Guidelines on Choice of Models and Model Parameters

1. Introduction

1.1     Assessing air quality for Environmental Impact Assessment (EIA) in Hong Kong follows a three-tier approach (see “Guidelines on Assessing the ‘TOTAL’ Air Quality Impacts”). A number of models / modelling system have been used for this purpose. Three models are listed in Schedule 1 in the Annex and are referred to as Schedule 1 models. These models are currently accepted by EPD for general use in EIA.

1.2     The first Schedule 1 model is of the Gaussian plume type (Caline4), the next is a steady-state dispersion model (AERMOD). These can be used in relatively straightforward assessments and usually cover impacts from the first two tiers (project induced; pollutant-emitting activities in the immediate neighbourhood). The PATH model (last version: PATH v2.1) is a grid-based system and operates on a set of comprehensive emission data after generating its own meteorological data. This covers the third tier (background contribution). The meteorological output from the PATH model can be used to drive the above mentioned Caline4 and AERMOD models while the air pollutant concentration output from PATH can be taken as Tier 3 background contribution.

1.3     The Gaussian plume and steady state dispersion models in Schedule 1 have their user’s guide in the open literature and these should be referred to in any application. The following guidelines supplement the standard user’s guide in focusing on areas that are of common concern in an EIA in Hong Kong for these two models. PATH is constituted from open source modules with documentations available in the open literature. Specific information on the configuration of the PATH system and the public accessible data model output for EIA application can be found on http://www.epd.gov.hk/epd/english/environmentinhk/air/guide_ref/guide_aqa_model.html.

1.4     Given that the default approach to air quality assessment using model is model-output-based (see Sections 3.1.2-3 of the ‘Guidelines on Assessing the 'TOTAL' Air Quality Impacts’), the rest of this guidelines only deals with issues arising from this approach.

2. Model input requirements for Tier 1 and Tier 2 emission sources

2.1.1 At least 1 year of recent meteorological data (including wind speed, wind direction, stability class, ambient temperature, cloud base height, cloud cover and mixing height) from PATH should be used to determine the highest short-term (hourly, 8-hour, daily) and long-term (annual) air quality impacts at identified air sensitive receivers in the assessment period.

2.1.2 Alternatively, the meteorological conditions listed below can be used to determine the worst case short-term impacts:

• Day time: stability class D; wind speed 1 m/s (at 10m height); worst-case wind angle; mixing height 500 m
• Night time: stability class F; wind speed 1 m/s (at 10m height); worst case wind angle; mixing height 500 m

Apart from the above, any alternative approach that will capture the worst possible impact values (both short term and long term) may also be considered.

2.1.3 When using PATH's meteorological output to drive models for Tier 1 and Tier 2 assessment, the meteorological parameter from PATH, including wind speed and direction, should be taken as uniform over the whole relevant PATH grid cell(s) (i.e. the 3-dimensional volume) without further adjustment. In cases an anemometer height is required as input to models for Tier 1 and Tier 2 assessment, it should be set at the mid-layer height of the relevant PAT grid cell(s).

2.1.4 An additional parameter, namely, the standard deviation of wind direction, σ_θ, needs to be provided as input to the CALINE4 model. Typical values of σ_θ range from 12^o for rural areas to 24^o for highly urbanised areas under 'D' class stability. For semi-rural areas such as new development, 18^o is more appropriate under the same stability condition. The following reference (or more up-to-date version) can be consulted for typical ranges of standard deviation of wind direction under different stability categories and surface roughness conditions.

Ref.(1): Guideline On Air Quality Models (Revised), EPA-450/2-78-027R, United States Environmental Protection Agency, July 1986.

2.2 Surface Roughness Height (only required in CALINE4)

This parameter is closely related to the land use characteristics of a study area and associated with the roughness element height. As a first approximation, the surface roughness can be estimated as 3 to 10 percent of the average height of physical structures. Typical values used for urban and new development areas are 370 cm and 100 cm, respectively.

The surface characteristics parameters should be determined according to the methods specified in AERMOD implementation guide published by US-EPA.  The current recommendation is excerpted below:

• The determination of the surface roughness length should be based on an inverse distance weighted geometric mean for a default upwind distance of 1 kilometer relative to the measurement site. Surface roughness length may be varied by sector to account for variations in land cover near the measurement site; however, the sector widths should be no smaller than 30 degrees.
• The determination of the Bowen ratio should be based on a simple unweighted geometric mean (i.e., no direction or distance dependency) for a representative domain, with a default domain defined by a 10km by 10km region centered on the measurement site.
• The determination of the albedo should be based on a simple unweighted arithmetic mean (i.e., no direction or distance dependency) for the same representative domain as defined for Bowen ratio, with a default domain defined by a 10km by 10km region centered on the measurement site.

The exact approach in determining these parameters should be fully justified and agreed by EPD.

2.4.1 The conversion of NO_X to NO₂ is a result of a series of complex photochemical reactions and determines the prediction of near field impacts of NO_X emissions. Three approaches are currently acceptable in the determination of NO₂:

1. Discrete Parcel Method (DPM, available in CALINE4); 
2. Ozone Limiting Method (OLM); or
3. Empirical relationship for annual NO2:NOx ratio.

2.4.2 When applying DPM and OLM to determine motor vehicle impacts, a conservative 28% of tailpipe NO_X can be assumed to be NO₂ (on a mixing ratio basis) for all vehicle types. If further refinements are desirable, individual tailpipe NO₂ and NO_X output are available in EMFAC-HK model.

2.4.3  For other non-road sources, initial NO₂/NO_X ratios reported in the Heathrow Airport EIA report (http://www.cerc.co.uk/environmental-software/assets/data/doc_validation/ADMS-Airport_Adding Capacity_Air Quality.pdf) can be adopted. Otherwise, project specific ratios can be proposed with strong supporting evidence and should be fully agreed with EPD.

2.4.4  The hourly background ozone concentration for using DPM and OLM can be taken from PATH.

2.4.5 As an alternative, the annual NO2 concentrations can be estimated with project specific empirical relationship derived using the latest available and representative data from EPD’s air quality monitoring stations (AQMS). The empirical relationship should be described by a fitted curve of the selected annual NO2 and NOx monitoring data (Jenkin 2004a, Environment Agency UK 2007). The annual NO2 concentrations can be determined from the fitted curve using the total NOx concentrations. The use of project specific empirical relationship and the calculation details should be approved by EPD on a case-by-case basis.

Ref.(2): Jenkin M E, 2004a. Analysis of sources and partitioning of oxidant in the UK – Part 1: The NOx-dependence of annual mean concentrations of nitrogen dioxide and ozone. Atmospheric Environment, 38, 5117-5129.

Ref.(3): Environment Agency UK 2007. Review of methods for NO to NO2 conversion in plumes at short ranges (https://www.gov.uk/government/publications/review-of-methods-for-no-to-no2-conversion-in-plumes-at-short-ranges)

In assessing odour impacts, a much shorter time-averaging period of 5 seconds is required due to the shorter exposure period tolerable by human receptors. Conversion of model-computed hourly average results to 5-second values is therefore necessary to enable comparison against the recommended standard. The hourly concentration is first converted to 3-minute average value according to a power law relationship which is stability dependent (Ref. 4) and a result of the statistical nature of atmospheric turbulence. Another conversion factor (10 for unstable conditions and 5 for neutral to stable conditions) is then applied to convert the 3-minute average to 5-second average (Ref. 5). In summary, to convert the hourly results to 5-second averages, the following factors can be applied:

Stability Category	1-hour to 5-sec Conversion Factor
A & B	45
C	27
D	9
E & F	8

 

Under 'D' class stability, the 5-second concentration is approximately 10 times the hourly average result. Note, however, that the combined use of such conversion factors together with the AERMOD results may not be suitable for assessing the extreme close-up impacts of odour sources.

Ref.(4): Richard A. Duffee, Martha A. O' Brien and Ned Ostojic, 'Odor Modelling - Why and How', Recent Developments and Current Practices in Odor Regulations, Controls and Technology, Air & Waste Management Association, 1991.

Ref.(5): A.W.C. Keddie, 'Dispersion of Odours', Odour Control - A Concise Guide, Warren Spring Laboratory, 1980.

These include traffic emissions from tunnel portals and any other similar openings and are generally modelled as volume sources according to the PIARC 91 (or more up-to-date version) recommendations (Ref. 6, section III.2). For emissions arising from underpasses or any horizontal openings of the like, these are treated as area or point sources depending on the source physical dimensions. In all these situations, the AERMOD model will have to be used instead of the CALINE4 model.

Ref.(6): XIXth World Road Congress Report, Permanent International Association of Road Congresses (PIARC), 1991.

These include discrete receptors representing all the identified air sensitive receivers at their appropriate locations and elevations and any other discrete or grid receptors for supplementary information. A receptor grid, whether Cartesian or Polar, may be used to generate results for contour outputs.

Guided by a Working Group consisting of experts in air quality modelling, PATH has gone through extensive testing. It was determined that PATH's output of RSP and FSP concentrations should be adjusted as follows before being applied for EIA to account for the limited information on pollutant emissions on a larger scale:

• 10^th highest daily RSP concentration: add 11.0* μg/m³.
• Annual RSP concentration: add 10.3* μg/m³
• 19th and 36th highest daily FSP concentration: Nil
• Annual FSP concentration: add 3.5* μg/m3

* The above figures are based on 2015 observation data.

These adjustments are only applicable for PATH v2.1 and the figures may be modified by EPD based on the latest information and assessments.

The highest short-term and long-term averages of pollutant concentrations at prescribed receptor locations are output by the models and appropriately summed up for comparison with the relevant air quality standards specified for the corresponding pollutant. Contours of pollutant concentration are also required for indicating the general impacts of emissions over a study area.

Copies of model files in electronic format should also be provided for EPD's reference.

Modelling Section, Air Science and Modelling Group
Environmental Protection Department
January 2023

 

Air Quality Models Generally Accepted by Hong Kong Environmental Protection Department for Regulatory Applications