Model-extrapolated Estimates of Airborne
Lead Concentrations at U*S* Airports
£%	United States
Environmental Protect
Agency

-------
Model-extrapolated Estimates of Airborne
Lead Concentrations at U*S* Airports
This technical report does not necessarily represent final EPA decisions or
positions. It is intended to present technical analysis of issues using data
that are currently available. The purpose in the release of such reports is to
facilitate the exchange of technical information and to inform the public of
technical developments.
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
4>EPA
United States
Environmental Protection
Agency
EPA-420-R-20-003
February 2020

-------
Table of Contents
Abbreviations	2
Summary	3
1.	Introduction	6
1.1	Use of Leaded Avgas in Piston-Engine Aircraft	6
1.2	Lead Concentrations in Air from Leaded Avgas Use in Piston-Engine Aircraft at Individual Airports 7
1.3	Characterizing Maximum Impact Area Lead Concentrations from Piston-Engine Activity at U.S.
Airports	8
2.	Air Quality Modeling of Lead from Piston-Engine Aircraft at a Model Airport	10
2.1	Overview of Air Quality Modeling at a Model Airport	10
2.2	Air Quality Model Performance at a Model Airport	11
2.3	Yearlong Air Quality Modeling to Develop AQFs at a Model Airport	13
3.	Method to Calculate Model-Extrapolated Lead Concentrations Nationwide	15
3.1	Calculation of AQFs for Piston-Engine Aircraft Activity and Lead Concentrations	15
3.2	National Analysis Methods	18
3.3	Evaluation of Airports for Potential Lead Concentrations Above the Lead NAAQS	36
3.3.1	Sensitivity Analysis of Airport-Specific Parameters that Influence Potential for Lead
Concentrations to be Above the NAAQS 	36
3.3.2	Airport-Specific Activity Data	38
3.3.3	Airport-Specific Criteria for Identifying Potential Lead Levels Above the NAAQS	41
3.4	Characterization of Uncertainty of Cross-Airport Parameters that Influence the Potential for Lead
Concentratins to be Above the NAAQS for Lead	48
4.	Model-Extrapolated Lead Concentrations: Results and Uncertainty Characterization	52
4.1	Ranges of Lead Concentrations in Air at Airports Nationwide	52
4.2	Airports with Potential Lead Concentrations Above the Lead NAAQS with Unrestricted Access
Within 50 m of the Maximum Impact Site	59
4.3	Quantitative Uncertainty Analysis of Concentrations of Lead in Air at Airports: The Influence of
Run-up Time and Avgas Lead Concentration	63
4.3.1	National Analysis and Airport-Specific Monte Carlo Results	63
4.3.2	Comparison of Model-Extrapolated Concentrations From the Airport-Specific Activity Analysis
with Monte Carlo Bounds to Monitored Concentrations in the Maximum Impact Area	65
4.4	Qualitative Characterization of Uncertainty and Variability in Model-Extrapolated Lead
Concentrations from National and Airport-Specific Activity Analyses	68
4.4.1 Meteorological Parameters	68

-------
4.4.2	AERMOD and AERSURFACE Parameters
4.4.3	Operational Parameters	
References	
70
71
72
Appendix A. Supplemental Information on Detailed Air Quality Modeling at a Model Airport
Appendix B. Supplemental Data for Piston-Engine Aircraft Activity and Model-Extrapolated Lead
Contraction Gradients
Appendix C. Uncertainty Characterization
Abbreviations
Air Quality (AQ)
Air Quality Factor (AQF)
Air Taxi (AT)
Air Traffic Activity Data System (ATADS)
Airport Cooperative Research Program (ACRP)
American Meteorological Society/Environmental Protection Regulatory Model (AERMOD)
Clean Air Act (CAA)
US Environmental Protection Agency (EPA)
US Federal Aviation Administration (FAA)
General Aviation (GA)
General Aviation and Air Taxi Activity Survey (GAATA)
Landing and take-off operations (LTOs)
Multi-Engine (ME)
National Ambient Air Quality Standard (NAAQS)
National Academies of Sciences (NAS)
National Emissions Inventory (NEI)
One hundred octane low lead (100LL)
Reid-Hillview Airport of Santa Clara County (RHV)
Santa Monica Municipal Airport (SMO)
Single-Engine (SE)
Terminal Area Forecast (TAF)
Tetraethyl lead (TEL)
Touch-and-Go (T&G)
2

-------
Summary
The main objective of the analyses presented in this report is to estimate the potential ranges
of lead concentrations at and downwind of the anticipated area of highest concentration at
airports in the US. To accomplish this objective, the relationship between piston-engine aircraft
activity and lead concentration at and downwind of the maximum impact site at one airport
was applied to piston-engine aircraft activity estimates for each US airport. This approach for
conducting a nationwide analysis of airports was selected due to the dominant impact of
piston-engine aircraft run-up operations on ground-level lead concentrations, which creates a
maximum impact area that is expected to be generally consistent across airports. Specifically,
these aircraft consistently take-off into the wind and typically conduct run-up operations
immediately adjacent to the take-off runway end, and thus, modeling lead concentrations from
this source is constrained to variation in a few key parameters. These parameters include: 1)
total amount of piston-engine aircraft activity, 2) the proportion of activity conducted at one
runway end, 3) the proportion of activity conducted by multi-piston-engine aircraft, 4) the
duration of run-up operations, 5) the concentration of lead in avgas, 6) wind speed at the
model airport relative to the extrapolated airport, and 7) additional meteorological, dispersion
model, or operational parameters. These parameters were evaluated through sensitivity
analyses across airports or using quantitative or qualitative uncertainty analyses.
Results of the national analysis show that model-extrapolated 3-month average lead
concentrations in the maximum impact area range from less than 0.0075 |-ig/m3 up to 0.475
l-ig/m3 at airports nationwide. The range of model-extrapolated concentrations in the maximum
impact area aligns with expectations from previous monitoring at airports that showed
exceedances of the lead NAAQS in the maximum impact area of some airports.1 Results of the
national analysis also demonstrate and quantify the gradient in lead concentrations with the
highest concentrations in locations closer to the maximum impact area than those further
downwind.
For the subset of airports where estimated lead concentrations could potentially be above the
lead NAAQS, the analysis was further refined using a set of sensitivity analyses and airport-
specific data. This airport-specific analysis identified some airports where model-extrapolated
lead concentration estimates suggest the potential for piston-engine aircraft activity to cause
lead concentrations above the lead NAAQS in the area of maximum impact with unrestricted
public access. Lead concentration estimates in this analysis should not be used to evaluate
attainment of the lead NAAQS.
Overall, comparisons of both national and airport-specific model-extrapolated concentrations
to monitored values show general agreement and suggest that the extrapolation method
1 For additional information on monitoring data collected at airports see: https://www.epa.gov/regulations-
emissions-vehicles-and-engines/airport-lead-inventories-air-qualitv-monitoring-air.
3

-------
presented in this report provides reasonable estimates of the range in concentrations of lead in
air attributable to peak activity periods of piston-engine aircraft at airports. Uncertainty in the
national and airport-specific activity analyses were evaluated using a Monte Carlo analysis,
which characterized how variability in run-up duration and avgas lead concentrations influence
model-extrapolated lead concentrations. Results showed that model-extrapolated lead
concentrations may increase at airports with average run-up durations that are longer than the
average run-up duration observed at the model airport, even if the avgas lead concentration is
lower than that used in the national analysis. Additional, qualitative analyses were used to
evaluate sources of uncertainty that were not addressed in sensitivity or Monte Carlo analyses.
Quantitative and qualitative evaluations of meteorological parameters that can impact model-
extrapolated concentrations focused on adjusting concentrations to reflect site-specific wind
speeds (See Section 3.2 for details) and evaluating changes in wind direction, mixing height, and
temperature. While the wind speed adjustment did not meaningfully impact the range of
concentrations in the maximum impact area of US airports, this adjustment does have an
important impact on model-extrapolated concentrations at individual airports, particularly at
those airports where wind speeds during the maximum activity period differ significantly from
those observed at the model airport. As discussed in Section 4.4.1, minimal uncertainty is
expected in model-extrapolated concentrations due to shifts in wind direction given that most
airports are built with the predominate runway facing into the wind. It is also anticipated that
mixing height has a minimal impact on uncertainty in model-extrapolated concentrations at the
maximum impact area, because of the dominant impact of the very localized run-up emissions
at this location and the fact that GA and AT aircraft activity occurs almost entirely during the
day when vertical mixing is greatest. At downwind locations, mixing height may play a larger
role and would be an important variable to examine when evaluating individual airports,
particularly those with mixing height characteristics significantly different from the model
airport. Finally, ambient temperature and other microclimate or meteorological variables are
not expected to meaningfully impact nationwide results, however, there is more uncertainty in
model-extrapolated concentrations at airports that have maximum activity periods during
meteorological conditions not observed at the model airport.
Additional sources of potential uncertainty that were evaluated qualitatively included
dispersion modeling inputs and operational parameters. While dispersion modeling inputs such
as surface roughness, Bowen Ration, and albedo may result in some uncertainty at downwind
locations, their impact on variability near the maximum impact site is mitigated due to
consistency in on-airport characteristics and land-use requirements immediately downwind of
runways based on landing and take-off safety requirements. As with meteorological
parameters, the appropriateness of dispersion modeling inputs used in this analysis for
individual airports with meaningful differences in land use of the areas immediately
surrounding a runway would need to be considered on a case-by-case basis. Differences in
operational parameters (e.g., piston/turboprop split and single-engine/multi-engine split,
distribution of aircraft engine types operating at the airport, diurnal activity patterns) are not
expected to contribute significantly to uncertainty in extrapolated concentration estimates for
4

-------
airports nationwide; however, in modeling individual airports, national fleet and operational
data should be supplemented with local data where available and feasible.
The model-extrapolated lead concentrations provided in this report reflect only lead
concentrations in air attributable to piston-engine aircraft activity and only at the area of
maximum concentration and downwind of that location. Additional analyses, which are outside
of the scope set by the objective of this report, would be necessary to evaluate concentrations
of lead in air at other areas at and near airports. In addition, to understand total lead
concentrations in air, other airborne sources of lead (e.g., nearby industrial sources, sources
contributing to local background concentrations) would need to be considered. Understanding
total lead exposure, which is relevant for understanding blood lead levels, would also need to
consider exposure to lead from additional media (e.g., soil, drinking water).
5

-------
1. Introduction
The United States (US) Environmental Protection Agency (EPA) is evaluating the air quality
impact of emissions of lead from piston-engine aircraft operating on leaded fuel. One
component of the evaluation includes conducting an analysis of concentrations of lead in air at
and downwind of airports. This analysis was conducted to provide an understanding of the
potential range in lead concentrations in air at the approximately 13,000 airports with piston-
engine aircraft activity in the US. This report describes the methods that the EPA used to
estimate these lead concentrations and presents the results of this analysis along with a
quantitative uncertainty analysis. Background information is presented immediately below in
order to provide a general understanding of the use of leaded fuel in aircraft, and the state of
the science on modeling concentrations of lead in air from aircraft emissions at individual
airports. Subsequent sections provide details on the analysis approach for airports nationwide.
1.1 Use :led Avgas in Piston-Engine Aircraft
Emissions of lead from aircraft operating on leaded aviation gasoline (avgas) are the largest
source of lead released into the atmosphere in the US, accounting for 62% of lead (456 tons) in
the 2014 National Emissions Inventory (NEI) (USEPA 2016a). Leaded avgas is used in piston-
engine aircraft, of which there are approximately 140,000 in the US (FAA 2014) . These aircraft
operate at most of the approximately 20,000 US airport facilities (approximately 13,000 of
which are airports, while the remainder are heliports, balloon ports, and other facility types)
(FAA 2017).2,3 Piston-engine aircraft conduct approximately 32 million landing and take-off
operations (LTOs) annually (USEPA 2011).4 Most piston-engine aircraft operations fall into the
categories of either General Aviation (GA) or Air Taxi (AT) activity. GA is defined as the
operation of civilian aircraft for purposes other than commercial, such as passenger or freight
transport, including personal, business and instructional flying; AT is scheduled or on-demand
services that carry limited payload and/or passengers (FAA 2012).
Piston-engine aircraft rely on lead as an additive to avgas to help boost fuel octane and prevent
engine knock, as well as prevent valve seat recession and subsequent loss of compression for
2	This report focuses on fixed-wing piston-engine airplane activity at airports. Facility types other than airports are
not included in this report; seaports and water runways at airports are both excluded from analyses in this report,
and rotorcraft operations at airports are not included in this report. Appendix B provides some information on
conducting additional rotorcraft analyses in the future.
3	Data on airport facilities was downloaded from FAA Air Traffic Activity Data System (ATADS) at
http://aspm.faa.gov/opsnet/svs/Airport.asp on 13 February 2014.
4	Piston-engine aircraft conduct two types of operational cycles, or cycle-types. These cycle-types include: 1) a full
landing-and-take-off operation (full LTO) during which the pilot conducts all pre-flight engine checks and
completes full take-off and landing operations, and 2) a touch-and-go operation (T&G) during which the pilot
briefly touches down on a runway before taking-off again almost immediately in order to practice take-off and
landing procedures. This is a training exercise most commonly performed by student pilots. Throughout this
report, "cycle-type" is used to refer to the full LTO and T&G categories, while "LTOs" is used to refer more
generally to all cycle-types (i.e., both full LTO and T&G).
6

-------
engines without hardened valves.5 Lead is added to the fuel in the form of tetraethyl lead (TEL)
along with ethylene dibromide, which acts as a lead scavenger to prevent lead deposits on
valves and spark plugs. Currently one hundred octane low lead (100LL), which contains up to
2.12 grams of lead per gallon, is the most commonly used type of avgas in the US, although FAA
survey data reports limited use of a leaded avgas containing 4.24 grams of lead per gallon,
known as "100 Octane," and unleaded avgas (FAA 2015). Lead is not added to jet fuel, which is
used in commercial aircraft, most military aircraft, and other turbine-engine aircraft.
1.2 L sncentrations in Air from Leaded Avgas Use in Piston-Engine Aircraft at
Individual Airports
Lead emissions from piston-engine aircraft operating on leaded avgas increase concentrations
of lead in air at and downwind of airports (Environment Canada 2000, Fine et al. 2010, Carr et
al. 2011, Anchorage DHHS 2012, Feinberg et al. 2016). Gradient studies evaluating lead
concentrations near airports where piston-engine aircraft operate indicate that concentrations
of lead in air are one to two orders of magnitude higher at locations proximate to aircraft
emissions compared to locations approximately 500- to 1000-meters downwind (Fine et al.
2010, USEPA 2010a, Carr et al. 2011, Feinberg et al. 2016). The most significant emissions in
terms of ground-based activity, and therefore ground-level concentrations of lead in air, occur
near the areas with greatest fuel consumption where the aircraft are stationary for a period of
time (USEPA 2010a, Carr et al. 2011, ICF 2014, Feinberg et al. 2016). For piston-engine aircraft
these areas are most commonly locations in which pilots conduct engine tests during run-up
operations prior to take-off (i.e., magneto checks during the run-up operation mode). Run-up
operations are typically conducted adjacent to the runway end from which aircraft take-off and
the brakes are engaged so the aircraft is stationary.6 As a result of the aircraft being stationary,
duration of run-up, and high fuel consumption rate, emissions from run-up activity are the
largest contributor to local maximum atmospheric lead concentrations; run-up emissions are
estimated to contribute over 80% of the lead concentrations at and immediately downwind of
the area where the run-up mode of operation occurs, even though this mode of operation does
not have the highest fuel consumption rate (Appendix A). Hence, the area adjacent to the
runway end at which run-up operations most frequently occur is identified here as the
maximum impact site for lead concentrations.7,8
5	Minimum octane requirements as well as other carefully controlled fuel parameters in avgas prevent the general
use of unleaded motor vehicle fuel in piston-engine aircraft.
6	A single "runway" has a magnetic heading designation for each "runway end" in order to distinguish which
direction the aircraft is taking off from or landing on to; we use "runway end" throughout this report.
7	For purposes of this report and the underlying analysis, the maximum impact site is defined as 15 meters
downwind of the tailpipe of an aircraft conducting run-up operations in the area designated for these operations
at a runway end. The maximum impact area is the approximately 50 meters surrounding the maximum impact site.
The downwind gradient is the approximately 500-meter area that extends from the maximum impact site.
Additional characterization of the maximum impact site, area, and downwind gradient is provided in Section 2.
8	While run-up operations are most frequently the location of the maximum impact site of aircraft lead emissions
at airports, at some airports other operations such as taxi or idling near the runway may result in a hotspot of
emissions. This report focuses on run-up as the location of the maximum impact site in an effort to characterize
concentrations of lead in air at the location of maximum impact for most US airports. Additional analyses would be
necessary to more specifically characterize concentrations of lead in air at individual airports.
7

-------
1.3 Characterizing Maximum Impact Area Le icentrations from Piston-Engine
Activity at U.S. Airports
The understanding of piston-engine aircraft lead emissions and resulting concentrations in air
was developed through detailed monitoring and modeling studies at individual airports.
However, conducting detailed air quality monitoring or modeling for lead at each of the 13,000
US airports is not feasible; thus, the analysis of concentrations of lead in air at and downwind of
airports nationwide is based on detailed air quality modeling at a representative, model airport.
The modeling results were used to develop factors that relate piston-engine aircraft activity to
concentrations of lead in air. The factors, termed Air Quality Factors (AQFs), were used in
conjunction with estimates of piston-engine aircraft activity at airports nationwide to calculate
model-extrapolated concentrations at and downwind of each US airport.
The rationale for this approach is based on the consistent set of parameters required for the
safe operation of a piston-engine aircraft. Specifically, piston-engine aircraft consistently
conduct run-up operations prior to take-off, and the run-up activity has the following
characteristics: 1) run-up operations require high fuel consumption rates while the aircraft is
stationary, and thus are the location of the maximum impact site for lead concentrations, 2) the
location of run-up activity occurs in a designated area proximate to the runway end from which
aircraft take-off, and 3) the runway end used for take-off, and hence the location of run-up
operations, can be identified using wind direction since piston-engine aircraft takeoff into the
wind.
This analysis focuses on the maximum impact areas at airports nationwide (i.e., the 50 meters
surrounding the maximum impact site adjacent to run-up operations). Notably, the maximum
impact area lead concentration estimates provided in this report are based on average values
for several key input variables; thus, the concentrations are not "worst-case" estimates (i.e.,
they do not reflect the use of the maximum values for all the key input parameters). For each
US airport, model-extrapolated lead concentrations are calculated as 3-month average values
to maintain consistency with the form of the National Ambient Air Quality Standard (NAAQS)
for lead (i.e., a maximum 3-month average of 0.15 |-ig/m3) (National primary and secondary
ambient air quality standards for lead 40 CFR 50.12, USEPA 2016b). Importantly, while model-
extrapolated concentrations are calculated and presented in a manner consistent with the lead
NAAQS, these results should not be used to determine attainment of the lead NAAQS at
individual airports. Information on the process that EPA, the states, and the tribes follow to
determine whether or not an area is meeting the NAAQS for lead is described on the EPA
website (USEPA). Lead concentration estimates presented in this report are provided to inform
an understanding of the potential range of impacts that lead emissions from piston-engine
aircraft alone may have on air quality in close proximity to this source of lead. Due to the
inherent uncertainties in extrapolating relationships between concentration and activity from
one well-characterized model airport to others, uncertainty and variability in model-
extrapolated lead concentrations is characterized.
8

-------
This document is organized to first provide the methods and results of detailed air quality
modeling of lead at a model airport (Section 2). Section 3 describes how the modeling results
were used to develop a quantitative relationship between piston-engine aircraft activity and
lead concentrations; this section further provides the methodology to estimate piston-engine
aircraft activity at airports nationwide, which is used to calculate lead concentrations at airports
nationwide based on the relationship between activity and lead concentrations. Section 3 also
presents methods to identify a subset of airports for more in-depth analyses using airport-
specific data. Section 4 presents the model-extrapolated lead concentrations that result from
combining piston-engine aircraft activity estimates with the relationship between activity and
lead concentrations in the maximum impact area and locations downwind at each airport
nationwide. In addition, Section 4 characterizes uncertainty and variability in these model-
extrapolated lead concentrations.
9

-------
2. Air Que lodeling of Lead from Piston-Engine Aircraft at a Model Airport
To characterize concentrations of lead in air at and downwind of the maximum impact area of
airports nationwide, EPA first conducted detailed air quality modeling at a model airport. The
results of this detailed air quality modeling were used to develop factors, known as AQFs, which
provide quantitative relationships between piston-engine aircraft activity and lead
concentrations at and downwind of the maximum impact site at the modeled airport. The AQFs
were subsequently applied to estimates of aircraft activity at other airports across the country
in order to calculate model-extrapolated lead concentrations at and downwind of the
maximum impact area of airports nationwide. In this section we briefly explain the overall
approach for the detailed air quality modeling at the model facility, summarize the model
performance, and then discuss how the air quality modeling was conducted to develop the
AQFs.
2.1 Overview of Air Quality Modeling at a Model Airport
In order to characterize local-scale air quality impacts of lead at a model airport, EPA applied
the air quality model that is used for EPA and Federal Aviation Administration (FAA) regulatory
analysis of near-field gradients of primary pollutants such as lead, namely the American
Meteorological Society (AMS)/EPA Regulatory Model (AERMOD).9 10 Since AERMOD had not
been previously applied to modeling lead emissions from piston-engine aircraft activity, EPA
developed the necessary model inputs and parameters, including: piston-engine aircraft
parameters (i.e., sub-daily time-in-mode activity, dispersion due to aircraft turbulent wake,
allocation of approach and climb-out emissions at altitude) and emissions characteristics of
non-aircraft sources (e.g., nearby roads) (USEPA 2010a, Carr et al. 2011). These model inputs
were developed and first applied at a GA airport (Santa Monica Airport, SMO) that was selected
due to the availability of previously collected lead monitoring data, which indicated elevated
concentrations of lead in air at and near the runway (Fine et al. 2010). Additional monitoring
data were collected in parallel to the development of AERMOD modeling inputs in order to
evaluate model performance. Details regarding the AERMOD inputs, model performance, and
results are published elsewhere (USEPA 2010a, Carr et al. 2011).
The foundational work to establish AERMOD inputs for modeling lead emissions from piston-
engine aircraft at SMO provided an understanding of the key characteristics of the relationship
between aircraft activity and concentrations of lead in air. Some of the key findings from this
work, included: 1) piston-engine aircraft operations increase ground-level concentrations of
lead, with the largest concentrations resulting from engine checks prior to take-off (i.e., run-up
operations), 2) lead concentrations attributable to piston-engine aircraft decrease with
9	AERMOD is a steady-state plume model that incorporates air dispersion based on planetary boundary layer
turbulence structure and scaling concepts, including treatment of both surface and elevated sources, and both
simple and complex terrain. Additional details about AERMOD are available at: https://www.epa.gov/scrarn/air-
qualitv-dispersion-modeline-preferred-and-recommended-models
10	The FAA inventory tool for air emissions and noise, Aviation Environmental Design Tool (AEDT), does not include
lead emissions (https://aedt.faa.gov/Documents/AEDT 2b NEPA Guidance.pdf).
10

-------
increasing distance from the run-up location, such that the maximum impact location is
immediately adjacent to the run-up area at a runway end, and 3) above-background lead
concentrations occur up to 900 and 450 meters downwind of the maximum impact location on
a daily and average 3-month basis, respectively (USEPA 2010a, Carr et al. 2011). The National
Academies of Sciences (NAS) Airport Cooperative Research Program (ACRP) subsequently
conducted a similar study of airport lead concentrations at three airports and similarly
identified run-up as a critical operation mode to evaluate when modeling the impact of piston-
engine aircraft lead emissions on ground-based lead concentrations (Heiken et al. 2014,
Feinberg et al. 2016). These findings presented a clear approach for conducting air quality
modeling at an airport, which would be used as a model facility for developing AQFs and
subsequently characterizing concentrations of lead in air at and downwind of airports
nationwide.
Reid-Hillview Airport of Santa Clara County (RHV) was selected as a representative GA airport
for use as the model airport.11 To apply AERMOD at the model airport, aircraft and
meteorological data, similar to those collected at SMO, were collected at RHV. Specifically, data
collected at this facility included: 1) number and type of piston-engine aircraft LTOs, 2) time in
each operating mode, 3) time-of-day and day-of-week patterns of aircraft activity, 4) the
concentration of lead in avgas, and 5) meteorological data (i.e., wind direction, wind speed,
mixing height, temperature). These inputs were collected first for a seven-day period in order
to characterize model performance at the model airport through comparisons of modeled and
monitored concentrations. After characterizing model performance, additional activity and
meteorology data were collected to model a yearlong period, which was then used to develop
AQFs. Information on model performance at the model facility is presented immediately below
in Section 2.2; information on the yearlong modeling is in Section 2.3. Appendix A provides
details on specific AERMOD inputs at the model airport study, as well as information regarding
the piston-engine aircraft modeled at the model airport compared to the national piston-
engine aircraft fleet.
2.2 Air Quality Model Performance at a Model Airport
Comparisons of modeled and monitored daily average concentrations at the model airport
were conducted over a seven-day period at three monitoring sites (upwind, 60 meters
downwind, and at the maximum impact site). The daily average was over 15 hours, from the
hours of 7 a.m. to 10 p.m. local time, representing the time when the airport was operational.
The overall R2 value across the three monitoring sites regressed against the paired modeled
concentrations was 0.83, as shown in Figure 1. At the maximum impact site, the model tended
to under-predict monitored concentrations for the seven days of comparison conducted, but
was generally within 20% of monitored values and was within the 2:1 and 1:2 lines for all but
11 RHV is considered generally representative of GA airports based on several factors, including: type of piston-
engine aircraft operations, runway configuration, fleet composition of piston-driven aircraft engine technology
types, and diurnal profile of piston-engine aircraft activity (see Appendices A and B for comparisons of RHV fleet
and diurnal profiles relative to other GA airports).
11

-------
one monitored value.12 The generally good agreement between modeled and monitored
concentrations was also observed in previous studies comparing AERMOD air quality dispersion
model output with on-site monitoring data for lead at airports (Carr et al. 2011; Feinberg et al.
2016). As observed in these other studies, modeled lead concentrations can be both slightly
over- and underestimates of on-site monitored values, and the performance observed for the
model airport is considered to be aligned with prior work. We focused on understanding
discrepancy between modeled and monitored concentrations on the few days when the
discrepancy was greater than 20%. For these days, sensitivity analyses were conducted to
identify possible reasons for the divergence. Details on the sensitivity analyses are presented in
Appendix A, but generally showed that run-up location, run-up duration, and relative levels of
multi-engine aircraft activity explained instances when the model under- or over-predicted
monitored concentrations; uncertainty and variability in monitored values are not evaluated
here, but also contribute to the divergence in these comparisons with modeled data. In
addition, variability in emission rates for a given engine and across engine types will also
contribute to variability in measured concentrations, as discussed in Section 4.4. The
application of a 3-month averaging time is expected to minimize the impact of individual days
in which the model may have over- or under-predicted lead concentrations. Comparisons
between model-extrapolated concentrations, based on the AQFs developed at the model
airport, and monitored concentrations at airports other than the model airport are presented in
Section 4.
12 Agreement with monitored concentrations within a factor of two is a common model evaluation criterion Chang,
J. and S. Hanna (2004). Air quality model performance evaluation. Meteorology and Atmospheric Physics, 87 (1),
167-196, Luecken, D., W. Hutzell and G. Gipson (2006). Development and analysis of air quality modeling
simulations for hazardous air pollutants. Atmospheric Environment, 40 (26), 5087-5096.
12

-------
Model-to-Monitor Comparison
0.60 t
O
¦M-
¦H
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Monitored Pb Cone., |ig/m3
O Maximum Impact Site
A Upwind
O 60m Downwind
	1:1
- -2:1 and 1:2
Figure 1. Comparison of modeled and monitored daily average concentrations at three sites at the
model airport during a 7-day period.
The model performance at the model airport confirmed previous work showing that a limited
set of parameters influence concentration in the maximum impact site, and supported moving
forward with the development of AQFs to characterize the relationship between piston-engine
aircraft activity and lead concentrations at and downwind of a maximum impact area.
2.3 Yearlong Air Quality Modeling to Develop AQFs at a Model Airport
This section provides general information used to model yearlong concentrations of lead in air
that were subsequently used to calculate 3-month average AQFs at the model facility. Details
regarding inputs to AERMOD including aircraft emission inventories, source parameterization,
meteorological inputs, and receptor placement are provided in Appendix A.
As noted above, air quality modeling for this work built on prior piston-engine aircraft modeling
in which aircraft- and airport-specific parameterizations were used in AERMOD to evaluate
near-field gradients in ambient lead concentrations. Inputs in the yearlong modeling included 1)
a detailed inventory for emissions of lead from piston-engine aircraft (i.e., aircraft activity,
source locations, and lead emission rates), 2) meteorological data, 3) a dense receptor grid, and
4) piston-engine aircraft characterization and parameterization. Using previously published
modeling methods, which are further described in Appendix A, Section 1.5, aircraft lead
emissions were modeled as volume sources. The parameterization of aircraft lead emissions at
the model airport included aircraft wake turbulence, and plume rise from ground-based aircraft
emissions. Specific values for the initial vertical and horizontal dispersion by operation mode
are provided in Appendix A.
13

-------
Aircraft activity data for the yearlong modeling at the model facility used on-site observations
in conjunction with on-site daily operations data collected by FAA.13 Hourly aircraft activity
profiles were developed from on-site observations for single-engine and multi-engine aircraft
conducting either full landing and take-off or touch-and-go operation cycles. Time spent in each
mode (i.e., start-up, idle, taxi, run-up, take-off and landing) was recorded during the days of
observation and was used along with fuel consumption rates by mode to calculate emissions by
mode. Source locations for all modes of aircraft activity (i.e., start-up, idle, taxi, run-up, take-off
and landing) are described in Appendix A; emissions at altitude were represented using volume
sources at 50-meter intervals up to approximately 500 meters and release heights for ground-
based activity were 0.5 meters.
Surface and upper-air meteorological data (from stations 10 km, and 55 km away from the
model facility, respectively) were processed using AERMOD's meteorological preprocessor,
AERMET, to produce hourly data on mixing heights, stability, wind direction, wind speed,
temperature, and precipitation. The wind direction data were used to identify the runway end
from which piston-engine aircraft took off during each hour of each day in the year of modeling
(2010). Surface characteristics and AERSURFACE parameterization are described in the
Appendix A.
To identify the spatial extent of elevated lead concentrations within the vicinity of the airport,
2,250 receptor locations were used, with the most densely located receptors placed at 50-
meter intervals at and near ground-based aircraft activity, as well as out to 1 km downwind
from run-up and take-off activity. Receptor spacing was at 100-meter intervals at other
locations within the 1 km perimeter of the runway centroid, and increased to 200 meters after
2 km.
Results of the yearlong model run provided daily lead concentrations at and downwind of the
maximum impact site that are attributable to piston-engine aircraft activity (i.e., do not include
background lead concentrations from other sources). These daily average lead concentrations
were used to calculate 3-month, rolling-average lead concentrations. As detailed in Section 3
below, the 3-month, rolling average lead concentrations were then used to calculate AQFs that
relate piston-engine aircraft activity over 3-month periods to lead concentrations at and
downwind of the maximum impact site. The combination of the AQFs and activity estimates at
other US airports provides model-extrapolated lead concentrations for a national analysis of
lead concentrations at and downwind of maximum impact areas at airports nationwide.14
13	As discussed in Section 3, FAA data does not indicate which aircraft operations are conducted by piston-engine
aircraft, compared to turboprop or other engine types. Rather activity is reported as specific to GA or AT, which
can be used to estimate activity specific to piston-engine aircraft based on national averages or airport-specific
data. For the model airport, data collected at the airport during the model-to-monitor comparison evaluation
provided inputs to appropriately allocate GA and AT aircraft activity to piston-engine activity. For additional
information see Appendix A.
14	As stated in Section 1 we define maximum impact site as the 15 meters immediately adjacent to run-up and the
maximum impact area as the 50 meters surrounding the maximum impact site. 'Maximum impact site' is used in
14

-------
3. Method to Calculate Model-Extrapolated Lead Concentrations Nationwide
In this section we discuss the methods for calculating model-extrapolated lead concentrations
at US airports. Section 3.1 provides the AQFs developed from the yearlong air quality modeling
at the model airport discussed above. Section 3.2 provides the methodology for estimating
activity at each airport and shows how we use activity estimates for each airport in
combination with the AQFs to develop a national analysis of model-extrapolated
concentrations of lead attributable to piston-engine aircraft at and downwind of the maximum
impact area at approximately 13,000 US airports. This national analysis uses US average
statistics for the fraction of GA and AT activity conducted by piston-engine aircraft. This analysis
is further refined using airport-specific data for a subset of airports as described in Section 3.3.
Section 3.4 then describes quantitative Monte Carlo uncertainty analyses for both the national
and airport-specific analyses.
3.1 Calculation of AQFs for Piston-Engine Aircraft Activity and Lead Concentrations
The AQFs were calculated for the different piston-engine aircraft cycle types and engine classes.
Specifically, piston-engine GA and AT aircraft perform two types of operational cycles: 1) full
LTOs, in which aircraft start or end the operation in a full stop outside of the active runway, and
2) T&Gs, in which aircraft land and take-off without coming to a full stop.15 Further, fixed-wing
piston-engine GA and AT aircraft can be subdivided into two classes, single-engine (SE) and
multi-engine (ME) planes. Due to differences in fuel consumption and time in each operational
mode between aircraft classes and cycle-types, respectively, an AQF was calculated specific to
each aircraft class (i.e., single- or multi-engine, SE or ME) and cycle-type (i.e., full LTO or T&G).
Accordingly, four different types of AQFs (i.e., SE full LTO, SE T&G, ME full LTO, ME T&G) were
calculated for nine specific receptor sites at and downwind of the maximum impact site, which
was the runway end at which LTOs most frequently occurred at the model airport facility. The
AQFs are calculated as the ratio of the average lead concentration over rolling 3-month time
periods to piston-engine aircraft LTOs at the most frequently used runway end over the same 3-
month period.16 For example, the SE full LTO AQF at the maximum impact site is the ratio of the
3-month average modeled lead concentration (|ag/m3) attributed to SE LTO at the model airport
maximum impact site and the number of full LTOs conducted by SE piston aircraft at the most
frequently used runway end in the same 3-month period (Equation l).17
the context of the model airport and 'maximum impact area' is used in the context of airports for which we
calculated model-extrapolated lead concentrations.
15	As noted in Footnote 3, for simplicity, both types of LTOs (i.e., full LTO and T&G) are referred to as LTOs, while
"cycle-type" is used to denote the categories of full LTO and T&G.
16	As noted in Section 1, this analysis uses 3-month average lead concentrations to allow for comparisons with the
3-month average concentration set for the lead NAAQS USEPA (2016b). Review of the National Ambient Air Quality
Standards for Lead EPA-HQ-OAR-2010-0108; FRL-9952-87-OAR.
17	Both full LTO and T&G AQFs include concentration attributable to emissions from aircraft operating in all modes
(e.g., taxi, take-off, run-up), with the exception that T&G AQFs do not include the lead concentration due run-up
emissions.
15

-------
3-month average modeled lead concentration ()
Eq. 1: SE full LTO AQF at maximum impact site =	^
# of full SE LTOs during 3-month period
The specific steps to calculate AQFs at and downwind of the maximum impact site are:
1.	Calculate average modeled daily lead concentrations at each of the nine receptor locations
over fourteen consecutive one-month periods separately for emissions from each aircraft
class and cycle-type (e.g., SE T&G, ME full LTO).
2.	Calculate rolling 3-month average modeled lead concentrations at each of the nine receptor
locations by averaging across monthly average concentrations attributable to each aircraft
class and cycle-type (e.g., SE T&G, ME full LTO).
3.	Sum piston-engine activity by cycle-type and aircraft class (e.g., SE T&G, ME full LTO) in the
3-month periods.
4.	Divide each 3-month average ambient lead concentration at each receptor site for each
cycle-type and aircraft class by the corresponding total number of LTOs separated by cycle-
type and aircraft class (e.g., ambient lead concentration from SE full LTO emissions at 50 m
during July - Sept. 2011 / # of SE full LTOs during July - Sept. 2011).
5.	Calculate the average AQF across the 12 rolling 3-month periods separately for each aircraft
class and operation-type pair at each of the nine receptor locations (e.g., average of the 12,
3-month AQFs for SE full LTOs at the 50-meter receptor site).
As Steps 1 through 4 above describe, for each aircraft class and operation-type pair 12 AQFs
were calculated for each set of 3 consecutive months in a 14-month period. The set of 12 AQFs
for each aircraft class and operation type were used to evaluate variability in AQFs due to
changes in meteorology over a 14-month period.18 In order to average across the largest range
in meteorology inputs to AQFs (e.g., wind speed), the resulting 12 AQFs were averaged to
provide a single 3-month AQF for each aircraft class, operation-type, and location combination
(Table 1). The extent to which meteorology variability included in the modeling to calculate
AQFs is representative of the range of meteorology at airports across the country is discussed
further in Section 4.
18 Variation in the rolling 3-month average AQFs for full LTOs is generally +/-25% of the mean across all 12 AQFs.
Specifically, rolling 3-month average AQFs for SE full LTOs vary from 28% greater to 14% less than the associated
mean AQFs. For ME full LTOs, the individual rolling 3-month AQFs vary from 23% greater to 13% less than the
associated mean AQFs. The variation is consistent across locations. While ME aircraft typically have two engines,
ME AQFs are more than double the equivalent SE AQFS due to greater fuel consumption of their engines and
differences in time-in-modes. The T&G AQFs are one to two orders of magnitude smaller than the full LTO AQFs in
the same location, and variability between AQFs is somewhat larger by percentage (46% greater to 16% less than
the associated mean AQFs) but smaller in absolute terms.
16

-------
Table 1. Average of the 12 rolling 3-month AQFs (ng Pb/m3/LTO) at and downwind of the maximum
impact site19
AQFs
Distance (meters)
Max
Impact
Site
50 m
100 m
150 m
200 m
250 m
300 m
400 m
500 m
SE Full
LTO
1.5x10 s
3.5x10 s
1.6x10 s
1.1x10 s
9.2xl0"7
7.6xl0"7
5.5xl0"7
4.0xl0"7
2.9xl0"7
SE T&G
1.7xl0"7
1.6xl0"7
1.7xl0"7
1.3xl0"7
1.2xl0"7
l.OxlO"7
8.0x10 s
6.1xl0"8
5.5xl0"8
ME Full
9.0xl0"5
2.3xl0"5
l.lxlO"5
8.2x10 s
6.6x10 s
5.5x10 s
4.0x10 s
3.0x10 s
2.2x10 s
ME T&G
6.8xl0"7
5.0xl0"7
4.5xl0"7
3.3xl0"7
2.7xl0"7
2.2xl0"7
1.7xl0"7
1.3xl0"7
1.2xl0"7
When each AQF is multiplied by the number of corresponding LTOs (full LTOs or T&Gs) that
occur at the most frequently used runway end during a 3-month period, the sum of the
products equals the lead concentration over the 3-month period at each of the nine locations.
The concentration of lead in air, [Pb]Air, is calculated by Equation 2, where Avgas[Pb] is the
concentration of lead in fuel and PA is piston activity for the given engine and operation type.
The next section describes how the number of piston-engine LTOs, specific to aircraft class and
operation-types, was estimated for each US airport in order to calculate 3-month average
model-extrapolated concentrations of lead in air at each airport.
Eq. 220-21:
[ Pbl Air=
Avgas[Pb]
^~pf [(PASe,fuNxAQFSe, fum)+(pase, t&gxAQFse, t&g)+(pame, fuNxAQFMe, fun)+(pame, t&gxAQFMe, t&g)]
2.12
gal
19	Additional information on the relationships between AQFs and distances downwind is available in Appendix C.
20	Per the description in the above text, the concentration of lead in air is calculated at nine distances starting
immediately adjacent to run-up out to 500 meters downwind.
21	The scalar for the concentration of lead in avgas is used to normalize the lead concentration to the ASTM
specification for 100 LL (ASTM International (2016). Standard Specification for Leaded Aviation Gasolines.
https://compass.astm.org/EDIT/html annot.cgi?D910+19). The impact of variability in avgas lead concentrations
on model-extrapolated lead concentrations is discussed in Section 3.4.
17

-------
3.2 National Analysis Methods
This section summarizes the approach and rationale for the national analysis of lead
concentrations at and downwind of the maximum impact area at US airports. At a high-level,
this approach entails estimating piston-engine aircraft activity at each runway end of each
airport, and then combining activity estimates from the most actively used runway end in a 3-
month period with the AQFs presented in the previous section. The following text describes, in
brief, the methods used to estimate 3-month maximum piston-engine aircraft activity at each
runway end for airports nationwide; the detailed methods for this analysis are provided in
Table 2.
Airport-specific piston-engine aircraft activity data are not collected by FAA or reported by
airports in a national data source. Rather, piston-engine aircraft activity is reported by FAA as
part of GA and AT activity, which can also include jet-engine aircraft activity. To estimate
piston-engine activity, we used national datasets as described in Appendix B and FAA survey
data regarding the national average for number of hours flown by piston-engine GA or AT
aircraft nationwide.22 Specifically, the percent of hours flown by piston-engine aircraft
categorized as GA (72%) and, separately, AT (23%) was used to estimate the number of LTOs
conducted by piston-engine aircraft at US airports that report GA and AT LTOs (e.g., if an airport
reports 100 GA LTOs and 10 AT LTOs, then 72 and 2 LTOs would be attributed to piston-engine
aircraft for each respective category). For airports that do not report LTOs conducted by GA and
AT, EPA expanded on an FAA method to estimate LTOs using data on the number of aircraft
based at the airport (i.e., aircraft that are air worthy and operational that are based at an
airport for the majority of the year, commonly referred to as "based aircraft").23 This approach
to estimate piston-engine LTOs is routinely applied in the EPA National Emissions Inventory and
is documented in full on the EPA website.24 The national analysis of lead concentrations at and
downwind of airports nationwide used these annual piston-engine LTO estimates to calculate
the number of piston-engine LTOs at each runway end of US airports over 3-month rolling
periods as described below (Figure 2).25 For this analysis, annual piston-engine LTO estimates
22	Data on hours flown by piston-engine aircraft is consistent with activity data (LTOs), but activity data are
reported as number of LTOs conducted by piston-engine aircraft in both GA and AT categories, whereas hours
flown data are reported for piston-engine aircraft in GA and, separately, AT categories. Piston-engine aircraft flew
65.8% of hours categorized as GA and AT combined compared to conducting 65.7% of LTOs categorized as GA and
AT combined. Piston-engine aircraft flew 72% of hours categorized as GA, and, separately, 23% of those
categorized as AT.
23	When airports do not report LTOs specific to GA and AT activity, then the number of aircraft that can use leaded
fuel (i.e., SE, ME, helicopters, and ultralight aircraft) that are based at a given airport was used to help estimate the
number of LTOs conducted by each category of activity (GA or AT) out of the total number of LTOs conducted at
that airport. Airports lacking data on both the number of LTOs and the number of based aircraft were assigned 1
LTO per year based on a review of available information. For more information, see Sections 4a and 4b of:
http://nepis.epa.eov/Exe/ZvPDF.cgi/P1009113.PDF?Dockev=P1009113.PDF.
24	See Sections 4 and 6a of: http://nepis.epa.gov/Exe/ZvPDF.cgi/P1009ll3.PDF?Dockev=P1009ll3.PDF
25	The method used to estimate piston-engine aircraft activity at specific runway ends has inherent uncertainty
from both underlying operational data and local airport traffic patterns. Nevertheless, comparisons of the
methodology presented here to airport-specific observations and data suggest that this method is appropriate for
estimating piston-engine specific activity (See Section 3.3). EPA acknowledges that there are other methods to
18

-------
from 2011 formed the basis of calculating activity at each runway end over 3-month rolling
periods. Additional discussion on piston-engine activity in 2011 compared to other recent years
is provided in Appendix B, Section 1. For a subset of airports, airport-specific data were used to
provide an additional estimate piston-engine LTOs, as detailed in Section 3.3.
Annual GA and, separately, AT piston-engine LTOs at each US airport were separated into the
four categories of the aircraft classes and cycle-types: SE full LTO, SE T&G, ME full LTO, and ME
T&G, based on FAA data for GA and AT activity. Next, annual LTOs in each of these four
categories at each airport were temporally allocated into daily and then hourly periods based
on a combination of daily activity data from FAA and observations of hourly activity patterns at
the model airport. The allocation of annual to daily piston-engine aircraft activity was
accomplished by calculating a daily fraction of activity (i.e., GA or AT LTOs on a given
day/annual GA or AT LTOs) for each airport. The daily fraction was then multiplied by the
number of piston-engine LTOs in each of the four aircraft class and cycle-type categories. The
resulting number of daily LTOs in each category was then allocated to each hour of each day
based on a diurnal profile (i.e., fraction of daily LTOs per hour) from the model airport
described in Section 2.2. Appendix B provides additional information on the diurnal profile
observed at the model facility compared to observations at other airports.
estimate piston-engine specific activity (Heiken et al. 2016), and that the national analysis focuses on activity
estimates during a single year (2011), which does not capture the annual variability in piston-engine aircraft
activity at each airport due to local circumstances or national trends.
19

-------
r
Annual (2011) jet and piston
«

LTOs per airport
B
L
(GAand AT)
M
>
re
c
0
E
1
m
aj
~o


V)
_0J
c

o
aj
4—1
N
ro


Q)
4-J
4—1
c
O
aj
ro
o
i»_
c
ro
o
_c
u
u

Bin by:
Engine type (piston/jet)
Aircraft class (SE/ME)
Cycle type (full LTO/T&G)
Fraction annual LTOs into daily
LTOs
Fraction daily LTOs into hourly
LTOs
(SE full LTO, SE T&G, ME full LTO, ME T&G)
Assign hourly LTOs to runways
(SE full LTO, SE T&G, ME full LTO, ME T&G)
Sum hourly LTOs to
daily LTOs per runway
(SE full LTO, SE T&G, ME full LTO, ME T&G)
Sum daily LTOsto
LTOs per 3-months per runway
(SE full LTO, SE T&G, ME full LTO, ME T&G)
ID most active runway during
3-month period
(SE full LTO, SE T&G, ME full LTO, ME T&G)
# of LTOs on most active runway
per 3-months *
(3-month average ng Pb/m3/ LTO)
3-month average ng Pb/m3
@ 9 distances from run-up
36 AQFs
(SE/ME, LTO/T&G
9 distances)
5
Figure 2. Overview of method to estimate piston-engine aircraft activity at airports nationwide. Center
rectangles represent main calculation steps, while colors denote different spatial granularity. Grey
cylinders represent input data sets. See Table 2 for details.
20

-------
With the number of piston-engine LTOs (categorized as SE full LTO, SE T&G, ME full LTO, ME
T&G) per hour at each airport, the next step was to assign LTOs to specific runway ends at each
airport. Hourly LTOs were assigned to the runway end at which piston-engine activity would
occur based on wind direction data since piston-engine aircraft take-off and land into the wind
(See Appendix B for additional information on runway assignment and wind direction data).26
Hourly LTOs per runway end were then summed to daily and, subsequently, rolling 3-month
totals (aircraft class and cycle-type categories were maintained when aggregating up to 3-
month LTOs). The total piston-engine LTOs per runway end in a 3-month period was then used
to identify the most active runway at each airport. Next, the number of 3-month LTOs on the
most active runway is multiplied by the appropriate AQF (e.g., number of 3-month SE full LTOs x
SE full LTO AQF at maximum impact site) (Figure 3). As depicted in Equation 2, summing across
the products from each of the four aircraft class and cycle-type categories provides a 3-month
average, model-extrapolated concentration of lead in the maximum impact area and eight
downwind locations for each of the approximately 13,000 airports. These model-extrapolated
3-month average lead concentrations are: 1) attributable to aircraft using leaded avgas, and 2)
located at each of the nine specified distances at each US airport.
Activity
Estimate
(# of 3-month LTOs
on most active
runway at each
airport)
AQFs
(maximum impact
site and eight
downwind distances)
Full	Full
SE-	* SE-
Piston	Piston
AQF	Activity
T&G T&G
SE- „ SE-
Piston Piston
AQF Activity


Full
Full
ME- „
ME-
Piston
Piston
AQF
Activity
T&G T&G
ME- * ME-
Piston Piston
AQF Activity
Estimated
Ambient [Pb]
Avgas [Pb]/
2.12
Figure 3. Visualization of approach for calculating extrapolated lead concentrations by multiplying
emission factors (AQFs) by activity estimates for each airport nationwide using Equation 2.
26 While piston-engine aircraft may conduct run-up and take-off on an alternative runway (i.e., not one facing into
the wind) due to activity levels, weather, noise restrictions, or other airport operational considerations, wind is the
primary driver of active runway selection Lohr, G. W. and D. M. Williams (2008). Current practices in runway
configuration management (RCM) and arrival/departure runway balancing (ADRB). NASA/TM-2008-215557 NASA.
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/2009001Q329.pdf. Therefore, prevailing wind direction is an
appropriate indicator for identifying which runway and direction piston-engine aircraft conduct take-off and
landing operations. Runways are built to allow the maximum possible days of flying by taking into account the
dominant wind direction(s) experienced at the airport; thus, the runway end(s) predominantly used for piston-
engine aircraft take-off can be identified.
21

-------
While several meteorological, geographical, and operational parameters may vary from
conditions at the model airport or from the national default parameters used across the
national analysis described above, wind speed is one meteorological parameter that clearly
affects local concentration profiles of atmospheric aerosols. The model-extrapolated
concentrations at and downwind of the maximum impact site as characterized in the approach
above can be adjusted to better consider meteorological conditions by using inverse wind
speed data over the 3-month maximum period. Specifically, the near-field concentration of a
non-reactive pollutant scales with , where u is wind speed and angled brackets imply a
time average (Barrett and Britter 2008). If the wind speed at the model airport is v and the wind
speed at a specific airport is u, then the wind-adjusted concentration would be the model-
extrapolated concentration estimated by the methodology detailed above multiplied by the
ratio of average inverse wind speeds /. If the wind speed at the specific airport is, in
general, higher than the wind speed at the model airport where the AQFs were derived, then
 would be less than  resulting in a lower concentration per activity at the specific
airport than the AQF. Utilizing the same wind data that was used to assign operations to
specific runways, model-extrapolated concentrations at airports nationwide can be adjusted for
wind-speed, thereby appropriately characterizing concentrations at airports with significantly
higher or lower wind speeds than the model airport. For the wind speed adjustment, wind
speeds from 6am to 11pm27 were averaged over the entire year at the model airport and for
the 3-month maximum activity period at each US airport. As the inverse of wind speed tends
toward infinity as wind speed tends toward zero, 0.5 m/s is chosen as a minimum allowable
wind speed; this choice also aligns with ASOS station wind detection limits. Further details of
the wind-adjustment approach are provided in Appendix A.
Results of the national analysis method and wind speed adjustment described here, and
detailed in Table 2, are provided in Section 4. Additional quantitative and qualitative
assessments of uncertainty from other potentially influential parameters, such as avgas lead
concentration and seasonality of operational profiles are discussed in Section 3.4.
27 These are the modeled hours from opening through one hour past closing for each airport, reflecting the times
when atmospheric lead concentrations are expected to be highest.
22

-------
Table 2. Steps to Calculate Airport Facility Specific Piston-Engine Aircraft Lead Concentrations
Step
#
Step28
Description
Rationale
Data
Source29
Steps 1-7 Objective: Estimate how much piston-engine activity occurred at each U.S. airport on an hourly basis, by engine, and
operation type.
1
Estimate how much
activity is
conducted by
piston-engine
aircraft annually
Estimate the annual number of piston-
engine LTOs30 in defined categories
(i.e., GA and AT)
Only piston-engine aircraft use leaded avgas,
thus we needed to estimate how much of the
total activity at an airport was specific to
piston-engine aircraft, rather than turbine-
engine aircraft.
While several data sources provide airport-
specific aircraft activity data (separately for
General Aviation (GA) and Air Taxi (AT)
activity), none specifically identify the
number of piston-engine aircraft LTOs that
occur at each U.S. airport facility.
2011 NEI
GA and AT
piston-
engine
annual
LTOs31
(USEPA
2011)
28	Each step in this table was carried out for the 13,153 airports in the US. Heliports and rotorcraft activity at airports were not included in this analysis; see
Appendix B for additional information. For each of the 13,153 airports included in the analysis, calculations were completed for each day of 2011 and January -
February 2012; however, annual estimates of piston-engine specific LTOs were only available for 2011, and thus estimates of piston-engine aircraft LTOs from
January - February 2011 were used as surrogate activity data in the first two months of 2012. Based on the 2010 FAA Terminal Area Forecast (TAF), GA activity
levels were similar between 2011 and 2012 (5% lower activity in 2012 than 2011) (https://taf.faa.gov/).
29	Additional information on available FAA data sources is presented in Appendix B.
30	An aircraft operation is defined as any landing or takeoff event, therefore, to calculate LTOs, operations are divided by two. Most data sources from FAA
report aircraft activity in numbers of operations. Our air quality factors (AQFs), described in step 13, are in units of concentration per LTO, therefore for the
purposes of this analysis, operations need to be converted to LTO events.
31	The EPA 2011 NEI estimates annual GA and AT piston-engine LTOs that occur at each airport nationwide. These estimates were the starting point for this
national analysis of lead concentrations at and downwind of maximum impact sites at airports nationwide. The general approach to estimate piston-engine
aircraft LTOs in the 2011 NEI is briefly outlined here with more details are available in Sections 1, 3, 4, and 6a of the NEI documentation USEPA. (2011). "2011
National Emissions Inventory (NEI) Data." 2017, from http://www.epa.gov/air-emissions-inventories/2011-national-emissions-inventorv-nei-data. In particular,
the 2011 NEI used based aircraft, reported as single- or multi-engine, to develop more airport-specific piston-engine LTOs at airports with the potential for lead
air emissions inventories greater than 0.50 tons per year. In the national analysis, based aircraft are similarly used to develop more airport-specific results for
airports with model-extrapolated concentrations in the upper range of those nationwide (see Section 3.3 for details).
23

-------
Step
Step28
Description
Rationale
Data
#



Source29
la
For GA activity
The national average percent of GA
Multiplying GA LTOs at an airport by the
2011 NEI


activity that was conducted by piston-
national average of GA LTOs conducted by
GA piston-


engines (72%), according to the 2010
piston-engine aircraft was necessary to
engine


FAA GAATA report, was multiplied by
estimate the annual number of GA piston-
annual


total GA LTOs at each airport.
engine LTOs that occurred at each airport.
LTOs & FAA
GAATA,
2010(FAA
2010)
lb
For AT activity
The national average percent of AT
Multiplying AT LTOs at an airport by the
2011 NEI


activity that was conducted by piston-
national average of AT LTOs conducted by
AT piston-


engines (23%), according to the 2010
piston-engine aircraft was necessary to
engine


FAA GAATA report, was multiplied by
estimate the annual number of AT piston-
annual


total AT LTOs at each airport.
engine LTOs that occurred at each airport.
LTOs &
(FAA 2010)
Result: Annual number of GA piston-engine LTOs and AT piston-engine LTOs at each U.S. airport
2
Estimate how much
Estimate the number of total annual
Different aircraft classes and cycle-types have


of the annual
piston-engine LTOs that are conducted
different fuel consumption rates, and


piston-engine
by specific aircraft classes (i.e., SE and
therefore different quantities of lead


aircraft activity is
ME for specific cycle-types (i.e., Full LTO
emissions.


conducted by each
and T&G) at each airport.



piston-engine




aircraft class,




performing




different cycle-




types



2a
For GA piston-
Multiply the annual number of GA
Fractioning GA piston-engine activity into 4
Step la &

engine LTOs
piston-engine LTOs (from Step la) by
the national fraction of annual GA
activity conducted by each aircraft class
and cycle-type (i.e., SE Full LTO, SE T&G,
ME Full LTO, ME T&G).
combinations of aircraft and cycle-types (i.e.,
68% SE Full LTO, 23% SE T&G, 8% ME Full
LTO, 2% ME T&G) allows us to categorize
LTOs by sub-type of GA piston-engine activity
which is important since each sub-type

24

-------
Step
Step28
Description
Rationale
Data
#



Source29



impacts the resulting concentrations
(FAA



differently.
2010)(Tabl
e 1.4)32
2b
For AT piston-
Multiply the annual number of AT
Fractioning AT piston-engine activity into
Step lb &

engine LTOs
piston-engine LTOs (from Step lb) by
4 combinations of aircraft classes and cycle-
(FAA 2010)


the national fraction of annual AT
types (i.e., 57% SE Full LTO, 0% SE T&G, 43%
(Table 1.4)


activity conducted by each aircraft class
ME Full LTO, 0% ME T&G) allows us to



and cycle-type (i.e., SE Full LTO, SE T&G,
categorize LTOs by sub-type of AT piston-



ME Full LTO, ME T&G).
engine activity, which is important since each
sub-type impacts the resulting concentrations
differently.

Result: Annual number of piston-engine LTOs at each U.S. airport categorized as: 1) GA SE Full LTO, 2) GA SE T&G, 3) GA ME Full LTO,
4) GA ME T&G, 5) AT SE Full LTO, 6) AT SE T&G, 7) AT ME Full LTO, 8) AT ME T&G.

3
At the U.S. towered
Approximately 500 airports have air
Steps 1-2 provide annual piston-engine
ATADS

airports, estimate
traffic control towers (i.e., are "towered
activity; however, aircraft activity varies by


what fraction of
airports") and therefore have daily
month, day, and hour. Because of this


annual activity
activity counts (separate for GA and
temporal variability, identifying the maximum


occurred on each
AT). At each of these airports we
3-month period of activity necessitates that


day of the analysis
developed separate GA and AT daily
we apportion the annual activity data to daily


(separately for GA
activity profiles, or fractions of annual
activity (this step) and subsequently (in the


and AT)33
activity that occurred during each day
of the analysis. These daily activity
following steps) further apportion daily data
to each hour of the day.

32	The 2011 FAA GAATA report was not published, therefore the 2010 FAA GAATA report was used for this step. Based on a comparison of the 2010 and 2012
FAA GAATA reports, engine and operation type splits were very similar between 2010 and 2012 (<1% difference in any category between 2012 than 2010)
(https://www.faa.gov/data research/aviation data statistics/general aviation/). See Section 4 for additional discussion on uncertainty and variability in data
used in this analysis. The full LTOs and T&Gs fractions were based on the number of hours flown for GA or AT activities where T&Gs were defined as the
percent of "instructional" hours and full LTOs were defined as the percent of all remaining hours (e.g., total GA hours flown - instructional hours). The amount
of instructional activity will vary by airport. For instance, T&G activity was 4.5 to 29% and 0 to 35% of total SE and ME LTOs, respectively at airports for which
EPA has conducted onsite observational surveys (see Appendix C for survey details).
33	For example, the number of GA operations at each towered airport on January 1, 2011 (from ATADS dataset) were divided by each airport's respective total
number of GA operations in 2011. All operational data were converted to LTOs by dividing by two (i.e., two operations is one LTO).
25

-------
Step
#
Step28
Description
Rationale
Data
Source29


profiles will later be applied to all U.S.
airports (see Step 5).


3a
For GA LTOs
At each towered airport, divide daily GA
LTOs for each day included in the
analysis by annual GA LTOs to reach the
daily fraction of GA LTOs at each
towered airport.
Dividing daily by annual GA activity produces
a daily GA activity profile for each towered
airport.
ATADS
3b
For AT LTOs
At each towered airport, divide daily AT
LTOs for each day included in the
analysis by annual AT LTOs to reach the
daily fraction of AT LTOs at each
towered airport.
Dividing daily by annual AT activity produces
a daily AT activity profile for each towered
airport.
ATADS
Result: Daily Activity Profiles, separately for GA and AT activity, at each towered airport for each day in the analysis.
4
For each non-
towered U.S.
airport, identify its
closest towered
airport
Use latitude/longitude data and a
distance formula to determine the
closest towered airport to each non-
towered U.S. airport.34 These data will
be used in combination with the daily
activity profiles calculated in step 3 to
estimate daily piston activity at each
U.S. airport.
Data to develop daily activity profiles are only
available for airports that report daily activity
data (i.e., towered airports). To apportion
each airport's annual activity to individual
days, we apply the daily profile from the
towered airport closest in distance to the
non-towered airport. To do so, we first
determine the closest towered airport for
each non-towered U.S. airport.35
FAA 5010
Result: Identification of the closest towered airport for each non-towered airport in the U.S.
34	For two airports with (latitude, longitude) pairs of (LatA, LongA) and (LatB, LongB), the distance between them will be:
distance (km) = R*arccos[cosd(LatA)*cosd(LatB)*cosd(LongB-LongA)+sind(LatA)*sind(LatB)] where R is the radius of the spherical approximation of Earth.
35	Airport towers at the 500 most active airports in the U.S. report the number of total operations on each day, which are recorded in the FAA ATADS database.
For airport facilities without ATADS data, we used activity data from the nearest ATADS facility as a surrogate for the airport facility without daily activity data
(distances between ATADS facility and surrogates: Mean 64 km, Max 672 km, 25th % 28 km, 75th % 79 km, 90th % 128 km, 95th % 169km, 99th % 292 km). The
closest towered airport to a towered airport will be itself. Note that primary airports (i.e., airports with mainly commercial jet activity) were not used as
surrogates since these airports likely have a distinctly different activity profile than GA airports. (See Appendix B for additional details on the ATADS database.)
26

-------
Step
Step28
Description
Rationale
Data
#



Source29
5
Estimate the
Multiply each airport's annual activity
The GA and AT daily activity profiles (step 3)


number of daily
(step 2) by the daily activity profile
allow us to apportion annual activity into


piston-engine LTOs
(step 3) for its closest towered airport.
daily activity.


at all U.S. airports
This is done separately for GA and AT.


5a
For GA LTOs
Multiply each airport's annual piston-
Daily activity data are only available for the
Steps 2a &


engine GA activity (for each of the 4
combined set of all GA aircraft engine &
3a


types: 1) GA SE Full LTO, 2) GA SE T&G,
operation types (i.e., SE Full LTO, SE T&G, ME



3) GA ME Full LTO, 4) GA ME T&G) by
Full LTO, ME T&G), thus, we use the same GA



the GA daily activity profile for its
daily activity profile for each of the 4 subsets



closest towered airport.
of GA activity at all airports.

5b
For AT LTOs
Multiply each airport's annual piston-
Similar to GA, daily activity data are only
Steps 2b &


engine AT activity (for each of the 4
available for all types of AT aircraft engine &
3b


types: 1) AT SE Full LTO, 2) AT SE T&G,
operation types (i.e., SE Full LTO, SE T&G, ME



3) AT ME Full LTO, 4) AT ME T&G) by
Full LTO, ME T&G) combined, thus, we use



the AT daily activity profile for its
the same AT daily activity profile for each of



closest towered airport.
the 4 subsets of AT activity at all airports.

Result: Number of daily piston-engine LTOs at each U.S. airport categorized as: 1) GA SE Full LTO, 2) GA SE T&G, 3) GA ME Full LTO, 4)
GA ME T&G, 5) AT SE Full LTO, 6) AT SE T&G, 7) AT ME Full LTO, 8) AT ME T&G.

6
Sum the number of
Sum the daily number of GA and AT
The concentration of lead emissions is related
Step 5

daily LTOs by
LTOs across aircraft engine and
to the type of aircraft engine and operation


aircraft engine type
operation type (i.e., SE Full LTO, SE
type, thus there is no distinction in terms of


& operation mode
T&G, ME Full LTO, ME T&G).
emissions between a SE Full LTO conducted
as GA vs. AT. Understanding levels of GA vs.
AT activity was necessary to appropriately
apportion annual GA and AT activity into
specific piston engine and operation types.

6a
For SE full LTO
Sum the # of GA SE full LTOs & # of AT
SE full LTOs for each day at each
airport.


6b
For SE T&G
Same as Step 6a but for SE T&G.


6c
For ME full LTO
Same as Step 6a but for ME full LTO.


6d
For ME T&G
Same as Step 6a but for ME T&G.


27

-------
Step
#
Step28
Description
Rationale
Data
Source29
Result: Number of daily piston-engine LTOs at each U.S. airport categorized as: 1) SE Full LTO, 2) SE T&G, 3) ME Full LTO, 4) ME T&G
7
Estimate the
number of LTOs
that occurred
during each hour of
each day (i.e., the
distribution of LTOs
across facility
operational hours
of the day)
For each day at each U.S. airport,
multiply the number of daily piston-
engine LTOs (separated into 1) SE Full
LTO, 2) SE T&G, 3) ME Full LTO, 4) ME
T&G) by the corresponding hourly
activity profile (i.e., % of daily aircraft
LTOs that occurred during each
operational hour) from the model
airport. There are separate profiles for
each engine type (1) SE Full LTO, 2) SE
T&G, 3) ME Full LTO, 4) ME T&G) by
weekday/weekend status.36
(e.g., If 30% of SE Full LTOs occurred
during Hour 5 on a weekday at the
representative facility, and 10 SE Full
LTOs occurred at a given facility on Day
1 (a weekday) of the analysis, then 3 SE
Full LTOs would be assigned to Hour 5
of Day 1 at the given facility).
Step 6 results in daily piston-engine activity;
however, aircraft activity varies by month,
day, and hour. Because of this temporal
variability, identifying the maximum 3-month
period of activity necessitates that we
apportion the daily activity data to hourly
activity (this step). Subsequently (in the
following step), we use wind direction data to
apportion the hourly data to specific runway
ends at each airport.
Model
airport (see
Section 2 &
Appendix
A) & Step 6
7a
For weekdays

Since data we collected suggests that the
distribution of piston-engine aircraft activity
can vary between weekend and weekdays,
we used an activity distribution
representative of weekday activity, and
separately, an activity distribution for
weekend activity.
Appendix A
& Step 6
36 For more information on the distribution of LTOs over operational hours at the model airport see Appendix A. We characterize the influence of using a
different distribution of LTOs across the day on estimates of ambient lead in Appendix B.
28

-------
Step
#
Step28
Description
Rationale
Data
Source29
7ai
ForSE Full LTO
Multiply % of SE Full LTOs that occurred
in each operational hour of a weekday
at a representative facility by the
number of daily SE Full LTOs for each
facility in the analysis; repeat for each
day in the analysis.


7aii
For SE T&G
Repeat Step 7ai for SE T&G.


7aiii
For ME Full LTO
Repeat Step 7ai for ME Full LTO.


7aiv
For ME T&G
Repeat Step 7ai for ME T&G.


7b
For weekends
Repeat Steps 7ai - 7aiv using the
distribution of LTOs across operational
hours on a weekend day.

Appendix A
& Step 6
Result: Number of hourly piston-engine LTOs that occurred on each dav of the analysis at each U.S. airport, categorized as: 1) SE Full
LTO, 2) SE T&G, 3) ME Full LTO, 4) ME T&G
Steps 8-12 Objective: Estimate how much piston-engine activity occurred on each runway end over each rolling 3-month period.
8
Identify the runwav
end at which
aircraft activity
likely occurred for
each hour of each
day in the analysis
Use wind direction data for each hour
that an airport is open (i.e., operational
hours)37 to identify the runway end on
which piston-engine aircraft LTOs were
conducted; repeat for each day in the
analysis.
Piston-engine aircraft take-off into the wind,
thus wind direction dictates the runway end
that is used; wind direction can change
throughout the day so we evaluate hourly
wind direction38 to identify the runway end
used predominantly for each hour.
ASOS wind
tower with
shortest
distance to
airport
8a
For each U.S.
airport, determine
Use latitude/longitude data and
distance formula39 to determine the
Hourly wind direction data was available at
the 938 ASOS stations, most of which are
ASOS and
FAA5010
37	Operational hours were defined as 6 a.m. to 10 p.m. for all airport facilities in the analysis. While some airport facilities may have slightly different
operational hours (e.g., open 6 a.m. to 11 p.m.), the operational hours selected for the analysis are likely representative of most airport facilities based on
review of operational hours at numerous airports (www.airnav.com).
38	The hourly wind direction data used in this analysis is the result of 1-min wind data having been processed by EPA's AERMINUTE into hourly wind data (see
section 4.6 of AERMINUTE User's Guide for averaging method: https://www3.epa.gov/ttn/scram/7thconf/aermod/aerminute userguide.pdf
39	See footnote 30 for distance formula.
29

-------
Step
Step28
Description
Rationale
Data
#



Source29

its closest ASOS
closest ASOS station to each U.S.
located at airports.41 To determine runway
(See

station
airport.40
usage based on wind direction data, we first
determined the closest ASOS station to each
U.S. airport.
Appendix B
for details)
8b
Use the hourly wind
direction data from
an airport's closest
ASOS station to
determine which
runway end was
used for each hour
of the analysis
See Appendix B for details.
In order to appropriately estimate the
location of the maximum lead concentration
from piston-engine activity, we use wind
direction data to identify where activity
occurred (i.e., which runway end).

Result: Location (i.e., runway end) of aircraft activity at each U.S. airport during each hour of each day in the analysis
9
Determine number
Assign piston-engine aircraft LTOs in
Merging information regarding the number of


of LTOs that
each hour (Step 7) to the runway end
hourly LTOs (Step 7) with our assessment of


occurred on each
that was active during each hour (Step
hourly runway usage (i.e., which runway end


runway end on an
8); repeat for each day in the analysis.
was used during each hour) allows us to


hourlv basis

quantify the hourly number of LTOs that
occurred on each runway end at each U.S.
airport for each day of the analysis.

9ai
ForSE Full LTO
Assign SE Full LTOs in each hour (Step 7)
to the runway end that was active
during each hour (Step 8); repeat for
each day in the analysis.

Steps 7 & 8
9aii
For SE T&G
Repeat Step 9ai for SE T&G.


9aiii
For ME Full LTO
Repeat Step 9ai for ME Full LTO.


9aiv
For ME T&G
Repeat Step 9ai for ME T&G.


Result: Number of piston-engine LTOs that occurred during each hour on each runway end during each day of the analysis at each
U.S. airport, categorized as: 1) SE Full LTO, 2) SE T&G, 3) ME Full LTO, 4) ME T&G

40	The closest ASOS station to an airport with an ASOS station will be its own station.
41	ASOS & Climate Observations Fact Sheet. November 2012. U.S. NOAA
30

-------
Step
#
Step28
Description
Rationale
Data
Source29
10
Determine the
number of LTOs
that likely occurred
on each runway
end on a daily basis
For each runway end at each airport,
sum the number of aircraft LTOs that
occurred during all operational hours
for a given day; repeat for each day in
the analysis.
To estimate the number and type of LTOs
that occurred at an airport on each runway
end over an entire day, we sum the hourly
LTOs, by runway end. In subsequent steps we
use this daily information to estimate activity
over 3-month time periods, which
corresponds to the lead NAAQS averaging
period.

lOai
ForSE Full LTO
For each runway end at each airport,
sum the number of SE Full LTOs that
occurred during all operational hours
for a given day; repeat for each day in
the analysis.
Summing all of the SE Full LTOs at an airport
that occurred at each runway end during
each operational hour of a day allows us to
estimate the number of SE Full LTOs that
occurred on each day of the analysis at each
runway at an airport.
Step 9
lOaii
For SE T&G
Repeat Step lOai for SE T&G.


lOaiii
For ME Full LTO
Repeat Step lOai for ME Full LTO.


lOaiv
For ME T&G
Repeat Step lOai for ME T&G.


Result: Number of piston-eng
Full LTO, 2) SE T&G, 3) ME Ful
ine LTOs that occurred during each dav on each runwav end at each U.S. airport, categorized as: 1) SE
LTO, 4) ME T&G
11
Sum daily# of LTOs
estimated to have
occurred on each
runway end by
rolling 3-month
period

We estimate the number and type of LTOs
that occurred on each runway end at each
airport over a rolling 3-month period using
the daily information generated in Step 10,
since the averaging time for the lead NAAQS
is a rolling 3-month averaging period (e.g.,
January - March, February - April, March -
May).42

42 At some airports available data suggest that the sum of LTOs in the 3-month period is less than one; this is predominantly due to the airport having fewer
than 5 LTOs per year, but in some cases, may be due to missing data (e.g., runway end identifiers). Low activity or a lack of data resulted in 2,095 out of the
13,000 airports nationwide with less than one LTO in the 3-month period. Model-extrapolated concentrations at these airports are thus less than 0.0075 ug/m3
31

-------
Step
Step28
Description
Rationale
Data
#



Source29
llai
ForSE Full LTO
For each runway end at each airport,
sum the number of SE Full LTOs that
occurred during each day of a 3-month
period; repeat for each rolling 3-month
period included in the analysis.

Step 10
llaii
For SE T&G
Repeat Step llai for SE T&G.


llaiii
For ME Full LTO
Repeat Step llai for ME Full LTO.


llaiv
For ME T&G
Repeat Step llai for ME T&G.


Result: Number of piston-engine LTOs that occurred during each rolling 3-month period on each runway end at each U.S. airport,
categorized as: 1) SE Full LTO, 2) SE T&G, 3) ME Full LTO, 4) ME T&G


12
Identify the runway
end with the
highest estimates
of piston-engine
aircraft activity
during any 3-month
period at each
airport

Piston-engine aircraft activity is a first-order
determinant of lead concentrations in the
maximum impact area in monitoring and
modeling studies, as described in Section 2,
and thus the period of maximum activity is
assumed to represent the period of
maximum concentration.43
Step 11
12a
For each runway
Sum Steps llai - llaiv by runway and
In addition to understanding how much
Step 11

end, sum the
by 3-month period for each U.S. airport.
piston-engine aircraft activity of specific


number of total

engine class & cycle types occurred at each


piston aircraft LTOs

runway end over rolling 3-month periods


that occurred

(which will be used in Step 13), we to need


during each 3-

identify the runway end at which the most


month period for all

piston aircraft activity of any type was


engine & operation

conducted over a rolling 3-month period.


types; repeat for

Identifying the runway end used most

(see Section 4.1 for results). Additional analyses outside the scope of this report would be needed to evaluate airborne lead concentrations at these individual
airports.
43ln some instances, meteorological parameters (e.g., low mixing height) may result in maximum concentrations during relatively lower activity periods.
Uncertainty and variability in meteorological parameters is discussed further in Section 4.3.
32

-------
Step
#
Step28
Description
Rationale
Data
Source29

each rolling 3-
month period
included in the
analysis

frequently by piston-engine aircraft allows us
to estimate ambient concentrations at the
location (i.e., runway end) with the most
piston-engine activity, and in turn the highest
lead emissions.

12b

Review number of piston-engine LTOs
conducted at each runway end during
each rolling 3-month period included in
the analysis and identify the runway
end with the most total piston-engine
LTOs during any 3-month period; repeat
for each airport facility in the analysis.

Step 12a
Result: Identification of the most active runway during any 3-month period at each airport facility included in the analysis
Steps 13 - 15 Objective: Estimate maximum 3-month lead concentrations from Piston-engine aircraft at each U.S. Airport
13
Estimate ambient
lead concentrations
from piston-engine
aircraft lead
emissions at the
runway end most
frequently used by
piston-engine
aircraft during the
most active rolling
3-month period
Multiply the number of LTOs that
occurred on the runway end most
frequently used by piston-engine
aircraft during the most active 3-month
period by corresponding air quality
factors; repeat for each facility in the
analysis.
In Steps 1-12 we estimate piston-engine
aircraft activity (i.e., how many LTOs of which
engine class and cycle type that occur when
and where) at each airport facility included in
the analysis. We then combine our activity
estimates with estimates of lead
concentrations associated with each type of
LTO in order to calculate total maximum 3-
month lead concentrations from piston-
engine aircraft. To do so, we use AQFs that
are specific to each engine class and cycle
type (SE Full LTO, SET&G, ME Full LTO, ME
T&G).

13ai
ForSE Full LTOs at
the most active
runway during the
Multiply the following:
1) the number of SE Full LTOs that
occurred at the runway end most
frequently used by piston-engine
As described in Section 3.1, AQFs are the
relationship of lead concentration per unit of
aircraft activity (with distinct AQFs for each
aircraft engine and operation type) and
Steps
11&12;
Model
airport (see
33

-------
Step
Step28
Description
Rationale
Data
#



Source29

most active 3-
aircraft during the most active 3-month
having units of average 3-month ng Pb/ m3/
Section 2 &

month period
period, by 2) the AQF for SE Full LTOs at
LTO. By multiplying each AQF by the level of
Appendix


the max impact site; repeat for each
activity we estimate the lead concentration
A)


facility in the analysis.
(pig Pb/ m3) associated with the number of
LTOs we estimated in Steps 1 - 12.

13aii
For SE T&G
Repeat Step 13ai for SE T&G.


13aiii
For ME Full LTO
Repeat Step 13ai for ME Full LTO.


13aiv
For ME T&G
Repeat Step 13ai for ME T&G.


13av
For all piston-engine
activity
Sum Steps 13ai - 13aiv.
We need to understand total lead
concentrations from all types of piston-
engine activity, which is the sum of Steps
13ai-13aiv.

13avi
Scaled by the lead
First, divide the ASTM standard for Pb
The AQFs were generated at a model airport


concentration in
concentration in avgas (2.12 g Pb/gal)
with a concentration of Pb in avgas that is


avgas
by the avgas Pb concentration at the
model airport (2.16 g Pb/gal). Second,
multiply the ratio of 2.12/2.16 by the
sum of lead concentration from all
types of piston-engine activity (Step
13av).
different from the ASTM maximum
specification for this fuel. Thus, we scale the
lead concentrations at each airport by the
ratio of the ASTM standard lead
concentration to the avgas lead
concentration at the facility used to develop
AQFs.44

Result: Ambient lead concentration estimates at the max impact site at the most active runway end during the most active 3-month
period for each airport facility included in the analysis


14
Estimate ambient
Repeat Step 13 with the appropriate
As discussed in Section 3.1, in addition to
Model

lead concentrations
AQFs for the 8 locations further
developing AQFs at the max impact site, we
airport (see

at locations further
downwind of the max impact site (50,
also developed AQFs at 8 locations downwind
Section 2 &

downwind from the
100, 150, 200, 250, 300, 400, 500 m);
of the max impact site (i.e., where piston-
Appendix

runway end
repeat for each facility included in the
analysis.
engine aircraft conduct run-up checks) in
order to provide estimates of how lead
concentrations change with distance. Similar
A)
44 We examine the influence that using the ASTM standard for avgas lead concentration has on our ambient lead concentration estimates in Section 4.
34

-------
Step
Step28
Description
Rationale
Data
#



Source29



to Step 13, we need to combine each




respective AQF with activity estimates in




order to estimate concentrations of ambient




lead at each distance for each airport




included in the analysis.

Result: Ambient lead concentration estimates at 8 locations downwind of the max impact site at the most active runway end during
the most active 3-month period for each airport included in the analysis


15
Estimate wind-
Scale the model-extrapolated ambient
As discussed in Section 3.2, wind speed has a
Appendix A

adjusted ambient
lead concentrations by the ratio of the
consistent and well-characterized impact on
and ASOS

lead concentrations
average inverse wind speeds at the
the near-field concentration of a passive
wind tower

using average
model airport to the average inverse
tracer under dispersion. Therefore, scaling
with

inverse wind speed
wind speeds recorded at the nearest
model-extrapolated lead concentrations to
shortest


ASOS wind tower.
consider wind speed will better characterize
local concentrations at airports nationwide,
particularly those airports where wind speeds
during the maximum activity period differ
significantly from those observed at the
model airport.
distance to
airport
Result: Ambient wind-adjusted lead concentration estimates at and downwind of the max impact site at the most active runway end
during the most active 3-month period for each airport included in the analysis

35

-------
3.3 Evaluation of Airports for Potential Lead Concentrations Above the Lead NAAQS
The national analysis methods described in Section 3.2 provided estimates of 3-month average
model-extrapolated lead concentrations in the maximum impact area and locations downwind
out to 500-meters for 13,153 airports. Within this large set of model-extrapolated
concentrations, we identified the subset of airports where lead concentrations were estimated
to potentially approach, within 10%, or to be above the lead NAAQS.45 To do this, we first
identified airports where model-extrapolated concentrations were above the NAAQS. Next, we
ran a series of sensitivity analyses to identify any additional airports where model-extrapolated
concentrations may be above or approach the NAAQS when considering the major drivers of
airport-to-airport variability and uncertainty. For this subset of airports, we then identified
additional, airport-specific data that could refine the estimates of piston-engine aircraft activity.
Finally, for this subset of airports we considered additional airport-specific criteria, such as the
unrestricted access within 50 meters of the maximum impact location. An overview and
rationale for the approach is provided in Section 3.3.1 followed by a description of how we
adjusted activity estimates for the identified subset of airports using airport-specific data in
Section 3.3.2. The full methodology for considering concentrations using airport-specific activity
data and additional criteria is presented in Section 3.3.3.
3.3.1 Sensitivity Analysis of Airport-Specific Parameters that Influence Potential for Lead
Concentrations to be Above the NAAQS
The first step to identify airports at which model-extrapolated concentrations are potentially
above the lead NAAQS was to evaluate which airport-specific parameters may result in
uncertainty or bias that would lead to underestimates in model-extrapolated concentrations
from the national analysis methods presented in Section 3.2. There is potential uncertainty
and/or bias from using national defaults for: 1) percentages of piston aircraft at an airport, 2)
percentages of piston operations performed by single- versus multi-engine aircraft, and 3)
assigning piston operations to runway ends. To address these sources of uncertainty and to
identify airports where lead concentrations may approach or be above the NAAQs, but would
not be identified by using national defaults, we conducted a series of sensitivity analyses. These
sensitivity analyses expand the number of airports that would be within 10% of the NAAQs by
using different assumptions for each of the three parameters outlined above that used national
defaults in the national analysis.
For the first two parameters, we accounted for the possibility that the percentage of activity
conducted by piston-engine aircraft and/or the percentage of piston-engine aircraft activity
conducted by multi-engine aircraft at each airport might be underestimated by national
averages. We did so by evaluating a scenario in which all GA and half of AT activity was
conducted by piston-engine aircraft at each airport (i.e., we substituted 100% and 50% for the
national average percentages of 72% and 23% piston-engine aircraft of total GA and AT,
45 The current NAAQS for lead is 0.15 ng/m3 as a 3-month rolling average. For this analysis, "approaching" the lead
NAAQS is defined as within 10% of the current standard, or 3-month average model-extrapolated concentrations
>0.14 ng/m3.
36

-------
respectively; see Step 1 of Table 2).46 Because AT operations are more often conducted by ME
aircraft, this sensitivity analysis impacts both the estimates of piston-engine aircraft activity and
the predominance of ME or SE piston-engine aircraft at an airport. We then identified airports
that had 3-month average model-extrapolated lead concentrations that were within 10% of the
lead NAAQS after accounting for the possibility that national averages might be under-
representations of piston-engine activity at some airports.
An additional sensitivity analysis was performed on the percentage of operations that occur at
the most-utilized runway during the maximum activity period (Step 12 of Table 2). Two factors
contribute to this percentage: the seasonal profile of operations and the allocation of
operations to different runways based on wind direction. For the airports that were identified
as having maximum 3-month concentrations above or approaching 0.15 ng/m3 through the
national analysis method presented in Section 3.2, the average percentage of annual activity
occurring at the maximum period runway end is 20%. However, this percentage ranges from
<6% at some airports, up to 45% at others. Reasons why an aircraft could take-off or land on a
runway end other than the one assigned in the extrapolation, or be active during another 3-
month period, include that the airport's seasonal profile of piston operations differs from that
of the nearest ATADS airport, or the airport has two runways with similar headings, such that
the dominant wind direction bisects them. These effects could bias estimates of operations and
therefore concentrations either high or low. To better understand if some airports could have
concentrations approaching or above the NAAQs that were not identified in the initial
nationwide analysis due to a runway assignment bias, a sensitivity analysis was performed;
airports that had less than 20% of their operations occurring at their maximum utilized 3-month
period runway end were changed to having 20% of operations occur at that runway during that
period.47
Additional sources of uncertainty in operational data that could impact the national analysis
results are discussed in Section 4.4 of this report. For example, there may be uncertainty in the
annual GA operations counts that underlie the piston operations data. However, changing the
total annual GA operations count effects the resulting maximum concentrations in the same
way that changing the percentage of GA operations that are conducted by piston aircraft
effects the maximum concentration (i.e., increasing total GA operations by 10% would be
analytically equivalent to keeping GA operation counts constant and increasing the percentage
46	The parameters presented in these sensitivity analyses, such as the 100% GA and 50% AT activity conducted by
piston-engine aircraft, were only used to identify airports for additional analysis; neither these parameters nor the
resulting maximum 3-month concentrations were used in the airport-specific activity analysis described below and
presented in Section 4.2.
47	This sensitivity analysis may not identify all airports where maximum concentrations have been under- (or over-)
estimated due to the operational profile and runway assignment methodology. For example, an airport that the
national analysis identifies as having 21% of operations occurring at the maximum runway end may in practice
have 35% of operations occurring at that runway end. However, initial analysis showed that model-extrapolated
concentrations estimated to be above the level of the lead NAAQS were mostly insensitive to operational shifts of
this scale. This suggests that the national analysis methodology is appropriate for identifying airports with the
potential for model-extrapolated concentrations to be above the lead NAAQS even considering this operational
uncertainty.
37

-------
performed by piston aircraft by 10%). Thus, the sensitivity analyses performed above may be
interpreted to account instead, at least in part, for independent uncertainty from these other
sources.
The airports identified in the national analysis or sensitivity analyses as having maximum 3-
month concentrations above or approaching the NAAQS were the focus of a more refined
assessment of piston-engine aircraft activity, as described below.
3.3.2 Airport-Specific Activity Data
The objective of the sensitivity analyses described above was to identify additional airports at
which it would be informative to evaluate airport-specific piston-engine aircraft activity data,
rather than national average data. The above sensitivity analyses applied alternative default
assumptions for two parameters to all 13,000 airports, while the analyses in this section apply
airport-specific data to the subset of airports identified through the sensitivity analyses and the
national analysis. The objective of the analyses in this section is to account for the fact that
national average activity estimates may potentially be improved by using airport-specific
activity surrogates. As described in Section 3.2, piston-engine aircraft activity is not reported for
individual airports, thus estimates of activity specific to piston-engine aircraft were calculated
using national averages for the fraction of total GA and AT LTOs conducted by piston-engine
aircraft. Similarly, national average fractions were used to estimate piston-engine LTOs
conducted by SE versus ME aircraft. Both of these parameters (piston-engine aircraft activity
and SE versus ME activity) particularly influenced monitored and modeled lead concentrations
attributable to piston-engine aircraft in previous analyses conducted by EPA and others (Fine et
al. 2010, Carr et al. 2011, Heiken et al. 2014, Feinberg et al. 2016). In these analyses, piston-
engine aircraft activity had a direct impact on lead concentration, where more piston-engine
aircraft activity (i.e., more LTOs) generally correlated with higher lead concentrations (Figure 4
provides one example of this relationship at Palo Alto Airport (PAO), which was included in EPA
NAAQS lead surveillance monitoring network).
Additionally, sensitivity analyses conducted at two GA airports (RHV and SMO), showed that the
amount of activity conducted by multi-engine piston aircraft had a disproportionately larger
impact on lead concentrations compared with single-engine aircraft activity (see Appendix B;
(Carr et al. 2011)). 48 Based on the important influence of these two parameters in previous
analyses, additional, airport-specific information was gathered to further characterize total
piston-engine aircraft activity and the percentage of activity conducted by single- versus multi-
piston-engine aircraft at each of the airports included in this refined, airport-specific activity
analysis.49
48	Multi-engine (ME) piston aircraft have a higher fuel consumption rate compared to single-engine (SE) piston
aircraft; thus, LTOs conducted by ME aircraft result in higher lead concentrations.
49	Total landing and take-off counts, the percentage split between piston and non-piston aircraft, and the runway
assignment method may also each contribute to uncertainty in counts of piston-engine aircraft LTOs at a given
runway end. The runway assignment method and its impact on LTO counts is discussed in Appendix B.
38

-------
0.16
0.14
0.12
___ 0.10
CO
E
*35 0.08
3
-Q
a.
0.06
0.04
0.02
0.00











• •








< *
••







• .«
•




•
•
<•





•
•
1
•
• ••
•




•
• ••
•





•













100 200 300 400 500 600 700 800
Number of Operations
Figure 4. Example of the relationship between monitored lead concentrations and piston-engine
aircraft activity.50
Specifically, based aircraft data (i.e., the number and class of aircraft that are parked at an
airport) were collected for airports included in this airport-specific activity analysis. Data from
previous EPA studies at six airports showed agreement within 10% between the number of SE
and ME aircraft based at an airport and onsite observations of piston-engine aircraft activity at
the airport (see Appendix B for study details).51 As such, the number and class of aircraft based
at each airport included in this airport-specific activity analysis was used to refine the national
average percentages for estimating the number of LTOs specific to piston-engine aircraft, and
then SE versus ME piston aircraft.
For each airport included in this analysis, the number of aircraft based at that airport was
collected from available data sources.52 Next, for each airport, the percent of total operations
511 The relationship between monitored lead concentrations and piston-engine aircraft activity is impacted by
several parameters including distance of the monitor from the area where aircraft conduct run-up checks, wind
speeds, the type of aircraft (multi-engine or single-engine), and the type of operation (full landing and take-off
versus touch-and-go. This figure does not analyze each of these influencing variables but is illustrative of the
general relationship between activity and lead concentration at a general aviation airport).
51	SE and ME aircraft based at an airport were considered piston-engine aircraft. While some SE and ME aircraft
based at an airport may be turboprop or other non-piston-engine aircraft, comparisons with onsite activity counts
suggest based aircraft data provide reasonable, airport-specific data and FAA considers based aircraft data to be a
reliable indicator of activity at small airports (FAA 2015).
52	A search was conducted for airport master plans or onsite studies on piston-engine aircraft activity, and in the
absence of such information, based aircraft data were used from airport master plans orAirnav.com.
39

-------
conducted by piston-engine aircraft was calculated using the number of SE and ME aircraft over
the total number of aircraft based at the airport (i.e., sum of SE and ME based aircraft over total
SE, ME, turboprop, jet, and helicopter based aircraft multiplied by 100). Similarly, the percent of
piston-engine operations conducted by SE versus ME aircraft was calculated using the numbers
of SE versus ME aircraft based at the airport (e.g., SE based aircraft over sum of SE and ME
based aircraft multiplied by 100). Table 3 presents a summary of the percent of LTOs allocated
to piston-engine aircraft, and separately SE versus ME piston aircraft, in the national analysis
compared to the allocation using data for aircraft based at the airports included in this airport-
specific activity analysis.
Table 3. Comparison of Piston-Engine Activity Estimates Using National Averages versus Airport-
Specific Data

National Analysis
National Averages
Airport-Specific
Based Aircraft Data53
Data Sources
% piston versus jet
operations
GA: 72%
AT: 23%
Unique to each airport
(%SE & ME based aircraft of
total based aircraft)
GA & AT Mean: 92%
GA & AT Range: 60 - 100%
National Analysis:
(FAA 2010, USEPA
2011)
Airport-specific:
Airport Master Plans
& Airport Master
Record Forms 5010-1
& 5010-2
% single-versus
multi-engine
operations
GA SE: 90%
GA ME: 10%
AT SE: 57%
AT ME: 43%
Unique to each airport
(%SE OR ME based aircraft of
SE AND ME based aircraft)
•	GA & AT SE:
o Mean: 89%
o Range: 58-99%
•	GA & AT ME:
o Mean: 11%
o Range: 0.02 - 42%
National Analysis:
(FAA 2010, USEPA
2011)
Airport-specific:
Airport Master Plans
& Airport Master
Record Forms 5010-1
& 5010-2
In general, for the airports evaluated here, using the number of piston-engine aircraft based at
the airports as a surrogate for activity suggests that piston-engine aircraft activity at these
airports is higher than indicated by the national average fraction (Table 3). The higher percent
53 In the national analysis, the percent of activity attributed to piston-engine vs. jet, and separately, multi- vs.
single-engine aircraft differed for GA vs. AT activity based on FAA data; however, based aircraft data do not
provide information on differences between GA and AT and thus the same percentages are used for both.
40

-------
of piston-engine aircraft activity at these airports is expected given that master plans and other
available information (e.g., airport websites) show that these airports are predominately GA
airports, which generally have higher levels of piston-engine aircraft activity compared to a
national average that includes activity at commercial and other larger airports with more jet
activity. For the percentage of piston-engine aircraft activity conducted by SE versus ME
aircraft, the number of SE and ME aircraft based at these airports suggest similar percentages
of aircraft activity are conducted by each aircraft class compared to the national average data
for GA activity. Conversely, the number of ME aircraft based at these airports generally suggest
ME activity is lower than the national average used to estimate ME piston aircraft activity from
total AT activity (Table 3). The airport-specific activity estimates calculated using aircraft based
at these airports were used to calculate refined model-extrapolated lead concentrations, per
the methods described in Table 2. These refined model-extrapolated concentrations are
compared with national analysis values, as well as relevant monitoring data, in Section 4.
3.3.3 Airport-Specific Criteria for Identifying Potential Lead Levels Above the NAAQ5
This section summarizes the approach and rationale for selecting airports included in the
airport-specific activity analysis of lead concentrations at the maximum impact area. At a high-
level, this approach entails identifying airports where the maximum 3-month average model-
extrapolated concentrations may be above or approach the NAAQS, characterizing model-
extrapolated maximum 3-month concentrations at these airports using airport-specific, refined
estimates of aircraft activity splits, and then evaluating each airport on local criteria such as the
proximity of the maximum-impact site to unrestricted public access. The detailed methods for
this analysis are provided in Table 4.
41

-------
Table 4. Steps for Identifying Airports Where Lead Concentrations May Be Above the Lead NAAQS
Step
#
Step
Description
Rationale
Data
Source
Steps 1-3 Objective: Identify a subset of airports where, considering sources of variability and uncertainty, model-extrapolated
atmospheric lead concentrations could be above or approach the NAAQS for Lead.
1
Identify airports
with maximum
model-extrapolated
concentrations
approaching or
above the NAAQS
Sum the contributions of single- and
multi-engine T&G and LTO operations
to atmospheric lead concentrations at
the maximum impact site for the
maximum activity period from the
national analysis described in Section
3.2. Identify all airports where the
maximum concentration is above or is
within 10% of 0.15 ng/m3.54
The primary and secondary National Ambient
Air Quality Standards for Lead are 0.15
micrograms per cubic meter lead in total
suspended particles as a 3-month average.
Because the AQFs relate operations to
average atmospheric lead concentrations
over the same timescale (3 months), the
results of the national analysis indicate
whether or not model-extrapolated lead
concentrations may approach or be above
the concentrations specified in the NAAQS
for lead when the inputs described in Section
3.2 are used.
National
Analysis
Step 14
and 40CFR
Part 50
2
Identify airports
with maximum
model-extrapolated
concentrations
approaching or
above the NAAQS
when all GA and
half of all AT
operations are
assumed to be
Scale the contributions of single- and
multi-engine T&G and LTO operations
to maximum impact area atmospheric
lead concentrations to characterize
these concentrations if 100% GA
operations and 50% of AT operations
were operated by piston-engine
aircraft.
As detailed in Steps la and lb in the national
analysis methods, the national analysis
assumed that 72% of GA and 23% of AT
operations are performed by piston-engine
aircraft. The current step identifies airports
where concentrations would be above or
approach the NAAQS if piston-engine aircraft
were a larger portion of activity at each
airport.
National
Analysis
Steps la
and lb and
GAATA
Survey
54 Aircraft activity for the most recent year available was evaluated at this stage; airports where overall activity decreased such that estimated lead
concentrations were no longer within 10% of 0.15 ng/m3 were excluded.
42

-------
Step
#
Step
Description
Rationale
Data
Source

piston aircraft
operations



2a
Scale concentration
contributions from
T&G operations
Scale lead concentration contributions
from GA operations by (1/0.72)
In the national analysis, all T&G operations
are assumed to be from GA flight activity.
Thus, as concentrations scale with
operations, both single- and multi-engine
concentrations can be scaled by the
proportional change in GA piston-engine
operations.

2b
Scale concentration
contributions from
full flight operations
Scale full flight lead concentration
contributions from AT operations by
(0.5/0.23), the ratio of new operational
cycles to old operational cycles for both
SE and ME concentration contributions.
Both GA and AT operate SE and ME full flight
operations.
National
Analysis
(FAA2010,
EPA 2011)
3
Identify airports
with maximum
model-extrapolated
concentrations
approaching or
above the NAAQS
when at least 20%
of operations occur
at the most-used
runway end during
the maximum 3-
month period
Scale model-extrapolated lead
concentrations by the ratio (0.2/X),
where X is the airport-specific fraction
of operations occurring at the most-
used runway end during the maximum
3-month period and X < 0.2.
For the airports that are identified as
potentially having lead concentrations
approaching or above the NAAQs for lead at
Step 1 of the airport specific analysis, the
average percentage of operations occurring
at the maximum period runway end is 20%.
This sensitivity analysis identifies airports
where operations at the most-used runway
end may have been underestimated due to
assumptions about wind direction, runway
assignment, and local seasonal operational
profile.
Airport
Specific
Analysis
Step 1
Result: Identification of a subset of airports as having model-extrapolated lead concentrations that could be above the NAAQS for
lead.
Steps 4-7 Objective: Refine model-extrapolated concentrations at the subset of airports identified in Steps 1-3 using airport-
specific activity data
4
Collect based-
aircraft data for the
Designate to each airport in the airport-
specific analysis counts of jet, single-
For the national analysis, national average
splits of piston/non-piston and subsequently
FAA Form
5010 Data
43

-------
Step
#
Step
Description
Rationale
Data
Source

subset of airports
identified in Steps
1-3
engine, and multi-engine aircraft from
reported based-aircraft numbers at that
airport.
SE/ME operations were applied to both GA
and AT operations. Because individual
airports may serve different aircraft
populations, an airport-specific activity
assessment may provide a refined
characterization of operational splits by
aircraft type. This assessment uses counts of
aircraft based at a particular airport as a
proxy for a representative sample of the split
of operations by aircraft type.

4b
Retain national
average splits of
operational cycles
for airports with no
based-aircraft data
in Form 5010.
Where airports have no reported
based-aircraft data55, retain the
national average splits of operational
cycles by SE/ME and Full/T&G for AT
and GA.
Where based-aircraft are not reported, the
national average percentage of SE/ME and
Full/T&G operational cycles remain the best
estimates of operational characteristics at
that individual airport.
National
Analysis
(FAA2010,
EPA 2011)
4c
Retain national
average splits of
operational cycles
for airports with
low based-aircraft
counts relative to
annual operations.
Where airports have an annual-
operations-to-based-aircraft ratio
greater than 73 0 56, retain the national
average splits of operational cycles by
SE/ME and Full/T&G for AT and GA.
As based-aircraft numbers are self-reported,
Form 5010 Data may be incomplete at some
airports. Further, at busy airports with
significant commercial or AT traffic, aircraft
based at the airport may not be
representative of all aircraft serving the
airport. The lower the ratio of operations-to-
based-aircraft, the more appropriate based-
aircraft is expected to be a proxy for
operational splits. We make the assumption
that annual operations-to-based aircraft
greater than 730 (2 operations per based
aircraft per day), is an upper limit above

55	For the airport-specific analysis presented in Section 4, 5.7% of airports have no based-aircraft data.
56	For the airport-specific analysis presented in Section 4,10.0% of airports have annual-operations-to-based-aircraft ratios above 730.
44

-------
Step
#
Step
Description
Rationale
Data
Source



which the based aircraft data are not a
suitable proxy for activity at an individual
airport.

5
Assign splits of GA
and AT piston/non-
piston operations
from based-aircraft
data
Characterize the number of operations
that would be performed by piston-
engine aircraft at each airport if the
non-jet aircraft based at the airport
were representative of the percent of
GA and AT operations performed by
piston-engine aircraft at that airport.
While several data sources provide airport-
specific aircraft activity data (separately for
General Aviation (GA) and Air Taxi (AT)
activity), none specifically identify the
number of piston-engine aircraft LTOs that
occur at each U.S. airport. In the national
analysis, a default percentage representative
of national averages was used to determine
piston-engine aircraft operations at each
airport; this analysis uses local airport-
specific information (namely based-aircraft)
to better characterize model-extrapolated
lead concentrations at those airports that
could have model-extrapolated
concentrations that approach, or be above
the NAAQS for lead as identified in Steps 1-3.
FAA Form
5010 Data
6
Assign splits of ME
and SE Full and
T&G operations
from based-aircraft
data
Characterize the percentage of piston
aircraft operations that would be
classified as SE Full, SE T&G, ME Full,
and ME T&G
In the national analysis, default percentages
of operational splits for AT and GA operations
by aircraft class (SE/ME) and operational
cycle type (Full/T&G) representative of
national averages were used to characterize
piston aircraft operations at each airport; this
analysis uses local airport-specific
information (namely based-aircraft) to better
characterize model-extrapolated lead
concentrations at those airports that could
have model-extrapolated concentrations that
approach, or are above the NAAQS for lead as
identified in Steps 1-3.
FAA Form
5010 Data
45

-------
Step
#
Step
Description
Rationale
Data
Source
6a
Determine
operational splits
for AT at each
airport
The percent of AT operational cycles
that are SE (or ME) full LTO matches the
percent of based-aircraft that are SE (or
ME).
All AT operations are considered to be full
LTOs.

6b
Determine
operational splits
for GA at each
airport
The percent of GA operational cycles
that are SE (ME) matches the percent of
based-aircraft that are SE (ME). Of the
GA SE operational cycles, 24% are
characterized as T&G consistent with
the national analysis. Of the GA ME
operational cycles, 20% are
characterized as T&G consistent with
the national analysis.
Both full LTO and T&G operational cycles are
performed by GA aircraft.

Result: Characterization of a r
for each of the airports identi
efined estimate of the number and type o
:ied in Steps 1-3.
operations performed by SE and ME piston-engine aircraft
7
Refine model-
extrapolated lead
concentrations
using updated
operational splits
For the airports identified in Steps 1-3,
estimate model-extrapolated lead
concentrations at and downwind of the
maximum impact site using the data
gathered in Steps 4-6 paired with the
methodology described in the National
Analysis (Table 2).

National
Analysis
Steps 3-14
Result: Lead concentration estimates at and downwind of the maximum impact site at the most active runway end during the most
active 3-month period for each airport identified in Steps 1-3 using airport-specific activity data.
Step 8 Objective: Identify whether there is unrestricted access to the area of maximum impact at airports identified at Step 7
8
Identify airports
where there is
unrestricted access
to the 50 m
perimeter around a
For the airports that have model-
extrapolated lead concentrations that
are above the lead NAAQS as identified
in Step 7, estimate the distance from
the run-up area at the most-utilized
The layout and footprint of many general
aviation airports is such that, aircraft run-up
areas and the maximum impact site may be
in close proximity to where people have
unrestricted access. We sub-select airports
Satellite
and street-
view
imagery
46

-------
Step
#
Step
Description
Rationale
Data
Source

maximum impact
site
runway end to the nearest unrestricted
access using satellite imagery.
where there was unrestricted access within
50m of the maximum impact site where lead
concentrations were estimated as potentially
above the lead NAAQS in Step 7.

9
Identify local
airport
characteristics that
may influence lead
concentrations at
the maximum
impact site
For the airports that have model-
extrapolated lead concentrations that
are above the lead NAAQS as identified
in Step 7, review satellite imagery and
airport documentation to determine if
there are any airport-specific conditions
or characteristics that could influence
lead concentrations at the maximum
impact site.
As all airports are unique, any airport may
have a layout, local characteristic, or
operational pattern that may differ from the
assumptions underlying the national analysis
and may impact resulting atmospheric lead
concentrations.
Satellite
imagery,
airport
master
plans
Result: Identification of airports that have model-extrapolated lead concentrations above the NAAQS for lead considering both
airport-specific activity data and unrestricted access to the maximum impact area.
47

-------
haracterization of Uncertainty of Cross-Airport Parameters that influence the
Potential for b ncentrations to Be Above the NAAQS for Lead
As discussed in Section 1, the goal of this work is to characterize lead concentrations at and
downwind of the maximum impact area at airports nationwide. The approach described in
Sections 3.2 and 3.3 was selected because of the consistent set of ground-based parameters
that are inherent to safe operation of piston-engine aircraft. Namely, that these aircraft take-
off into the wind and conduct pre-flight engine checks adjacent to the take-off runway end.
These parameters are consistent across airports, and thus constrain the uncertainty and
variability that might be associated with results based on combining information from one
model airport with activity estimates at airports nationwide. The limited set of key parameters,
which influenced maximum impact area ground-level air lead concentrations in previous
modeling by EPA and others, were: 1) the duration of run-up, where longer run-up times results
in higher concentrations, 2) the concentration of lead in the fuel, where higher avgas lead
concentrations results in higher concentrations, 3) activity, where more piston-engine aircraft
activity increases lead concentrations, 4) the percent of activity conducted by ME piston-
aircraft, where more ME activity results in higher lead concentrations due to the higher fuel
consumption rates of these aircraft relative to SE aircraft, and 5) meteorological factors and
local topography (including wind speed, wind direction, mixing height, atmospheric stability,
and surface roughness) (Section 2; Appendix A) (Carr et al. 2011, Feinberg et al. 2016).
Parameters 3 and 4 (activity estimates and SE/ME aircraft splits) were evaluated for a subset of
airports for which uncertainty in the extent to which national average fractions represented the
individual airport would most influence whether or not model-extrapolated concentrations are
above the lead NAAQS, as described in Section 3.3. The uncertainty from these two parameters
and the fifth parameter (meteorological and other local factors) are additionally assessed
qualitatively in Section 4.4.
The duration of run-up operations and the concentration of lead in avgas were both found to
be highly influential in ground-level 3-month average lead concentrations in air attributable to
piston-engine aircraft. Run-up emissions accounted for 82% of the 3-month average lead
concentration attributable to piston-engine aircraft in EPA air quality modeling at a model
facility, and was a primary contributor to emissions in modeling conducted by Feinberg et al.
(Section 2, Appendix A) (Feinberg et al. 2016). Moreover, variation between the 5th and 95th
percentiles of average run-up times observed in EPA modeling resulted in an almost 8-fold
variation in concentration attributable to only run-up emissions (Appendix C). Similarly,
Feinberg et al. found greater variation in the duration of run-up than that of other modes of
operation in the LTO cycle (e.g., landing and take-off time in mode), and variation in run-up
time led to variation in concentrations downwind (Feinberg et al. 2016).
Similarly, the concentration of lead in avgas has a direct impact on atmospheric lead
concentrations attributable to piston-engine aircraft activity, where higher levels of lead in fuel
result in greater lead emissions and hence concentrations of lead in air. The ASTM standard for
48

-------
the maximum lead concentration in 100LL was used in the national analysis; however, the
amount of lead in the fuel can vary across fuel suppliers and by batch. The concentrations of
lead in air attributable to aircraft are expected to directly scale with the concentration of lead in
avgas; thus, the lead avgas concentration was used as a scalar in the calculation of model-
extrapolated concentrations at airports nationwide (see Equation 2, Section 3.1). Based on the
important influence of these two parameters (run-up time and avgas lead concentration) in
modeling 3-month average lead concentrations attributable to piston-engine aircraft activity,
additional information was gathered to further characterize each parameter in results from
both national and airport-specific activity analyses.
Information on average run-up times was collected from a series of studies that observed run-
up operations at five airports (Appendix C) (USEPA 2010a, Heiken et al. 20 14).57 The average
run-up time from each airport was used to develop a distribution of average run-up times.58
This distribution of run-up times provided a way to evaluate model-extrapolated lead
concentrations based on observations at a larger number of airports compared to the run-up
times used in the national analysis, which were based on observations at the model airport. The
distribution of average run-up time across the five airports was lognormally distributed with an
average of 70 seconds, compared to the 40- or 63-seconds used for SE or ME aircraft,
respectively, in the national analysis (Table 5). The relationship between variation in run-up
time and concentrations of lead in air at and downwind of the maximum impact area was not
characterized in the additional studies used to develop the distributions of average run-up
times, and thus observations at the model airport were used to characterize how changes in
run-up time impacted changes in lead concentrations in the maximum impact area and
downwind (See Appendix C for details).
The distribution of average run-up times combined with an understanding of the relationship
between run-up time and downwind lead concentrations attributable to piston-engine aircraft
provided the necessary inputs for conducting a Monte Carlo analysis. The objective of the
Monte Carlo analysis was to characterize the impact of variation in the 3-month average run-up
time at a given airport on 3-month average model-extrapolated lead concentrations.
Conceptually, the Monte Carlo analysis entailed repeatedly selecting a run-up time value from
the distribution of average run-up times, and then adjusting the model-extrapolated lead
concentration based on the difference between the selected run-up time and the run-up time
used in the national analysis. For example, if an average run-up time of 70 seconds was selected
from the distribution of average run-up times, then the national model-extrapolated
concentration for SE piston aircraft would be adjusted up to account for the 30 second
difference between the time used in the national analysis (40 seconds) and the time selected in
the Monte Carlo draw. The amount of increase in concentration in this example would be based
57	One airport was included in two different studies, so while four unique airports were included in the studies
referenced here, a total of five observational periods is included in the combined dataset.
58	The use of average run-up times was selected as more representative of run-up times over a 3-month period,
the time period of the model-extrapolated concentrations, than the variability observed in the raw run-up time
data. For consistency with the national analysis, the median, rather than mean, run-up time at RHV was retained in
the distribution of run-up times across the five airports included here.
49

-------
on the relationship observed between run-up time and concentration at each distance
downwind at the model airport, such that the concentration of lead in air would increase more
at the maximum impact site than locations downwind (see Table C-l in Appendix C). The
resulting model-extrapolated concentration at each location, which accounted for the change in
run-up time, would then be used to adjust the model-extrapolated concentration resulting from
the national analysis. The Monte Carlo analysis used 10,000 iterations (i.e., 10,000 average run-
up times were selected from the distribution and used to adjust the model-extrapolated
concentration at each airport, at each downwind distance, which produced 10,000 adjusted
concentrations that then provided a range of potential concentrations at each airport, at each
downwind distance, based on variation in run-up time).
A similar approach was used to characterize the impact of variation in avgas lead
concentrations on 3-month average model-extrapolated atmospheric lead concentrations.
Available data from FAA and EPA reporting lead concentrations in avgas samples had an
average lead concentration of 1.79 g/gal and were normally distributed within the range
specified for 100LL (i.e., 1.70 to 2.12 g/gal) (see Appendix C for details on avgas lead data and
their distribution). A Monte Carlo analysis was used to characterize variation in 3-month
average model-extrapolated lead concentrations based on variation in avgas lead
concentration. As with run-up time, a value was selected from the distribution of avgas lead
concentrations (Table 5), and then used to scale a model-extrapolated concentration. For
example, if an avgas lead concentration of 1.80 was selected from the distribution, a model-
extrapolated concentration would be scaled by 0.85 (i.e., 1.80/2.12) to decrease extrapolated
concentration and account for a lower concentration of lead in fuel. The Monte Carlo analysis
was conducted 10,000 times. Results of the avgas lead and run-up time Monte Carlo analyses
were combined per Equation 3 to provide model-extrapolated concentrations that account for
variation in each parameter at and downwind of the maximum impact area at each US airport
(see Appendix C for details).
Eq. 3:
Monte Carlo Adjusted Lead Concentration, [Pb]Mc = LMC/2.12^(YnXCn)
Where:
LMc= concentration of lead in avgas (g/gal) from Monte Carlo analysis of avgas lead distribution
Yn= model-extrapolated concentration from national analysis at location n
Cn= %difference change in concentration at location n due to change in run-up time (see Equation C-l)
N= location at or downwind of maximum impact (i.e., 0, 50, 100, 150, 200, 250, 300, 400, 500 meters)
50

-------
Table 5. Monte Carlo analysis inputs for characterizing variability in key AQF parameters
Variable
National Analysis
Monte Carlo Analysis
Assumptions
Data
Source
Value
Data Source
Mean
(SD)
Range
Distribution
Shape
Run-up Time
(seconds)
Model
facility
(Appendix
A)
SE: 40
ME: 63
(USEPA
2010a, Carr
etal. 2011,
Feinberg and
Turner
2013)(Appen
dix A) (n=5)
Model
Airport
(Appendix A)
SE& ME:
70 (21)59
Min: 49
Max: 91
Log-normal
(Time in
Mode)
Exponential
(distance)
We assume that the log-normal distribution of data from the
five airports noted in text is representative of the distribution
of piston aircraft run-up times nationwide since these are the
only data in the literature reporting this information. We
assume that bounding the distribution by one sigma above and
below the logarithmic mean is representative of average run-
up times over a 3-month period.
The lead concentration attributable to run-up decreases as a
negative power law with distance from the maximum impact
site. As such, increases or decreases in run-up time compared
to an average influences lead concentration more at 0 or 50 m
from run-up than at 500 m meters for run-up. Our modeling
suggests an exponential curve describes the relationship
between run-up time and variability in lead concentration
estimate (see Appendix Cfor details).
Avgas Lead
Concentration
(g/gal)
ASTM
standard
2.12
EPA & FAA
fuel samples
(n=116)
1.79
(0.27)
Min: 1.70
Max: 2.12
Normal
We assume that the normal distribution of data from EPA and
FAA fuel samples is representative of the distribution of avgas
lead content at all US airports. The EPA fuel data were
collected during modeling studies discussed in Section 2. FAA
published a study reporting the lead concentration of avgas
fuel samples which was also used in this analysis.
We bounded the distribution based on the ASTM fuel
specifications for 100 octane Very Low Lead avgas (100VLL)
which has a lead concentration of 1.70 g/gal, and 100 Low Lead
(100LL) which has a maximum lead concentration of 2.12 g/gal.
59 As noted in the text, the average run-up times observed in four studies were used in combination with the median run-up time observed at the model
airport, and used in the national analysis, to develop a distribution of average run-up times. As such, the standard deviation here is the SD of average values.
51

-------
4, Model-Extrapolated Le.< j <. ncentrations: Results an <, x ertainty
Characterization
In this section we present results of the national analysis and the evaluation of individual
airports with the potential to be above the lead NAAQS described in Sections 3.2 and 3.3,
respectively, as well as the results of our methods to characterize uncertainty and variability in
model-extrapolated concentrations of lead from piston-engine aircraft operating at US airports.
Section 4.1 provides results of the national analysis; we then further evaluate of the impact of
the wind speed, and, separately, multi-engine aircraft activity on lead concentrations at the
maximum impact site. Lastly, Section 4.1 characterizes performance of the model-extrapolation
methodology through a comparison of results to monitored concentrations. Section 4.2
provides results of using airport-specific data to refine concentration estimates at airports with
the potential for lead concentrations to be above the lead NAAQS, and similarly characterizes
performance through comparisons of results with monitored concentrations. Section 4.3
discusses the results of the quantitative uncertainty analysis on variability in run-up durations
and avgas lead concentrations. Finally, Section 4.4 discusses qualitative uncertainty analyses for
results from both national and airport-specific activity analyses.
4.1 Ranges of Lead Concentrations in Air at Airports Nationwide
The national analysis methods described in Section 3.2 produced estimates of 3-month average
model-extrapolated lead concentrations at and downwind of maximum impact areas at 13,153
airports nationwide. These model-extrapolated concentrations are calculated for 3-month
periods of peak activity at each airport, and are attributable only to piston-engine aircraft
activity.60 Recall that model-extrapolated concentrations should decrease with increasing
distance from maximum impact area, based on the AQFs used in the analysis (Table 4), and that
concentrations across all sites should generally correlate with estimates of piston-engine
aircraft activity given the relationship between activity and concentration described in Section
3.3.2. Table 6 shows that indeed model-extrapolated concentrations decrease as distance from
the maximum impact area increases (left to right in table), and higher levels of piston-engine
activity (i.e., LTOs) generally correlate with higher model-extrapolated concentrations (top to
bottom in table). The decrease in model-extrapolated concentrations with increasing distance
from the maximum impact area has also been observed in lead monitoring data near airports
servicing piston-engine aircraft (Environment Canada 2000, Fine et al. 2010, Anchorage DHHS
2012), as well as lead modeling work conducted by others (Feinberg et al. 2016), and conforms
to near field concentration gradients for other primary pollutants.
60 As discussed in Section 2, since model-extrapolated lead concentrations are attributable to piston-engine aircraft
activity only, these lead concentrations may not reflect the total lead concentration (i.e., local emissions other
than aircraft as well as local background lead concentrations are not included in the estimates provided in Table 6).
52

-------
Table 6. Ranges of Piston-Engine LTOs and 3-month model-extrapolated lead concentrations at and
downwind of maximum impact areas at airports nationwide during 3-month peak activity61'62
LTOs
Model-Extrapolated Concentrations of Lead (ng/m3) at and Downwind of the Maximum
Impact Area
Max
Site
50 m
100 m
150 m
200 m
250 m
300 m
400 m
500 m
3,616 -
26,816
0.155-
0.475
0.038-
0.116
0.018-
0.054
0.013-
0.040
0.011-
0.032
0.009-
0.027
0.006-
0.019
0.005-
0.014
0.003-
0.010
2,579 -
8,814
0.100-
0.154
0.024-
0.038
0.011-
0.018
0.008-
0.013
0.007-
0.011
0.006-
0.009
0.004-
0.006
0.003-
0.005
0.002-
0.003
1,783 -
5,728
0.075-
0.100
0.018-
0.025
0.009-
0.012
0.006-
0.009
0.005-
0.007
0.004-
0.006
0.003-
0.004
0.002-
0.003
0.0017-
0.0023
1,275 -
4,302
0.050-
0.075
0.012-
0.018
0.006-
0.009
0.004-
0.006
0.003-
0.005
0.003-
0.004
0.002-
0.003
0.0015-
0.0023
0.0011-
0.0017
160-
2,889
0.0075-
0.050
0.002-
0.012
0.001-
0.006
0.001-
0.004
0.001-
0.004
0.0004-
0.003
0.0003-
0.002
0.0002-
0.0016
0.00002-
0.001
<1-446
< 0.0075
< 0.002
< 0.001
< 0.001
< 0.001
< 0.0004
< 0.0003
< 0.0002
< 0.0002
The relationship between piston-engine aircraft activity and model-extrapolated concentrations
is discussed further below; this relationship is influenced by a few key factors that include the
fraction of SE and ME piston-engine aircraft, and wind speed at a given airport. Looking
specifically at model-extrapolated concentrations at maximum impact areas, results show a
range of <0.0075 to 0.475 |-ig/m3 at airports nationwide, depending on aircraft activity levels
(Table 6). Inspecting the ranges of activity and model-extrapolated concentrations reveals that
there is a wide range of activity that could result in model-extrapolated concentrations above
the lead NAAQS. The airports with comparatively higher lead concentrations and 3-month
maximum activity levels between 3,616 and 26,816 LTOs represent a mix of airports, some of
which are dominated by SE aircraft activity and some of which have a mix of SE and ME aircraft
activity. As noted earlier, SE activity results in lower lead concentrations per LTO compared with
ME activity. Figure 5 presents a plot of the relationship between 3-month average
concentrations and activity, with the relative amount of ME depicted in shades of blue. As
indicated in Figure 5, more activity occurs at an airport dominated by SE aircraft to result in
lead concentrations similar to those at other facilities where there is a mix of ME and SE
aircraft. The mix of SE and ME activity, along with other characteristics of airports with model-
extrapolated concentrations above the lead NAAQS is explored further in Section 4.2.
61	As discussed in Section 3.2, model-extrapolated concentrations in Table 6 are attributable to piston-engine
aircraft activity and do not include local background lead concentrations.
62	In monitoring 3-month average lead concentrations at airports, concentrations in ng/m3 are typically presented
out to two decimal places. Additional decimal places and/or significant figures are shown in this table and in select
other figures either to demonstrate the trend of lead concentrations further downwind of the maximum impact
location or at airports with few operations.
53

-------
Relationship Between 3-Month Lead Concentration and 3-
Month Piston-Engine Aircraft Activity
(n = 12,932 airports)
_ 0.5
"J 0.45
s 0.4
Lo
S 0-35
Q.
- 0.3
E
| 0.25
X
2 0.2
-t-J
ra
e 0.15 +
0.1 --

-------
Range of LTOs Associated with Lead Concentrations at the
Maximum Impact Site
£
o
c
01
v
~
0
u
ID
TO
01
>0.15
0.10-0.15
0.075-0.10
0.050-0.075
0.0075 - 0.050
0-0.0075
5,000 10,000 15,000 20,000 25,000 30,000
Number of LTOs
B Non-Adjusted ~ Wind-Adjusted
Figure 6. Average 3-month model-extrapolated concentrations versus the number of piston-engine
LTOs during the same 3-month period at the maximum impact area runway end. Concentrations are
generally categorized relative to the lead NAAQS (e.g., greater than the standard of 0.15 ng/m3, less
than half the standard, 0.075 ng/m3, less than concentrations generally detected by monitors, 0.0075
ng/m3, etc).
Across all airports, the effect of the wind adjustment ranges from a 45% decrease in
concentration to a 210% increase in concentration; however, 48% of airports have
concentrations that change by less than 10%. The impact of the wind adjustment on maximum
impact site concentrations for all airports is shown in Figure 7. In absolute difference, the 3-
month maximum concentration at the maximum impact site changes by less than 0.01 |-ig/m3 at
most airports. At airports with concentrations greater than half the lead NAAQS, the absolute
concentration change from wind adjustment tends to be higher, from -0.06 to 0.16 |-ig/m3, as
shown in Figure 8. Overall, results of adjusting for wind speed show that while this parameter is
influential at individual airports, it does not meaningfully impact the range of concentrations in
the maximum impact area at airports nationwide. In turn, individual airports with the potential
to have concentrations above the lead NAAQS are evaluated more closely in Section 4.2.
55

-------
Percent Difference in Maximum Impact Site
Concentration Between Non-Adjusted and Wind-
Adjusted Concentrations at All Airports
2500
2000 -
>•
= 1500 -
 <3° ^ ^ C?' cS3 cy* "O'
,Ov £>v Ov O Q>v Ov O ^ O- o-
Concentration ng/m3
Figure 8. The absolute change in 3-month maximum concentration at the maximum impact site from
accounting for average inverse wind speed at airports with concentrations greater than % the NAAQS
for Lead.
The model-extrapolated concentrations from the national analysis presented above can be
evaluated through a comparison to monitored concentrations. Such an evaluation would ideally
be informed by monitored data that corresponds spatially and temporally with the model-
extrapolated concentrations. However, as detailed below, monitored lead concentrations are
only available at a subset of airports and none of these data are spatially and temporally
56

-------
consistent with model-extrapolated data. Nevertheless, a coarse comparison of model-
extrapolated to monitored concentrations is feasible for a subset of airports at which monitors
were placed proximate to the maximum impact area, or downwind, as part of evaluating
attainment of the lead NAAQS.63 In evaluating these comparisons, it is noteworthy that in
addition to spatial differences, monitored and model-extrapolated concentrations differ in
temporal periods and scope. Model-extrapolated concentrations were calculated for 2011
while monitored concentrations were collected over different 1-year periods depending on the
airport.64 As described in Section 3.1, while 2011 is expected to be generally representative of
piston-engine aircraft activity during monitored periods, differences in the volume and type of
piston-engine activity (i.e., SE vs. ME, full LTO vs. T&G) and meteorological conditions would be
expected to impact the comparisons presented here. In addition, model-extrapolated
concentrations are specific to aircraft lead emissions, while monitored concentrations include
background lead from other sources. Other factors could influence lead concentrations in air
from year-to-year as well, and both monitored, and model-extrapolated concentrations also
have inherent variability and uncertainty. With the characteristics of each dataset in mind,
Figure 9 provides a coarse comparison of national model-extrapolated to monitored
concentrations at three airports with monitors placed proximate to the maximum impact area
or downwind locations.65 Each panel presents the monitored NAAQS design value (i.e.,
maximum 3-month average concentration during monitored time period) along with model-
extrapolated concentrations. Across these airports, model-extrapolated and monitored
concentrations generally align when considering both the downwind gradient, and horizontal
transport of lead emissions at the maximum impact area.
63	Logistical considerations (e.g., aviation safety clearance regulations for siting fixed objects near the landing and
take-off area, and availability of power in these locations) typically prevented placement of lead monitors in the
maximum impact area.
64	Monitoring agencies were required to measure the maximum lead concentration in ambient air resulting from
specific lead sources, including a subset of airports USEPA (2010b). Revisions to Lead Ambient Air Monitoring
Requirements.; these monitoring data are part of the lead surveillance network that is used to evaluate attainment
of the NAAQS for lead (https://www3.epa.gov/ttnamtil/pb-monitoring.html). A summary of monitored data is
available on the EPA website USEPA. (2017a). "Airport Lead Monitoring and Modeling." 2017, from
https://www.epa.gov/regulations-emissions-vehicles-and-engines/airport-lead-inventories-air-aualitv-monitoring-
air.
65	Among the 17 airports where lead surveillance monitoring was conducted, eight NAAQS monitors were sited in
locations proximate to or downwind of the maximum impact area. Four are presented in this section with the
remaining four presented in Section 4.3. In two instances NAAQS monitors were sited particularly close to model-
extrapolated locations, which supported an extended comparison of monitored to model-extrapolated
concentrations, also in Section 4.3.
57

-------
Direction
Direction
, Max Impact
0.34 ng/m3
Monitor
0.07 |ig/m
Max Impact 50 m
0.20 ng/m3 0.05 ng/m3
Monitor:
0.07 (ig/m
Airport A	Airport B
Monitor
0.06 jig/m
Max Impact •
y 0.38 |ig/m3
Wind
Direction
Airport C
Satellite Image Source: Google Earth
Figure 9. Coarse comparison of monitored to model-extrapolated lead concentrations at airports with
NAAQS monitors sited proximate to the maximum impact area or locations downwind. Red dots
represent approximate monitor placement, while yellow dots represent approximate locations of
model-extrapolated concentrations from national analysis methods (Section 3.2). Blue arrows denote
the prevailing wind direction at each airport. As noted above, the year in which monitored
concentrations were collected varies by airport, while model-extrapolated concentrations represent
2011. All locations are based on scientific judgment of the alignment of model-extrapolated locations
58

-------
from the expected maximum impact area. The max impact concentrations represented in the figure are
not wind speed adjusted. The wind speed adjusted concentrations at max impact for airports A, B, C are
0.36, 0.23, and 0.44 ng/m3 respectively.
4.2 Airports with Potent!	)ncentrations Above the Lead NAAQS with
Unrestricted Access Within 50 m of the Maximum Impact Site
As described in Section 3.3, a series of sensitivity tests were performed to identify a subset of
airports beyond those identified in the national analysis where model-extrapolated lead
concentration estimates were above the NAAQS for lead. Additional data were then identified
to calculate airport-specific activity estimates for each airport in this subset.66 Next, the airport-
specific activity estimates for each airport were used to calculate updated model-extrapolated
lead concentrations for that airport with a focus on concentrations in the maximum impact
area. In addition, for each of these airports, satellite imagery was utilized to assess if there was
unrestricted access within 50 meters of the maximum impact site. The results of this screening
analysis are presented in Table 7.
Each column in Table 7 represents the outcome of analysis steps presented in Section 3.3 and
described in Table 4: the first column identifies the airport, the second column indicates the
lead concentration at the maximum impact site relative to the lead NAAQS using national
default parameters (Section 3.2); the third column adjusts the national default concentrations
based on average inverse wind speed (Section 3.2); the fourth and fifth columns present the
outcomes of airport-specific parameters that influence the potential for lead concentrations to
be above the NAAQS for lead; the sixth column shows the results of the airport-specific-activity
analysis before adjusting for average inverse wind speed; and the seventh column shows the
results using both airport-specific activity and airport-specific wind speed data. Black filled
circles indicate model-extrapolated concentrations are above the NAAQS for lead and white
unfilled circles indicate model-extrapolate concentrations that are more than 10% below the
NAAQS for lead. The potential impacts of additional local characteristics (e.g., mixing height,
local terrain) on airport-specific estimates of lead concentration are discussed qualitatively in
Section 4.4.
66 As described in Section 3.3, airport-specific data consist of the number of SE and ME piston-engine aircraft based
at an airport. Airport-specific activity estimates were calculated using the following steps. First, the number of
LTOs specific to piston-engine aircraft was estimated by summing the number of SE and ME piston-engine aircraft
based at an airport and dividing the sum by the total number of aircraft based at an airport, then multiplying the
fraction by total LTOs at the airport. Next, the fraction of piston-engine aircraft LTOs conducted by SE piston
aircraft was calculated by dividing the number of SE based aircraft by the total number of SE and ME based aircraft
at an airport. The same approach was used to calculate the fraction of piston-engine aircraft LTOs conducted by
ME piston aircraft. For airports where no based aircraft data were available or for where based aircraft numbers
represented fewer than one aircraft for every 730 operations, national default splits were used for the airport-
specific activity estimates.
59

-------
Among the airports in Table 7, air quality monitoring has been conducted at RHV at a location
approximately 60 m downwind from the maximum impact site. Lead concentrations at RHV
measured 60 m downwind were above half the level of the lead NAAQS.67
Table 7. Airports with Model-Extrapolated Lead Concentrations Potentially Above the Lead NAAQS at
the Maximum Impact Area With Unrestricted Areas Within 50 Meters.
Airports68 National Wind % Piston Runway Based Based
Defaults Adjusted Adjusted Shift Aircraft Aircraft
Wind Adj.
52F
•
•
•
•
•
•
RHV
•
•
•
•
•
•
ORS
O
•
•
o
•
•
WHP
o
o
•
o
•
•
For the airports identified in Table 7, model-extrapolated concentrations increase when using
airport-specific data to estimate piston-engine aircraft activity; the magnitude of the increase
varies based on the difference between the airport-specific fleet and operational characteristics
compared with the national average values used for piston-engine aircraft activity. The
percentage of piston-engine activity estimated as SE versus ME also influences the magnitude
of change between airport-specific and national analysis results. As described previously and in
greater detail in Section B.4, the use of based aircraft to estimate piston activity, as well as SE
and ME splits in activity was evaluated by comparing on-site observations with based aircraft at
a subset of airports and reasonable agreement was observed (within 10%) between based
aircraft and on-site observations. Additional factors that influence model-extrapolated
concentrations (e.g., run-up time, avgas lead concentration) are discussed in Section 4.3. Figure
10 presents the model-extrapolated lead concentrations in the maximum impact area from
both the airport-specific analysis and the national analysis at individual airports where lead
concentrations at the maximum impact site with unrestricted access may potentially be above
the lead NAAQS.
67	See the program overview titled Airport Lead Monitoring:
https://nepis.epa.eov/Exe/ZvPDF.cgi/P100L)DW.PDF?Dockev=P100LIDW.PDF
68	Airport codes are commonly used to identify airports; the name and location of airports in this table is provided
in Appendix B.
60

-------
c
o
0.50
c
(L)
U
c
o
u
T3
ro ,

I
QC
Individual Airport
~ National Default Wind-Adjusted ~ Based Aircraft Wind-Adjusted
Figure 10. Comparison of model-extrapolated lead concentrations from the wind-speed adjusted
national default parameters (orange squares; Section 3.2), and wind-speed adjusted airport-specific
activity analysis (blue diamonds; Section 3.3) at airports that have the potential for maximum impact
site concentrations to be above the NAAQSfor lead with unrestricted access.
Similar to national analysis results, results of the airport-specific activity analysis can be
evaluated through a comparison to monitored data. Of the airports included in the airport-
specific activity analysis, four had NAAQS surveillance monitors located proximate to or
downwind from the maximum impact area. Figure 11 presents the comparison of monitored
and model-extrapolated concentrations at these airports. As discussed in Sections 3.4 and 4.1,
the coarse comparison presented in Figure 11 has attendant uncertainties (e.g., spatial and
temporal differences between monitor and model-extrapolated data). Despite these
uncertainties, monitored data suggest that model-extrapolated concentrations which use
airport-specific activity estimates generally align with monitored concentrations. A more in-
depth comparison of model-extrapolated to monitored concentrations is presented in the
context of additional uncertainty analysis in Section 4.4.
61

-------
Wind
Direction
Direction
• Max Impact
0.334J.g/mS.
Max impact#
Monitor
0.12 ng/m3
•50 m
^0.14 ng/m
Monitor
0.06 ng/m
Airport D	Airport E
Direction
• Max Impact
0.23 ug/m3
100m
0.02 ng/m3
Max Impact Q
0.20 [ig/m3
Monitor
0.10 ng/m3
Monitor
0.17 ng/m
Wind
Direction
Airport F	Airport G
Satellite Image Source: Google Earth
Figure 11. Coarse comparison of monitored to model-extrapolated airport-specific lead concentrations
at airports with NAAQS monitors sited proximate to the maximum impact area or locations
downwind. Red dots represent monitor location, while yellow dots represent approximate locations of
model-extrapolated concentrations from airport-specific activity analysis (Section 3.3). Blue arrows
denote the prevailing wind direction at each airport. Locations for model-extrapolated lead
concentrations depicted here were based on approximated location of the dominant run-up location. The
max impact concentrations represented in the figure are not wind speed adjusted. The wind speed
adjusted concentrations at max impact for airports D, E, F, and G are 0.58, 0.31, 0.26, and 0.24 fig/m3
respectively.
62

-------
4.3 Quantitative Uncertainty Analysis of Concentrations of L« Air at Airports: The
Influence of Run-up Time «>*< as Le •>. oncentration
As with any analysis of this scope in which estimates of pollutant concentrations at facilities
nationwide are developed using an extrapolation approach, there is inherent uncertainty and
variability in the estimates. The focus here is on two key parameters that have been
demonstrated in previous studies to impact lead concentrations at and downwind from the
maximum impact area at airports: run-up time and avgas lead concentration. Run-up time and
avgas lead concentrations are not constrained by the functional role of a given airport, but
rather vary across airports independently of airport attributes. These two parameters were
thus the focus of a quantitative variability evaluation using a Monte Carlo analysis, which is
discussed in Section 3.4 above. Additional meteorological and local considerations may
contribute to uncertainty at individual airports; the uncertainty from these parameters is
discussed qualitatively in Section 4.4.
4.3.1 National Analysis and Airport-Specific Monte Carlo Results
Figure 12 shows the national analysis results with Monte Carlo bounds around each model-
extrapolated concentration for the airport with the highest, and, separately, the airport with
the lowest model-extrapolated concentration at the maximum impact site and downwind
locations. As the Monte Carlo bounds show, variability in run-up duration and avgas lead
concentrations add uncertainty to the exact range of model-extrapolated concentrations
nationwide (i.e. exact value of the highest and lowest model-extrapolated concentration in the
maximum impact area and downwind locations of US airports); however, the quantitative
uncertainty shown in the Monte Carlo is small enough such that it does not obscure meaningful
differences between model-extrapolated concentrations at different US airports.
Further, Monte Carlo results consistently show the potential for higher model-extrapolated
concentrations than the national analysis results (compare black or blue dots to upper error
bars in Figure 12). The potential for higher model-extrapolated concentrations is due to the
difference in observed run-up times at the model airport compared to run-up times observed at
airports included in the Monte Carlo analysis. As noted in Section 3.4 the deterministic national
analysis used 3-month median run-up times for SE and ME, separately, which were measured at
the model airport at which AQFs were developed, while the Monte Carlo analysis included
observations of longer run-up times from studies at additional airports (Table 5). The increase
in model-extrapolated concentrations due to the potential for longer durations of run-up at
airports nationwide compared to that observed at the model airport, generally aligns with a
sensitivity analysis conducted at the model airport. The sensitivity analysis showed that
increasing run-up time from the 5th (16 seconds) to 95th (121 and 160 seconds for SE and ME
respectively) percentiles resulted in approximately an order of magnitude increase in 3-month
average modeled concentrations (i.e., 5th to 95th percentiles of 3-month average modeled
concentrations increased from 0.043 to 0.322 |-ig/m3 and from 0.005 to 0.035 |-ig/m3 for SE and
ME, respectively) (Appendices A and C). The average run-up time at a given airport may be
impacted by a number of factors (e.g., the number of pilots in training); however, the use of
63

-------
average run-up time from airports with available data provides relevant information to
characterize the potential range of concentrations at airports nationwide in a manner
consistent with the approach laid out in Section 3.
While the concentration of lead in avgas is also included in the Monte Carlo analysis, this
parameter influences results less than run-up duration for two reasons. First, the range of lead
in avgas is smaller than the range of average run-up times used in the analysis (Table 5).
Second, the impact of longer run-up durations is additive, whereas the impact of lower avgas
lead concentrations is incremental (i.e., each additional second of run-up compared to the
median value used in the national analysis contributes the same amount to downwind lead
concentrations, whereas fuel with 2.10 g/gal lead rather than the 2.12 g/gal contributes
0.02 g/gal less to emissions). The difference in the influence of these parameters helps explain
why the uncertainty analysis for model-extrapolated concentrations consistently demonstrates
higher values compared with the point estimate.
Extrapolated Lead Concentration at Maximum Impact and Locations
Downwind
E
=±
o

a;
o
CO
0.1
0.01
ai 0.001
00
0.0001
£
i
£
£
I
i i
I
1
	1	1	1	1	1	1	1	1	1
Max. 50 m 100 m 150 m 200 m 250 m 300 m 400 m 500 m
Impact
Site
Distance Downwind from Max. Impact
Figure 12. The range of model-extrapolated lead concentrations at and downwnind of the maximum
impact area based on national analysis results. Black diamonds represent the maximum and blue
squares represent the minimum model-extrapolated concentration at each location for the 13,153
airports included in the national analysis. Error bars are the concentrations at the 97.5th percentile of
Monte Carlo results, which account for potential ranges in run-up time and avgas lead concentrations
across airports.
Similar to the Monte Carlo bounds around national analysis results, the model-extrapolated
concentrations from the airport-specific activity analysis are consistently at or near the 2.5th
percentile of the Monte Carlo bounds while the 50th percentiles and 97.5th percentiles of the
Monte Carlo analysis are on average 38% and 91% higher than the model-extrapolated
concentrations from the airport-specific activity analysis. As discussed above, this observation is
64

-------
primarily the result of having used a shorter run-up time in developing the model-extrapolated
lead concentrations in the national analysis compared with run-up times that have been
observed at other airports, which were used in the Monte Carlo analysis (Table 5). In addition,
the greater influence of run-up time versus lead concentrations in avgas on ground-based
atmospheric lead concentrations, leads to changes in run-up time dominating the potential
range of concentrations observed in the Monte Carlo results (ICF 2014, Feinberg et al. 2016).
The uncertainty results presented here are sensitive to the choice of input distributions for
avgas lead concentration and run-up time.
4.3.2 Comparison of Model-Extrapolated Concentrations From the Airport-Specific Activity
Analysis with Monte Carlo Bounds to Monitored Concentrations in the Maximum Impact Area
To evaluate the approach for calculating airport-specific model-extrapolated concentrations
with Monte Carlo bounds, results from the approach were compared to relevant monitoring
data. Comparisons between model-extrapolated and monitored lead concentrations are most
informative when the model-extrapolated and monitor concentrations are in the same
approximate location. Two airports had monitors located in close proximity to the location of
the model-extrapolated concentrations; however, monitoring at each airport was conducted
during different time periods than the time period of national analysis. Thus, model-
extrapolated concentrations were adjusted to reflect activity and meteorological data from the
monitored time periods. The same national analysis data sources were used to update activity
and meteorology in model-extrapolated concentrations to monitored time periods (See Section
3.2, Table 2 for data source details). In addition, as with the airport-specific activity analysis,
onsite observational survey data or data on the number and class of aircraft based at the
airport were used to calculate piston-engine aircraft activity, as well as SE and ME activity at
each airport.69
Figure 13 compares the rolling 3-month average model-extrapolated concentrations at the two
airports with monitored data in similar locations.70 At the airport in Panel A, two lead monitors
were co-located proximate to the maximum impact area; the primary monitor is identified with
a blue dot, the co-located monitor with a black dot, and the model-extrapolated concentrations
(based on the lower run-up time estimates) are identified with green dots. Model-extrapolated
lead concentrations at this facility are consistently lower than lead concentrations measured at
the primary monitor with the difference ranging from 12% to 52% yet the Monte Carlo bounds
reflecting potential variation in model-extrapolated values due to variability in run-up duration
and avgas lead concentrations consistently include the primary monitored value. Model-
extrapolated concentrations at the airport in Panel A identified the majority of 3-month
monitored concentrations that exceeded the lead NAAQS (noted by the red line).
69	The following percentages were used to allocate total LTOs given observational survey or based aircraft data: 70
and 86% piston-engine, 73 and 98% SE, 27% and 2% ME for each airport, respectively. See Appendix C for details
on observational survey data; based aircraft data are from Airnav.com (May 2016).
70	The time period of rolling 3-month average is used here for comparison with the lead NAAQS. Model
extrapolated values presented in Figure 13 are not wind-speed adjusted.
65

-------
Similarly, model-extrapolated concentrations appropriately reflect attainment of the lead
NAAQS at the airport in Panel B. In this instance, both model-extrapolated (green dots) and
monitored (blue dots) concentrations are below the NAAQS. In addition to providing an
example of model-extrapolation performance below the NAAQS, Panel B, also provides an
example of a location further downwind than the maximum impact area. At this airport, the
monitor was located approximately at the 50-meter downwind model-extrapolation site, which
along with activity and other parameters discussed in previous sections, explains the lower
concentrations relative to the airport in Panel A.
66

-------
Panel A
ff}

e





c

o
ro
°4«#
QJ
e
L.
m
c
t!
CD
TO
U
C
o
Q.
E
u
"*¦**
"O
E
ra
3
Jli
E
0>
X
m
m
S
£

c


M7%.
¦M* j%22%
»
_1_

o
E
CO
Mar-May Apr-Jura
2012 2012
10098
tTOs
11612
ITOs
May-Jui
2012
12174
I.TOS
Ifjl
/i't;
LI' h
A\ ft ui« Sep-No\
"i" 2012
I5 .41
10788
tTOs
CM Mm
n~u« ;i w
i Jh 'Ml '
f I K
v. n\
h'b
.»C » I
8<'bl
4R VU"
2013
UVMh
tTOs
feb-Apr
2013
10500
UOs
S Extrapolated 3-month Concentration
|
I
—, m
c* Q)
-.I
E
3
C .1
oi x
0	m
§ E
uf5
"O
.£
9J 3
60 C
5
_ o
> -O
1	Si
o g
E ?
eh °
m in
"8
JU
OJ
sa
1
Panel B
JtsS-Sep
2011
Aug-Oct
2011
3639110s
110%
Sep-Nov
201}
4229ITGs
OcC-Dec
2011
45811.TCK
Nov201Man
2012
3839 LTOs


Dec 2011-Feb
201,2
3821ITOs
I
msm
kv%
I Primary Monito
Figure 13. Comparisons of model-extrapolated (green dots) to monitored (blue and black dots)
concentrations at the two airports with monitors placed proximate to model-extrapolated locations.
The airport in Panel A had both a primary and co-located monitor (blue and black dots, respectively) in
the maximum impact area. The airport in Panel B had a monitor approximately 50 meters downwind
of the maximum impact site. The red line denotes the NAAQS for lead (i.e., rolling 3-month average of
0.15 ng/m3).
67

-------
4.4 Qualitative Characterization of Uncer	inability in Model-Extrapolated
b ncentrations from National and Airport-Specific Activity Analyses
As discussed in Sections 1 and 2, emissions from piston-engine aircraft during run-up is the
single largest contributor to the maximum impact area concentrations for lead from this source,
and there is consistency in how and where these run-up operations are conducted across
airports. The run-up emissions are released near the surface while the aircraft is stationary,
occur in a flat terrain that is required for landing and take-off, and predominately impact
receptor sites nearby (i.e., up to 500 meters downwind) (Carr et al. 2011, Feinberg et al. 2016)
(Appendix A). While the consistent nature of piston-engine aircraft run-up emissions results in a
straight-forward dispersion modeling scenario that can be used to extrapolate to other airports,
key parameters impart uncertainty on the model-extrapolated results. This section qualitatively
discusses additional sources of uncertainty that were not addressed in previous sections,
namely uncertainty from meteorological, dispersion modeling, and operational parameters.
4.4.1 Meteorological Parameters
Several meteorological parameters affect modeled concentrations that result from dispersion
modeling of pollutant emissions released at surface level. These parameters include wind speed
and direction, mixing height, atmospheric stability, and ambient temperature since they directly
relate to conditions of atmospheric turbulence, thermal buoyancy, as well as resulting vertical
and lateral dispersion.
Low wind speeds disperse emissions less rapidly compared with high wind speeds, resulting in
higher concentrations near the emissions source. Conversely, higher wind speeds result in
lower concentrations near the emissions source. Specifically, as discussed in Section 3.2 and
demonstrated in Appendix A, the near-field concentration of a non-reactive pollutant
approximately scales with , where u is wind speed and angled brackets imply a time
average (Barrett and Britter 2008). Three-month average inverse wind speeds varied -23% to +
21% from the annual average wind speed. The range of inverse wind speeds at the model
airport results in 3-month AQFs that vary +23% to -15% from the annual average.
Approximately 51% of airports have 3-month average inverse wind speeds during the 3-month
period of maximum piston-engine aircraft activity at a single runway end that fall within the
range of 3-month average inverse wind speeds at the model airport.71 Thus, we do not expect
wind speed to be a significant source of uncertainty nationwide as sensitivity to wind speed will
be captured by the wind speed scaling technique applied, and 3-month AQFs were only
sensitive to wind speed by approximately +/-20% at the model airport. For individual airports at
the extremes of high and low wind speed, we recognize there is more uncertainty in the
extrapolated concentrations. 72 However, we do not expect significant GA activity during winds
below 2.6 m/s or above 10.3 m/s as FAA safety recommendations state that these may be
71	Wind speed data is from the nearest ASOS station to each airport. See Appendix A for additional information on
data sources.
72	At very low wind speeds, the inverse wind speed tends toward infinity and the wind speed scaling approach is
limited by the choice of modeled minimum wind speed and the resolution of the wind monitor data.
68

-------
conditions under which it is particularly challenging for a general aviation aircraft to fly (FAA
2006).
At both high and low wind speeds, significant variability in wind direction can result in
additional uncertainty. When wind direction shifts significantly, airport operators may or may
not initially change the runway end from which piston-engine aircraft take-off due to
considerations of cross-winds and operational consistency. As noted in Section 1, airports are
built such that one runway-end faces directly into the predominate wind direction, which limits
the likelihood of runway-end variability. Further, Section 3.3 discusses a sensitivity analysis that
evaluated the impact of shifting piston-engine aircraft operations to a specific runway-end,
which addresses instances such as when wind direction variability leads to differences between
the active runway-end and wind direction.
Mixing height is another meteorological condition that can influence atmospheric lead
concentrations both independently and in conjunction with wind conditions. When mixing
heights are very low, as is often the case overnight, then pollutants released at the surface
remain trapped in the shallow surface layer, resulting in higher concentrations. Higher mixing
heights occur when there is substantial surface mixing, which more rapidly disperses pollution
away from the surface and result in lower surface-level concentrations. An unstable
atmosphere where the mixing height is changing rapidly will also affect the concentration of
lead at the maximum impact site. Previous air quality modeling conducted by EPA at individual
airports characterized the influence of mixing height on modeled aircraft lead concentrations
(Section 2; Appendix A) (Carr et al. 2011, Feinberg et al. 2016). At the model airport, there is a
strong relationship between the 3-month average wind speeds and mixing heights (Figure 14),
making it difficult to separately calculate the influence of mixing height on the AQFs. However,
because run-up is the largest contributor to lead concentrations at the maximum impact site,
the AQF at the maximum impact site is not expected to be sensitive to local mixing height.
Concentrations at sites downwind may be more sensitive to mixing height and atmospheric
stability, particularly during long periods of atmospheric inversion or at airports that have
mixing height characteristics significantly different from the model airport.
69

-------
850
M. 800
-M
w) 750
a>
x
QJD 700
c
X
^ 650
> 600
<
P 550
¦t 500
m
450
Mixing Height vs Wind Speed
O ....
0..--6
Cu
O-'O

..•O
.-•o
o'
y = 278.2x- 248.81
R2 = 0.9803
2.6	2.8	3.0	3.2	3.4	3.6
3-Month Avg. Scalar Wind Speed (m/s)
3.8
Figure 14. 3-month average mixing height at the model airport as a function of 3-month average
scalar wind speed at the model airport over the same period.
Microclimate conditions and other meteorological parameters may contribute to some
variability in the relationship between aircraft operations and resulting atmospheric lead
concentrations. For example, near-source maximum primary pollutant concentrations have
shown some dependence on ambient air temperature, but to a lesser extent than wind speed
(Liang et al. 2013). A preliminary analysis of 3-month AQFs at the model airport showed that
temperature was a significant variable (p-value =0.001046) when controlling for average
inverse wind speed; however, because average 3-month temperature varied by less than +/-2%
at the model airport, maximum impact and downwind concentrations were not sensitive to
ambient temperature. Thus, while results nationwide are not expected to be particularly
sensitive to microclimate conditions and other meteorological variables, there is more
uncertainty in model-extrapolated concentrations at airports that have maximum activity
periods during meteorological conditions not observed at the model airport.
4.4.2 AERMOD and AERSURFACE Parameters
Modeling parameters in AERMOD may be a source of both aleatoric and epistemic uncertainty.
73 Near-field surface and geographic characteristics may have an impact on lead concentrations
at and downwind of the maximum impact site. The calculation of AQFs included a fixed
parameterization of surface roughness, Bowen Ratio, and albedo as described in Appendix A,
73 Uncertainty can be classified into aleatoric uncertainty and epistemic uncertainty. Aleatoric uncertainty is often
characterized as natural randomness that is often difficult to measure. Epistemic uncertainty is typically
characterized as uncertainty due to the lack of data (e.g., data that could be collected but the methods may be
prohibitive).
70

-------
but downwind surface characteristics may differ at airports nationwide. Other research has
suggested that, at certain receptors, modelled AERMOD concentrations are sensitive to
changes in surface roughness length but indifferent to albedo and Bowen Ratio variation
(Grosch and Lee 2000, Karvounis et al. 2007). Further, the modeling approach does not
necessarily account for complex airflow around or near buildings and other obstructions. While
these factors may cause uncertainty at downwind concentrations, their impact on variability
near the maximum impact site is mitigated by requirements for on-airport characteristics and
land-use immediately downwind of runways due to landing and take-off safety requirements,
which results in some consistency nationwide. Where obstructions such as noise barriers or
fences may impact atmospheric lead concentrations near the maximum impact site,
extrapolated concentrations and their associated uncertainty should be considered on a case-
by-case basis. Finally, the aircraft were modeled as volume sources with fixed horizontal and
vertical plume extents, which may introduce uncertainty at airports with aircraft and engines
that differ significantly from those at the model airport. Details on the modeling approach for
aircraft sources, information on prior modeling work, and a comparison between piston-engine
aircraft included in the model airport modeling with those active at airports nationwide is
provided in Appendix A.
4.4.3 Operational Parameters
As discussed throughout the report, the availability, resolution, type, and detail of operational
data available at airports nationwide can contribute to uncertainty in the estimated lead
concentrations. The impact of airport-specific fleet heterogeneity (i.e. piston/turboprop split
and single-engine/multi-engine split) was explored through the use of airport-specific data for a
subset of airports in Section 4.2. However, other local fleet characteristics (e.g. distribution of
aircraft engine types operating at the airport) are not accounted for in the analysis and may
also contribute to uncertainty at specific airports that have distinct local characteristics. The
nature of piston engines means that there is also a great deal of variability in their emissions,
even for the same pilot operating the same airplane (Yacovitch et al. 2016); however, the
sensitivity of atmospheric lead concentrations to this variability should be minimized by
averaging concentrations over a 3-month period. Similarly, the diurnal profile of aircraft activity
may influence local lead concentrations over short timescales, but is not expected to be a
sensitive parameter in determining 3-month average concentrations as discussed in Appendix
B. Regional, local, and seasonal differences in daily operational patterns may contribute
additional uncertainty to that discussed in Appendix B. However, given the insensitivity of
average concentrations to different diurnal patterns in sensitivity analysis modeling, these are
not expected to contribute significantly to uncertainty in extrapolated concentration estimates
for airports nationwide. In modeling individual airports, national fleet and operational data
should be supplemented with local data where available and feasible.
71

-------
References
Anchorage DHHS (2012). Merrill Field Lead Monitoring Report. Municipality of Anchorage Department of
Health and Human Services. Anchorage, Alaska. December 2012.
http://www.muni.ore/Departments/health/Aclmin/environment/AirQ/Documents/Merrill%20Field%20L
ead%20Monitoring%20Studv 2012/Merrill%20Field%20Lead%20Studv%20Report%20-%20final.pdf.
ASTM International (2016). Standard Specification for Leaded Aviation Gasolines, D910.
Barrett, S. R. H. and R. E. Britter (2008). Development of algorithms and approximations for rapid
operational air quality modelling. Atmospheric Environment, 42 (34), 8105-8111. DOI:
http://doi.Org/10.1016/i.atmosenv.2008.06.020.
Carr, E., M. Lee, K. Marin, C. Holder, M. Hoyer, M. Pedde, ... J. Touma (2011). Development and
evaluation of an air quality modeling approach to assess near-field impacts of lead emissions from
piston-engine aircraft operating on leaded aviation gasoline. Atmospheric Environment, 45 (32), 5795-
5804. DOI: http://dx.doi.Org/10.1016/i.atmosenv.2011.07.017.
Chang, J. and S. Hanna (2004). Air quality model performance evaluation. Meteorology and Atmospheric
Physics, 87 (1), 167-196.
Environment Canada (2000). Airborne Particulate Matter, Lead and Manganese at Buttonville Airport.
Conor Pacific Environmental Technologies for Environmental Protection Service. Ontario.
FAA (2010). General Aviation and Part 135 Activity Surveys - CY 2010. F. A. Administration.
FAA (2012). General Aviation Airports: A National Asset
FAA (2014). General Aviation and Part 135 Activity Surveys - CY 2014. F. A. Administration.
FAA (2015). General Aviation and Part 135 Activity Surveys - CY 2015. F. A. Administration.
FAA (2017). FAA Form 5010, Airport Master Record.
Feinberg, S. and J. Turner (2013). Dispersion Modeling of Lead Emissions from Piston Engine Aircraft at
General Aviation Facilities. Transportation Research Record: Journal of the Transportation Research
Board,{2325), 34-42.
Feinberg, S. N., J. G. Heiken, M. P. Valdez, J. M. Lyons and J. R. Turner (2016). Modeling of Lead
Concentrations and Hot Spots at General Aviation Airports. Transportation Research Record: Journal of
the Transportation Research Board, 2569, 80-87. DOI: 10.3141/2569-09.
Fine, P., A. Polidori and S. Teffera (2010). General Aviation Airport Air Monitoring Study. South Coast Air
Quality Management District.
Grosch, T. G. and R. F. Lee (2000). Sensitivity of the AERMOD air quality model to the selection of land
use parameters.
Heiken, J., J. Lyons, M. Valdez, N. Matthews, P. Sanford, J. Turner and N. Feinberg (2014). Quantifying
Aircraft Lead Emissions at Airports. ACRP Report 133. http://www.nap.edu/cataloe/22142/quantifvine-
aircraft-lead-emissions-at-airports.
ICF (2014). Final Report: Modeling Analysis of Air Concentrations of Lead from Piston-engine Aircraft. ICF
International.
Karvounis, G., D. Deligiorgi and K. Philippopoulos (2007). On the sensitivity of AERMOD to surface
parameters under various anemological conditions.
Liang, M. S., T. C. Keener, M. E. Birch, R. Baldauf, J. Neal and Y. J. Yang (2013). Low-wind and other
microclimatic factors in near-road black carbon variability: A case study and assessment implications.
Atmospheric environment (Oxford, England: 1994), 80, 204-215. DOI: 10.1016/j.atmosenv.2013.07.057.
Lohr, G. W. and D. M. Williams (2008). Current practices in runway configuration management (RCM)
and arrival/departure runway balancing (ADRB). NASA/TM-2008-215557 NASA.
http://ntrs.nasa.eov/archive/nasa/casi.ntrs.nasa.eov/20090010329.pdf.
72

-------
Luecken, D., W. Hutzell and G. Gipson (2006). Development and analysis of air quality modeling
simulations for hazardous air pollutants. Atmospheric Environment, 40 (26), 5087-5096.
National primary and secondary ambient air quality standards for lead (40 CFR 50.12). Amended Nov.
12. 2008.
USEPA (2010a). Development and Evaluation of an Air Quality Modeling Approach for Lead Emissions
from Piston-Engine Aircraft Operating on Leaded Aviation Gasoline. EPA-420-R-10-007.
https://nepis.epa.eov/Exe/ZvPDF.cei/P1007H4Q.PDF?Dockev=P1007H4Q.PDF.
USEPA (2010b). Revisions to Lead Ambient Air Monitoring Requirements.
USEPA. (2011). "2011 National Emissions Inventory (NEI) Data." 2017, from http://www.epa.gov/air-
emissions-inventories/2011-national-emissions-inventorv-nei-data.
USEPA. (2016a). "2014 National Emissions Inventory (NEI) from http://www.epa.gov/air-emissions-
inventories/2014-national-emissions-inventorv-nei-data.
USEPA (2016b). Review of the National Ambient Air Quality Standards for Lead EPA-HQ-OAR-2010-
0108; FRL-9952-87-OAR.
USEPA. (2017a). "Airport Lead Monitoring and Modeling." 2017, from http://www.epa.gov/regulations-
emissions-vehicles-and-engines/airport-lead-monitoring-and-modeling.
USEPA. (2017b). "Learn About Lead Designations." 2018, from http://www.epa.gov/lead-
designations/learn-about-lead-designations#process.
Yacovitch, T., Z. Yu, S. C. Herndon, R. Miake-Lye, D. Liscinsky, W. Knighton, ... P. Pringle (2016). Exhaust
Emissions from In-Use General Aviation Aircraft (ACRP Report No. 164).
73

-------
Appendices

-------
Appendix A: Supplemental information on Detailed Air Quality Modeling at a Model
Airport
As described in Section 2, factors that relate aircraft activity to resulting air lead concentrations,
referred to as Air Quality Factors (AQFs), are used to estimate atmospheric lead concentrations
at airports nationwide by extrapolating the relationship between the number of piston-engine
aircraft landing-and-take-off operations (LTOs) and the resulting atmospheric lead
concentrations. The AQFs were developed through detailed air quality modeling at a model
airport. This appendix provides details on the air quality modeling used to develop the AQFs
and the model airport at which they were developed. Specifically, details on three topic areas
are included in the sections below: A.l) the air quality modeling setup, input data, and
parameters; A.2) characterization of the air quality model performance through model-to-
monitor comparisons at the model airport; A.3 and A.4) characteristics of the model airport
fleet composition and the fuel consumption rates that are incorporated into the AQFs; and A.5)
details of the wind speed adjustment methodology to scale AQFs based on average inverse
wind speed during.
A.l Details Regarding Air Quality Modeling at the Model Airport
This section provides information used to conduct detailed air quality modeling at the model
airport. Specifically, the subsections below present: 1) input data and methods to develop
aircraft emissions inventories, 2) non-aircraft emissions inventory data, 3) meteorological
inputs, 4) model receptors, and 5) the characterization of emission sources and receptors.
A.1.1 Aircraft Activity, Source Locations, and Emissions
Two aircraft emissions inventories were developed for this analysis: a seven-day inventory to
facilitate model-to-monitor comparison, and an annual operations inventory used to generate
the 3-month average AQFs. Both inventories were developed from a combination of published
Federal Aviation Administration (FAA) Air Traffic activity data and on-site surveys. This section
presents the underlying data and methods used to develop both inventories. The annual
emissions inventory and its use in developing the AQFs at the model airport are further
described in Section 2.3 of the report.
A ircra ft A ctivity Surveys
Surveyors collected aircraft operations data at RHV for the following ten days in 2011 from
10 a.m. until 7 p.m. local time.1
1 Bolded dates correspond to those when aircraft surveys were conducted and when lead air concentration
monitoring was conducted. Both aircraft activity and lead monitoring data were collected on the predominantly
active runway (Runway 31R). Monitor data was collected for 8/20, southerly winds resulted in operations
occurring predominantly on Runway 13L; therefore, model-to-monitor comparisons were not used for this day.
A-l

-------
8/14
Sunday
8/17
Wednesday
8/20
Saturday
8/23
Tuesday
8/26
Friday
8/28
Sunday
9/1
Thursday
9/3
Saturday
9/5
Monday
9/7
Wednesday
The collected data included runway location, aircraft tail fin number (N-Number), LTO mode
(taxi-out, run-up, take-off, climb, approach, landing, or taxi-in), duration of LTO mode (time-in-
mode), and other details about the aircraft activity (e.g., touch-and-go, altitude at approach,
altitude at departure). Whenever possible the surveyors visually identified the aircraft type.
Where aircraft type was not visually identified, the recorded tail fin numbers were matched to
the FAA tail number registry to obtain the type and number of engines. To match local typical
airport flight patterns, surveyors attempted to record the timing of approaches beginning at
1,100 ft (335 m) by listening to the control tower broadcast; the surveyors stopped timing
climbs at the same height. For those altitudes recorded lower or higher, adjustments were
made to normalize the time-in-mode for climb and approach to 1,100 ft. 5.6% of survey entries
were flagged as invalid, largely due to missing or incorrectly recorded time data.
Hourly and Daily Aircraft Activity Estimates
Hourly activity profiles for each aircraft class [single engine (SE), multi-engine (ME) and
rotorcraft (R)] and each operation-cycle type [full landing and take-off (LTO) and touch-and-go
(T&G)] were developed from the 10 days of survey data described above2. For morning and
evening hours when survey data was not available, the percentage of total daily flights
occurring in each hour was taken from a prior survey of piston-aircraft operations at another
airport (Carr et al. 2011). Total daily operations were estimated by adjusting the surveyed
operational counts to match FAA Air Traffic Activity Data System (ATADS) operation counts for
itinerant-general aviation and local-civil activity to account for operations that may have been
missed or mis-categorized by surveyors. The adjustment factor equation is given below in
Equation A-l. The adjustment factors for the ten days ranged from 1.02 to 1.23.
2 As described in Section 4.3 a Monte Carlo analysis was conducted to evaluate variation run-up duration, since as
discussed in Section 2, this mode of operation has the most significant impact on downwind lead concentrations.
The Monte Carlo analysis draws from additional airport studies and incorporates a range of run-up durations that
can account for variation due to a variety of factors (e.g., seasonal changes, regional differences, individual airport
characteristics). See Section 4.3 for additional details on the Monte Carlo analysis.
A-2

-------
'Daily, ATADS
I] Aon Hours, Survey
I] A off Hours
(Equation A-l)
Where:
= the adjustment factor, used ensure estimated aircraft activity counts
based on surveyed data matched daily ATADS counts;
Aon Hours, Survey = the hourly aircraft activity counts when surveys were conducted,
according to the survey data. This is the sum of all surveyed counts of
the aircraft and operation types selected for modeling;
For fixed-wing, multi-engine aircraft the relative scarcity of data led to some hours containing
very small operational counts. Thus, for these aircraft types, instead of scaling operational
counts for each hour, operational counts were scaled for the entire day and the incremental
increase in operations was divided evenly across all hours with non-zero survey counts. This
incremental adjustment led to fractional operations being modeled for multi-engine aircraft for
some hours on some days. For all aircraft types, the adjusted hourly activity counts for each day
were used in three of the seven days in the emissions inventory developed for model-to-
monitor comparisons. These three days corresponded to those with overlapping monitor and
survey data. For the remaining four days, the inventory used the average of the adjusted
activity profiles across all ten survey days.
Aircraft Emission Rates
Piston engines operating on leaded fuel can emit lead in both the gaseous and particulate form.
Aviation fuels containing tetraethyl lead also contain ethylene dibromide as an additive to
prevent lead from depositing within the engine. Lead reacts with the ethylene dibromide to
form brominated lead compounds. These brominated lead compounds are exhausted as vapors
but quickly cool and condense to solid particles. In contrast, organic lead emissions remain as
vapors after cooling to ambient temperatures (USEPA 2013b). A fraction of lead is retained in
the engine, engine oil, and/or exhaust system, which is estimated in this work at 5% (USEPA
2013a).
^Off Hours
= the hourly LTO count when surveys were not conducted; and
Aoaiiy, atads = the daily operations from ATADS. The ATADS data reports
"operations" which sums arrivals and departures. Because activity is
modeled as landing and take-off cycles (LTOs), operations are divided
by 2.
A-3

-------
Equation A-2 was used to calculate lead emission rates as a function of the mode specific fuel
consumption rate.
Time(s)xfuelconsumed(g))xf!^)x»x(l-RetRate)
EmissionsCg-s"1) =	—^; 1 Hour M gal J 		
/fuel weight(g)\ /Seconds\
\ gal / \ Hour /
(Equation A-2)
Where:
Time(s)
fuel consumed(g)
s
Engines
Hour
Pb(g)
gal
RetRate
fuel weight (g)
gal
Seconds
Hour
= the time (seconds) in mode;
= the amount (grams) of fuel consumed per second for one engine in a
given LTO mode;
= the number of engines operating each hour;3
= the concentration (g/gal) of Pb in avgas (2.16 g/gal, the average Pb
concentration measured in avgas from RVH);
= the fraction of Pb retained in the engine after fuel consumption
(0.05);4
= the weight (g) of avgas fuel per gallon (2,730.6 g/gal); and
= the number of seconds per hour (3,600 s);
Aircraft emissions profiles for each hour were developed by calculating hourly fuel consumption
for each aircraft type by operational mode (taxi, run-up, take-off, climb, approach, landing).
Total fuel consumption is a function of the time spent in each operational mode and the fuel
consumption rate of the aircraft's engine(s) during that mode. Aircraft class-specific (SE, ME, R)
median times-in-mode for each operational mode were developed for each hour for each of
the 10 survey days. Use of median times-in-mode avoided biasing fuel consumption high for
activities such as run-up which had occasional aircraft with unusually long activity durations or
biasing fuel consumption low for activities such as approach where surveyors may have
recorded short durations as a result of not knowing of an aircraft approach until it was on final
approach and well below the 1,100 feet nominal height. For hours when surveys were not
conducted, the inventory assumes the median of all recorded data for that activity mode. For
3	Two engines for multi-engine fixed-wing aircraft, one for single-engine fixed-wing aircraft
4	The information used to develop this estimate is from the following references: (a) Todd L Petersen,
Petersen Aviation, Inc, Aviation Oil Lead Content Analysis, Report # EPA 1-2008, January 2, 2008,
available at William J. Hughes Technical Center Technical Reference and Research Library at
http://actlibrary.tc.faa.gov/and (b) E-mail from Theo Rindlisbacher of Switzerland Federal Office of Civil
Aviation to Bryan Manning of U.S. EPA, regarding lead retained in engine, September 28, 2007.
A-4

-------
the 3 days of overlapping monitor and survey data, the corresponding day-specific hourly
median times-in-mode were used in the emissions inventory. For the remaining 4 days, each
day was assigned the same median hourly TIM values.
Published fuel consumption data was used to develop engine- and mode-specific fuel
consumption rates. Fuel consumption data by operational mode was available for 18 engines
from FAA's Emission and Dispersion Modeling System and the Swiss Federal Office of Civil
Aviation report on piston-engine emissions (FAA 2007, SFOCA 2007). These data spanned the
range of engine technology groups (fuel injected, turbocharged, carbureted, radial) observed at
the airport. Where observed aircraft had specific engine types (e.g., 4-stroke radial, 2-stroke
fuel injected) not in the fuel consumption database, fuel consumption for that aircraft was
modeled using the engine type with the closest rated horsepower in the same engine
technology group. Further details on the mode-specific fuel consumption rates are given in
Section A.4.
Aircraft Emission Inventories
The above calculation of emission rates was used to develop two separate emissions
inventories. As noted above, both a seven-day and annual emissions inventory were developed
for two distinct purposes, but using similar methods. The seven-day emissions inventory was
used to facilitate model-to-monitor comparison for the days in which survey data and on-site
monitoring were conducted concurrently, and as such used the highly resolved hourly
operational data from on-site survey data described above. The annual inventory was used to
calculate 3-month average concentrations (and derived AQFs), and thus required emissions
modeling for 14 months to understand 12 consecutive, 3-month rolling-average concentrations.
Absent detailed on-site monitoring data for aircraft and engine types and hourly operations for
each day of the 14 months, the annual emissions inventory used the average operational
profile, median time-in modes and fuel consumption rates, and ATADs operations data as
described in Section 2.3 of the main report. Aircraft emissions for months 13 and 14 were taken
from months 1 and 2 respectively so that rolling-average concentrations represented
concentrations from a consistent year of emissions. The total lead emissions from aircraft (in
tons) are given by month and operational mode in Table A-l. (Lead emissions included in the
modeling from other sources are described in Tables A-2 through A-6.)
Table A-l. Monthly and annual lead emissions from aircraft (tons)




Aircraft Mode



Total of all
Aircraft
Modes
Activity
Month
Taxi-out
Run-Up
Take-off
Climb

Approach
Landing
Taxi-in

ME Full
LTO
01
1.16E-04
5.85E-05
3.28E-05
3.82E-05
1.61E-05
7.05E-06
1.99E-05
2.88E-04
02
9.40E-05
4.52E-05
2.73E-05
2.94E-05
1.34E-05
6.15E-06
1.59E-05
2.32E-04
03
1.23E-04
6.00E-05
3.53E-05
3.90E-05
1.74E-05
7.85E-06
2.08E-05
3.03E-04
A-5

-------
Total of all
Aircraft
Aircraft Mode Modes
Activity Month Taxi-out Run-Up Take-off Climb Approach Landing Taxi-in

04
9.85E-05
4.61E-05
2.87E-05
2.99E-05
1.41E-05
6.55E-06
1.64E-05
2.40E-04
05
1.15E-04
5.55E-05
3.31E-05
3.62E-05
1.63E-05
7.40E-06
1.94E-05
2.83E-04
06
1.24E-04
5.85E-05
6.20E-05
3.57E-05
3.80E-05
1.76E-05
8.10E-06
2.07E-05
3.02E-04
07
1.29E-04
3.71E-05
4.02E-05
1.83E-05
8.30E-06
2.17E-05
3.16E-04
08
1.31E-04
6.25E-05
3.77E-05
4.05E-05
1.86E-05
8.50E-06
2.19E-05
3.20E-04
09
1.11E-04
5.35E-05
3.18E-05
3.48E-05
1.56E-05
7.10E-06
1.87E-05
2.72E-04
10
1.06E-04
5.15E-05
4.16E-05
3.04E-05
3.36E-05
1.50E-05
6.75E-06
1.79E-05
2.61E-04
11
9.15E-05
2.69E-05
2.68E-05
1.32E-05
6.30E-06
1.51E-05
2.21E-04
12
7.20E-05
3.18E-05
2.13E-05
2.04E-05
1.05E-05
5.05E-06
1.17E-05
1.73E-04
Annual
Total
1.31E-03
6.27E-04
3.78E-04
4.07E-04
1.86E-04
8.51E-05
2.20E-04
3.21E-03
MET&G
01
-
	
-
2.08E-05
1.31E-05
-
-
3.38E-05
02
-
-
1.81E-05
1.22E-05
-
-
3.03E-05
03
-
-
-
2.31E-05
1.53E-05
-
-
3.84E-05
04
-
-
-
1.94E-05
1.35E-05
-
-
3.29E-05
05
-
	
-
2.19E-05
1.46E-05
-
-
3.65E-05
06
-
-
2.39E-05
1.63E-05
-
-
4.02E-05
07
-
-
-
2.45E-05
1.65E-05
-
-
4.10E-05
08
-
-
-
2.51E-05
1.69E-05
-
-
4.20E-05
09
-
-
-
2.09E-05
1.39E-05
-
-
3.48E-05
10
-
-
-
1.99E-05
1.32E-05
-
-
3.31E-05
11
-
-
-
1.86E-05
1.33E-05
-
-
3.19E-05
12
-
-
-
1.50E-05
1.10E-05
-
-
2.60E-05
Annual
Total
0.00E+00
-
-
2.51E-04
1.70E-04
-
-
4.21E-04
SE Full
LTO
01
8.20E-04
3.62E-04
3.09E-04
3.81E-04
3.12E-04
6.65E-05
2.23E-04
2.48E-03
02
7.15E-04
3.16E-04
2.70E-04
3.20E-04
2.66E-04
5.75E-05
1.96E-04
2.14E-03
03
9.15E-04
4.03E-04
3.38E-04
3.44E-04
4.13E-04
3.42E-04
7.35E-05
2.49E-04
2.74E-03
04
7.70E-04
2.89E-04
3.38E-04
2.83E-04
6.20E-05
2.11E-04
2.29E-03
05
8.65E-04
3.81E-04
3.25E-04
3.89E-04
3.22E-04
6.95E-05
2.36E-04
2.59E-03
06
9.45E-04
4.17E-04
3.56E-04
4.20E-04
3.51E-04
7.60E-05
2.59E-04
2.83E-03
A-6

-------
Total of all
Aircraft
Aircraft Mode	Modes
Activity Month Taxi-out Run-Up Take-off Climb Approach Landing Taxi-in

07
9.70E-04
4.27E-04
3.65E-04
4.35E-04
3.61E-04
7.80E-05
2.65E-04
2.90E-03
08
9.90E-04
4.37E-04
3.73E-04
4.42E-04
3.68E-04
8.00E-05
2.71E-04
2.96E-03
09
8.25E-04
3.64E-04
3.47E-04
3.11E-04
3.72E-04
3.09E-04
6.65E-05
2.26E-04
2.48E-03
10
7.90E-04
2.96E-04
3.56E-04
2.95E-04
6.35E-05
2.15E-04
2.36E-03
11
7.35E-04
3.24E-04
2.77E-04
3.18E-04
2.69E-04
5.90E-05
2.02E-04
2.19E-03
12
5.95E-04
2.61E-04
2.24E-04
2.53E-04
2.15E-04
4.77E-05
1.64E-04
1.76E-03
Annual
Total
9.94E-03
4.37E-03
3.74E-03
4.43E-03
3.69 E-03
8.00E-04
2.72E-03
2.97E-02
SET&G
01
--
--
--
2.86E-04
2.79E-04
--
--
5.65E-04
02
--
--
--
2.44E-04
2.42E-04
--
--
4.86E-04
03
--
--
--
3.13E-04
3.09E-04
--
--
6.20E-04
04
--
	
--
2.60E-04
2.59E-04
--
--
5.20E-04
05
--
--
2.96E-04
2.92E-04
--
--
5.90E-04
06
--
--
--
3.21E-04
3.20E-04
--
--
6.40E-04
07
--
--
--
3.31E-04
3.28E-04
--
--
6.60E-04
08
--
	
--
3.38E-04
3.35E-04
--
--
6.70E-04
09
--
--
2.83E-04
2.80E-04
--
--
5.60E-04
10
--
--
--
2.70E-04
2.67E-04
--
--
5.35E-04
11
--
	
--
2.46E-04
2.48E-04
--
--
4.94E-04
12
--
--
1.97E-04
2.00E-04
--
--
3.97E-04
Annual
Total
--
--
--
3.39E-03
3.36E-03
--
--
6.75E-03
Annual Total
1.20E-02
5.00E-03
4.25E-03
8.48E-03
7.40E-03
8.99E-04
3.16E-03
4.12E-02
Aircraft Source Locations
Both the seven day and annual emissions inventories are spatially allocated at the model
airport to characterize resulting atmospheric concentrations. A three-dimensional
representation of the modeled aircraft source locations is provided in Error! Reference source
not found. A-l for northwest take-offs and in Figure A-2 for southeast take-offs. All ground-
level release heights were set to 0.5 meter to represent the approximate aircraft exhaust
height. For taxi activity, emission sources were placed approximately 50 meters apart along two
taxiways - one directly adjacent to the terminal and hangars, and one along a separate taxiway
to the west of the first taxiway. For run-up activity, two run-up locations were modeled near
the terminus of both taxiways. The two run-up locations were approximately 8 meters apart at
A-7

-------
both ends of the taxi-way. For modes at altitude (i.e., climb and approach), emissions sources
had 50 meters horizontal spacing and ascent/descent angles of 4.7 and 3.8 degrees
respectively, angles similar to those used in the SMO study (Carr et al. 2011).5 Airport noise
ordinances dictate that aircraft using runway 31R (northwest takeoff) make a 30-degree right
turn after departing the runway and after climbing to an altitude of at least 500 feet above
ground level.
5 This also includes a consideration to account for the wake turbulence created by the forces that lift the aircraft.
High pressure air from the lower surface of the wings flows around the wing tips to the lower pressure region
above the wings. A pair of counter-rotating vortices is shed from the wings where the right and left wing vortices
rotate. It is within this region of rotating air behind the aircraft where wake turbulence occurs. To account for this
effect, the effective emission height was adjusted for the angle of climb (takeoff) and glide slope angle for landing.
This adjustment lowers the effective emission height to approximate the maximum downward extent of the
aircraft's trailing wake. This results in an angle of climb-out for take-off of approximately 4.7 degrees, while for
landing this was 3.8 degrees.
A-8

-------
Satellite Image Source: ESRI Prime Imagery 3D
The aircraft release heights are at each colored sphere. Symbol shadings correspond to release heights,
where green values are close to the surface and dark red values are approximately 145 m above the surface.
Satellite photography and terrain features are shown. Vertical terrain height is exaggerated. The point of
view is elevated, approximately 2.3 km away and facing northeast.
Figure A-l. Modeled aircraft emission source locations for northwest take-offs, in three dimensions
A-9

-------
Satellite Image Source: ESRI Prime Imagery 3D
The aircraft release heights are at each colored sphere. Symbol shadings correspond to
release heights, where green values are close to the surface and dark red values are
approximately 145 m above the surface. Satellite photography and terrain features are
shown. Vertical terrain height is exaggerated. The point of view is elevated, approximately
2.3 km away and facing northeast.
Figure A-2. Modeled aircraft emission source locations for southeast take-offs, in three dimensions
A-10

-------
A.1.2 Emissions from Sources Other than Aircraft
Data on emissions from sources other than aircraft operating at the model airport was collected
in order to include these sources in the modeled concentrations that would be compared to
monitored concentrations, the latter of which of course included atmospheric lead from any
source. For purposes of modeling, these non-aircraft emissions include rotorcraft at the model
airport, point sources, nearby road sources, mobile and area sources, and background lead
within 20 km of the airport.
Rotorcraft
Piston-engine Rotorcraft activity estimates at the model airport were generated using the same
methodology described above for fixed-wing single- and multi-engine aircraft. Daily surveys
were combined with activity counts from ATADS data to develop hourly rotorcraft activity
profiles. Rotorcraft parking was assumed to be near the model airport terminal, and take-offs
occurred at a 4.7 degree trajectory from Runway 31R and Runway 13L, or vertically from a
helicopter practice area (also called the haypatch). As with fixed-wing aircraft at the model
airport, rotorcraft at the model airport were modeled as volume sources as described in section
A.1.5 with 2.3 vertical meters between source points in landing and take-off as shown in Figures
A-l and A-2. Monthly and annual emissions from rotorcraft are given in Table A-2. Nearby
heliports (i.e. rotorcraft landing and takeoffs not occurring at the model facility) were modeled
as area sources.
Table A-2. Monthly rotorcraft emissions at the model airport (tons)
Location
Month
Taxi-out
Take-off
Landing
Taxi-in
Total

01
3.38E-05
4.88E-06
1.29E-06
9.10E-06
4.90E-05

02
2.78E-05
4.75E-06
1.03E-06
8.15E-06
4.17E-05

03
3.61E-05
5.85E-06
1.35E-06
1.03E-05
5.35E-05

04
2.92E-05
5.30E-06
1.07E-06
8.80E-06
4.43E-05

05
3.39E-05
5.65E-06
1.26E-06
9.80E-06
5.05E-05

06
3.64E-05
6.35E-06
1.35E-06
1.08E-05
5.50E-05
31R/13L
07
3.79E-05
6.40E-06
1.41E-06
1.10E-05
5.65E-05
08
3.84E-05
6.60E-06
1.43E-06
1.13E-05
5.75E-05

09
3.25E-05
5.35E-06
1.21E-06
9.35E-06
4.84E-05

10
3.11E-05
5.05E-06
1.16E-06
8.90E-06
4.62E-05

11
2.72E-05
5.30E-06
9.85E-07
8.55E-06
4.20E-05

12
2.14E-05
4.43E-06
7.65E-07
6.95E-06
3.36E-05

Annual
Total
3.85E-04
6.59E-05
1.43E-05
1.13E-04
5.78E-04

01
3.38E-05
4.88E-06
--
9.10E-06
4.77E-05
A-ll

-------
Location
Month
Taxi-out
Take-off
Landing
Taxi-in
Total

02
2.78E-05
4.75E-06
--
8.15E-06
4.07E-05

03
3.61E-05
5.85E-06
--
1.03E-05
5.20E-05

04
2.92E-05
5.30E-06
--
8.80E-06
4.33E-05

05
3.39E-05
5.65E-06
--
9.80E-06
4.93E-05

06
3.64E-05
6.35E-06
--
1.08E-05
5.35E-05

07
3.79E-05
6.40E-06
--
1.10E-05
5.50E-05
Haypatch
08
3.84E-05
6.60E-06
--
1.13E-05
5.65E-05

09
3.25E-05
5.35E-06
--
9.35E-06
4.72E-05

10
3.11E-05
5.05E-06
--
8.90E-06
4.51E-05

11
2.72E-05
5.30E-06
--
8.55E-06
4.10E-05

12
2.14E-05
4.43E-06
--
6.95E-06
3.28E-05

Annual
Total
3.85E-04
6.59E-05
--
1.13E-04
5.64E-04
Point Sources
Sixteen point sources of lead emissions within approximately 20 km of the model airport were
modeled based on emissions data from the 2008 National Emissions Inventory (version 1.5)
(USEPA 2011). Excluded from the point source inventory were approximately 122 facilities that
each emit less than 1E-05 US Tons per year (TPY) of lead (totaling 1.8E-04 TPY or 0.36 Ibs/yr),
which is more than a factor of 100 times smaller than the model airport aircraft emissions. The
hourly emissions profiles for point sources were approximated using the temporal codes that
the Bay Area Air Quality Management District (BAAQMD) used for Community Air Risk
Evaluation (CARE) modeling (BAAQMD 2011). Point source facility descriptions and emissions
magnitudes are given in Table A-3.
Table A-3. Industries corresponding to the 16 modeled facilities within 20 km of the Model Airport
Source Description
Number of
Facilities
Modeled Pb
Emissions (TPY)a
Emissions Percentage
of Total Point
San Jose Airport - Piston-engine emissions
l
1.82x10 1
98.86%
Heliports
4
1.41xl0"3
0.77%
Nonferrous Metal (except Aluminum)
Production and Processing
1
2.87xl0"4
0.16%
Aerospace Product and Parts Manufacturing
1
1.29xl0"4
0.07%
Computer Storage Device Manufacturing, Data
Processing, Hosting, and Related Services
2
7.25xl0"5
0.04%
Crematorium
2
5.40xl0"5
0.03%
A-12

-------
i Number of i Modeled Pb i Emissions Percentage
Source Description i Facilities j Emissions (TPY)a j of Total Point
Dry cleaning and Laundry Services
1
4.75xl0"5
0.03%
Commercial and Service Industry Machinery
Manufacturing
1
2.70xl0"5
0.01%
Water, Sewage and Other Systems
1
2.70xl0"5
0.01%
General Medical and Surgical Hospitals
1
2.50xl0"5
0.01%
Recyclable Material Merchant Wholesalers
1
2.10xl0"5
0.01%
TOTAL
16
1.84X101
100%
a San Jose Airport emissions are representative of 2015. All other emissions in this Table are representative of
2008.
Nearby Road Sources
Lead emissions were modeled from three roadways in close proximity to the airport. Time-
varying emission rates were calculated from a combination of diurnal traffic count data, the
area of each roadway, and a mobile lead emissions per mile traveled estimate. The hot
stabilized summer emission factor of 0.002 mg of Pb/mile was used for gasoline vehicles, and
an emission factor of 0.00724 mg/mile was used for diesel fueled trucks and buses (USEPA
2006). The total annual average lead emissions from all three adjacent roadways was 4.3xl0~5
TPY. The location of the roadway emissions sources is shown in Figure A-3, and roadway
characteristics are given in Table A-4.
A-13

-------
Figure A-3. Explicitly modeled road sources adjacent to the model airport
Table A-4. Collected input data for explicitly modeled road sources adjacent to the model airport
Street
Length of Road (m)
Area of Road (m2)
Average Daily Traffic
(ADT)
Year of ADT Data
East Capitol Expressway
1223.10
50661.24
40,700
2008
Tully Road
643.74
22457.86
33,676
2010
Ocala Avenue
740.30
12355.21
10,867
2009
Gridded Area, On-Road Mobile, and Non-Road Mobile Sources
Lead emissions from area,6 non-road mobile/ and on-road mobile sources8 within
approximately 20 km of the model airport were modeled in lxl-km grid cells using an annual
6	Area (non-mobile) sources included agricultural and livestock waste, cooking, wind erosion, mining, and open
burning and other fires.
7	Non-road mobile sources included military aircraft, commercial aircraft, airport support vehicles, construction
and other road dust, agricultural and commercial off-road vehicles, railroad equipment, and marine and pleasure
craft.
8	On-road mobile sources included lead emissions based on California's Emission Inventory Reporting System PM
speciation profiles database which includes tire and brake wear from light- and heavy-duty vehicles and diesel
fueled vehicle exhaust emissions. The speciation profile has zero lead emissions from gasoline exhaust. Emissions
A-14

-------
year 2015 projection inventory developed previously by a local agency (BAAQMD 2011). The
total modeled emissions are shown in Table A-5 (for on-road mobile) and Table A-6 (for non-
road mobile plus area).
Table A-5. Modeled on-road mobile gridded Pb emissions within 20 km of the model airport
Source Description
Pb Emissions (TPY)
Emissions Percentage of Total
On-Road Mobile
Highway Vehicle - Tire Wear
2.27xl0"5
59%
Highway Vehicle - Brake Wear
1.06xl0"5
27%
Highway Vehicle - Diesel Exhaust
5.30xl0"6
14%
TOTAL
3.86xl0"5
100%
Table A-6. Modeled non-road mobile and area gridded lead emissions within 20 km of the model
airport
j Pb Emissions i Emissions Percentage of Total
Source Description i (TPY) i Non-Road Mobile and Area
Construction: Industrial, Commercial, Institutional,
Residential, Road Construction
3.37xl0"3
50.96%
Paved Roads
1.84xl0-3
27.85%
Food and Kindred Products: Commercial Cooking and
Miscellaneous
7.10xl0"4
10.73%
Agriculture Production - Livestock
2.10xl0"4
3.17%
Military and Commercial Aircraft
2.08xl0"4
3.14%
2.03%
GeogenicWind Erosion
1.34xl0"4
Residential Oil and Wood Burning
5.60xl0"5
0.85%
Mineral Processes: Concrete, Gypsum, Plaster
3.39xl0"5
0.51%
0.39%
Unpaved Roads
2.59xl0"5
Forest Wildfires, Prescribed Burns, Motor Vehicle Fires
1.24xl0"5
0.19%
Off-highway Vehicle Diesel (incl. Airport Ground
Support Equipment)
5.38xl0"6
0.08%
Agriculture Production - Crops
3.15X106
0.05%
Railroad Equipment
1.82xl0"6
0.03%
from the nearby roadways were explicitly modeled. Because of the differences in inventory years, fuel emission
rates, and accounting for break and tire wear in these two inventories, the combined gridded emissions may result
in some small double-counting or undercounting of lead emissions from brake and tire wear and vehicle exhaust in
the modeling domain.
A-15

-------
Source Description
Pb Emissions
(TPY)
Emissions Percentage of Total
Non-Road Mobile and Area
Open Burning
7.90xl0"7
0.01%
Agriculture Production - Crops - as nonpoint
4.30xl0"7
0.01%
Pleasure Water Craft
5.61xl0"9
0.0001%
TOTAL
6.61xl0"3
100%
Background Emissions
The background ambient lead concentration was set to 0.5 ng/m3, the pristine background
ambient lead concentration used by the EPA Office of Air Quality Planning and Standards
Regulatory Impact Analysis for the lead National Ambient Air Quality Standard (NAAQS) (USEPA
2008b).
A.1.3 Meteorology
Meteorological Data
Surface and upper-air meteorological data were processed using AERMOD's meteorological
preprocessor, AERMET, to produce hourly data on mixing heights, stability, winds, temperature,
and precipitation.9,10 We used the twice-daily upper-air data from the radiosonde station at the
Metropolitan Oakland International Airport (OAK), which is the nearest upper-air station
(approximately 55 km north-northwest of the model airport). For the seven-day modelling used
in the model-to-monitor comparison, we used on-site hourly surface wind data collected
concurrently with air concentration data. As described in Section 2.3, for the annual modeling,
we used the 1-minute ASOS wind data and the hourly surface meteorological data from SJC
(Norman Y. Mineta San Jose International Airport; SJC; WBAN ID 23293; approximately 10 km
northwest of RHV). Review of the SJC ASOS station data showed its flow directions were similar
to those measured concurrently at RHV, and sensitivity analyses showed that model
performance using the SJC 1-minute wind data was equivalent to using the on-site monitored
data for the seven days of monitoring data.
AERSURFACE Parameterization
AERMET requires three surface characteristics for the area around the study site (RHV) in order
to estimate turbulence and mixing heights. These characteristics are albedo, Bowen ratio (ratio
of sensible heating to latent heating), and surface roughness length. The AERSURFACE11
preprocessor estimates these surface characteristics based on land cover and user inputs that
describe the site and its climatology. Albedo and Bowen ratio values are calculated within 10
9	AERMOD: AERMIC Model, where AERMIC = American Meteorological Society/EPA Regulatory Model
Improvement Committee.
10	AERMET: AERMIC Meteorological Processor. Available:
http://www.epa.gov/ttn/scram/metobsdata procaccproes.htm#aermet
11	AERSURFACE: AERMOD Surface Characteristics Processor. Available:
http://www.epa.gov/ttn/scram/dispersion related.htmtfaersurface
A-16

-------
km of the study site, and 1 km is the recommended default radius for calculating surface
roughness (USEPA 2008a, USEPA 2009).
While AERSURFACE only accepts version 1992 of the Multi-Resolution Land Characteristics
Consortium (MRLC) and the National Land Cover Database (NLCD92), two more recent versions
of the NLCD (2001 and 2006) are available. Between 1992 and the date of the most recent
version, 2006, the proportion of land that was developed within 10 km of RHV increased
approximately 15 percent12, the developed land became more "intensely" developed,13 and
grasslands shrank. Reductions in grasslands and increases in development increase Bowen ratio
(by up to 100-200 percent) and surface roughness (by up to two orders of magnitude), which
lead to greater turbulence and higher mixing heights.
Given these important differences in land cover between 1992 and 2006, and given that the
AERSURFACE only accepts NLCD92 data; Geographic Information Systems (GIS) software was
used to reproduce the functionality of AERSURFACE while using NLCD06 data. For surface
roughness lengths, land cover class fraction was determined within approximately 1 km of the
meteorological site for each 30 degree wedge over 12 directional sectors. For simplicity, the 1-
km distance was determined using a lxl-km square centered on the meteorology site, so the
distance was 1 km from the site perpendicularly to each side and approximately 1.4 km to each
corner. For Bowen ratio and albedo calculations, the number of each land cover class was
counted within a lOxlO-km area centered on the meteorology site (the default area specified in
(USEPA 2008a)). The land cover and lOxlO-km squares for the model airport and the ASOS
station site (SJC) are shown in Figure A-4.
12	From approximately 56 percent to approximately 71 percent.
13	The proportion of developed land that was medium- and high-intensity increased from approximately 18
percent in 1992 to approximately 66 percent in 2006.
A-17

-------
12 Kilometer!
¦	Pasture/Hay
Cultivated Crop*
¦	Woody Wetlands
¦	Emergent Herbac
Land Cover image Source: National Land Cover Database http://www.mrlc.gov/nlcd2006.php
Figure A-4. The lOxlO-km squares used to evaluate albedo and Bowen ratio, with NLCD06 base map
Next the surface characteristics were paired with the climate conditions (i.e., season
definitions, wetness, and aridity) of each site based on 1981-2010 NCDC climate normal14 data
for SJC. Monthly season assignments were subjective and based on monthly average
temperatures and precipitation. The aridity determination used annual average precipitation
data (where the average annual rainfall at SJC is 15 inches). Wetness values were determined
by comparing monthly precipitation data to the climate normal monthly precipitation data,
where months receiving less than half the normal precipitation amount were considered dry
and months receiving over twice the normal precipitation amount were considered wet.
Monthly climate statistics at SJC are shown in Table A-7.
14 Climate normals are the three-decade averages of climatological variables including temperature and
precipitation. These are updated every ten years with the most current period from 1981-2010. Available:
http://www.wrcc.dri.edu/cgi-bin/cliMAIN.pl7ca7821
A-18

-------
Table A-7. Monthly climate statistics at the SJC ASOS Station

Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
TEMPERATURE (F):
30-year normal
50.1
53.3
56.2
58.9
63.4
67.5
70.0
70.1
68.5
63.2
55.1
50.0
PRECIPITATION
(in.): 30-year
normal
3.07
3.11
2.54
1.18
0.51
0.10
0.02
0.02
0.18
0.80
1.68
2.61
SEASON
ASSIGNMENT FOR
ALL MODELING:
Fall
Fall
Spring Spring
Sum.
Sum.
Sum.
Sum.
Sum.
Sum.
Fall
Fall
PRECIPITATION
(in.): 2010
4.58
2.12
1.94
3.10
0.35
0
0
0
0
0.02
1.76
2.58
PRECIPITATION
(in.): Ratio
2010 to 30-year
normal for month
1.49
0.68
0.76
2.63
0.69
0
0
0
0
0.03
1.05
0.99
WETNESS
ASSIGNMENTS
AVG
AVG
AVG
WET
AVG
DRY
DRY
DRY
DRY
DRY
AVG
AVG
PRECIPITATION
(in.): 2011
	
	
	
	
	
	
	
0
	
	
	
	
PRECIPITATION
(in.): Ratio
2011 to 30-year
normal







0




a Color shading is arbitrary and is for visualization purposes only.
Finally, the above information was combined with the surface characteristic lookup tables
(USEPA 2008a), to report the surface characteristics at each site by month and sector (the latter
only for surface roughness length). The values of albedo and Bowen ratio are shown in Table A-
8 for the on-site meteorology at the model airport (only for August 2011, as used in the model-
to-monitor comparison) and for the meteorology site at SJC (for 2010, as used in the annual
modeling). The albedo is 0.17 at both sites, reflecting the predominantly residential land cover
within 10 km of both sites. The Bowen ratio values vary by the season and wetness value
determination, from 0.78 to 2.42 (with 2.07 the August 2011 value at the model airport).
Table A-8. Albedo and Bowen ratio values within approximately 10 km of the meteorology sites
Met. Site
Month
Albedo
Bowen
Ratio
T&B at model
airport (year 2011)
8
0.17
2.07
SJC (year 2010)
1
0.17
1.24
A-19

-------
Met. Site
Month
Albedo
Bowen
Ratio

2

1.24

3

1.12

4

0.78

5

1.15

6

2.42

7

2.42

8

2.42

9

2.42

10

2.42

11

1.24

12

1.24
After completing the 2010 modeling results, it was discovered that the 1- and 10-km squares
used at the SJC site were not oriented in a typical north-south alignment but were rotated
approximately 18° counterclockwise from the north. Rotating to a north-south alignment had
no effect on albedo values. On average, the Bowen ratio values were 6 percent larger than if
the area was oriented north-south—this is mostly because the north-south alignment
encompassed less open water and more developed land compared to the rotated square. The
effect of the rotation on surface roughness length values was mixed. Compared to the values
using the north-south alignment, roughness length values were unaffected for most or all
months in two sectors (the 60-90 and 150-180 degree sectors), were an average of 8 percent or
4 cm smaller for all months in six sectors (the 0-30, 90-120, 210-240, 240-270, 300-330, and
330-360-degree sectors), and were an average of 13 percent or 5 cm too large for all months in
4 sectors (the 30-60, 120-150,180-210, and 270-330 degree sectors). The sectors that have the
largest impact on the modeling results, in terms of dominant wind directions, are the 90-120,
120-150, and especially the 270-300 and 300-330 degree sectors. On average during airport
operation hours, the cumulative effect of this rotation was to raise the mixing height by less
than 1 percent (less than 5 m), with similarly small positive or negative effects on sensible heat
flux, surface friction velocity, convective velocity scale, and stability. Test runs in AERMOD
suggested that these differences would have no more than a 1-percent effect on 3-month
average modeled air concentrations and depositions.
Urban Setting
Urban boundary layers were parameterized with AERMOD's urban setting option, along with
the estimated 2009 population of the local area, Santa Clara County (1,785,000).15
15 At the time of this development the 2010 census data had not yet been released. Subsequently, the 2010 census
has become available and the reported total population of Santa Clara County was 1,781,642. Available:
http://www.census.gov
A-20

-------
A. 1.4 Receptors
To fully define the spatial extent of elevated lead concentrations within the vicinity of the
airport, 2,250 receptor locations were used, as shown in Figure A-5. Also shown are the
locations of the three ambient air monitoring stations. Inside the airport property and around
the airport boundary, receptors were placed at 50 m intervals, with approximately 365
receptors placed across the airport on taxiways, airport hangars, access roadways, buildings,
and the airport meteorology station. Beyond the airport boundary, the 50 m grid spacing
continued along the northwest-southeast orientation axis which is parallel to the runways; for
the other areas within 1 km of the facility boundary a 100 m grid spacing was implemented; for
areas out to 2 km, 200 m grid spacing was implemented.
Freeways and Highways
Secondary Roads
Receptors
X Monitor 1
X Monitor 2
Monitor 3
A Onsite Met. Stn.
Hangars, Taxiways, etc.
RHV Boundary (50-m)
¦ Parks and Rec. Areas
X	Schools
•	Grid (50-m)
•	Grid (100-m)
•	Grid (200-m)
Road Image Source: ESRI U.S. Major Roads version 9.3.1
Figure A-5. Receptor field for ambient air quality concentration analysis
A.1.5 Emission Source Characterization and Parameters
Aircraft
We modeled aircraft as volume sources within AERMOD, which is the method recommended by
EPA for modeling a "line source" representing a moving object (USEPA 2004), and is the same
approach used in previous air quality modeling at a general aviation airport (USEPA 2010, Carr
et al. 2011). The modeled coordinates and release heights above the surface represent the
center of the volume. The equations below, as shown in Table A-9, were used to calculate initial
sigma-y and sigma-z values (horizontal and vertical plume extents).
A-21

-------
Sigma — z(m) = [l.8 + (o.ll X	X (Equation A-3)16
Sigma — z(m) = F + G + Eq. A-3	(Equation A-4)17
Sigma — y or z(m) = -	(Equation A-5)18
Sigma — y(m) = C + D + E	(Equation A-6)19
Table A-9. Sigma-y and sigma-z values of modeled aircraft volume sources
LTO
Sigma-y (m)
Sigma-z (m)

23.26
2.64
Taxi
Used Eq. A-5 where:
A = 50 m (horizontal spacing
between source points);
B = 2.15
Used Eq. A-3, where:
W2 = 11.43 m (width of the wider
taxiway (22.86) divided by 2);
U = 2.54 m/s (mean wind speed during
the modeling period during RHV
operational hours)

6.4
3.48
Run-up
Used Eq. A-6, where:
C = 4.96 m (typical wingspan (10.67
m) divided by 2.15);
D = 0.6 m (horizontal momentum of
the exhaust);3
E = 0.85 m (propeller turbulence
wake)b
Used Eq. A-4, where:
F = 0.5 m (release height);
G = 0.65 m (exhaust buoyancy);0
W2 = 5.34 m (typical wingspan (10.67
m) divided by 2);
U = 2.54 m/s (mean wind speed during
the modeling period during RHV
operational hours)
16	Benson, P. E. (1979). Abridged User's Guide for CAUNE-3 - A versatile dispersion model for predicting air
pollutant levels near highways and arterial streets. O. o. T. Laboratory. Section 5.2.
17	USEPA (2010). Development and Evaluation of an Air Quality Modeling Approach for Lead Emissions from Piston-
Engine Aircraft Operating on Leaded Aviation Gasoline. EPA-420-R-10-007.
https://nepis.epa.gov/Exe/ZyPDF.cgi/P100yH4Q.PDF?Dockey=P10QyH4Q.PDF ., Section 4.2, for the run-up mode
of fixed-wing piston-fired aircraft.
18	USEPA (2004). User's guide for the AMS/EPA regulatory model - AERMOD. Office of Air Quality Planning and
Standards. September 2004. https://www3.epa.gov/ttn/scram/models/aermod/aermod usereuide.pdf., Table 3-
1, for either a single volume source or a line source represented by separated volume sources, depending on the
LTO.
19	USEPA (2010). Development and Evaluation of an Air Quality Modeling Approach for Lead Emissions from Piston-
Engine Aircraft Operating on Leaded Aviation Gasoline. EPA-420-R-10-007.
https://nepis.epa. gov/Exe/Zy PDF. cgi/P1007H4Q.PDF?Dockey=P1007H4Q. PDF . Section 4.2
A-22

-------
LTO

Sigma-y (m)
Sigma-z (m)
Moving
non-taxi
(take-off,
climb,
approach,
landing)
23.26
Used the same method as taxi (see
above).
2.33
Used Eq. A-3, where:
W2 = 5.34 m (typical wingspan (10.67
m) divided by 2);
U = 2.54 m/s (mean wind speed during
the modeling period during RHV
operational hours)
a The 0.6 m horizontal exhaust momentum was based on a sensitivity test with SCREEN3 with and
without the typical exhaust flow of 100 ft3/min.
b The 0.85 m propeller turbulence wake was calculated by dividing the typical 1.83 m propeller
size by 2.15.
c The 0.65 m exhaust buoyancy was based on a sensitivity test with SCREEN3 with and without an
exhaust temperature of 573 K.
Point and Area Sources
SJC, a nearby airport, was modeled as an area source using the approximate airport boundary
line, while all other data in the point source file were modeled as point sources. For the SJC
area source, the North American Industrial Classification System (NAICS) code was used to
identify the average release height from California lead emitters in the 2005 NEI v2, and this
average release height was used in lieu of a sigma-z. For the point sources, the same NAICS
methodology was used from the 2005 NEI v2 for release height, stack diameter, and exit gas
temperature and velocity. All release heights were between 8 and 13 meters, stack diameters
were less than 1 meters, exit gas temperatures were between 300 and 600 K, and exit gas
velocities were between 5 and 25 m/s.
Nearby Road Sources
The lengths and widths of the area sources used to model nearby road sources were
determined using aerial photos. A release height of 2 meters and an initial sigma-z of 2.15 m
were applied to all roadways which accounts for the mix of light and heavy-duty vehicles as well
as vehicle induced turbulence.
Cridded Area, On Road Mobile, and Non-Road Mobile Sources
Consistent with previous air quality modeling at a General Aviation airport (USEPA 2010, Carr et
al. 2011), gridded area, on-road mobile, and non-road mobile sources were modeled as area
sources with a release height of 2 m and a sigma-z of 2.15 meters.
A.2 Sensitivity Analysis of Model-to-Monitor Comparison at the Model Airport
As discussed in Section 2.2 of the report, model-to-monitor comparisons at the model airport
over a seven-day period showed modeled concentrations were well within a factor of two of
monitored concentrations, which is a typical metric for robust model performance. The
sensitivity analysis described in this section was conducted in order to evaluate key input
A-23

-------
parameters for their potential impact on modeled concentrations. Lead monitoring data also
have uncertainty and variability that may contribute to differences between monitored and
modeled concentrations.20 This appendix presents information regarding only the modeling
sensitivity analyses.
The sensitivity analyses focused on the subset of parameters that previous modeling suggested
most strongly influence concentrations: 1) the location that pilots use for run-up activities, 2)
the duration that pilots conduct run-up activities, and 3) whether pilots are using a SE- or ME
plane.21 Regarding the first parameter, run-up activities are conducted in a designated area
immediately adjacent to a runway end; however, the area is generally large enough for several
aircraft to park in, and thus the location of run-up relative to a monitor or model receptor can
vary by several meters depending on where an aircraft is within the designated run-up area.
Run-up activities that are modeled to occur closer to a model receptor than actually occurred,
would be expected to result in the model over-predicting concentrations when compared with
monitored concentrations. Conversely, run-up activity that is modeled to be further away from
the location at which an aircraft conducted this activity would be expected to result in the
model under-predicting concentrations compared with monitored data. To evaluate this
distance parameter, a series of supplemental model receptors were setup in three concentric
rings around the monitor adjacent to the maximum impact location on days with under- or
over-prediction as shown in Figure A-6. The use of supplemental model receptors is analogous
to moving the emissions source in the model (i.e., aircraft conducting run-up), but is more
feasible (i.e., requires less modifications to the model runs and can be completed in a single
model run).
20	EPA's Data Quality Objective Goals for lead is defined as follows: Measurement quality objectives for precision
will be 20% for a 90% confidence limit coefficient of variation and an overall absolute bias upper bound goal of
15%. Goals will be assessed on 3 years of data at the PQAO level of aggregation. EPA-545/B-14-002
September 2014. Available at:
https://www3.epa.gov/ttnamtil/files/ambient/pb/PbPEPHighVolumeSamplineSOP2Q14Revision.pdf
21	The concentration of lead in avgas is also a key parameter impacting ground-level concentrations of lead. The
concentration of lead in fuel supplied at RHV during this time period was analyzed and found to be slightly higher
than the maximum specification for avgas and is therefore considered the highest concentration likely. The
analytical value determined for avgas supplied at RHV was used in the air quality modeling and a sensitivity
analysis was not conducted. It is possible that aircraft had fueled at other airfields and the lead content in the
avgas being burned at RHV could have been lower than the maximum specification, potentially accounting for
cases in which the modeled lead concentration over-estimated the monitored value.
A-24

-------
oo0°o
oo0oo
40 Metersl
LJ
O	"Tiered" Runup Sensitivity Sources
(^)	Default Runup Sources
X	Monitors 1 and 2
•	5-m Buffer Receptors
#	10-m Buffer Receptors
O	20-m Buffer Receptors
Satellite Image Source: ESRI
Figure A-6 Concentric rings of receptors around two downwind monitoring locations
Regarding the second parameter, the duration of run-up varies from pilot to pilot. The median
run-up observed at RHV was 40 and 63 seconds for SE and ME aircraft, respectively. These
median run-up times were used to develop modeled concentrations for SE and ME; however,
the 5th and 95th percentiles of run-up times were 16 and 121 seconds for SE aircraft, or 16 and
160 seconds for ME aircraft at RHV.22 If one or more pilots run-up for longer or shorter periods
than the median used in modeling, then modeled concentrations could over- or under-predict
monitored concentrations, respectively.
Finally, the third parameter of SE versus ME activity can influence modeled concentrations
based on the fact that ME aircraft have higher fuel consumption due to the fact that ME aircraft
have two, or more, engines rather than the one engine of a SE aircraft. As such, more ME
activity than included in the modeling for a given day could result in model over-prediction,
while fewer ME aircraft conducing run-up could result in model under-prediction.
All three parameters were examined for days during which the model over- or under-predicted
monitored concentrations. On the one day of over-predication, shifting the run-up location (i.e.,
using supplemental model receptors) to a more southerly run-up location, which was further
from the maximum impact location monitor, resulted in the median value of concentrations at
22 The distributions of run-up duration at the model airport and at other airports are shown in Appendix C.
A-25

-------
supplemental model receptors falling just below the monitored concentrations, such that the
monitored value was within the third quartile of supplemental receptor concentrations (Figure
A-7). The remaining difference between modeled and monitored concentrations could result
from less ME aircraft activity occurring on this day than was recorded by observers, or from
shorter than median run-up durations. While other parameters could contribute to some of the
remaining difference between monitor and modeled concentrations, this evaluation focused on
those parameters shown to most strongly impact concentrations.
Pb Concentrations from Piston-Engine
Aircraft Emissions at Maximum Impact Site
Monitor
(Overprediction Day)
0.7 --
0.6
do 0.5 --
g.
~ 0.4 +
c
Cl)
u
c
o
u
0.3 --
0.2
0.1 --
~ Third Quartile
Second Quartile
~ Modeled Value at
Monitor
~ Monitored Value

111

1


0 -1-
Figure A-7. Range of modeled lead concentrations from piston-engine aircraft during airport operating
hours at supplemental receptor sites on day that model over-predicted monitored concentrations.
Whiskers represent first and fourth quartiles. Supplemental model receptors were placed in concentric
circles of 5,10, and 20 meters from the monitor to mimic a change in run-up location.
Figure A-8 presents the range of concentrations at the supplemental model receptors on one of
the days that the modeled lead concentration under-predicted the monitored concentration.
The range in concentrations at supplemental model receptor sites still falls below the
monitored concentration, suggesting that a shift in run-up location did not account for the
difference in concentrations. Longer run-up time and higher fuel consumption on this particular
day (compared with the median observed at the model airport) are possible explanations for
the difference between modeled and monitored concentrations. Thirty-percent of the
A-26

-------
difference in concentrations would be accounted for if 10% of pilots conducted run-up for 2
mins, as opposed to the 40 or 63 second median values for single- or multi-engine aircraft,
respectively. This combined with higher levels of ME activity (i.e., higher fuel consumption)
could result in a modeled concentration much closer to the monitored value, similar to model
performance on the majority of model to monitor comparison days.
Pb Concentrations from Piston-Engine
Aircraft Emissions at Maximum Impact Site
Monitor
(Underprediction Day)
0.8
0.7 -
0.6 -
M 0.5 -
~ 0.4 -I
c
Cl)
u
c
o
u
0.3 -
0.2 -
0.1
0 J
~	Third Quartile
~	Second Quartile
~ Modeled Value at
Monitor
~ Monitored Value















Figure A-8. Range of lead concentrations from piston-engine aircraft during airport operating hours at
supplemental receptor sites on one day that the model under-predicted monitored concentrations.
Whiskers represent first and fourth quartiles. Supplemental model receptors were placed in concentric
circles of 5,10, and 20 meters from the monitor to mimic a change in run-up location.
Another possible cause for the underestimation of the modeled concentration is a possible
overestimate of the initial sigma-y and initial sigma-z. As described in Section A.l, we used an
average wingspan of 10.67 m (35 feet) for our initial sigma-y and sigma-z calculation. If the
average aircraft wingspan is smaller, then the initial lateral and vertical dispersion may be
overestimated, leading to an underestimate of the peak concentration.
One additional factor of note is a challenge commonly faced when modeling mobile source
emissions, namely modeling concentrations closely proximate to the emission source can result
in receptors being placed within the modeling exclusion zone. The exclusion zone is the region
A-27

-------
2.15 X sigma-y + 1 meter from the center of the volume source. When the source is
parameterized as a volume source (the approach used for aircraft emissions in this work),
AERMOD reports concentrations of zero for receptors that are inside the exclusion zone of that
particular volume source. This results in concentrations that are biased low. In the sensitivity
analysis presented here, several receptors were within the exclusion zone and, therefore, the
concentration of lead reported at these receptors is biased low. Note that for the year-long
modeling analysis presented in Section A.l, there were no receptors inside the exclusion zone.
A3 Fleet Composition Between the Model Airport and the National Fleet
Section A.3 presents an evaluation of two key parameters that underlie the development of the
AQFs (as presented in Section 3 of the report) in order to understand similarities and
differences between piston-engine aircraft operating at the model airport and those in the
national fleet. Specifically, this appendix compares the aircraft classes and engine types active
at the model airport to a national database of registered piston-engine aircraft since fuel
consumption rates differ across aircraft class and engine type which in turn impacts lead
concentrations. As described in Section 2 of the report, AQFs were developed using modeled
concentrations at a model general aviation airport. As input to the air quality model, aircraft
were observed at the model airport for a period of ten days. The counts of unique aircraft by
aircraft class (SE/ME) and engine type were then compared to a national database of all US
registered piston-engine aircraft. Of the 403 piston aircraft observed at the model airport, 377
(92.0%) were single-engine aircraft. In the national registered piston database, 225,697 of
245,665 (91.9%) aircraft were single-engine. Thus, the differences in the aircraft class
populations are not statistically significant (x2 =1.617, p = 0.2035).
Tables A-10 and A-ll present the total number of piston aircraft organized by engine
technology group (4-stroke horizontal carbureted [carb], 4-stroke horizontal fuel injected [fi], 4-
stroke horizontal spark turbocharged [turbo], and 2-stroke horizontal) and horsepower for the
national database of registered piston aircraft and the observed aircraft at the model airport.
Aircraft were categorized as 'Missing' where either engine technology type or engine size data
was unavailable. Radial engines are not presented in Tables A-10 and A-ll as they span a
broader range of horsepower, but account for only 3.6% and 2.5% of engines in the national
database and observed at the model airport, respectively.
Table A-10. Number of piston-engine aircraft in the national fleet by technology group and
horsepower
Tech.
Group
Horsepower (hp)
hp<100
100
-------
Tech.
Group
Horsepower (hp)
hp<100
100
-------
injected engines are generally more fuel efficient and turbocharged engines are generally less
fuel efficient than carbureted engines (Heiken et al. 2014), there is an expectation that any high
or low bias in fuel consumption from the high prevalence of one of these engine technology
types would be offset by the similar high prevalence of the other. In addition, the prevalence of
fuel injected and turbocharged engines across the 10 survey days suggest that these engine
technology groups may make up a slightly disproportionate percentage of total operations at
the model airport compared to the national fleet. Importantly, due to the fact that not all
aircraft registered in the national database may be routinely operated, differences between the
observed aircraft operating at the model airport and aircraft in the registered database may not
be indicative of differences between aircraft operating at the model airport and aircraft
operating at other airports. Overall, results suggest that differences between the observed fleet
at the model airport and the national piston-engine aircraft fleet generally balance out (i.e.,
turbocharged versus fuel-injected), or are not statistically significant.
Table A-12. Fleet composition by aircraft engine technology
Engine Technology
Group
Model
Airport
Composition
National
Fleet
Composition
Observed Aircraft
at Model Airport
Expected Aircraft
at Model Airport
Given National
Composition
4-Stroke,
Carbureted
58.1%
65.7%
210
237
4-Stroke, Fuel
Injected
29.1%
21.7%
105
79
4-Stroke,
Turbocharged
10.0%
7.8%
36
28
2-Stroke
0.3%
1.2%
1
4
4-Stroke Radial
2.5%
3.6%
9
13
A.4 Mode-Specific Fuel Consumption
In the air quality modeling approach used to develop AQFs, total fuel consumption was
modeled using engine-specific fuel consumption rates for 18 engine types. The 18 engine types
span the five engine technology groups (4-stroke horizontal carbureted, 4-stroke horizontal fuel
injected, 4-stroke horizontal turbocharged, 4-stroke radial, and 2-stroke horizontal) for piston-
engine spark ignition aircraft engines and range from 64-575 horsepower. For each engine type,
fuel consumption rates were prescribed for each landing-and-take-off (LTO) mode based on the
FAA's Emission and Dispersion Modeling System (EDMS) Version 5.0.2 or the Swiss Federal
Office of Civil Aviation (SFOCA), "Aircraft Piston Engine Emissions Summary Report" supporting
data (FAA 2007, SFOCA 2007).
A-30

-------
The approach laid out in Heiken, et al., 2014 included calculating fuel consumption by using
default brake specific fuel consumption (BSFC) rates for each of the five spark ignition piston-
driven aircraft engine technology groups (Heiken et al. 2014). BSFC is the mass of fuel
consumed per unit work done by the engine and is a measure of engine efficiency. BSFC data
for each LTO mode for 29 unique engines were compiled from several sources including the
FAA's EDMS model and the SFOCA piston engine summary report. Where BSFC data were not
available for a LTO mode for an engine, mode-specific BSFC were estimated by applying a
scaling factor to the takeoff BSFC for that engine. Engines were sorted by technology group,
and the default mode-specific BSFCs were taken as the mean BSFC for each technology group
and LTO mode.
A comparison between the resulting fuel consumption rates used to characterize lead
concentrations at the model airport (labeled as MA Fuel Consumption) and in Heiken et al.,
2014 (labeled as Heiken Fuel Consumption) is shown in Figure A-9. For each LTO mode, the fuel
consumption rate for each of the 18 engines used in the AQF approach is plotted against the
fuel consumption rate that would be assumed for an identical engine using the mode-specific
default BSFC approach.
A-31

-------
(b)
12 x
Taxi
-10 -¦
C
o
+J
Q.
E
3
c
o
u
"aj
Ll_
C
(D
6 --
4 --
Q>
10
#>•
0
_ 50
Is45
c 40
o
£35
E 30
c 25
o
u 20
S 15 +
c 10
aj
^ 5
QJ
1 0
-+-
+
+
+
-+-
2 4 6 8 10
MA Fuel Consumption (g/s)
(c)
Climb
12
:: 
c 80
0
~ 70
Q.
E 60
1	50
o
O 40
Takeoff
30
20
C
(D
^ 10
(D
3= 0

cP
_20
j» 18
c 16
o
a14
E 12
C
o
u
"aj
Ll_
C
(D
10 --
-h
-h
-h
0	50	100	150
MA Fuel Consumption (g/s)
(d)
200
Approach
<6*
10 20 30 40
MA Fuel Consumption (g/s)
50
50 -r
-40 +
O
30 --
c
o
u
"aj
Ll_
C
a;
20 --
10 --
^ 0
0
Run-Up
8-
—i-
o o
-+
H
15
5	10
MA Fuel Consumption (g/s)
• Carbureted • Fuel Injection OTurbocharged
O Radial	•2-Stroke
Figure A-9. Fuel consumption rates by engine technology group and LTO mode (Heiken et al. 2014)
A-32

-------
For taxi, takeoff, climb, and approach modes, fuel consumption rates between the two studies
strongly correlate (R = 0.993 for takeoff to R = 0.745 for taxi). For each of these modes, a single
radial engine was an outlier with model airport fuel consumption greater than the fuel
consumption rate reported by Heiken et al. Two radial engines were in the database used to
determine the radial engine default BSFC in Heiken et al. one (Wright R-1820) with similarly
high fuel consumption rates to the single engine value used in the model airport fuel
consumption database and one radial engine with an undisclosed manufacturer and model
number. Averaging the mode-specific BSFC across the two radial engines results in significantly
lower predicted fuel consumption rates. Thus, using the default BSFC methodology from Heiken
et al. may under-predict fuel consumption rates by not accounting for engine-to-engine
variation within an engine technology group. Notably, the impact of the radial engine fuel
consumption modeling assumption on resulting lead concentrations is expected to be small as
radial engine aircraft account for only 3.6% of the national piston aircraft fleet.
Aircraft run-up fuel consumption rate data is sparser than data for other LTO modes. For the
model airport, fuel consumption rates for the run-up mode were taken directly from EDMS and
SFOCA where available. Only 4 unique run-up fuel consumption rates were identified. In the
approach used in Heiken et al. run-up fuel consumption rate was defined as 52% of the
maximum fuel consumption rate based on a survey of seven engine manuals that suggested
individual run-up fuel consumption rates range from 43-68% of the maximum fuel consumption
rate. Despite the lack of granularity in data, the models still show weak to moderate correlation
(R2 = 0.28). Fuel consumption rates used at the model airport were higher than those reported
by Heiken et al. for 2-stroke carbureted engines and engines less than 100 horsepower. Aircraft
with these engines account for less than 3% of the national piston aircraft fleet, as seen in Table
A-10. In contrast, relative to the Heiken et al. approach, the data used at the model airport
generally underestimates run-up fuel consumption rates for 4-stroke horizontal engines greater
than 200 horsepower, about 24% of the national piston fleet. The fuel consumption rate for the
radial engine is again an outlier; however, where higher estimates of radial engine fuel
consumption rates were reported for other LTO modes, the run-up fuel consumption rate used
for this engine at the model airport was lower than the fuel consumption rate relative to the
Heiken et al. study. Overall, while methods used to calculate fuel consumption rates differ
between the two studies, a comparison of the resulting rates suggest that the values are largely
similar for modes other than run-up. Within the run-up mode, the results are moderately
consistent, but there is more variability in the results compared with other modes of operation.
Additional data would be necessary to further characterize and evaluate fuel consumption rates
during the run-up mode for piston-engine aircraft.
A-33

-------
A.5 Wind Speed Adjustment Methodology
Meteorological, geographical, and operational parameters may vary from conditions at the
model airport or from the national default parameters used in the national analysis. Wind
speed is one meteorological parameter that effects local concentration profiles of atmospheric
aerosols. Specifically, the near-field concentration of a non-reactive passive tracer scales with
, where u is wind speed and angled brackets imply a time average (Barrett and Britter
2008). Thus, the model-extrapolated concentrations at and downwind of the maximum impact
site can be adjusted to better consider meteorological conditions at individual airports by using
inverse wind speed data over the 3-month maximum period. The methodology for deriving and
applying the wind-speed adjustment is described below.
If the wind speed at the model airport is v and the wind speed at a specific airport is u, then the
wind-adjusted concentration would be the model extrapolated concentration estimated by the
methodology detailed in Section 3 of the main report multiplied by the ratio of average inverse
wind speeds /. If the wind speed at the specific airport is, in general, higher than the
wind speed at the model airport where the AQFs were derived, then  would be less than
<1/_1> resulting in a lower concentration per unit of activity (i.e., 1 LTO) at the specific airport
than the AQF.
The effect of the wind-adjusted AQF is demonstrated and validated using the model airport
data. First, we evaluated the range of weighted 3-month AQFs versus the range of rolling 3-
month, inverse wind speeds. As shown in Figure A-10, the average inverse wind speed for the
average weighted AQF23 is 0.426 s/m with 3-month average inverse wind speed ranging from
0.325 to 0.523 s/m.
23 The 'weighted AQF' is a composite of the maximum SE (89%) and ME (11%) full AQF, thereby reflecting expected
operational splits between SE and ME aircraft. The approach is valid and holds for each of the individual (SE/ME,
full/T&G) AQFs.
A-34

-------
3.5E-05
~	3.0E-05
E	2.5E-05
Ss
~ 2.0E-05
LL
<	1.5E-05
"S	1.0E-05
¦If 5.0E-06
O.OE+OO
Weighted AQF vs Inverse Wind Speed
CO
8	*
o
y = 0.00003941x+ 0.00000645
R2 = 0.79923908
0.25 0.30 0.35 0.40 0.45 0.50
Rolling 3-Month  (s/m)
0.55
0.60
Figure A-10 Weighted AQF vs inverse wind speed for each 3-month rolling AQF at model airport
modeled, weighted AQF for each of the 3-month periods with the average AQF (or the AQF
applied in the main analysis without considering wind adjustment) and the wind-adjusted AQF.
Figure A-ll shows that the average AQF is always within +/-20% of the specific 3-month
average modeled AQF. The wind-adjusted AQF performs even better, staying within 6% of the
modeled AQF for all modeled periods.

3.50E-05
o
3.00E-05


	1

m*""
F
2.50E-05


3
2.00E-05
LL
a

<
1.50E-05
TJ

Q)

_C
1.00E-05
op

*0)

5
5.00E-06

0.00E+00
+
o
~
AQF by 3-Month Period at RHV
£~~~~~
* ^ i ^
a
+
o
A
+
o O O
Add
+ Modeled AQF
¦ Avg AQF
O Wind-Scaled AQF
^ sT Vs" A ^ v*
if
^	S*
x'	(.
0°

-------
Utilizing the same wind data that was used to assign operations to specific runways, model-
extrapolated concentrations at airports nationwide can be adjusted for wind-speed, thereby
appropriately characterizing concentrations at airports with significantly higher or lower wind
speeds than the model airport. For the wind speed adjustment, wind speeds from 6am to
11pm24 are averaged over the entire year at the model airport and for the 3-month maximum
activity period at the model airport. As the inverse of wind speed tends toward infinity as wind
speed tends toward zero, 0.5 m/s is chosen as a minimum allowable wind speed to account for
ASOS station wind detection limits.
24 These are the modeled hours from opening through one hour past closing for each airport, reflecting the times
when atmospheric lead concentrations are expected to be highest.
A-36

-------
References
BAAQMD (2011). 2015 Toxics Modeling to Support the Community Air Risk Evaluation (CARE)
Program. 201101-008-TX. Bay Area Air Quality Management District. San Francisco. January,
2011.
Barrett, S. R. H. and R. E. Britter (2008). Development of algorithms and approximations for
rapid operational air quality modelling. Atmospheric Environment, 42 (34), 8105-8111. DOI:
http://doi.Org/10.1016/i.atmosenv.2008.06.020.
Benson, P. E. (1979). Abridged User's Guide for CALINE-3 - A versatile dispersion model for
predicting air pollutant levels near highways and arterial streets. O. o. T. Laboratory.
Carr, E., M. Lee, K. Marin, C. Holder, M. Hoyer, M. Pedde,... J. Touma (2011). Development
and evaluation of an air quality modeling approach to assess near-field impacts of lead
emissions from piston-engine aircraft operating on leaded aviation gasoline. Atmospheric
Environment, 45 (32), 5795-5804. DOI: http://dx.doi.Org/10.1016/i.atmosenv.2011.07.017.
FAA (2007). EDMS 5.0.2: Emission and Dispersion Modeling System Software. Washington, DC,
Federal Aviation Administration. 5.0.2.
Heiken, J., J. Lyons, M. Valdez, N. Matthews, P. Sanford, J. Turner and N. Feinberg (2014).
Quantifying Aircraft Lead Emissions at Airports. ACRP Report 133.
http://www.nap.edu/catalog/22142/quantifying-aircraft-lead-emissions-at-airports.
SFOCA (2007). Aircraft Piston Engine Emissions Summary Report. Bern, Switzerland, Swiss
Federal Office of Civil Aviation (SFOCA).
USEPA (2004). User's guide for the AMS/EPA regulatory model - AERMOD. Office of Air Quality
Planning and Standards. September 2004.
https://www3.epa.gov/ttn/scram/models/aermod/aermod userguide.pdf.
USEPA (2006). Air Quality Criteria for Lead - Volume 1. EPA/600/R-05/144aF. U.S.
Environmental Protection Agency. Research Triangle Park, NC. October 2006.
USEPA (2008a). AERSURFACE users' guide. EPA-454/B-08-00. Office of Air Quality Planning and
Standards. January 2008.
USEPA (2008b). Regulatory impact analysis of the proposed revisions to the National Ambient
Air Quality Standards for Lead. EPA Office of Air Quality Planning and Standards. October 2008.
USEPA (2009). AERMOD implementation Guide. Office of Air Quality Planning and Standards.
March 2009.
http://www.epa.gov/ttn/scram/7thconf/aermod/aermod implmtn guide 19March2009.pdf.
A-37

-------
USEPA (2010). Development and Evaluation of an Air Quality Modeling Approach for Lead
Emissions from Piston-Engine Aircraft Operating on Leaded Aviation Gasoline. EPA-420-R-10-
007. https://nepis.epa.gov/Exe/ZyPDF.cgi/P1007H4Q.PDF?Dockey=P1007H4Q.PDF.
USEPA. (2011). "2011 National Emissions Inventory (NEI) Data." 2017, from
http://www.epa.gov/air-emissions-inventories/2011-national-emissions-inventory-nei-data.
USEPA (2013a). Calculating Piston-Engine Aircraft Airport Inventories for Lead for the 2011
National Emissions Inventory. EPA-420-B-13-040. September 2013.
USEPA (2013b). Integrated Science Assessment for Lead. EPA-600-R-10-075F. National Center
for Environmental Assessment - RTP Division Office of Research and Development. June 2013.
A-38

-------
Appends ><: Supplemental Data for Piston-Engine Aircr f; tivity and Model-
Extrapolate d Contraction Gradients
This appendix provides supplemental information on data sources and their application in
methods detailed in Section 3 of the main report, which characterizes 3-month average model-
extrapolated lead concentrations at and downwind of the maximum impact site for the runway-
end with the most piston-engine aircraft activity in a 3-month period for airports nationwide.
Section B.l briefly describes the sources of airport activity data noting the extent of the data, its
resolution, the data collection method, and the location of publicly available information.
Section B.2 provides a detailed description of the diurnal profile used in the national and
airport-specific activity analyses (Sections 3.2 and 3.3 of the main report); this profile is based
on observations at the representative airport facility and is utilized in the AQFs presented in
Section 3.1 of the main report. Section B.3 describes in detail the method for assigning
operations to a specific runway using hourly wind data. Section B.4 provides the detailed
aircraft and operational data for six airports where airport-specific data was available, which
was utilized in Sections 3.3 and 4 of the main report. Section B.5 discusses piston-engine
rotorcraft activity and provides information to support future analyses.
B.l Airport Operations Data Sources
FAA provides a number of data sources related to aircraft activity at airports. The data sources
relevant to this report are presented below and generally organized from the most to least
detailed. As a reference, Table B-l lists of FAA location identifiers (LID) with airport names and
locations for airports identified by FAA LID in this report.
Table B-l. List of FAA location identifiers, airport names, and location by state/territory for airports
identified in this report.
Airport Name
LID K
Location
52F	Northwest Regional Airport
ORS	Orcas Island Airport
RHV	Reid-Hillview Airport
WHP	Whiteman Airport
TX
WA
CA
CA
B-l

-------
Air Traffic Activity System (ATADS)1
The Air Traffic Activity [Data] System (ATADS) contains the official National Airspace System air
traffic operations data available for public release.2 Approximately 500 US airports have either
an FAA air traffic control tower or an FAA contract tower. ATADS provides daily operational
data at the airport as reported by the control tower categorized by itinerant and local
operations and separated by air carrier, air taxi, general aviation, and military. ATADS is
updated monthly. While ATADS activity data is the most up-to-date and offers the finest
temporal resolution of the FAA datasets, it is only available at the approximately 500 towered
airports and does not specify activity by engine-type (e.g., piston- vs. jet-engine activity) within
general aviation and air taxi.
National Plan of Integrated Airport Systems
The National Plan of Integrated Airport Systems (NPIAS) identifies nearly 3,400 existing and
proposed airports that the FAA considers significant to national air transportation and are thus
eligible to receive Federal grants under the Airport Improvement Program. The NPIAS includes
all ATADS airports. Because these airports are eligible to receive Federal grants, they are subject
to data reporting requirements, including reporting based-aircraft by tail number.
Terminal Area Forecast
The Terminal Area Forecast (TAF) is the official FAA forecast of aviation activity for US airports.
Yearly reports detail historical annual activity data and projected operations for future years at
the airport level for NPIAS airports.3 The TAF also provides data and future projections of based
aircraft for NPIAS airports.
FAA 5010 Report (Airport Master Record)
An Airport Master Record (Form 5010) is an electronically generated file for each airport facility
that details General Information, including ownership, management, and location data; Services
& Facilities including available fuel types and flight services; Based Aircraft & Operations;
Runway Information; and Remarks, including special instructions and updates. A complete Form
5010 can be generated for an individual airport through the online AirportlQ DataCenter.4
The FAA Office of Airport Safety & Standards (AAS-100) provides access to airport facilities and
runway data from Form 5010 for all public-use and private-use facilities (including ATADS
airports, non-ATADS NPIAS airports, and non-NPIAS airports) available for download through
the FAA website.5 Based aircraft counts and annual activity levels are available through the
Airport Facilities Data database. Operations data is reported for 12-month periods and are
partitioned by Air Carrier, Air Taxi, General Aviation Local, General Aviation Itinerant, and
Military. Operations at non-FAA airports are estimated by FAA inspectors or are based on
1	ATADS activity data and other operational data (such as delay) are now available through the FAA Operations
Network (OPSNET). https://aspm.faa.eov/
2	https://aspm.faa.gov/opsnet/svs/Main.asp
3	https://aspm.faa.gov/main/taf.asp
4	https://www.airportiq5010.com/5010web/
5	https://www.faa.gov/airports/airport safetv/airportdata 5010/
B-2

-------
information provided by airport managers, state aviation activity surveys, or other sources.
Airport location and runway orientation are available through the Airport Runways Data
database.
Airport data in the 5010 report is derived from the National Airspace System Resources
Database (NASR). The NASR Database contains extensive aeronautical information on all US
airports including safety critical data, navigational aid data, and airport configuration data. The
NASR is updated every 56 days. The NASR Database is provided by the FAA's Aeronautical
Information Services Group (AJV-5) through the National Flight Data Center (NFDC).6 The NFDC
is responsible for the collection, validation, and quality control of aeronautical information
detailing the physical description, geographical position, and operational characteristics of
airport facilities.
General Aviation and Part 135 Activity Survey/General Aviation and Air Taxi Activity (GAATA)
Surveys
The General Aviation and Part 135 Activity Survey, also known as the General Aviation and Air
Taxi Activity Survey (GAATA), is an annual data report produced from surveys of US pilots. The
GAATA enables the FAA to monitor the general aviation fleet for purposes of anticipating and
meeting demand, evaluating initiatives and regulatory changes, and measuring the safety of the
GA community. The GAATA provides data as to the composition of the general aviation and air
taxi fleet, including national data on fleet composition and operational hours by type of aircraft
(e.g. piston-driven, turboprop, turbojet) and aircraft class (i.e. SE vs ME). The data collected are
also used by other government agencies, the general aviation industry, trade associations, and
private businesses to pinpoint safety problems and to form the basis for critical research and
analysis of general aviation issues. Tabular data from annual GAATA reports are publicly
available for download on the FAA website.7
General Aviation Manufacturers Association (GAMA) Statistical Databook
GAMA is an international trade association representing more than 90 of the world's leading
manufacturers of general aviation airplanes and rotorcraft, engines, avionics, components, and
related services. GAMA publishes an annual statistical databook and an annual industry
outlook, containing diverse data from operations at airports to sales. While the GAMA
databook is not used as a primary data source for modeling in the main report, it is a useful tool
for validating data assumptions, such as GA operation trends.
The trends in total piston operations over six years for a sample subset of the 50 most active GA
airports, as listed in the General Aviation Statistical Databook (GAMA 2016), are taken from
ATADS data and are shown in Figure B-l. The median year-on-year change in operations at the
top 50 GA airports ranges from -4% to 4% for a given year, and the interquartile of airport year-
on-year operational changes is between +/-10% for all years. However, individual airport year-
on-year changes range from -25% to +44%.
6	https://nfdc.faa.gov/nfdcApps/
7	https://www.faa.gov/data research/aviation data statistics/general aviation/
B-3

-------
on-Year Airport Fractional Change in GA Operations
—O— Median
	Interquartile
	5th, 95th Perc.
Min, Max
-0.2
-0.4
-0.6
-0.8
-1
2011	2012	2013	2014	2015	2016
Figure B-l. Change in annual GA aircraft operations at the 50 most active airports by GA traffic from
2011-2016s
Year-
i
0.8
0.6
0.4
B.2 Diurnal profile
Section 2,2 of the main report details piston-engine aircraft emissions modeling at a GA airport
facility (RHV). Part of this work addressed the fact that aircraft activity is not constant over the
course of a day. Specifically, a detailed distribution of aircraft operations by hour was
developed from on-site observations and survey data and applied in the modeling approach.
Figure B-2 shows the distribution of operations across operational hours (i.e., diurnal profile) at
RHV by aircraft class (SE vs. ME) and operational cycle-type (Full LTO vs. T&G). Distinct diurnal
profiles were observed for weekdays (Figure B-2a) and weekends (Figure B-2b), with ME
operations showing more variation between the two-day types.
8 For this comparison, GA operations are the sum of itinerant general aviation and local civil aviation operations as
reported in ATADS. Overflight and air taxi operations are not included.
B-4

-------
(a) RHV Diurnal Profile - Weekdays
0.35
0.3
¦£ 0.25
0.2
~ 0.15
0.1
0.05
0
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Hour
•	.SELTO 	-SET&G 	ME LTO MET&G
(b) RHV Diurnal Profile - Weekends
0.35
0,3
~ 0.25
0,2
~ 0.15
K o.i
0.05
0
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Hour
•	-SELTO 		.SET&G 	ME LTO	MET&G
Figure B-2. Diurnal profile of aircraft operations by aircraft class, cycle-type, and day-type from RHV
monitoring and survey data
Detailed hourly operational data similar to that presented in Figure B-2 is unavailable for each
of the 13,153 airports in the national analysis. Yet, as described in Section 3 of the main report,
applying the AQFs to characterize lead concentrations at airports nationwide requires detailed
operational data as an input for each airport. For each hour, aircraft activity and wind direction
data are used to assign operations to specific runways at each airport. As noted in Table 2 of
Section 3, we used the diurnai operational profile shown in Figure B-2 to calculate hourly
B-5

-------
operations at each airport. We recognize that the distribution of operations can vary between
facilities. Figure B-3 shows the hourly operational profiles at four airports. The profiles for RHV
(weekday and weekend) represent the same operational profiles shown in Figure B-2, albeit not
separated out by aircraft class and cycle-type. The profile for SMO was developed by on-site
observation and survey data and underlies the lead modelling development study described in
Section 2.1 (Carr et al. 2011). Two additional profiles developed from on-site observation and
monitoring at the Richard Lloyd Jones airport (RVS) and Centennial airport (APA) are also shown
as taken from the National Academies of Sciences Airport Cooperative Research Program
Report Quantifying Aircraft Lead Emissions at Airports9 (Heiken et al. 2014).
0.14
0.12
s —
0.1
£ 0.08
£ 0.06
0.02
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Hour
	RHV-weekday	RHV-weekend	SMO	RVS	APA
Figure B-3. Diurnal profiles of aircraft operations at four airports: RHV, SMO, RVS, and APA
A comparison of the diurnal profiles across these four facilities shows the same basic features: a
ramp-up of activity in early morning, peaks in activity in late morning and early afternoon, and
decreasing operations in the evening. For the National Analysis (Section 3 of main report), we
selected the RHV profile as it provides the most detailed information in terms of how activity
may vary over the course of a day based on aircraft class, operation types, as well as day type
(weekday vs. weekend). In modeling studies, monthly average concentration has been shown
to be insensitive to diurnal profile choice while holding daily operation count and runway
assignment constant (Feinberg and Turner 2013). In characterizing concentrations using AQFs in
the main report, the operational diurnal profile is an important parameter because runway
assignment is based on predominant hourly wind direction as described in Section B.3 of this
appendix.
9 Quantifying Aircraft Lead Emissions at Airports also presents a diurnal operational profile for SMO; however, data
was not available for all operational hours.
B-6

-------
Because take-offs contribute more significantly to concentrations at the maximum impact
location, a possible source of uncertainty is using the same diurnal profile for modeling landings
and take-offs. To understand this sensitivity, the impact of using a generic "operational" diurnal
profile vs. a "landing" or "take-off" was examined. For this choice of diurnal to be impactful on
three-month averaged concentrations, two factors would need to occur: the difference
between the diurnal profile of landings and the diurnal profile of take-off would need to be
significantly different, and the average wind direction at the time of over-estimated take-offs
would need to be significantly different than the average wind direction at the time of under-
estimated take-offs.
Given that piston aircraft do not typically operate at night and that an aircraft must first take-
off for it to land, there is an expectation that take-offs will (on average) occur earlier than
landings. However, piston-engine aircraft typically perform short operational missions. Thus,
while at the margins, landings should occur later than takeoffs, we do not expect the profile of
landings and takeoffs to differ significantly. Airport surveys at six airport reported counts of
landing and take-offs during operating hours or a subset of hours for between three and six
days of operation. Operational survey data were excluded for any day that did not have survey
data covering at least 80% of operational hours or for any days where both landing and take-off
data were not available. Figure B-4 shows the difference in percentage points of the landing
and take-off diurnal profiles at each of these airports as reported in survey data. The data
confirms the expectation that, in the first (last) hour of operations monitored, take-offs were
relatively more (less) prevalent than landings, but that profiles were otherwise similar over the
day. The figure shows that, on average the difference between a landing diurnal profile and a
takeoff diurnal profile is 2.6 percentage points. Further, 95% of examined hours show a
difference of less than 6 percentage points between the landing diurnal profile and the take-off
diurnal profile. Thus, using a generic "operation (LTO)" profile will, on average, over or under
predict takeoffs by 1.3% in any given hour.
B-7

-------
RHV
•MRI
CRQ
DCU
DVT
VNY
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure B-4. Difference between landing diurnal profile and takeoff diurnal profile from 3 to 6 days of
survey data at each of 6 airports
In the main report, a +/-10% maximum 3-month concentration is considered in addressing
airports where lead concentrations may approach or exceed 0.15 (J,g/m3 (in addition to
considering variation in other more sensitive parameters such as the expected split between
multi- and single-engine aircraft), which far exceeds the +/-1.3% uncertainty from differences
in landing and take-off diurnal profile.
Further, while the average difference between a generic operation diurnal profile and a take-off
only diurnal profile is 1.3%, the actual uncertainty in resulting 3-month average concentration
may be even smaller. Since aircraft are assigned to runway primarily based on wind direction, a
modeled difference in operations would require the wind direction to change significantly from
the time in which takeoffs were overestimated to the time in which they are underestimated.
Figure B-5 shows the average wind speed at 938 ASOS stations nationwide for each hour of the
day. Wind direction is normalized such that the wind direction at 00:00 is 0° at all stations. Each
ASOS station is plotted along a circle of different unit radius, and each hour is modeled by a dot
where angle represents difference in wind direction from 00:00 and color represents time of
day. The figure shows that across all ASOS stations, 86% of all hours have average wind speeds
that fall within 90° of the initial recorded wind direction. Therefore, for example, at a single
runway airport, even if the diurnal profile were moving 2% of operations from the morning to
the afternoon, there is an expectation that, averaged over a three-month period, those
operations would still generally be assigned to the same runway end given consistent wind
directions.
B-8

-------
180 -180
00:00-01:00
Figure B-5. Average wind direction by hour of the day at 938 ASOS stations
Given the evidence that 3-month average concentrations are insensitive to diurnal profile, the
selection of the RHV diurnal profile is appropriate for the national analysis of lead
concentrations.
B.3 Runway Assignment
The methodology for characterizing lead concentrations at airports nationwide presented in
Section 3 of the main report requires every aircraft operation10 at every airport be assigned to
an active runway. Piston-engine aircraft typically take-off and land into the wind, and wind is
the primary driver for selecting active runways (Lohr and Williams 2008). In the national and
airport-specific analyses, hourly local wind direction data were used to identify on which
runway piston-engine aircraft conduct take-off and landing operations. The active runway is
determined using the minimum degree difference approach - identify the runway that has the
smallest difference between the direction of the prevailing wind in that hour and the runway's
heading in degrees and assign operations to that runway end. Where wind or runway headings
are given with reference to magnetic north, directions are corrected to true north to maintain
consistency across data sets.
In the national analysis presented in Section 3 of the main report, an active runway was
selected for each airport for each hour of the day. The minimum degree difference approach
10 One LTO cycle consists of two operations (takeoff and landing). Thus, the number of operations is divided in half
to calculate LTOs per runway in the National Analysis.
B-9

-------
was applied to 99.1% of airport-hour cases. In the remaining 0.9% of cases, there was no
measurable wind, or wind direction data was missing. In these cases, operations were assigned
assuming the most active runway configuration was consistent with the remaining hours of the
day as there is an expectation that, absent other factors, an airport would maintain operational
consistency.11
Of the 99.1% of airport-hour cases where airport-hour pair12 data existed, there is only one
runway that meets the minimum degree difference requirement for the overwhelming majority
of cases (98.4%). However, in the remaining 1.6% of airport-hour pairs, the minimum degree
difference approach was insufficient for assigning an active runway either because there were
two or more runways with identical runway headings (parallel runways) or because the wind
angle bisected two or more runway headings.
Section B.3.1 details the active runway configuration selection method for these scenarios, as
well as the minimum degree difference scenario. Section B.3.2 describes the wind source data
underlying the runway assignment methodology and discusses sources of uncertainty and
caveats of the runway assignment method.
Section B.3.1 Runway Assignment Algorithm
This section describes the runway assignment algorithm. When hourly wind data is available
and the predominant wind direction aligns with only one airport runway, the active runway end
is assigned based on the minimum degree difference approach (Scenario 1). For cases where
wind data is unavailable for a given hour or more than one runway aligns with the predominant
wind direction, the choice of active runway end is dependent upon the layout of the airport
runway system. These cases are described in Scenarios 2 through 5.
Scenario 1: Minimum Degree Difference Approach
Figure B-6. Example Scenario 1 Configurations. Black lines represent runways, green arrows represent
wind direction, and red dots represent the runway-end identified as active
Description: The minimum degree distance approach results in only one preferred available
runway.
Occurrence: 98.4%
11	In the few cases where wind direction data was missing for an entire day, operations were assigned to each
runway end equally.
12	An airport-hour pair is a matched pair of operation cycle (LTO/T&G) data and wind data for one airport over one
specific hour of the year.
B-10

-------
Selection Decision: All operations for the hour are assigned to the runway identified by the
minimum degree distance approach.
Scenario 2: Non-Parallel Bisection
Figure B-7. Example Scenario 2 Configurations. Black lines represent runways, green arrows represent
wind direction, and red dots represent the runway-end identified as active
Description: The prevailing wind bisects the heading of one or two non-parallel runways
resulting in two options for an active runway, based on the minimum direction distance
approach.
Occurrence: 0.15%
Selection Decision: All operations are assigned to the most active runway during the remaining
hours of the day, based on the minimum direction distance approach. If more than one
runways have been equally active during the remainder of the day, then operations for
the hour are split evenly between the runway options.
Scenario 3: Parallel Runways
Figure B-8. Example Scenario 3 Configurations. Black lines represent runways, green arrows represent
wind direction, and red dots represent the runway-end identified as active.
Description: The minimum direction distance approach identifies two or more parallel runways
options for an active runway
Occurrence: 1.44%
Selection Decision: 90% of operations are assigned to a primary runway and 10% of operations
are assigned to a second runway. Any 3rd or 4th parallel runway is assumed to not serve
piston-engine aircraft. GA airports with multiple parallel runways will often have a
preferential runway for operations, while airports serving GA and commercial operations
may have a designated GA runway or a preferred take-off runway for all operations.
Thus, the selection decision assumes a preferred operational runway. However,
B-ll

-------
operations can still occur on the non-preferred runway(s).13 Thus, allocating 100% of
operations to the preferred runway would over-estimate lead concentrations in the case
where some operations occur on alternative runways.
Scenario 4: Combination Parallel and Bisection
Figure B-9. Example Scenario 4 Configurations. Black lines represent runways, green arrows represent
wind direction, and red dots represent the runway-end identified as active
Description: The minimum direction distance approach identifies both parallel and non-parallel
runways
Occurrence: < 0.1%
Selection Decision: Parallel runways are first treated as runway groups. The approach for non-
parallel bisection (Scenario 2) is applied between runway groups. The approach for
parallel runways is then applied to the operations assigned to each runway group.
Scenario 5: Additional Multi-Runway Airports
Description: There are 15 airports in the national dataset with 5 or more runways. For these
multi-runway airports, a subset of runways that serves piston-engine operations was
identified from available operational, planning, or capacity information from the airport
website or the FAA. These airports are described in more detail below.
Selection Decision: Only runways serving piston-engine operations are considered. Operation
assignment rules then follow the selection criteria in Scenarios 1-4 noted above unless
otherwise noted.
(a) Runways with only one identified piston-aircraft runway.
The following multi-runway airports have a single identified runway serving as the primary
runway for piston-engine aircraft operations.
ATL: Based on the ATL Master Plan,14 General Aviation hangars and other infrastructure are
located on the north side of the airport. Therefore, piston-engine operations are assigned
to runway 8L/26R following the approach laid out for Scenarios 1 and 2.
13	For example, at RHV, an airport with two parallel runways, most take-offs and landings occur on the eastern-
most runway (31R/13L). However, the western-most runway is used frequently for touch-and-go operations.
14	https://www.atl.com/wp-content/uploads/2016/12/ATL ExecSumm 2015 101415 Spreads.pdf
B-12

-------
DFW: Based on the DFW Master Plan, runway 17L/35R15 is the preferred runway for General
Aviation. Therefore, piston-engine operations are assigned to runway 17L/35R following
approach laid out for Scenarios 1 and 2.
IAH: Based on the IAH Master Plan,16 General Aviation exclusively uses runway 15R/33L.
Therefore, piston-engine operations are assigned to runway 15R/33L following the
approach laid out for Scenarios 1 and 2.
TCS: TCS has one asphalt runway (13/31) and four gravel runways. Therefore, piston-engine
operations are assigned to runway 13/31 following the approach laid out for Scenarios 1
and 2.
TX99: TX99 is a private use airport with 5 turf runways. Piston-engine operations are assigned
to runway 8R/26L following the approach laid out for Scenarios 1 and 2.
(b) Airports with two non-parallel identified piston-aircraft runways.
The following multi-runway airports have two identified non-parallel runways serving as the
primary runways for piston-engine aircraft operations.
BOS: Based on the BOS Tower Standard Operating Procedures,17 and information provided by
Massport, the operator of the airport,18 runway 14/32 is used exclusively for props and
small jet aircraft while runway 4L/22R (purple) is not used for jets. Therefore, piston-
engine operations are assigned to runways 14/32 and 4L/22R following the approach laid
out for Scenarios 1 and 2.
DEN: Two runways were identified as serving General Aviation at DEN, runway 17R/35L is the
preferred GA runway and runway 8/26 was selected as the preferred runway during a
crosswind due to its proximity to the GA hangars. Therefore, piston-engine operations
are assigned to runways 17R/35L and 8/26 following the approach laid out for Scenarios
1 and 2.
DTW: DTW has two non-parallel runways in close proximity to the general aviation area.
Therefore, piston-engine operations are assigned to runways 3R/21L and 09/27L
following the approach laid out for Scenarios 1 and 2.
FST: FST has two asphalt runways (12/30 and 3/21) and three turf runways in poor condition.
Therefore, piston-engine operations are assigned to runways 12/30 and 3/21 following
the approach laid out for Scenarios 1 and 2.
MDW: At MDW, General Aviation use runway 4L/22R during normal operations and runway
13R/31L during crosswinds as noted in the MDW Noise Compatibility Study.19 Therefore,
piston-engine operations are assigned to runways 4L/22R and 13R/31L following the
approach laid out for Scenarios 1 and 2.
15	https://dfwairport.com/development/masterplan/index.php
16	http://www.flv2houston.com/about-master-plans, 'Master Plan Volume V
17	http://www.bvartcc.eom/Portals/0/Air%20Traffic%20Control/ATC%20Documents/SQP/BVA KBOS.pdf
18	https://www.massport.com/environment/environmental-reporting/noise-abatement/runwav-use/,
https://www.massport.com/environment/environmental-reporting/noise-abatement/how-logan-operates/
19	https://www.faa.gov/airports/planning capacity/profiles/media/LGB-Airport-Capacity-Profile-Appendix-A-
2014.pdf
B-13

-------
MKE: Based on the MKE Master Plan Update Study,20 piston aircraft primarily use runway
1L/19R and 7R/25L. Therefore, piston-engine operations are assigned to runways 1L/19R
and 7R/25L following the approach laid out for Scenarios 1 and 2.
MWH: Based on the MWH Master Plan,21 General Aviation is primarily assigned to runway
18/36 and runwayl4R/32L. Therefore, piston-engine operations are assigned to runways
18/36 and 14R/32L following the approach laid out for Scenarios 1 and 2.
(c) Other Runway Configurations
The following multi-runway airports have unique identified traffic patterns. Runways identified
as serving piston-engine aircraft for these airports are shown in Figure B-10.
LGB: Based on the LGB Planning Capacity Profile from the FAA,22 "smaller aircraft" use runways
25R/7L and 25L/7R. In addition, an Environmental Impact Report23 showed that runways
16R/34L and 16L/34R are also used for general aviation. According to the EIR the parallel
runways are used evenly (i.e., when 25R/7L and 25L/7R are used, 50% of activity is on the
north runway and 50% is on the south runway. The same holds for runways 16R/34L and
16L/34R). Therefore, treating the parallel runways as runway groups, piston-engine
operations are assigned to a runway group following the minimum degree difference
approach. Within a runway group (i.e. for the identified parallel runway pair), operations
are split evenly.
NRQ: NRQ has 4 sets of parallel runways (8 total runways). For each pair, one runway was
selected as being eligible for piston-engine activity. Therefore, piston engine operations
are assigned to the four non-parallel runways 04L/22R, 09L/27R, 13L/31R, and 18L/36R
following the approach laid out for Scenarios 1 and 2.
ORD: Runway 9L/27R is used primarily for GA aircraft. In addition, runways 14R/32L and 4R/22L
were identified as runways that could serve piston-engine aircraft when operations were
prohibited on Runway 9L/27R during a crosswind. Therefore, piston-engine operations
are assigned to the three non-parallel runways 9L/27R, 14R/32L, and 4R/22L following
the approach laid out for Scenarios 1 and 2.
20	https://www.mitchellairport.com/files/9213/0988/8039/MKEMasterPlanComplete.pdf
21	http://moseslake.airportstudv.com/files/2012/12/MWH.Chl ,DF .6.23.14.pdf
22	https://www.faa.gov/airports/planning capacitv/profiles/media/LGB-Airport-Capacitv-Profile-Appendix-A-
2014.pdf
23	http://lbflving.com/files/ASNreport2009-05.pdf
B-14

-------
LGB
NRQ
ORD
Figure B-10. Multi-runway airports with unique operational profiles. Black lines represent runways not
serving piston-engine aircraft and blue lines represent runways serving piston-engine aircraft
Section B.3.2 Wind Data Sources and Uncertainty
Wind direction data are available from Automated Surface Observing System (ASOS)
meteorological stations. ASOS units are automated sensor suites that are designed to serve
meteorological and aviation needs. These systems report visibility and meteorological data
including wind speed and direction at approximately hourly intervals or when conditions
change rapidly and cross aviation thresholds. Data is available in an online database through
the National Centers for Environmental Information (NCEI).24 At some airports, particularly
those with higher levels of activity, an ASOS station is located on airport property, however, for
other airports the nearest ASOS station may be several kilometers (km), or more, away. Of the
13,153 airports in the national analysis, 6.6% (872) have ASOS stations onsite (<1 km) (Figure B-
11). Among the top 5% of airports by total piston-engine aircraft traffic, 48% have ASOS stations
located within 1 km. In the airport-specific activity analysis presented in Section 3.3, 37% of
airports (10) have an ASOS station onsite.
24 ftp://ftp.ncdc.noaa.gov/pub/data/noaa/
B-15

-------
6000
5000
g. 4000
<
0	3000
01
£ 2000
1000 -
12
10
o
Q.
<
M-
O
100 km
Distance to ASOS Station
(A)

>1 km 1-5 km 5-10 km 10-15 km 15-20 km 20-25 km >25 km
Distance to ASOS Station
(B)
Figure B-ll. Distribution of the number of airports by distance to the closest ASOS station for airports
in the national analysis (A), and airport-specific activity analysis (B). In both analyses, ASOS stations
provided wind direction data that was used to identify the active runway end for piston-engine
aircraft. Results from both analyses have greater uncertainty for airports with a longer distance to an
ASOS station compared to airports closer to an ASOS station.
The relationship between ASOS distance and uncertainty in assignment of aircraft activity to a
specific runway will depend on the distance to the nearest ASOS station, the number of
available runways, and the geographic area in which the airport and ASOS station are located.
For single runway airports, a relatively large shift in wind direction (an average shift of 90
degrees) is required for a shift in the active runway end for piston-engine aircraft, and thus such
B-16

-------
a shift would likely be detected at weather stations at even slightly longer distances (e.g., 10
km). At multi-runway airports where runway headings differ by fewer than 90°, small changes
in predominant wind direction could change the active runway selection from hour to hour in
the minimum-degree distance approach while the airport in practice is more likely to maintain
a consistent operational pattern (Lohr 2008). Alternatively, prevailing wind direction may align
with more than one runway in a given hour. In these case, the minimum-degree distance
approach is likely to underestimate maximum lead concentrations as it predicts fewer
operations at the dominant runway.
In areas with relatively consistent wind direction and minimal topographic perturbations (e.g.,
coastal regions), wind direction is likely to remain constant between an airport and an ASOS
station even when the airport and ASOS stations are separated by some distance. As such, in
these areas ASOS stations that are quite distant from an airport may provide accurate wind
direction data for the purposes of identifying the active runway-end, which increases
confidence in the approach used to estimate activity at the maximum impact area. In areas
where geographic structure (e.g., mountainous regions, river valleys) creates uncertainty in the
applicability of the wind direction data from an ASOS station to a different location, there is less
confidence in the approach used to identify and quantify activity at the maximum impact area.
At airports with less representative ASOS station wind data, model-extrapolated concentrations
may be higher or lower than actual concentrations due to more or less piston-engine aircraft
activity occurring at one runway end, versus others at the airport.
While wind direction is the primary driver for determining an airport's active runway, other
factors may be important including operational restrictions, airport infrastructure location,
runway length, and total airport capacity. For airports with significant operational restrictions
or for multi-runway commercial airports that have designated GA or piston-aircraft runways,
these operational considerations are an additional source of uncertainty.
B.4 Airport-Specific Observed Aircraft
As noted previously, operations conducted by piston-engine aircraft specifically are not
reported. As described in Section 3.3 of the main report, the number and type of aircraft based
at an airport (i.e., based aircraft data) were used to calculate airport-specific activity estimates
for a subset of airports included in the national analysis. In the national analysis, national
average fractions were used to partition activity estimates into piston-engine and non-piston-
engine aircraft, and separately partition piston-engine aircraft activity in single- and multi-
engine (SE and ME). For the airport-specific activity analysis, based aircraft data were selected
as an alternative to the national average fractions for two reasons: 1) based aircraft data were
available, unlike airport-specific counts of piston-engine LTOs, and 2) a comparison of
observations of aircraft activity at six airports and based aircraft at those airports showed that
based aircraft fractions are a reasonable proxy for activity fractions. The data used in that
comparison are provided below.
B-17

-------
Airport-specific information on piston-engine aircraft operations was collected for six airports
from on-site observations and surveys. Further, four of these airports (CRQ, MRI, RHV, and
VNY) had supplemental aircraft count data from noise-specific studies completed within the
past ten years. For each of these six airports, data were collected on the type and operational
characteristics of the observed General Aviation and Air Taxi fleet. Table B-2 shows the
observed aircraft counts at each airport, as well as the comparison between the percentages of
activity attributed to piston-engine aircraft, and then SE versus ME piston-engine aircraft, based
on observational counts or the number of aircraft based at each airport.
Table B-2. Observed piston-engine aircraft from noise studies and onsite observational surveys
Airport
Number
of
Observ.
Days
Number of
Aircraft
Piston-engine
Single-Engine
Multi-Engine
Based
Aircraft
(%)
Observ.
LTOs
(%)
Based
Aircraft
(%)
Observ.
LTOs
(%)
Based
Aircraft
(%)
Observ.
LTOs
(%)
CRQ
14
2,163
63
66
90
93
10
7
MRI
5
827
98
95
95
94
5
6
PAO
7
1,268
98
95
90
99
10
1
RHV
7
2,209
99
88
92
97
8
3
SQL
7
1,018
96
86
90
98
10
2
VNY
30
15,809
59
52
77
95
23
5
Supplemental Sources:(URS 2005, Mead & Hunt 2007, LAWA 2008, LAWA 2011, HMMH 2013)
Table B-3 further summarizes the aircraft classes and operating modes at these six airports.
Aircraft operational activity is broken out by jet and piston-engine aircraft, and piston-engine
activity is further categorized into SE and ME aircraft activity. The share of piston-engine only
activity by operational cycle-type (full LTO, T&G) is also shown.
B-18

-------
Table B-3. Aircraft class and operation mode survey data
Airport

General Aviation and Air Taxi Activity
Piston Engine Aircraft and








Activity Type


AT
AT
GA and
GA and
GA and
GA and
ME
ME
SE
SE T&G

Jet
ME
AT Jet
AT ME
AT SE
AT Heli
LTO
T&G
LTO
(%)

(%)
(%)
(%)
(%)
(%)
(%)
(%)
(%)
(%)

CRQ
94.9
5.1
12.1*
4.4*
65.9*
17.6*
97.4
2.6
93.7
6.3
MRI


2.8
5.4
90.0
1.8
93.8
6.3
84.1
15.9
PAO


5.3
1.0
93.7
0.0
100.0
0.0
92.1
7.9
RHV


11.0
4.0
84.0
1.1
64.1
35.9
71.0
29.0
SQL


9.9
1.9
84.5
3.7
100.0
0.0
95.5
4.5
VNY


44.9
2.8
49.1
3.2
98.9
1.1
88.5
11.5
* These values are for GA aircraft only.
These studies showed that the fraction of SE and ME based aircraft (i.e., sum of SE and ME
based aircraft over total based aircraft) was generally within 10% of observed piston-engine
LTOs (Table B-2), and that use of based aircraft can reveal airport-specific fleet and operational
characteristics (Tables B-2 and B-3). The general agreement between the number of based
aircraft and observed activity data suggest that the fraction of SE and ME based aircraft could
be used to estimate airport-specific piston-engine aircraft activity as a refinement of national
average estimates. Of course, there are inherent uncertainties in based aircraft data, including
the fact that some SE and ME based aircraft may be turboprop or other non-piston-engine
aircraft; however, the comparisons with onsite activity counts suggest based aircraft provide
reasonable, airport-specific data and FAA considers based aircraft data to be a reliable indicator
of activity at small airports (FAA 2015).
B.5 Piston-Engine Rotorcraft
Piston-Engine Rotorcraft Activity at Heliports
Piston-engine rotorcraft operate at airports and heliports, and contribute a growing fraction of
the activity conducted by rotorcraft (FAA 2011). Data on the activity of these aircraft is limited
with activity data available from fewer than 100 of the 5,000 heliports having (FAA 2011). For
the purposes of this report, fixed-wing aircraft are the focus of evaluation; however, methods
similar to those applied here could be applied to estimate lead concentrations at and near
heliports.
Piston-Engine Rotorcraft Activity at Airports
At airports, piston-engine aircraft include both fixed-wing airplanes and rotorcraft (i.e.,
helicopters). While fixed-wing aircraft take-off and land in consistent locations based on wind
direction, rotorcraft may take-off from and land in multiple locations at an airport facility. As
discussed in Section 1, the analysis in this report focuses on lead concentrations at and
downwind of the maximum impact site attributable to fixed-wing piston-engine airplane
B-19

-------
activity at specified run-up locations adjacent to runway ends. Rotorcraft have not been
included in this analysis.
If future analyses were to focus on calculating model-extrapolated concentrations for
rotorcraft, then the scaling factors provided in Table B-4 below could be used with similar
methods to those provided in Section 3 of the main report. For additional details on methods to
develop the scaling factors or the underlying air quality modeling, see Sections 2.1 and 3.1 of
the main report.
Table B-4. Rotorcraft air quality factor data
Operation Mode
Air Quality Factor
(|ig/m3 per LTO)
Distance Description
Rotorcraft All Operational Modes
3.57 x 10"5
Approximately 433 meters north of
haypatch, alongside hangers near terminal
Rotorcraft Climb & Landing Only
6.07 x 10"7
Approximately 18 meters southeast of
haypatch
B-20

-------
References
Carr, E., M. Lee, K. Marin, C. Holder, M. Hoyer, M. Pedde,... J. Touma (2011). Development
and evaluation of an air quality modeling approach to assess near-field impacts of lead
emissions from piston-engine aircraft operating on leaded aviation gasoline. Atmospheric
Environment, 45 (32), 5795-5804. DOI: http://dx.doi.Org/10.1016/i.atmosenv.2011.07.017.
FAA (2011). FAA Aerospace Forecast Fiscal Years 2011 - 2031.
FAA (2015). Evaluating the Formulation of the National Plan of Integrated Airport Systems
(NPIAS). US Department of Transportation.
https://www.faa.eov/airports/planning capacity/npias/media/evaluatioe-formulatioo-npias-
report-to-congress.pdf.
Feinberg, S. and J. Turner (2013). Dispersion Modeling of Lead Emissions from Piston Engine
Aircraft at General Aviation Facilities. Transportation Research Record: Journal of the
Transportation Research Board,(2325), 34-42.
GAMA (2016). 2016 General Aviation Statistical Data book & 2017 Industry Outlook.
Heiken, J., J. Lyons, M. Valdez, N. Matthews, P. Sanford, J. Turner and N. Feinberg (2014).
Quantifying Aircraft Lead Emissions at Airports. ACRP Report 133.
http://www.nap.edu/catalog/22142/quantifying-aircraft-lead-emissions-at-airports.
HMMH (2013). Noise Exposure Map Update: Merrill Field Airport. Municipality of Anchorage.
LAWA (2008). Van Nuys Airport Noisier Aircraft Phaseout Draft Environmental Impact Report
Los Angeles World Airports. September 2008.
http://www.lawa.org/welcome VNY.aspx?id=1076.
LAWA (2011). Van Nuys Airport Updated 14 C.F.R. Part 150 Noise Exposure Maps. Los Angeles
World Airports, http://www.lawa.org/welcome VNY,aspx?id=6645.
Lohr, G. W. and D. M. Williams (2008). Current practices in runway configuration management
(RCM) and arrival/departure runway balancing (ADRB). NASA/TM-2008-215557 NASA.
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/2009001Q329.pdf.
Mead & Hunt (2007). Reid-Hillview Airport Master Plan Update. County of Santa Clara.
URS (2005). McClellan-Palomar Airport FAR Part 150 Study Update. 1 & 2. McClellan-Palomar
Airport. San Diego, CA.
B-21

-------
Appenc	;ertainty Characterization
Section 4 of the report presents a quantitative and qualitative characterization of variability and
uncertainty in lead concentration at and downwind of the maximum impact site at airports
nationwide. This analysis focused on parameters identified in previous sections as particularly
influential on the atmospheric concentration of lead from piston-aircraft operations. This
appendix provides details on the data underlying this uncertainty analysis. Section C.l describes
the distribution of avgas lead concentrations used in the Monte Carlo analysis. Section C.2
describes the data underlying the evaluation of variation in run-up time on atmospheric lead
concentration, as a function of downwind distance from the maximum impact site, which was
also evaluated in the Monte Carlo analysis.
C.l Avgas L mcentrations
As described in Section 4.3 of the report, the method for characterizing piston-engine aircraft
attributable atmospheric lead concentration as a function of avgas lead concentration utilized a
Monte Carlo analysis by treating avgas lead concentration as a bounded stochastic parameter
based on ASTM standards for 100LL (1.70 - 2.12 g/gallon). Fuel samples may be spatially and
temporally heterogeneous. EPA and FAA analyzed separate samples of avgas for lead. In total
118 samples were tested (2 samples were removed from further analysis due to likely data
transcription errors). While these concentration data present the range of lead concentrations
in individual samples, the range in average fuel lead concentration of total fuel consumed at an
individual airport over a three-month period is likely more constrained. The average avgas lead
concentration at a given airport over 3-months, the period over which the AQFs were
developed, will likely represent an average of several fuel batches due to aircraft fueling at
other airports and multiple fuel deliveries to the airport. By the central limit theorem, there is
an expectation that the distribution of mean three-month fuel concentrations will approach a
normal distribution. Thus, in the absence of spatial or temporal data on fuel lead concentration
and given the sample size of n=116, a normal distribution was fit to the 116 fuel lead samples
shown below in Figure C-l.
C-l

-------
Distribution of Avgas Lead Concentrations
35
cy O' o' v v v v v v 
-------
report given the relatively small volume of usage compared to 100LL (i.e., all operations are
modeled as using 100LL).
C.2 Run- ne-in~Mode Distributions and Their Relationship to Atmospheric Lead
Concentrations at and Downwind from the Maximum Impact Site
The AQFs presented in Section 3 of the report show the relationship between concentrations at
the maximum impact site and at downwind locations given an operation cycle type (full
LTO/T&G) and aircraft class (SE/ME). While the concentration gradients for each of the AQFs
generally decrease monotonically with distance, two factors influence this relationship, namely:
spacing of model receptors, and time-in-mode data. The AERMOD receptor grid spacing was 50
m for 1km up- and downwind of the airport along the axis of the runway, as shown in Appendix
A. The AQFs are calculated using the nearest receptor location to the respective AQF distance.
ME and SE operations have different default Times-ln-Modes (TIM) based on data collected at
the model airport. The TIM data will influence the timing and location of emissions. For
instance, T&G operations have different spatial and temporal patterns than full LTOs, and do
not include certain operational modes like run-up.
First, this section describes the distribution or run-up times observed at General Aviation
airports. These observations inform the distribution of run-up times used in the Monte Carlo
uncertainty analysis in the main report. Next, this section briefly describes the relationship
between run-up time and atmospheric lead concentration at the model airport. Finally, this
section describes the modeled relationship between run-up time and the maximum 3-month
concentration at and downwind of the maximum impact site as applied in the Monte Carlo
uncertainty analysis.
C.2.1 Run-Up Time-in-Mode Distributions
Time spent in run-up mode will vary from pilot to pilot as a function of personal preference,
aircraft design, and training. Further, the distribution of run-up time-in-mode across all
operations may vary from airport to airport dependent upon the fraction of pilots in training at
that airport, airport run-up regulations, and local characteristics such as seasonal changes.
Information on average run-up times was collected from studies that observed run-up
operations (Carr et al. 2011, Heiken et al. 2014) and observations made at the model airport as
described in this report (Section 2 of the main report and Appendix A). Six sets of run-up
observations are represented across the three studies. One airport was surveyed separately by
both Heiken et al. (2014) and Carr et al. (2011) while run-up distributions were separately
surveyed for single-engine and multi-engine aircraft at the model airport in this report. The
resulting 6 distributions (representing survey data from 4 unique airports) are shown in Figure
C-2. More information on the surveying methods are available in Carr et al. 2011, Heiken et al.
2014, and Appendix A of this report. While the study designs, study durations, survey methods,
and quality assurance approaches vary across the three studies, 40 to approximately 100
observations were collected across peak and off-peak hours and multiple days at each airport.
For all airports where full data distributions were available, run-up duration observations
C-3

-------
showed consistent right-skewed distributions with median values between 49 and 73 seconds
and means 5% to 30% higher than their respective medians.
600 t
500
~ 400

i 200
DC
100 -t
0
Run-Up Time-ln-Mode
-Min —5th ~Median —95th -Max XMean
RVS (1)
X
APA (1)
SMO (1)
SMO (2)
RHV [SE]
(3)
RHV [ME]
(3)
Figure C-2 Distribution of observed run-up duration observations across different studies. (1) Heiken et
at. 2014, (2) Carr et ai. 2011, (3) this report. The survey of run-up duration for this report separately
characterized run-up for single-engine (SE) and multi-engine (ME) aircraft.
C.2.2 Relationship Between Run-Up Time and Atmospheric Lead Concentration
As discussed in the main report, previous analyses have identified that atmospheric lead
concentrations are sensitive to run-up operation characteristics (Carr et al. 2011, Feinberg and
Turner 2013, Heiken et al. 2014). For example, Feinberg and Turner (2013) found run-up
emissions to be the single largest contributor to ground-level lead concentrations, while only
accounting for about 11% of airport lead emissions. Their modeling found that changing the
emissions attributable to run-up from 3% of modeled emissions to 5% of modeled emissions
resulted in a 34% increase in annual atmospheric lead concentrations.
A sensitivity analysis was conducted at the model airport as described in Section 2 of the main
report. The sensitivity analysis examined the influence of run-up duration on 3-month average
lead concentrations at and downwind of the maximum impact site for one 3-month period
(January-March). The analysis was run for three run-up durations for each aircraft class: 16, 40,
and 121 seconds for SE aircraft and 16, 63, and 160 seconds for ME aircraft, which correspond
to the 5th percentile, median, and 95th percentiles of SE and ME run-up times observed at the
model airport, respectively. The concentration from only run-up emissions at the maximum
impact site receptor was 0.034 |-ig/m3 for the 5th percentile, 0.257 |-ig/m3 for the 95th percentile,
and 0.092 |-ig/m3 for the default run-up duration. Lead concentrations attributable to run-up
emissions alone exceeded the urban background lead concentration at a downwind distance of
up to 275 m using the 95th percentile run-up duration and 75 m using the 5th percentile run-up
duration. January-March 3-month average lead concentrations from run-up emissions alone are
C-4

-------
shown in Figures C-3 (5th percentile run-up duration) C-4 (default run-up duration) and C-5 (95th
percentile run-up duration).
Concentration (uQ/m3)
0.03 - 0.0525
0.0075 - 0.03
0.0015 - 0.0075
0 6 Kilometer
DETAILS:
Single- and Multi-Engine Fixed-Wing Sources
Run-up Activity Only
5th Percentile Run-up Times-in-mode
Runways 31R and 13L, Both Takeoff Directions
(Includes All Overnight Hours)
January-March Average
Figure C-3 January-March 3-month average lead air concentrations (jjg/m3) at the model airport from
run-up mode alone using the 5th percentile run-up duration.
0 6 Kilometer
Single- and Multi-Engine Fixed-Wing Sources
Run-up Activity Only
Default Run-up Times-in-mode
Runways 31R and 13L, Both Takeoff Directions
(Includes All Overnight Hours)
January-March Average
Figure C-4 January-March 3-month average lead air concentrations (ng/m3) at the model airport from
run-up mode alone using the default run-up duration.
C-5

-------
Figure C-5 January-March 3-month average lead air concentrations (ng/m3) at the model airport from
run-up mode alone using the 95th percentile run-up duration.
C.2.3 Characterization of the relationship between run-up time-in-mode and atmospheric
concentration as a function of distance from the maximum impact site
In order to quantitatively evaluate the variability in lead concentrations with run-up time for SE
and ME aircraft at each receptor site (described in Section 4 of the report), individual
relationships were developed as described here. The impact of run up-time on lead
concentration was characterized at the model airport by modeling atmospheric lead
concentrations from piston-engine aircraft and varying the run-up time while holding all other
parameters constant.2 SE and ME aircraft were modeled separately as SE and ME aircraft have
differing emission rates and different time-in-mode distributions for run-up, as shown in Figure
C-l. The analysis was run using the 5th percentile, median, and 95th percentile run-up TIMs
observed at the model airport. The resulting maximum 3-month average lead concentrations
are shown in Figure C-6. The rows of Figure C-6 correspond to SE and ME aircraft results
respectively, and the columns present results at the maximum impact site and 400m downwind
from the maximum impact site respectively.
2 The characterization of lead concentrations at the model airport and the associated parameter assumptions are
highlighted in Section 2 of the report.
0 6 Kilometer
DETAILS:
Single- and Multi-Engine Fixed-Wing Sources
Run-up Activity Only
95th Percentile Run-up Times-in-mode
Runways 31R and 13L, Both Takeoff Directions
(Includes All Overnight Hours)
January-March Average
C-6

-------
Maximum Impact Site
_Q

0.350
Q_
m

_C
¦4—»
j=
0.300
c
o
CuO
zi.
0.250



m
C
O
0.200
£
¦4—1


05
L_
0.150
E
x
+->
c
CD
0.100
ro
U


c
o
0.050
LU
u

CO

0.000
.a 0.040 T
CL	^
.c	"I- 0.035 --
c	<
0	60 0.030 --
:>	3
"T	V 0-025 --
m	^
E	~ 0.020 --
^	£ 0.015 --
1	g 0.010 --
^	O 0.005 --
LU	(J
^ 0.000 —
50	100
Run-Up Time (s)
(c)
Maximum Impact Site
o
—i-
50
+
-+-
100 150
Run-Up Time (s)
150
200
c
o
m
E
3
E
'x
tc
2.00E-03 --
c
1.50E-03 +
ro
c 1.00E-03 +
(L)
c 5.00E-04 +
o
u
0.00E+00
_q	3.50E-04 T
o_	^
£	"g 3.00E-04 --
°	2.50E-04 --
rn	g 2.00E-04 --
I	ro 1.50E-04 --
E	₯
¦x	£ 1.00E-04 --
CD	(J
^	§ 5.00E-05 --
LU	(J
0.00E+00
(b)
400m Downwind
-a 3.00E-03 t
2.50E-03 --
50	100
Run-Up Time (s)
(d)
400m Downwind
150
	1	1
100	200
Run-Up Time (s)
Figure C-6 Relationship between run-up duration and lead concentration at the model airport
The relationship between run-up time and resulting concentration is linear at both the
maximum impact site and at each downwind location.3 Unique linear equations were derived
for each concentration site for both SE and ME operations, where the slope represents the
sensitivity of the total concentration to run-up duration. Consistent with a decreasing
concentration gradient downwind of the maximum concentration site, the slopes of the linear
relationships between run-up time and concentration decrease from one site to a site further
downwind. The equations for the maximum impact site and each of the downwind locations
are given in Table C-l where the independent variable x is run-up time and the dependent
variable y is the resulting maximum 3-month average concentration.
3 Some of the SE sites demonstrated a linear relationship between run-up time and resulting concentration with a
negative intercept. Further, some of the ME sites, while approximately linear between 16 and 163 seconds,
suggest a relationship that may also be characterized as logarithmic. These results indicate that, while
characterizing the relationship between run-up and resulting concentration as linear may be appropriate for the
modeled run-up times, further work is necessary to extrapolate these results far beyond these modeled run-up
times.
C-l

-------
Table C-l Relationship between run-up time-in-mode and maximum 3-month average lead
concentration as a function of distance from the maximum impact site.
Site
Single-Engine Relationship
Multi-Engine Relationship
Maximum Impact
Site
y = (2.66x - 0.0290) X 10-3
y = (1.99x + 42.8) x 10~4
50m Downwind
y = (4.30x + 0.0081) x 10-4
y = (3.21x + 68.6) x 10-5
100m Downwind
y = (1.42x + 0.0115) x 10-4
y = (1.05x + 22.5) x 10-5
150m Downwind
y = (9.09x — 0.0672) X 10-5
y = (6.87x + 148) x 10~6
200m Downwind
y = (6.78x + 0.0120) X 10-5
y = (4.94x + 105) x 10~6
250m Downwind
y = (5.19x + 0.0250) x 10-5
y = (3.89x + 82.8) X 10-6
300m Downwind
y = (3.31x — 0.0039) X 10-5
y = (2.46x + 51.9) x 10-6
400m Downwind
y = (2.18x — 0.0132) x 10-5
y = (1.65x + 35.2) x 10-6
500m Downwind
y = (1.41x + 0.0016) x 10-5
y = (1.07x + 23.0) x 10-6
The relationship between run-up time [ x ] and resulting concentration [ y ] is linear for a
relevant range of run-up times, and for a known run-up time [ xinitiai ], the concentration at each
downwind location is characterized at each airport [ y(xinitiai)] as presented in Section 4 of the
report. Thus, the percentage change in lead concentration from a change in run-up time can be
calculated as:
o/oDiff =
(y(x) y(Xmedian))
y(Xmedian)
(Equation C-l)
Equation C-l can be used to vary resulting concentrations as a function of the aircraft run-up
time-in-mode and allows for the characterization of atmospheric concentrations while varying
additional parameters such as fuel lead concentration. Therefore, Equation C-l is used to
understand the uncertainty and potential variability of lead concentrations in the Monte Carlo
analysis presented in Section 4 of the report.
C-8

-------
References
Carr, E., M. Lee, K. Marin, C. Holder, M. Hoyer, M. Pedde, ... J. Touma (2011). Development and
evaluation of an air quality modeling approach to assess near-field impacts of lead emissions from
piston-engine aircraft operating on leaded aviation gasoline. Atmospheric Environment, 45 (32), 5795-
5804. DOI: http://dx.doi.Org/10.1016/i.atmosenv.2011.07.017.
FAA (2015). General Aviation and Part 135 Activity Surveys - CY 2015. F. A. Administration.
Feinberg, S. and J. Turner (2013). Dispersion Modeling of Lead Emissions from Piston Engine Aircraft at
General Aviation Facilities. Transportation Research Record: Journal of the Transportation Research
Board,{2325), 34-42.
Heiken, J., J. Lyons, M. Valdez, N. Matthews, P. Sanford, J. Turner and N. Feinberg (2014). Quantifying
Aircraft Lead Emissions at Airports. ACRP Report 133. http://www.nap.edu/catalog/22142/quantifving-
aircraft-lead-emissions-at-airports.
C-9

-------