversion.

       8.1.4 Selecting a Baseline Level Using Monitoring Data – This new guidance includes a similar approach for
        selecting 1-hour NO2 baseline. Consistent with the latest Guideline, use of a single value based on the 98th
        percentile of daily 1-hour maximum values is appropriate for Level 2 and 3 Assessments. However, this new
        guidance also provides further options for refinement including time-varying baseline NO2 levels. This document
        also includes a detailed study of available NO2 and NOX monitoring data in the province.

             o    As a change from the latest Guideline, NO2 baseline should be applied after applying ARM. This approach
                  is consistent applying U.S. EPA Ambient Ratio Method 2 (ARM2) if using AERMOD’s built-in ARM2 option.

             o    Clarification is provided in this new guidance that NO2 baseline should also be applied after applying OLM
                  or PVMRM.

       8.2.1 100% conversion – No substantive changes.

       8.2.2 ARM – This new guidance applies the same general approach to use ambient NO 2 and NOX data to determine
        an NO2 to NOX ratio curve. However, there are some key changes:

             o    The form of the equation is updated to follow the U.S. EPA ARM2 approach (6th order polynomial rather
                  than an exponential equation).

             o    The updated method also yields a curve with a maximum ratio of 90% and a minimum ratio of 20%.

             o    ARM2 curves are provided in Appendix A for different categories of areas (coastal, industrial, rural, and
                  urban). Applicants may still develop a project-specific curve based on representative ambient NO2 and
                  NOX data on a case-by-case basis.

             o    As a change from the latest Guideline, the NO2 baseline level is applied after the application of the ARM2
                  equation.

       8.2.3 OLM– This new guidance aligns with the latest Guideline with the following exceptions:

             o    The OLM equation is changed to use the ISR, rather than use 0.1 as the minimum NO 2/NOX ratio:

                  Equation from latest Guideline:       NO2 = 0.1 × NOX + the lesser of (O3 or 0.9 × NOX) + baseline NO2

                  Revised Equation:                     NO2 = ISR × NOX + the lesser of (O3 or (ER-ISR) × NOX) + baseline NO2

                  Where:

                           ▪   ISR stands for the in-stack ratio.

                           ▪   Information is provided to determine ISRs for each source.

                           ▪   ER stands for equilibrium ratio, and the default is 0.9. This is a change from the latest
                               Guideline equation, which used an equilibrium ratio of 1.0.




Version: July 2022                                                                                                     Page 5
                             Guidance on NO2 Dispersion Modelling in British Columbia


              o    More information, flexibility, and guidance are provided for using hourly O3 datasets rather than
                   “maximum hourly O3 concentration measured from one year of representative monitoring data.”

              o    Note that OLM may be used within AERMOD or a separate post-processing step.

        8.2.4 PVMRM (AERSCREEN and AERMOD Only) – This new guidance aligns with the latest Guideline with the
         following exceptions (similar to OLM):

              o    For AERMOD and AERSCREEN, the latest Guideline recommended PVMRM. This document clarifies that
                   OLM may be used in AERMOD.

              o    Information is provided to determine ISRs for each source.

              o    More information, flexibility, and guidance are provided for using hourly O3 datasets rather than
                   “maximum hourly O3 concentration measured from one year of representative monitoring data.”

        8.2.5 RIVAD/ARM3 and RIVAD/ISORROPIA Chemical Transformation – This new guidance maintains the same
         guidance as in the latest Guideline, not recommending the use of RIVAD/ARM3 and RIVAD/ISORROPIA for NO/NO2
         conversion.

The latest Guideline states that advanced NO2 modelling techniques are only acceptable for Level 3 Assessments. This updated
guidance allows advanced NO2 modelling techniques for any assessment level.

The following table highlights the specific sections of this document that offer further detail and clarity regarding NO2 model
techniques as compared to the latest Guideline:

Table 2-1: Summary of Sections Containing NO2 Modelling Guidance Changes


         Section                     Topic                                                Notes

 3.2.1.2 &                                               This document provides additional clarity on both how and when to
                           ARM2
 Appendix A                                              implement ARM2 for a Tier 2 NO2 model project.
                           Post-processing tools         Additional details for the available post-processing techniques for
 3.3.1
                           for NO2 conversion            refining NO2 model assumptions are provided.
                                                         This document offers a detailed study of available NO2 baseline data
                           Baseline NO2
 3.3.2                                                   in BC and provides guidance for obtaining and properly assessing
                           concentrations
                                                         available NO2 baseline data for use in an NO2 modelling project.
                                                         This document includes a comprehensive list of the files which
                           Materials for modelling
 4                                                       applicants must include when submitting a modelling project to the
                           project submissions
                                                         Ministry.
                                                         In addition to providing clarity for the use of ambient O3 datasets,
                                                         this document provides a detailed analysis of available O3
 Appendix C                                              monitoring data throughout the province, describes the O3 datasets
                           Ozone datasets
 Appendix D                                              developed by the Ministry for use in regions lacking representative
                                                         O3 data, and provides applicants with appropriate methods for
                                                         filling in gaps in O3 data for representative datasets.
                                                         Appendix B offers ISR values for a variety of source types, which can
                                                         be used by applicants to accurately take into account the
 Appendix B                In-stack ratios
                                                         characteristics of a given source type and the resulting effect on
                                                         NO2 conversion chemistry.




Version: July 2022                                                                                                         Page 6
                             Guidance on NO2 Dispersion Modelling in British Columbia



     NO 2 MODELLING METHODS

        AIR DISPERSION MODELS

This section introduces the preferred models listed in Section 2 of the latest Guideline, provides brief descriptions of the models
and underlying theories, and gives context to when each model would be most appropriate for a given project. For further
information regarding the theory and implementation of these air dispersion models, applicants should refer to the latest
Guideline.


         AERMOD AND AERSCREEN


3.1.1.1 AERMOD

AERMOD is a regulatory straight-line, steady-state plume modelling system that consists of three components: AERMOD
(calculates the concentrations), AERMET (prepares the meteorological input) and AERMAP (prepares the terrain input).

Recommended use:

        sources in an industrial complex (single or multiple point, area, line, volume sources)

        situations where a straight-line, steady state model applies (non-complex flow)

        single or multiple buildings with or without building downwash

        gas and particle depositions

        constant or time-varying emissions

        rural or urban areas

        transport distances less than 50 km (depends on terrain)

        concentration estimates for all terrain locations, except in areas located on the lee-side of the topographic features


3.1.1.2 AERSCREEN

AERSCREEN is a screening version of the refined model AERMOD. AERSCREEN is a shell that runs AERMOD with an input
matrix of meteorological conditions that represent a wide range of possible conditions.

Recommended use:

        maximum 1-hour, 3-hour, 8-hour, 24-hour and annual average concentrations

        single point, capped stack, horizontal stack, flare, area (circular or rectangular), volume sources

        building wake effects on point, capped stack, horizontal stack, and flare sources

        flat and elevated terrain

        urban and rural areas

        transport distances of less than 50 km (depends on terrain)




Version: July 2022                                                                                                         Page 7
                               Guidance on NO2 Dispersion Modelling in British Columbia



            CALPUFF


3.1.2.1 CALPUFF

CALPUFF is a curved trajectory, Gaussian puff model that can account for time and space-varying meteorological conditions,
different source configurations and contaminants, and chemical transformations. The CALPUFF modelling system is
comprised of three components: CALMET (meteorological model), CALPUFF (calculates concentration and deposition
output), and CALPOST (analysis and display of output).

Recommended use:

           complex flow: non-steady-state meteorological conditions (calms, time and space variability in wind and
            turbulence fields) such as found in complex terrain and coastal situations

           local scale up to long range transport (>50 to 200 km)

           multiple sources, source types (point, area, volume) and building(s)

           gaseous and particulate deposition

           wet and dry Sulphur and Nitrogen deposition

           PM2.5 secondary formation

           visibility assessments for regional (>10 km) and long-range transport distances (>50 km to 200 km)

           constant or time varying source conditions of gaseous and particulate contaminants

           fogging, icing and odour effects

           NO 2 MODEL TECHNIQUES GUIDANCE


            RECOMMENDED NO TO NO 2 CONVERSION TECHNIQUES

The following subsections outline recommended model techniques for refining predictions of NO2 concentrations from
modelled NOX emissions. These techniques are provided in order of increasing complexity of implementation. While each
subsequent technique requires more complex inputs or calculations, it is not necessarily a more refined model that will
yield lower NO2 concentrations. For example, a model that implements the ARM2 method will not necessarily yield higher
modelled NO2 concentrations than a model that implements OLM. Rather, specific scientific considerations and relative
availability of data will determine the appropriate method of NO2 model refinement for a given project. Modellers that are
unsure about the appropriate NO2 model technique for a given project are encouraged to consult with ENV regarding which
modelling techniques are appropriate.

Tier 1, Tier 2, or Tier 3 NOX conversion techniques (described in the following subsections) may be used with any level of
dispersion modeling (Level 1 to Level 3 described in Section 1.3.1).5 However, Tier 1 for total conversion must be presented
as part of all model reports so that reviewers can understand the upper-bound NO2 model results.




5
    See Section 4.2 for decision criteria for selecting the NO to NO2 conversion technique.


Version: July 2022                                                                                                    Page 8
                            Guidance on NO2 Dispersion Modelling in British Columbia



3.2.1.1 TIER 1: TOTAL CONVERSION

The simplest technique for modelling NO2 to implement is to assume that NOX emissions are converted entirely to NO2 in
the atmosphere. For this technique, models assume that 100% of NOX emissions resulting from a given project are
converted to NO2. In reality, the quantity of NOX converted to NO2 in an equilibrium state is dependent on numerous
factors, including O3 concentrations, NO2 to NOX ISRs for a given source, baseline NO2 concentrations, and more. Assuming
complete conversion of NOX to NO2 is a conservative screening technique to determine worst-case NO2 concentrations from
a source. This total conversion must be presented as part of all model reports so that reviewers can understand the upper-
bound NO2 model results.

This method, by nature, will provide a conservative estimate of NO 2 modelled concentrations and can therefore often serve
as an appropriate initial screening assumption. Beyond initial, preliminary assessments this method is also appropriate for
smaller sources with limited anticipated NO2 model concentrations.


3.2.1.2 TIER 2: AMBIENT RATIO METHOD

The ARM method applies a ratio of NO2 to total NOX in the atmosphere to calculate NO2 concentrations based on modelled NOX
concentrations. In 2013, RTP Environmental Associates, Inc. prepared the ARM2, a model refinement for NO2 1-hour modelling
in AERMOD, for the American Petroleum Institute and the U.S. EPA.6 The ARM2 methodology refines the U.S. EPA ARM
approach by applying a varying ambient ratio of NO2 to NOX as a function of the total NOX concentration in the atmosphere (the
ratio decreases with increasing NOX concentrations). The U.S. EPA ARM2 curve was based on a 10-year dataset of monitoring
data from the entire continental United States and is built into the U.S. EPA model, AERMOD. The U.S. EPA recommends ARM2
as the preferred NO2 Tier 2 technique rather than its previous ARM, which used a fixed ambient ratio of NO2 to total NOX (i.e.,
0.75 for the annual averaging period and 0.8 for the 1 hour averaging period7,8). ARM2 represents a method of refinement that
is often preferred when compared to Tier 3 methods (discussed in the next section) due to its shorter run times and fewer input
data requirements (thereby requiring less case-by-case review and approval). A plot of the measured NO2/NOX ratios for all of
the studied BC monitoring stations is provided in Figure 3-1 below, including the BC-specific ARM2 curve developed based on
the data (see Appendix A for details).




6
  RTP Environmental Associates, Inc. “Ambient Ratio Method Version 2 (ARM2) for use with AERMOD for 1-hr NO2 Modeling
– Development and Evaluation Report.” September 20, 2013.
https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf
7
  https://www.epa.gov/sites/production/files/2020-10/documents/no2_clarification_memo-20140930.pdf
8
  Note that under the existing guidelines, applicants may alternatively use a single conversion factor based on a range of
expected concentrations with ENV approval.


Version: July 2022                                                                                                     Page 9
                            Guidance on NO2 Dispersion Modelling in British Columbia




Figure 3-1: British Columbia ARM2 Curve with Associated Monitoring Data

The data included in the figure above represents 10 years of monitoring data from a total of 36 monitoring stations in BC. A
summary of the data is provided in Appendix A.

Trinity Consultants Inc. (Trinity) developed province-specific ARM2 curves using monitoring data obtained from the BC Air Data
Archive Website9 based on the methodology used by RTP Environmental Associates, Inc.10 Details of the equations used to
determine NO2 concentrations based on NOX concentrations, including specific datasets and ARM2 curves developed for
different site categories (coastal, industrial, rural, and urban) for BC are included in Appendix A of this document.

An applicant can apply the appropriate BC ARM2 curve provided in Appendix A based on the category that best fits the facility
location. In addition to those ARM2 curves provided in Appendix A, site-specific ARM2 curves can be developed if adequate
hourly NOX measurements are available (e.g., ten complete calendar years representative of current conditions). However, the
province-wide curves are generally preferred. An applicant will need to justify why a site-specific curve is needed and consult
with ENV to ensure an adequate dataset is used. Applicants should also consider anticipated future changes affecting conditions




9
 British Columbia Ministry of Environment Air Data Archive Website, Station Reports. https://envistaweb.env.gov.bc.ca/
10
  RTP Environmental Associates, Inc. “Ambient Ratio Method Version 2 (ARM2) for use with AERMOD for 1-hr NO2
Modeling – Development and Evaluation Report.” September 20, 2013.
https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf


Version: July 2022                                                                                                     Page 10
                            Guidance on NO2 Dispersion Modelling in British Columbia


(e.g., new NOX sources in the airshed). Applicants should consult with ENV before taking this approach and should review the
curves with ENV through the Dispersion Modelling Plan.

The NO2 baseline concentration (see Section 3.3.2) should be added to modelled NO2 concentrations after applying the ARM2
equation for consistency with the NO2 monitoring values used for comparing to the ambient objectives or standards. If AERMOD
is used, the ARM2 option in AERMOD may be selected as a screening approach because the U.S. EPA ARM2 curve is more
conservative (higher NO2/NOX ratio) at all NOX concentrations (see Figure A-1). The approach and curve selection with
justification should be outlined in the Dispersion Modelling Plan and approved by ENV.

Note that when applying ARM2 to determine 1-hour NO2 values, the same ARM2 approach should be used to determine annual
average values (by applying ARM2 to all 1-hour model results and calculating or outputting the annual average after ARM2 is
applied).


3.2.1.2.1 CONVERSION BETWEEN PPB AND ΜG/M 3

Ambient monitoring data is usually measured and recorded in units of parts per billion (ppb) by volume, while models typically
provide output in concentrations in micrograms per cubic metre (μg/m3). As such, it is important to appropriately convert
ambient monitoring data to units that are consistent with model output or vice-versa, using consistent conversion assumptions.
The ARM2 curves provided in Appendix A are in units of ppb. Therefore, model output should be converted or vice versa, so that
both are in the same units before applying the ARM2 equation.

When developing the AERMOD ARM2 curve, the following equation was used to convert from ppb to μg/m3:11

         C [μg/m3] = (ppb/1000) x MW x 40.8727

         Where MW is the molecular weight of the pollutant in grams/mole.

In the case of NOX and NO2 specifically, the molecular weight of 46 g/mole12 is used, and this equation is therefore
simplified to the following:

         C [μg/m3] = ppb * 1.880

Applicants should also use the above conversion when converting from ppb to μg/m3 or vice versa to maintain consistency.


3.2.1.3 TIER 3: OZONE LIMITING METHOD AND PLUME VOLUME MOLAR RATIO METHOD

Tier 3 NO2 modelling techniques, including OLM and PVMRM, incorporate ambient O3 concentrations when determining the
level of conversion of NO to NO2. The U.S. EPA states that “PVMRM is most appropriate with relatively isolated and




11
   https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf
These factors are developed for standard conditions using a standard temperature and pressure of 25 degrees C and 760
millimeters of mercury.
12
   For NOX, even though NO and NO2 have different molecular weights, the convention for NOX mass emission rate input to
dispersion models and source testing is to represent NO X “as NO2” (using the molecular weight of NO2 of 46 g/mole).


Version: July 2022                                                                                                     Page 11
                            Guidance on NO2 Dispersion Modelling in British Columbia


elevated sources, while OLM is more appropriate for area sources, near-surface releases, or scenarios with multiple sources
where plume overlap is likely to occur.”13


3.2.1.3.1 OZONE LIMITING METHOD

The OLM assumes that O3 is the limiting reagent in the photochemical reaction of O3 and NO to form NO2. As a result of this
assumption, at all receptor locations or any given model hour, the ambient concentration of O 3 along with an equilibrium
constant determines the amount of NO2 formed by the reaction of NO and O3 in the atmosphere. The amount of NO
emitted from a modelled source is determined using the NO2 to NOX ISR, with the remaining percentage of NOX emissions
assumed to be NO.

The OLM can be applied with the approval of the Ministry. The NO2 concentration can be determined using the following
equation:14

         NO2 = ISR × NOX + the lesser of (O3 or (ER-ISR) × NOX) + baseline NO2

Where:

         “ISR” refers to the in-stack ratio of NO2 to NOX. This value varies by source type, and the value should be identified
         with suitable justification in the Dispersion Modelling Plan. ISRs are described in further detail, along with a table
         of ISR values for various source types provided in Appendix B of this document. The NO2 baseline concentration
         (see Section 3.3.2) should be added to modelled NO2 concentrations after applying the OLM equation because the
         OLM equation is based the concentrations of NOX within the plume. When applying the equation above for
         multiple sources, ISR should be based on the source with the highest ISR. Alternative methods (e.g., weighted
         average based on contribution to ambient concentrations of NOX by source at receptors used for decision-making)
         15
            may be acceptable with approval from ENV.

         “ER” refers to the equilibrium ratio. The default ER is 0.9. Applicants should contact ENV if proposing to use an
         alternative value.

The above equation assumes the concentrations for all constituents are in vol/vol units (e.g., ppmv). If there are multiple
plumes that have merged, OLM is applied to the combined plume NOX concentration (OLMGROUP ALL option in AERMOD).
If use of the maximum ISR among each individual source is believed to be too conservative for a given model, applicants
may calculate the weighted average ISR based on contribution to ambient concentrations of NOX by source at receptors
used for decision-making. Applicants should contact ENV if they wish to pursue an alternative other than using of the
maximum ISR among individual sources.




13
   U.S. EPA’s memo on Clarification on the Use of AERMOD Dispersion Modeling for Demonstrating Compliance with the
NO2 National Ambient Air Quality Standard (2014)).
14
   Cole, H. S., & Summerhays, J. E. (1979). A Review of Techniques Available for Estimation of Short-Term NO2
Concentrations. Journal of the Air Pollution Control Association, 29(8), 812-817.
15
   Different ISRs may be entered in AERMOD and the weighted average of concentration is used to determine the hour and
receptor-specific NO2/NOX ratio and NO2 concentration. For CALPUFF modelling, when there are overlapping plumes, apply
OLM based on the weighted average of concentrations of NOX by source at receptors. An OLM post-processor for CALPUFF
will be available from ENV in the near future that can be used to assess NO2 concentrations from sources with different
ISRs.


Version: July 2022                                                                                                    Page 12
                            Guidance on NO2 Dispersion Modelling in British Columbia


For plumes that do not merge, OLM is applied to each plume separately. For AERSCREEN and AERMOD applications, OLM
can be applied by selecting the OLM option within the model. In most cases the OLMGROUP ALL option in AERMOD should
be specified to allow for competition for O3 in overlapping plumes. Sources at a single facility within a typical footprint (e.g.,
within 1 km in length) may have overlapping plumes, and the OLMGROUP ALL option is appropriate.16

The required additional inputs are:

        Background O3 concentrations – site-specific. See Appendix C.

        NO2 to NOX ISR – source-specific (can apply different ISR for each source). See Appendix B. If a source type is not
         listed in Appendix B and site-specific data are not available, a default ISR of 0.5 should be applied.

        NO2/NOX ambient ER – should be kept at the default of 0.9.

Detailed descriptions of how to select appropriate O3 datasets, ISRs, and filling in missing data gaps for those O3 datasets
are provided in the Appendices of this document. If selected, the OLM approach, O3 background, and ISR for each source
with justification should be outlined in the Dispersion Modelling Plan and approved by the Ministry.


3.2.1.3.2 PLUME VOLUME MOLAR RATIO METHOD

The PVMRM method builds on the same reaction chemistry used in OLM, calculating the ratio of NO2 to NOX based on the moles
of emitted NOX and ambient O3 contained in the volume of plume located between the receptor and the source. PVMRM
estimates the amount of O3 entrained in the dispersion plume of a source to determine the amount of O3 that is available for
oxidation of NO to form NO2, then applies a limiting factor approach.17

For AERSCREEN and AERMOD applications, use of the PVMRM for NO X conversion represents another potential Tier 3 NO2
modelling technique.18 This method limits the conversion of NO to NO2 based on the amount of O3 available in the plume
and accounts for the changing plume volume due to dispersion. The method also accounts for merging plumes in multi-
source modelling scenarios. The three additional inputs are the same as needed for OLM within AERMOD:

        Background O3 concentrations – site-specific. See Appendix C.

        NO2 to NOX ISR – source-specific (can apply different ISR for each source). See Appendix B. If a source type is not
         listed in Appendix B and site-specific data are not available, a default ISR of 0.5 should be applied.

        NO2/NOX ambient ER – should be kept at the default of 0.9.

Detailed descriptions of how to select appropriate O3 datasets, ISRs, and filling in missing data gaps for those O3 datasets
are provided in the Appendices of this report. If selected, the PVMRM approach, O3 background, and ISR for each source
with justification should be outlined in the Dispersion Modelling Plan and approved by the Ministry.




16
   https://www.epa.gov/sites/production/files/2015-07/documents/appwno2_2.pdf
17
   https://www.epa.gov/sites/production/files/2021-01/documents/no2_modeling_techniques_white_paper.pdf
18
   Hanrahan, P. L. (1999). The Plume Volume Molar Ratio Method for Determining NO 2/NOX Ratios in Modelling. Part II:
Evaluation Studies. Journal of the Air and Waste Management Association, 49, 1332-1338.


Version: July 2022                                                                                                      Page 13
                             Guidance on NO2 Dispersion Modelling in British Columbia



          ALTERNATIVE NO 2 MODELLING TECHNIQUES

Additional modelling techniques include:

         The RIVAD/ARM3 and RIVAD/ISORROPIA chemical transformation schemes within CALPUFF are able to provide
          predicted NO2 concentrations as part of the output suite of parameters. However, consistent with the latest
          Guideline, the use of these modules directly for local and regional scale NO2 estimates has not been evaluated.
          Until more evidence emerges on the use of CALPUFF for this specific purpose, ENV does not recommend the use of
          RIVAD/ARM3 and RIVAD/ISORROPIA for NO/NO2 conversion.

         Photochemical grid models (e.g., the Community Multiscale Air Quality Model (CMAQ) or the Comprehensive Air
          Quality Model with Extensions (CAMx)) and photochemical puff models (e.g., SCICHEM) estimate concentrations of
          NO2 and other pollutants by simulating complex atmospheric photochemical transformations. These models are
          generally used for estimating O3 or secondary particulate matter during episodic events. While photochemical
          models may include reactions that more accurately represent actual conditions, the photochemical models’
          sophistication, data requirements and cost are generally not warranted for model predictions of NO 2.

         The Atmospheric Dispersion Model System (ADMS) 19 Method applies a similar approach as PVMRM including
          plume entrainment.20 It also adds a reaction rate based on solar radiation and travel time from source to receptor.
          While this method has shown better performance in at least one study,21 this approach is an alternative approach
          that must be approved by the Ministry because ADMS is not an approved model listed in Section 2 of the
          Guideline.

This document does not provide detailed guidance on these techniques. Applicants should contact ENV prior to use of
alternative models if proposing that alterative model techniques are necessary to adequately demonstrate expected NO2
concentrations for a given modelled source.

         MODEL POST-PROCESSING

The following sections provide descriptions of post-processing methods for model outputs.


          CONVERSION OF NO TO NO 2 USING POST-PROCESSING TOOLS

The Tier 2 and Tier 3 techniques described in the previous section can be used by selecting the appropriate options in AERMOD.
CALPUFF does not include built-in algorithms for these same techniques, but CALPOST includes an option to specify short-term
and long-term ambient NO2/NOX ratio conversion tables with key words CNOX_S, TNO2NOX_S, CNOX_L, and TNO2NOX_L. The
tables include 24 values, so they can be used to specify a step function that corresponds to the province-specific curves.
Applicants must specify these 24 values based on the representative ARM2 curve to use CALPOST’s ARM-related option. Care
should be taken to ensure the step functions fall to the right of the province-specific curves. The 24 values should be listed in the
Dispersion Modelling Plan. The NO2 baseline should be added to the model result after applying the ARM2 equation. The ARM2
equations can also be applied to the model output using other post-processing methods such as running post-processing scripts.




19
   http://www.cerc.co.uk/environmental-software/ADMS-model.html
20
   https://www.cerc.co.uk/environmental-software/assets/data/doc_techspec/P18_02.pdf
21
   Carruthers, D.J., Stocker, J.R., Ellis, A., Seaton, M.D. and Smith, S.E. (2017). Evaluation of an explicit NO X chemistry
method in AERMOD. Journal of the Air & Waste Management Association, 67(6), pp.702- 712.


Version: July 2022                                                                                                         Page 14
                            Guidance on NO2 Dispersion Modelling in British Columbia


It is important to note that for 1-hour NO2 results, the ARM2 equation should be applied to each hour and receptor rather than
only to the summary result because the hour and location of the 98th percentile result can change depending on the outcome of
the ARM2 equation.

CALPOST does not include a specific option to apply Tier 3 techniques like OLM or PVMRM; however, applying the OLM
equation can be achieved using POSUTIL or another post-processing tool. The NO2 baseline should be added to the model result
after the OLM equation is applied. The OLM equation could also be applied using simpler tools such as Excel. Again, it is
important to note that for 1-hour NO2 results, the OLM equation should be applied to each hour and receptor rather than only
to the summary result because the hour and location of the 98th percentile result can change depending on the outcome of the
OLM equation.


         BASELINE NO 2 CONCENTRATIONS

The baseline concentration is the concentration added to model results to account for sources not explicitly modelled
(usually existing during the baseline period). The baseline concentration is typically applied as a single constant value
throughout the spatial and temporal model domain. Section 8.1 of the latest Guideline provides general baseline-related
recommendations (e.g., the baseline should be based on the most recent years of representative hourly monitoring data
with 75% completion in each quarter; the monitoring data can be pre-screened to exclude any periods when an
intermittent source has a significant influence on the monitor.). When modelling NO2 for a given facility, the baseline
concentrations of NO2 selected for an analysis can have a substantial effect on the resulting combined NO2 concentrations
at a given location.

        For Tier 2 (ARM2), NO2 baseline concentrations should be applied after applying the ambient NO2/NOX ratio to
         modelled total NOX concentrations in order to produce results consistent with monitored NO2 concentrations.

        For Tier 3 (OLM and PVMRM), NO2 baseline concentrations should be applied after OLM or PVMRM is
         implemented.

The temporal variability of NO2 baseline datasets is an important consideration when selecting an appropriate NO2 baseline
for a model. While an NO2 baseline is often represented as a single value applied to the entirety of the model’s time
domain, temporally varying NO2 baselines can offer a more refined approach for 1-hour averaging period NO2 models. This
refinement can range in complexity from using annual hour-of-day values (which take into account fluctuations in NO2 over
the course of a day) to datasets that vary by both hour-of-day and by month.

1-hour NO2 baseline datasets can vary widely in format. The following methods are presented in order of increasing
refinement. Applicants should progress sequentially through the following list as necessary to develop a refined 1-hour NO2
modeling baseline that is representative of the conditions at the modelled facility.22

        Level 1 Assessments should use the most conservative approach for consistency with the latest Guideline,
         requiring use of the maximum monitored concentration (100th percentile).
        For Level 2 and 3 Assessments, the total concentration (baseline + modelled concentration) sequential calculation
         options are provided below:




22
   Considerations to determine whether conditions are representative of the conditions at the modelled facility are included
in Section 4.2.


Version: July 2022                                                                                                   Page 15
                            Guidance on NO2 Dispersion Modelling in British Columbia


              o    98th Percentile of Daily 1-Hour Maximum (D1HM) Baseline from Monitoring Data
                   + 98th Percentile of D1HM Modelled Concentration
                   [For complete years, the 8th highest D1HM from monitor + the 8th highest D1HM model.]

              o    98th Percentile of D1HM of:
                   (Baseline 98th Percentile of Annual Hour-of-Day from Monitoring Data
                   + Hourly Modelled Concentration)
                   [For complete years, the 98th percentile baseline dataset is based on the 8th highest of each hour of day
                   over year from monitor, each hour-of-day averaged over each year of a multi-year monitor dataset.]

              o    98th Percentile of D1HM of:
                   (Baseline 98th Percentile of Monthly or Seasonal Hour-of-Day from Monitoring Data
                   + Hourly Modelled Concentration)
                   [For complete years, the 98th percentile baseline dataset is based on the 1st highest of each hour of day
                   over month or the 3rd highest of each hour of day over season, averaged over each year of a multi-year
                   monitor dataset.23 Note that the winter season is defined to include January and February of the year
                   under review and December of the previous year.]

The annual average baseline concentration is the annual mean calculated from screened or unscreened hourly data over a
year.

For sites without NO2 monitoring data, applicants should select a representative monitor based on the monitor categories
below. These categories provide a straight-forward approach to identify the most similar baseline concentrations based on
the types of NOX sources that may exist in the surrounding area.

        Rural Sites

        Coastal Sites

        Urban Sites

        Industrial Sites

The baseline NO2 concentration should reflect the NO2 concentrations in the area from existing emission sources.
Additionally, the ambient NO2 concentration is also heavily affected by the local climatology, including background O3
concentration and terrain, which could affect how quickly NO emitted from combustion sources converts to NO 2. If a
proposed site is located in a remote area without a monitor, the applicant should evaluate whether the location is
considered coastal (with nearby marine activity), industrial (with nearby industrial operations emitting NO X), or rural (with
no nearby notable NOX sources). If a proposed site will be located in an area that fits multiple categories, the applicant
should discuss the most representative baseline with ENV when preparing the Dispersion Modelling Plan. See Appendix A
for further explanation of how monitoring stations are categorized.




23
  The use of the 3rd highest to represent the seasonal 98th percentile is consistent with the California Air Pollution Control
Officers Association (CAPCOA) guidance on modelling compliance with the U.S. Federal 1-hour NO2 air quality standard.
When assessing values for seasonal hour-of-day datasets, use of the 2nd highest would effectively “ignore” only four total
values for the entire year, rather than the seven total values “ignored” when determining the 98th percentile (8th highest)
for a yearly dataset. Therefore, use of the 2nd highest from a given season would be more conservative, and the 3 rd highest
should therefore represent the 98th percentile for seasonal hour-of-day.


Version: July 2022                                                                                                    Page 16
                           Guidance on NO2 Dispersion Modelling in British Columbia


Once the category of the proposed site is determined, the most representative monitor can be selected based on
commonly applied considerations, such as proximity, land cover similarity, and population similarity. After the
representative monitor is selected, the baseline value(s) can be determined.

For example, oil and gas facilities in BC are primarily located in northeast of the province. Urban areas are not present for
most of the locations where the oil and gas facilities are situated. Based on the proximity of the continuous NOX monitoring
stations, Fort St. John Key Learning Centre, Farmington Community Hall, and Taylor Townsite are available options.

        The Fort St. John Key Learning Centre station represents a small urbanized residential area which is not influenced
         by point sources.24 The NO2 monitor data from this station could be used to estimate the NO2 baseline condition in
         areas further away from NO2 emitting sources but with human activities (e.g., Fort Nelson).

        The Farmington Community Hall station is an industrial-owned monitoring station located in a rural area. No
         nearby notable NOX sources are present. The closest NOX emission source is an oil and gas extraction site owned by
         Tourmaline Oil Corp. (approximately 3 km away). The NO2 monitor data from this station could be used to estimate
         the NO2 baseline condition in rural areas without notable NOX sources.

        The Taylor Townsite station is located approximately 1 km from Spectra Energy’s natural pipeline station, which
         reported over 1000 tonnes of NOX emissions in 2017. The NO2 monitor data from this station could be used to
         estimate the NO2 baseline condition in areas with notable NOX sources (e.g., areas close to Ikkuma Resource Corp’s
         Sierra Gas Plant which reported 1,200 tpy in 2017).

The baseline approach and dataset selection with justification should be outlined in the Dispersion Modelling Plan and
approved by the Ministry.


3.3.2.1 MONTE CARLO METHOD FOR BASELINE

The Monte Carlo method of analysis further refines baseline by using a statistical analysis to address variability in
monitored concentrations and how the concentrations are paired with model results. Prior approval from ENV is required
to use this approach, and the simpler baseline approaches described in the previous section should be applied first. A
Monte Carlo approach can be used for NO2 baseline by following the steps in the list below.

        Daily 1-hour maximum NO2 concentrations based on appropriate monitoring dataset are entered as inputs for this
         method. The monitoring dataset should cover 3 years in order to match the form of BC’s 1-hour NO2 Ambient Air
         Quality Objective (AAQO).

        For each receptor, the daily 1-hour maximum modelled NO2 concentration and ambient NO2 concentrations are
         randomly combined so that a combined value is generated for each day of the year (365 values). For a 5-year
         AERMOD modelling example, the sampling process first randomly selects thirty-one daily 1-hour maximum
         modelled NO2 concentrations from all modelled January days (155 in total for 5 years of modelled results). The
         sampling process then randomly select thirty-one daily 1-hour maximum monitored NO2 values from all monitored
         January days (93 in total for 3 years of monitoring data). The randomly-selected modelled NO2 concentrations and
         the monitored NO2 values are then combined to represent the thirty-one daily total January NO2 results. This




24
  Based on the NAPS Classification for Fort St. John Key Learning Centre monitoring station, provided by Li Huang (BC ENV)
on October 21, 2019 to Hui Cheng (Trinity).


Version: July 2022                                                                                                  Page 17
                           Guidance on NO2 Dispersion Modelling in British Columbia


         process is performed for each month and 365 values are generated. The 98 th percentile of the 365 resulting values
         for a given year is then taken as the output of the random sampling process.

              o   The number of years of monitoring data can be different than the number of years of model data without
                  making any adjustments to use the datasets.

        The above random sampling process is repeated 1,000 times, and the median of those 98th percentile values is
         taken to be the most likely total concentration in the form of BC’s 1-hour NO2 AAQO for each modelled receptor.

This method is applied using the Monte Carlo script, which was developed using the software “R” by the Washington State
Department of Ecology and adapted in conjunction with Idaho Department of Environmental Quality’s Pao Baylon and
Kevin Schilling for use with background concentrations.25 CALPOST outputs should be reformatted into AERMOD-output
format to allow the Monte Carlo R script to read the outputs.




25
  Baylon, P. and Schilling, K (Idaho Department of Environmental Quality). “Revisiting the Monte Carlo method for
combining modeled pollutant concentrations with monitored values: Case study in Idaho.” Guideline on Air Quality Models:
Planning Ahead, Paper #MO23. Durham, NC. March 19-21, 2019.


Version: July 2022                                                                                                Page 18
                            Guidance on NO2 Dispersion Modelling in British Columbia



     MATERIALS REQUIRED FOR A MODEL PROJECT APPLICATION

        FILES INCLUDED IN SUBMITTAL

The following list outlines the files that applicants should include in an application package when submitting a model
project. A project application will not be considered complete unless the following materials are provided:

        Input files

              o    Meteorological input and output files including prognostic data files (e.g., WRF .m3d files) if applicable

              o    Model control input files

              o    Building downwash files

              o    O3 data files (as applicable)

              o    NO2 baseline data files (as applicable)

        Output files

              o    Model output files

              o    Relevant plot files

              o    CALPOST or other post-processing files

              o    Data used to assess model quality (as applicable)

        DISPERSION MODELLING PLAN

Each modelling project that incorporates NO2 techniques must submit a completed Dispersion Modelling Plan that provides
justification for model parameters, assumptions made regarding NO2/NOX chemistry, and selection of any needed ambient
monitoring datasets for O3 or NO2. The Dispersion Modelling Plan should address the following specific to NO2 techniques:

        NO2 model technique/Tier

              o    Which NO2 modelling technique was selected and why? Factors considered may include availability of
                   representative data.

              o    What considerations were given to ambient concentrations in the model location, characteristics of the
                   modelled sources, and availability of relevant monitoring data?

        For Tier 2 ARM, what dataset is used for the ARM2 curve and why?

              o    If CALPOST is used, provide the 24 values used for the step function.

        For Tier 3 OLM and PVMRM, what ISR was used for each source and why?

              o    If multiple NOX sources are modelled using an OLM post-processing tool, applicants should provide
                   justification for how the single ISR is selected.

        For Tier 3 OLM and PVMRM, what O3 dataset is selected and why?




Version: July 2022                                                                                                    Page 19
                              Guidance on NO2 Dispersion Modelling in British Columbia


                 o   Applicants should provide justification for O3 datasets used in the model, taking into account NO2
                     chemistry, location and proximity of urban regions relative to the modelled source, proximity of nearby
                     large industrial or transportation sources of NOX, etc.

                 o   For representative hourly O3 datasets corresponding to meteorological data, the Dispersion Modelling
                     Plan also needs to describe the data substitution methods followed.

           NO2 baseline

                 o   Applicants should specify the NO2 monitoring dataset used and why the dataset is representative of the
                     facility site considering NO2 chemistry, location and proximity of urban regions relative to the modelled
                     source, proximity of nearby large industrial or transportation sources of NO X, etc.

                 o   Applicants should also specify the approach used to determine a single or time-varying baseline level
                     based on the dataset (see Section 3.3.2).

           If any of the advanced processing methods (e.g., the Monte Carlo method) are applied, the computer code used
            to generate results must be included in the submittal.

           QUALITY ASSURANCE AND MODEL PERFORMANCE

Good modelling practice involves the examination of the input files to ensure that specific data treatments have been
applied properly (missing data, calms, formats, and units). Section 9.1 of the Guideline provides guidance on the application
of QA/QC for CALPUFF as this model can involve more professional judgment for its proper application than the other
recommended models. Model output should also be reviewed to identify unexpected model behavior.

If sufficient historical emissions data and corresponding ambient monitoring data are available for a site conducting Level 3
Assessments, applicants are encouraged to conduct an assessment of model performance as quality assurance for the
model project. As detailed in Section 9.2 of the latest Guideline, demonstrations of model performance can vary in form,
each providing a different perspective on model performance. A robust model performance evaluation will provide added
confidence in the ability of the selected model to provide accurate predictions. Examples of model performance
assessments comparing modelled concentrations to monitored concentrations include but are not limited to the following:

           Comparison of summary concentrations

                 o   Maximum 1-hour, 98th percentile of daily 1-hour maximum, annual average, etc.

                 o   Paired in space, not time other than annual average

           Comparison of frequency statistics

                 o   Number of exceedances of a certain level such as ½ the air quality objective

                 o   Paired in space, not time

           Comparison of robust highest concentrations (RHC)

                 o   RHC represents a smoothed estimate of the highest concentrations, based on a tail exponential fit to the
                     upper end of the concentration distribution.26




26
     https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/aermod_mep.pdf (pp. 13-14)


Version: July 2022                                                                                                    Page 20
                              Guidance on NO2 Dispersion Modelling in British Columbia


                o    Paired in space, not time

          Quantile-Quantile (Q-Q) plots

                o    Q-Q plots are created by sorting by rank the predicted and the observed concentrations from a set of
                     predictions initially paired in time and space. The sorted list of predicted concentrations is then plotted
                     by rank against the observed concentrations also sorted by rank.

                o    These concentration pairs are no longer paired in time or location. However, the plot is useful for
                     answering the question, “Over a period of time and over a variety of locations, does the distribution of
                     the model predictions match those of observations?”15

                o    A similar Q-Q plot approach can be used by maintaining paring in space (separate datasets for each
                     monitor) and plotting each monitor’s dataset separately.

          Timeseries plots

                o    Paired in time and space to visualize the variation in concentrations over time.

          Bias, error, and correlation

                o    Bias, error, and correlation compare model concentrations to monitor concentrations for each hour
                     modelled (paired in time and space).

                o    These comparisons often show high error and weak correlations because air dispersion models “are
                     reasonably reliable in estimating the magnitude of highest concentrations occurring sometime,
                     somewhere in the area” but “much less reliable” when looking at “estimates of concentrations that occur
                     at a specific time and site.”27

                o    That said, the bias and error statistics can be helpful when comparing performance of two or more
                     similar models.




27
     https://www.epa.gov/sites/production/files/2020-09/documents/appw_05.pdf (pp. 66246)


Version: July 2022                                                                                                       Page 21
                            Guidance on NO2 Dispersion Modelling in British Columbia



APPENDIX A: NO 2 AND NO X DATA FOR ARM2 IMPLEMENTATION

In 2013, RTP Environmental Associates, Inc. prepared the ARM2, a model refinement for NO 2 1-hour modelling in
AERMOD, for the American Petroleum Institute and the U.S. EPA.28 This method expands on the previously used ARM,
for a Tier 2 model for NO2 concentrations with a 1-hour averaging period. While ARM previously used a fixed ambient
ratio of NO2 to total NOX of 0.8 for the 1 hour averaging period, the ARM2 methodology refines the ARM approach by
applying a varying ambient ratio of NO2 to NOX as a function of the total NOX concentration in the atmosphere. In
other words, ambient monitoring data demonstrate that the ratio of NO 2 to NOX changes as a function of total NOX
concentration in the atmosphere (the ratio decreases with increasing NO X concentrations). ARM2 represents a
method of refinement that is often preferred when compared to Tier 3 methods such as OLM and PVMRM due to its
shorter run times and fewer input data requirements (thereby requiring less case-by-case review and approval). Note
that when applying ARM2 to determine 1-hour NO2 values, the same ARM2 approach should be used to determine
annual average values (by applying ARM2 to all 1-hour model results and calculating or outputting the annual average
after ARM2 is applied).

Trinity developed province-specific ARM2 curves using monitoring data obtained from the BC Air Data Archive
Website29 based on the methodology used by RTP Environmental Associates, Inc. for the following specific categories:

        Industrial Sites

               o     Monitoring stations are considered industrial when they are located within 5 km of a large
                     industrial source that emits 500 tonnes per year or more of NOX.

        Urban Sites

               o     Urban sites are located in or near urban centres that influence the baseline NOX concentrations in
                     the ambient air.

        Rural Sites

               o     Rural sites are not located in or near urban centres or industrial areas (defined above).

        Coastal Sites

               o     These sites are near the Pacific Coast, and the NOX concentrations are not primarily driven by urban
                     centres or large industrial sources.

        All Sites

Each of the categories and their associated monitoring sites can be found in Table A-1, below. If a site is classified
under multiple categories, the data from the site are included in both sub-sets of data used to develop the category-
specific curves.




28
   RTP Environmental Associates, Inc. “Ambient Ratio Method Version 2 (ARM2) for use with AERMOD for 1-hr NO2
Modeling – Development and Evaluation Report.” September 20, 2013.
https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf
29
   British Columbia Ministry of Environment Air Data Archive Website, Station Reports.
https://envistaweb.env.gov.bc.ca/


Version: July 2022                                                                                                Page 22
                        Guidance on NO2 Dispersion Modelling in British Columbia


Based on the curves developed for the province, the AERMOD default ARM2 equation is an appropriate and
conservative equation for use with the ARM2 option:

               y=-5.176E-16x6 +1.005E-12x5 -7.288E-10x4 +2.296E-07x3 -1.981E-05x2 -5.148E-03x+1.244
                  (Where x is the NOX concentration in ppb, and y is the calculated ambient ratio)

Applicants wishing to further refine assumptions for NO2 conversion while using this Tier 2 method may consider
using the province- and category-specific equations developed by Trinity and provided in detail below. A comparison
of the equations and the corresponding curves for the province-specific and U.S. EPA AERMOD default ARM2
equations is provided in Figure A-1 below.




Version: July 2022                                                                                         Page 23
                             Guidance on NO2 Dispersion Modelling in British Columbia


Table A-1: Summary of Monitoring Site Categories


                      Site                                Category/Categories                     Monitoring Period
     Abbotsford Central                           Urban                                    2009-2018
     Burnaby Kensington Park                      Urban                                    2009-2018
     Burnaby Mountain                             Urban                                    2009-2018
     Burnaby South                                Urban                                    2009-2018
     Chilliwack Airport                           Urban                                    2009-2018
     Colwood City Hall                            Urban                                    2009-2018
     Coquitlam Douglas College                    Urban                                    2009-2018
     Duncan Cairnsmore                            Urban                                    2009-2018
     Fort St. John Key Learning Centre            Industrial                               2015-2018
     Hope Airport                                 Rural                                    2009-2018
     Kamloops Federal Building                    Urban                                    2009-2018
     Kelowna College                              Urban                                    2009-2018
     Langdale Elementary                          Rural/Coastal                            2009-2018
     Langley Central                              Urban                                    2009-2018
     Maple Ridge Golden Ears School               Urban                                    2009-2018
     Nanaimo Labieux Road                         Urban                                    2009-2018
     North Delta                                  Urban                                    2009-2018
     North Vancouver Mahon Park                   Urban                                    2009-2018
     North Vancouver Second Narrows               Urban/Industrial                         2009-2018
     Pitt Meadows Meadowlands School              Urban                                    2009-2018
     Port Moody Rocky Point Park                  Urban                                    2009-2018
     Powell River Cranberry Lake                  Rural                                    2009-2018
     Prince George Plaza 400                      Urban                                    2009-2018
     Quesnel Senior Secondary                     Industrial                               2009-2018
     Richmond South                               Urban                                    2009-2018
     Rolla 213 Road                               Industrial                               2016a - 2017
     Smithers St Josephs                          Rural                                    2009-2018
     Surrey East                                  Urban                                    2009-2018
     Taylor Townsite                              Industrial                               2018
     Tsawwassen                                   Coastal                                  2009-2018
     Vancouver Intl Airport 2                     Urban                                    2009-2018
     Vancouver Robson Square                      Urban                                    2009-2018
     Vernon Science Centre                        Urban                                    2009-2018
     Victoria Topaz                               Urban/Coastal                            2009-2018
     Whistler Meadow Park                         Rural                                    2009-2018
     Williams Lk Columneetza School               Industrial                               2009-2018
a.
  Rolla 213 Road data began in December 2015. Listing 2016 as start year because less than one month of data were recorded in
2015.

These curves, with their associated equations, can be used to refine NO 2 model concentrations based on the
characteristics of the area where a given facility or project is located. These curves are provided in Figure A-1 for
comparison, with the coefficients of the 6th order equations provided in Table A-2. The following list summarizes a few
key notes regarding the development of the curves:

          In order to address noise in data at low concentrations, a lower-bound NOX threshold of 20 ppb was used to
           filter data.



Version: July 2022                                                                                                   Page 24
                         Guidance on NO2 Dispersion Modelling in British Columbia



        Based on a study conducted by Chu and Meyers, the ARM2 equation is constrained to a maximum value of
         0.9 and a minimum value of 0.2.30

        The NO2/NOX ratio at a given NOX concentration is calculated as the 98th percentile value of all NO2/NOX
         values for a given “bin” of measurements. “Bins” of 10 ppb in range (or 20 ppb for measurements greater
         than 200 ppb) are used to divide the data. This method results in a curve that is conservative relative to the
         data obtained from the site(s).31

        A valid data point for these curves is any data point with non-negative NO2 and NO measurements and a
         calculated total NOX concentration of 20 ppb or greater.

              o    For data obtained later than 2017, a measured total NOX value is given preference over the sum of
                   the measured NO and NO2 concentrations, otherwise the sum is used for the total NO X.32




30
   RTP Environmental Associates, Inc. “Ambient Ratio Method Version 2 (ARM2) for use with AERMOD for 1-hr NO2
Modeling – Development and Evaluation Report.” September 20, 2013.
https://gaftp.epa.gov/Air/aqmg/SCRAM/models/preferred/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf
31
   Ibid.
32
   Total NOX measurements prior to January 2017 are not validated, per email correspondence between Trinity and Li
Huang on Wednesday, October 9, 2019.


Version: July 2022                                                                                              Page 25
                           Guidance on NO2 Dispersion Modelling in British Columbia




Figure A-1: Comparison of ARM2 Curves Developed for Various Monitoring Station Categories

* Note that there are some monitoring stations that are included in multiple categories for the purposes of the ARM2 curve
development, hence the discrepancy in site count between groups.




Version: July 2022                                                                                                     Page 26
                       Guidance on NO2 Dispersion Modelling in British Columbia


Table A-2: ARM2 Curve 6th Order Polynomial Equation Coefficients (μg/m3)


   Polynomial                                BC-Specific ARM2 Curves                          U.S. EPA
     Order             All           Urban           Rural        Industrial    Coastal       Default

       0            1.4217E+00     1.4081E+00       7.0908E-01     9.7054E-01   1.4097E+00    1.2441E+00
       1           -9.0043E-03     -8.4309E-03      1.8014E-02     2.7563E-03   -8.6617E-03   -2.7383E-03
       2             2.8689E-05     2.2008E-05     -4.0219E-04    -8.0316E-05   2.6443E-05    -5.6062E-06
       3           -5.1310E-08     -1.8692E-08      3.1248E-06     4.4204E-07   -5.9049E-08   3.4555E-08
       4             6.2556E-11    -1.4082E-11     -1.1639E-08    -1.0885E-09   1.3159E-10    -5.8345E-11
       5           -5.5299E-14      2.9761E-14      2.0910E-11     1.2663E-12   -2.0684E-13   4.2795E-14
       6             2.4169E-17    -1.1526E-17     -1.4534E-14    -5.6578E-16   1.3132E-16    -1.1723E-17




Version: July 2022                                                                               Page 27
                         Guidance on NO2 Dispersion Modelling in British Columbia



APPENDIX B: IN-STACK RATIOS OF NO 2 AND NO X

The in-stack NO2/NOX ratio or ISR is one of two critical inputs required for refined NO 2 modelling using the OLM or
PVMRM options (the other is background O3 levels). These OLM and PVMRM options are available in the U.S. EPA
AERMOD model. OLM may also be used with CALPUFF or any other model output data in a post-processing step. The
ISR defines the ratio of NO2 to total NOX within the stack. Many combustion sources have a low ISR of approximately
0.1, meaning approximately 10% of the NOX in the exhaust within the stack is in the form of NO2, with the remainder
in the form of NO. The OLM and PVMRM options estimate the ratio of NO 2 in the plume at the modelled receptor and
use the ISR as the minimum ratio (i.e., the minimum fraction of NO 2 begins at the stack and increases as more NO
converts to NO2 in the atmosphere).

ISR data have been collected formally by the U.S. EPA for several years, and the Alaska Department of Environmental
Conservation provides a dataset33 last updated in August of 2013. The U.S. EPA ISR webpage provides two different
datasets;34 a beta dataset with complete information collected as part of the formal collection effort and an alpha
dataset which contains NO2 ISR values collected by various Regional, State, and Local air permitting offices prior to the
formal collection initiated by U.S. EPA. Additionally, the U.S. EPA MOVES model includes several ISRs for different
categories of vehicles based on a dataset of 292 tests on a variety of vehicles. 35 The available ISR data from these
datasets are used to develop the recommended ISR values summarized in Table B-1 below.

The MOVES vehicle ISR dataset was reviewed for the most meaningful source categories. Based on the review, the
fuel type rather than equipment class (e.g., light duty vs heavy duty vehicles) plays a more important role in
determining the appropriate ISR. Therefore, the vehicle categories are differentiated by fuel type rather than vehicle
class.

The recommended ISR is the 75th percentile of each category. The 75th percentile is chosen rather than the average,
maximum or a higher percentile, to avoid a small number of high values skewing the value. ENV is collecting
additional source-specific ISRs through permitting applications and/or volunteering program. Source-specific ISR data
provided to ENV for use in NO2 models should include the following:

        Site details and contact information

        Equipment information, including:

              o   Equipment class

              o   Equipment descriptions

              o   Fuel type

              o   Manufacturer and model




33
   Alaska Department of Environmental Conservation, “NO2/NOX In-Stack Summaries.”
https://dec.alaska.gov/media/10228/no2-nox-instack-ratios-from-source-tests-082313.xlsx
34
   U.S. EPA Support Center for Regulatory Atmospheric Modeling (SCRAM), “NO2/NOX In-Stack Ratio (ISR) Database.”
https://www3.epa.gov/scram001/no2_isr_database.htm
Accessed October 2020. Current (October 2021) database available at: https://www.epa.gov/scram/nitrogen-
dioxidenitrogen-oxide-stack-ratio-isr-database
35
   Data extracted from the U.S. EPA Motor Vehicle Emission Simulator (MOVES2014b).
https://www.epa.gov/moves/latest-version-motor-vehicle-emission-simulator-moves


Version: July 2022                                                                                             Page 28
                        Guidance on NO2 Dispersion Modelling in British Columbia


             o    Equipment capacity

             o    Any control devices installed on the equipment

        Test information, including:

             o    Equipment load (% of total capacity)

             o    Exhaust flow rate

             o    Operating temperature

             o    Average NO2 and NO concentrations

             o    % O2 from sample gas

        Any additional comments that may impact the resulting ISRs.

The recommendations on ISR values will be updated once additional ISR data are available and reviewed.




Version: July 2022                                                                                       Page 29
                          Guidance on NO2 Dispersion Modelling in British Columbia


Table B-1: Summary of ISR Values for Various Source Types

                                                        Number of Data
 Equipment Class c         Fuel Type                                            Standard Deviation        Recommended ISR a
                                                           Points
                           All                                886                       0.053                     0.065
                           Natural Gas                        876                       0.050                     0.065
 Turbine
                           Diesel/Kerosene                      3                       0.112                     0.242
                           Fuel Oil Combustion                  5                       0.088                     0.177
                           All                                3349                      0.162                     0.182
                           Biogas                              96                       0.003                     0.004
 Reciprocating IC
                           Process Gas b                        9                       0.147                     0.314
 Engine
                           Natural Gas                        3099                      0.165                     0.187
                           Diesel/Kerosene                    139                       0.119                     0.083
                           All                                137                       0.154                     0.304
                           Fish Oil and Diesel
                                                                12                      0.005                     0.013
                           (70:30)
 Boiler/Heater             Process Gas b                        58                      0.109                     0.407
                           Natural Gas                          53                      0.042                     0.100
                           Diesel/Kerosene                      10                      0.016                     0.035
                           Coal, unspecified                     4                      0.006                     0.015
 Marine Vessels            All                                  85                      0.083                     0.113
                           All                                 292                      0.113                     0.156
                           Gasoline                            105                      0.043                     0.041
                           Ethanol (E-85)                      105                      0.043                     0.041
 Vehicles
                           Diesel Fuel                          80                      0.142                     0.398
                           Compressed Natural
                                                                2                         NA                      0.127
                           Gas

a.   The ISR is the in-stack-ratio of NO2/NOX and can vary from 0 to 1. A value of 1 indicates 100% of NOX is NO2.
b.   For equipment firing process gas, applicants may propose an alternative ISR based on the U.S. EPA database or other
     reference information, subject to approval as part of the Dispersion Modelling Plan.
c.   If a category is not listed in Table B-1, proposing an ISR based on the U.S. EPA database can by accepted on a case-by-case
     basis.




Version: July 2022                                                                                                        Page 30
                         Guidance on NO2 Dispersion Modelling in British Columbia


The ISR source categories in the previous section were also selected based on the industry or source sectors’
contribution to NOX based on the latest copy of the province-wide NOX emission levels extracted from Canada’s 2019
Air Pollutant Emission Inventory.36 The following list, ordered by which sectors contribute NOX from the most to the
least, provides cross references for which ISR source categories would be appropriate for each emission inventory
industry/source sector.

        Marine transportation – covered under marine vessels.

        Upstream oil and gas industry – most sources are covered under reciprocating engines and combustion
         turbines.

        Heavy-duty vehicles – covered under diesel fuel vehicles.

        Air transportation – representative data unavailable.

        Light-duty vehicles – covered under diesel or gasoline vehicles.

        Off-road vehicles and equipment - most sources are covered under reciprocating engines.

        Rail transportation – covered under reciprocating engines.

        Pulp and paper industry – some covered under wood fired boilers.

              o   Process specific units (e.g., recovery furnaces, lime kilns) should be based on site-specific data.

              o   NOTE: ISR data for this category is not currently available in the U.S. EPA datasets.

        Commercial/residential/institutional – covered under boilers/heaters.

        Mining and rock quarrying – some covered under reciprocating engines.

              o   Process-specific units should be based on site-specific data.

        Cement and concrete industry – some covered under heaters/boilers.

              o   Process-specific units (e.g., cement kilns) should be based on site-specific data.

        Wood products – some covered under wood fired boilers.

              o   Process-specific units (e.g., direct fired dryers) should be based on site-specific data.

              o   NOTE: ISR data for this category is not currently available in the U.S. EPA datasets.

        Electric power generation (utilities) – covered under boilers and combustion turbines.

        Downstream oil and gas industry – some covered under boilers/heaters, reciprocating engines, combustion
         turbines, and incinerators/thermal oxidizers.

              o   Process-specific units (e.g., refinery catalytic crackers, reformers, process heaters)37 should be
                  based on site-specific data.

              o   NOTE: ISR data for thermal oxidizers is not currently available in the U.S. EPA datasets.




36
   The emission inventory is developed by Environment Canada and Climate Change Strategy (ECCC), and it is available
on ECCC’s website (https://www.canada.ca/en/environment-climate-change/services/air-
pollution/publications/emissions-inventory-report-2019.html).
37
   Note many process heaters, particularly indirect-fired heaters could justify using a default ISR for heaters.


Version: July 2022                                                                                              Page 31
                         Guidance on NO2 Dispersion Modelling in British Columbia



APPENDIX C: OZONE DATA

Modellers have two options for developing and using O3 datasets:

        Option 1: Hourly representative data can be used, as long as hourly O3 data (with appropriate data gap filling
         methods described in this report) correspond to the complete meteorological data being used in the model.
         The applicant must also verify the O3 dataset is representative of the facility location.

        Option 2: A dataset developed by BC ENV or a dataset developed by the applicant using the methods
         outlined in this appendix.38

OZONE DATA AVAILABLE IN BRITISH COLUMBIA

BC and Metro Vancouver operate 43 O3 monitoring stations able to measure compliance against the O3 air quality
standard for the 2015-2017 period, which requires three years of complete data. 39 As shown in Figure C-1, majority of
the monitors (23) are located in the lower Fraser Valley Air Zone, where O3 concentrations are highest.40 Extending
the dataset period to 10 years (2010 – 2019), 49 stations recorded three or more years of data. The map in Figure C-1
below obtained from ENV’s website shows divisions of each air zone within the province, and the following list
summarizes the O3 data in each air zone.

        The Northwest Air Zone does not include any O3 monitoring stations; however, minimal industrial activity
         that would require O3 data for refined NO2 modelling takes place in this air zone.

        While the Coastal Air Zone contains only one O3 monitoring station, the coastal zone contains lots of
         industrial activity, with the most activity located relatively near the established Terrace monitor and the
         newer or shorter-term monitors in Kitimat and Prince Rupert.

        The Southern Interior and Central Interior air zones do have industrial activity that may require O3 data for
         refined NO2 modelling. Each of these air zones has four O3 monitors spread across the zone; as such, a
         representative O3 monitoring dataset is expected to be available for most or all applicant projects within
         these air zones.

        Representative O3 monitoring data are also expected to be available for any location within the Lower Fraser
         Valley and Georgia Strait air zones with their higher density of monitors (23 and 7 for 2015-2017,
         respectively).

        In contrast to the Southern and Central Interior air zones, the three O3 monitors in the Northeast Air Zone are
         clustered near Fort St. John, while much of the industrial activity is removed from urban areas (100 km away
         or further).




38
   ENV has developed an O3 dataset for use in rural sites located in Northeast BC. Additional O3 datasets may be
developed in the future.
39
   The statistical form of the Canadian Ambient Air Quality Standard O3 metric is the 3-year average of the annual 4th-
highest daily maximum 8-hour rolling average concentration.
http://www.env.gov.bc.ca/soe/indicators/air/ozone.html
40
   The highest measured concentration in the form of the standard for the 2015-2017 period was 64 ppb at the Hope
station.
http://www.env.gov.bc.ca/soe/indicators/air/ozone.html


Version: July 2022                                                                                              Page 32
                      Guidance on NO2 Dispersion Modelling in British Columbia




Figure C-1: BC Ozone Monitors, 2015-2017




Version: July 2022                                                               Page 33
                           Guidance on NO2 Dispersion Modelling in British Columbia



OZONE DATA IN NORTHEAST BRITISH COLUMBIA

Based on the available representative O3 monitoring data in relation to existing and potential industrial sites, the
Northeast Air Zone requires further review to evaluate O3 monitoring data alternatives to using a nearby
representative hourly O3 monitoring data, including:

          Analyzing the data quality and representativeness of all O3 monitors in the zone (even those with less than
           three years of data);

          Evaluating whether data from Fort St. John or other Northeast monitors can be used to represent all regions
           within the Northeast Air Zone; and

          Assessing whether a dataset should be developed that applies some conservative assumptions to represent
           the regions of the air zone without monitoring data and if so, how the dataset should be developed and
           whether wildfire years should be excluded.

Figure C-2 below shows the air quality monitors in Northeast BC. 41 The monitors that measure O3 are highlighted in
yellow. The Blueberry River monitor only operated from June 2016 to November 2017, but this monitor provides a
helpful indication of the amount of variation at sites north of Fort St. John. Based on the summary data, the site
measured similar O3 compared to the other northeast stations:

          Rolla 213 Road: 65 ppb (1-hour maximum) and 28.2 ppb (average December 2015–June 2017)

          Taylor Lone Wolf Golf Course: 67.2 ppb (1-hour maximum) and 24.9 ppb (average January 2016–June 2017)

          Blueberry River First Nation School: 60.7 ppb (1-hour maximum) and 26.5 ppb (average June 2016–June
           2017)




41
     http://www.env.gov.bc.ca/epd/bcairquality/readings/ne-stations-map.html


Version: July 2022                                                                                              Page 34
                         Guidance on NO2 Dispersion Modelling in British Columbia




Figure C-2: Northeast British Columbia Ozone Monitors

In order to evaluate whether some years should be excluded due to wildfire influence, annual charts were developed.
Figure C-3 below shows the annual trends of the summary statistics at the two northern sites with the longest history.
While some influence of the intense wildfire seasons in 2017 and 2018 may be seen at Prince George Plaza (Central
Interior), minimal influence is observed at Fort St. John in the Northeast. Therefore, excluding O3 data due to wildfire
influence is not warranted for the 2017-2018 period in the Northeast Air Zone.




Version: July 2022                                                                                             Page 35
                                            Guidance on NO2 Dispersion Modelling in British Columbia




                                                           Prince George Plaza
                        60.0

                        50.0




   Hourly Ozone (ppb)
                        40.0

                        30.0

                        20.0

                        10.0

                         0.0
                            2009     2010    2011   2012     2013    2014    2015        2016   2017   2018   2019   2020

                                    Percentiles      50        75       90          95          98




                                                              Fort St John
                        60.0

                        50.0




   Hourly Ozone (ppb)
                        40.0

                        30.0

                        20.0

                        10.0

                         0.0
                            2009     2010    2011   2012     2013    2014    2015        2016   2017   2018   2019   2020

                                   Percentiles       50        75       90          95          98


Figure C-3: Annual Trends for the 50th, 75th, 90th, 95th, and 98th Percentiles of Hourly Ozone Data

OZONE DATASETS FOR USE IN NO 2 MODELS

Identification of an appropriate and representative O3 monitoring station should consider the following, among any
other factors that may impact O3 concentrations at a given location:

                        Proximity to urban areas of the province.

                        Proximity to large sources of NOX emissions.

                        Proximity to major roadways with frequent traffic.




Version: July 2022                                                                                                          Page 36
                         Guidance on NO2 Dispersion Modelling in British Columbia


Rural areas tend to have higher O3 than urban areas because NO from vehicle traffic can reduce O3. In addition, O3
takes time to form, so it tends to be highest in rural areas 30+ km from the largest metropolitan areas. In order for a
site to be considered urban, it must be located within 10 km of any city or town with a population of 5,000 or more.
All other sites should be considered rural, but urban can be considered on a case by-case basis if the site is 10 km to
20 km from a larger city.

If a representative hourly O3 monitoring dataset is not available for a specific facility (Option 1), applicants should
contact the Ministry to discuss use of a BC ENV-prepared O3 dataset for the model’s region or development of a new
dataset. An example of a rural dataset is provided below for the Northeast Air Zone. Datasets for both rural and urban
O3 can be developed based on either the 1st or the 2nd highest O3 concentration for each hour of the day by month,
depending on the amount of data available. If fewer than 3 years of monitoring data are used to develop the O 3
dataset for use in an NO2 model, then the 1st highest value for each hour of the day by month should be used. For
example, the January noon hours (31 total) are ranked, and the 1 st highest January noon value represents the noon
hour for the month of January. If 3 or more years of monitoring data are used, then the 2nd highest value for each
hour of the day by month should be used. Therefore, one data point will exist for each hour of the day for a given
month (i.e., the ENV-developed datasets include 24 values for each month, 288 values for the year).




Version: July 2022                                                                                             Page 37
                            Guidance on NO2 Dispersion Modelling in British Columbia


Table C-1: Rural Ozone Dataset (ppb) for Northeast British Columbia

                                                                Month
      Hour
                  Jan       Feb    Mar    April    May      June      July     Aug      Sep      Oct      Nov      Dec
     24:00 AM    43.9       46.7   55.0   51.8     54.0     50.5      39.0     36.0     39.3     42.0     41.0     43.1
      1:00 AM    43.3       46.9   57.0   52.2     51.0     49.1      39.0     36.8     41.2     41.9     41.1     43.6
      2:00 AM    44.0       46.7   54.5   53.8     50.0     45.0      37.0     34.1     38.1     41.3     41.1     43.9
      3:00 AM    45.0       46.7   52.0   53.7     48.0     46.0      34.0     37.0     36.7     44.0     42.0     43.5
      4:00 AM    44.9       47.3   51.0   51.0     47.0     45.0      33.4     36.0     35.7     45.0     42.0     43.6
      5:00 AM    46.2       47.4   47.5   49.3     40.5     41.8      29.5     31.4     34.6     39.7     40.2     42.3
      6:00 AM    45.0       48.2   49.0   50.0     44.1     44.9      35.0     34.0     35.0     47.4     42.0     42.6
      7:00 AM    46.1       47.9   49.0   51.0     45.4     45.8      34.0     32.0     35.0     45.0     41.5     42.5
      8:00 AM    45.9       46.9   46.8   54.1     48.6     47.0      35.2     34.4     34.9     42.0     42.0     42.6
      9:00 AM    45.0       48.1   47.1   56.0     51.8     48.4      38.0     36.0     36.7     42.0     41.6     42.4
     10:00 AM    45.0       48.2   47.3   54.9     54.1     61.0      41.0     43.2     38.0     42.0     41.7     42.3
     11:00 AM    44.9       48.4   49.0   55.1     60.8     61.0      42.9     45.6     40.7     41.3     41.6     43.0
     12:00 PM    44.9       48.5   51.7   59.0     64.0     63.0      43.0     47.0     42.4     46.0     42.2     43.5
      1:00 PM    45.0       48.7   55.8   57.3     71.0     60.3      45.0     47.0     48.1     48.0     42.3     44.1
      2:00 PM    45.2       48.5   59.0   59.0     69.0     61.0      49.0     46.3     50.5     46.0     42.4     42.6
      3:00 PM    44.5       47.8   60.0   59.0     69.0     62.0      50.0     47.0     49.4     44.8     42.6     43.7
      4:00 PM    45.0       47.4   62.0   60.0     67.0     65.0      50.0     48.2     53.0     44.0     42.4     43.8
      5:00 PM    44.0       47.0   62.1   60.0     66.0     62.0      46.0     53.3     51.0     46.0     42.3     43.4
      6:00 PM    44.9       46.5   60.0   60.0     65.0     60.0      50.0     47.0     52.0     45.0     41.4     44.8
      7:00 PM    44.0       46.2   58.0   58.3     62.4     57.0      51.0     45.0     46.0     45.0     41.0     43.3
      8:00 PM    43.9       47.0   56.8   57.8     60.0     56.0      46.4     43.0     46.7     43.3     41.0     43.4
      9:00 PM    43.9       47.0   58.3   55.7     56.0     53.5      46.6     40.0     46.0     42.0     41.0     42.6
     10:00 PM    43.5       47.1   58.4   53.9     55.0     52.0      47.0     38.4     41.0     41.0     41.5     42.3
     11:00 PM    44.0       46.2   55.1   52.8     54.0     51.9      49.0     39.0     39.7     41.4     41.5     42.8

The dataset above is intended for use in NO2 models located in the Northeast region of BC, where O3 monitoring data
availability is limited. The O3 values represent the 2nd highest value among all O3 data points for each given hour and
month. The monitoring stations used to develop this dataset are the following, which are considered rural stations
based on the qualifications detailed above:

          Blueberry Hill

          Farmington Community Hall

          Rolla 213 Road

          Beaverlodge42




42
  The Beaverlodge monitoring station is located in Alberta but is considered appropriate for inclusion in this rural
dataset for the Northeast Air Zone on the basis of its proximity to Northeast British Columbia and its similarities with
other rural datasets in the region.


Version: July 2022                                                                                               Page 38
                           Guidance on NO2 Dispersion Modelling in British Columbia


Two to five years of data are available from each of the four stations, resulting in approximately 150 datapoints per
value at each station and 600 datapoints total per value. Since more than 3 years of data are available, the 2nd highest
value for each given hour and month is used. Using the 2nd highest value in this instance excludes only the top 0.16%.

The rural dataset is displayed in the plot below.




Figure C-4: Rural Ozone Dataset for Northeast British Columbia 43

The 2nd highest value for a given hour and month is selected for use in the O3 background dataset to avoid outliers
that would otherwise overestimate typical O3 concentrations throughout the year. When comparing the 2 nd highest
O3 value to the maximum for a given hour and month, there are hours when the difference is substantial, indicating
the maximum values could represent an overestimate. Additionally, the 2 nd highest values follow the maximum
concentrations closely during periods of lower O3 concentrations, indicating the use of the 2nd highest values is likely
not an underestimate of typical O3 concentrations. Figure C-5 below illustrates the differences in the two possible
values for each given hour and month, for comparison.




43
     Data presented in this figure represents the 2nd highest value for a given hour and month to exclude any outliers.


Version: July 2022                                                                                               Page 39
                         Guidance on NO2 Dispersion Modelling in British Columbia




Figure C-5: Comparison Between Maximum and 2nd Highest Ozone Concentrations for the Rural Northeast British
Columbia Background Ozone Dataset

Note that some remote rural locations may have so little influence from O3 precursors that their O3 levels could be
considerably lower than a typical rural location. For example, Trinity examined the O3 monitoring data available at
three remote monitoring stations located in the Northwest Territories. Table C-2 below compares summary data at
these remote locations with the summary data for stations used in the rural dataset developed above. While this
comparison indicates that O3 levels are lower for more remote rural locations, it is important to note that the stations
are considerably farther north of the comparison rural sites in Northeast BC, and the higher latitude may contribute to
the lower O3 concentrations. Therefore, use of these remote monitoring stations should be considered on a case-by-
case basis for locations in the far north of BC.




Version: July 2022                                                                                            Page 40
                      Guidance on NO2 Dispersion Modelling in British Columbia


Table C-2: Ozone Data Comparison – Rural and Remote Monitoring Stations

                                                                                     75th         98th
         Site           Minimum        Maximum        Average        Median
                                                                                  Percentile   Percentile
                              Remote Monitoring Stations, Northwest Territories
 Fort Smith                0.00          71.00          24.01         24.60         30.10        41.00
 Inuvik                    0.00          68.40          24.03         24.50         30.50        40.50
 Norman Wells              0.50          53.50          21.91         21.60         28.50        41.40
                                  Monitoring Stations Used in Rural Dataset
 Blueberry Hill            0.00          60.70          25.83         26.40         34.90        48.30
 Farmington
                           0.00          63.90          27.87         28.70         36.90        50.20
 Community Hall
 Beaverlodge               0.00          96.00          28.29         29.00         36.44        50.00
 Rolla 213 Road            0.00          65.00          27.47         27.70         35.10        47.00




Version: July 2022                                                                                Page 41
                         Guidance on NO2 Dispersion Modelling in British Columbia



APPENDIX D: SUBSTITUTION METHODS FOR MISSING OZONE DATA

O3 datasets for Tier 3 NO2 methods (OLM and PVMRM) require a complete dataset with no missing hours. However,
O3 datasets have the potential to include gaps where data are missing for individual hours or groups of hours. These
gaps can result from the absence of monitoring activity, temporary equipment malfunction, or from QA/QC
procedures where requirements are not met. In the event that data are missing from O3 datasets, the following
methods should be used to fill in gaps in available data.

If single hour values are missing, missing O3 concentrations should be filled using linear interpolation between the
concentrations for the hour before and the hour after. The values for the hours before and after are added, and the
sum is divided by two.

In the event that multiple hours are missing, there are a variety of methods that may be appropriate for a given
dataset. Provided that sufficient data are available (75% data completeness for each quarter), the preferred method
of filling data would be to use the monthly maximum hourly concentration to fill in any remaining data gaps. For
example, for every multi-hour block of missing data in January, the data would be filled with the maximum O3
concentration among all monitored values in January across all years of the model domain.

In the event that the data completeness is not sufficient, the following list represents available gap-filling options that
are relatively simple to implement and result in a conservative estimate of the O3 concentrations:

        Maximum Hourly Concentration

              o    Select the highest hourly concentration over the entirety of the modelled period and use this value
                   to fill in all remaining missing hours.

        Maximum Annual Hourly Concentration

              o    Select the highest hourly concentration over each individual year of the modelled period and use
                   this value to fill in all remaining missing hours for the corresponding year in the model.

        Maximum Annual Average Hourly Concentration

              o    Select the highest hourly concentration over each individual year of the modelled period and use
                   the average of those values to fill in all remaining missing hours.

Applicants are encouraged to contact the Ministry if there is uncertainty about the quality of a background O3 dataset,
particularly if the use of a Ministry-prepared O3 dataset may be more appropriate for a given model.




Version: July 2022                                                                                                Page 42
