Quantifying Aircraft Lead Emissions at Airports (2015)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5. AIRBORNE LEAD AND AIRCRAFT ACTIVITY DATA COLLECTION AT AIRPORTS During this phase of the project, field studies were conducted at three airports to generate data sets of airborne PM-Pb concentrations to evaluate the emission inventory methodology through dispersion modeling. Detailed aircraft activity data were collected to allow for the development of spatially and temporally resolved emission inventories for each of the three airports. While such fine-grained activity data are not routinely available, this approach was needed for a robust evaluation of emission inventory methodology and sensitivity studies to determine the data collection elements that are most important for an accurate inventory. In order to select the three airports at which field studies were to be performed, general aviation airports nationwide were systematically evaluated for consideration as a field study site. Desired attributes were identified and criteria were developed to screen and ultimately rank candidate airports. Desired attributes included, but were not limited to, a large Pb emissions load (based on the 2008 NEI), a large share of non-carrier operations and specifically a large share of piston-engine aircraft activity, and favorable meteorology (high wind direction persistence with few calms). A prioritized list of airports was generated and airport operators were contacted to determine their willingness to participate. Listed below are the three airports selected for field studies and the dates the studies were performed. • Richard Lloyd Jones Jr. (RVS), Tulsa, OK; March 27 to April 28, 2013 • Centennial Airport (APA), Englewood, CO; May 15, 2013 to June 10, 2013 • Santa Monica Airport (SMO), Santa Monica, CA; July 3, 2013 to July 30, 2013 These airports have distinctive characteristics. RVS and APA are among the busiest general aviation airports nationwide and have relatively large footprints with multiple runways. However, the spatial distribution of run-up and LTO activity patterns are quite different because of the runway layouts, and wind directions were more variable at APA than RVS. SMO is a much smaller airport but with concentrated run-up and LTO activity patterns and a history of being the subject of special PM-Pb studies. -45-

5.1 Data Collection Overview As indicated above, the field studies—each nominally one month in duration—were conducted at each of the three airports by staff from Washington University in St. Louis (WUSTL). A consistent data collection strategy was used across the three field studies, and the data collection plan was reviewed by the Project Panel prior to the first deployment. In this section, generic features of the data collection are summarized; this is followed by a summary for each airport that includes the airport-specific features and select results. 5.1.1 Aircraft Activity Data Collection Detailed aircraft activity data were collected to inform the development of a spatially and temporally resolved emissions inventory for the model-to-monitor comparisons. These data have been processed and compiled into databases (e.g., MS Excel spreadsheets). The key data collection elements are summarized below. • Landing and Takeoff Operations (LTOs) – Daily PM-Pb sampling was conducted during the 12-hour daytime period with highest aircraft activity. Video cameras were used to continuously record LTOs during each PM-Pb sampling event. The videos were played back to document takeoff, landing, and touch-and-go operations by runway at 10-minute intervals and these data were rolled up to one- hour periods. LTO data were collected for all fixed-wing aircraft at each airport and at SMO the piston-engine aircraft fraction was also directly measured. At RVS and APA, the piston-engine aircraft fraction was not directly measured from the video data because aircraft in the video images were often too small to be conclusively identified as either piston engine or jet. • Aircraft Fleet Inventory – LTOs were photographed for 30 hours at each airport. The data collection schedule was generated using a quasi-random process to populate a 2D matrix with dimensions of time of day and day of week (Weekdays / Saturdays / Sundays). The matrix was weighted towards data collection during hours with higher activity and to ensure adequate data collection on weekends. Photographs were reviewed to develop a time-stamped inventory of LTO activities by tail ID. The FAA Registry (http://registry.faa.gov/aircraftinquiry/) was used to identify the aircraft and engine characteristics for each recorded tail ID. Data were collected for all aircraft, not just piston-engine aircraft, to provide information about the distribution of activities between piston-engine airplanes and jets. Some aircraft were observed multiple times over the 30 hours of data collection. Given the objective to inventory the fleet from an operations perspective, each observation was an independent entry into the database. Each database record includes the observation time stamp; aircraft type, manufacturer, model, year, and number of engines; engine type, manufacturer, model, and horsepower; and number of times the aircraft was identified in the one-hour observation period and in the overall data set. Tail ID numbers are decoupled from the final database. -46-

• Time in Mode for Run-up – Run-up operations were manually observed for 15 hours at each airport. Data collection was scheduled to capture a range of conditions (time of day, day of week) and included the time aircraft spent in a run-up area (visual observation), the duration of the magneto test (audible changes in engine noise during run-up), and the aircraft tail ID. Some planes bypassed the run-up area prior to takeoff and such instances were recorded. In some cases, the magneto test duration could not be determined because of confounding sources of noise. Each record in the database includes the data collection hour, total run-up time, magneto test time, and the aircraft attributes listed above for the aircraft fleet inventory. Tail ID numbers are decoupled from the final database. • Time in Mode for Other Activities – Additional piston-engine aircraft activities such as taxiing, takeoffs, and landings were manually observed for 15 hours at each airport. Data collection was scheduled to capture a range of conditions (time of day, day of week). Observation points were chosen to maximize viewing of the entire airport footprint. Activities were tracked by aircraft and recorded by runway or taxiway. For example, a taxi-back would consist of the following data: landing time (time on runway between wheels down and turning onto taxiway); time taxiing and idling on each taxiway; and takeoff time (time on runway between starting rollout and wheels-up). Approach and climb-out times could not be adequately captured because of the difficulty in establishing aloft locations for the start of approach and end of climb-out. Instead, wheels-up and wheels-down locations on the runways were recorded to inform the development of TIM estimates for climb-out and approach and to spatially allocate runway emissions. TIM for touch-and-go operations was recorded as the time between wheels down for the landing portion and wheels-up for the takeoff portion. Each record in the database includes a plane identifier (arbitrary), activity (e.g., landing, takeoff, taxiing, idling), and location (e.g., runway ID, taxiway ID). Activity data processing was conducted in coordination with the Sierra Research and KB Environmental Services staff. TIM data were processed by the WUSTL field operator (Neil Feinberg) with QA/QC performed by the WUSTL lead investigator (Jay Turner). Most of the LTO video and fleet inventory photographs were processed by other WUSTL staff and in these cases initial QA/QC was performed by the WUSTL field operator with additional QA/QC by the WUSTL lead investigator. 5.1.2 Airborne PM-Pb Data Collection Airborne PM samples were collected daily and analyzed for Pb. At each airport, four PM sampling sites were selected based on the location of piston-engine aircraft activities, historical winds data, and Pb concentration fields generated from preliminary dispersion modeling. PM-Pb hot spots were predicted downwind of run-up areas and such locations were given high priority. Relatively flat terrain was desired, and it was necessary to stay clear of FAA-restricted areas; for SMO, the siting of samplers in previous studies was also considered. At each airport, the four sampling sites included two “primary” sites and two “secondary” sites. The primary sites were a location downwind of a run-up/takeoff area for prevailing winds, and a location chosen to capture background PM-Pb levels for -47-

prevailing winds. Characteristics for the two secondary sites are described in the airport-specific summaries. Up to four PM samplers were operated during each sampling event. A PM2.5 sampler was always operated at each of the two primary sites (except for collocated TSP data collection events to establish TSP-Pb measurement precision) and the remaining two PM samplers were used in one of the three configurations: (i) collocated PM2.5 sampling at the primary sites (to establish PM2.5-Pb measurement precision); (ii) TSP sampling at the primary sites; or (iii) PM2.5 sampling at the secondary sites. PM samples were collected using Model PQ100 portable samplers (BGI, Waltham, MA). The PQ100 is an EPA Federal Reference Method (FRM) for PM10 sampling; for this study, the samplers were used with BGI Very Sharp Cut Cyclones (VSCC) to achieve PM2.5 cutpoints. A louvered inlet with PM10 impactor—the standard configuration for ambient PM10 sampling—was used upstream of the PM2.5 cyclone. TSP samples were collected using PQ100 samplers with BGI TSP inlets. The TSP inlets have been previously characterized (Kenny et al. 2005). The design flow rate is 16.7 liters per minute (LPM) for these various inlets. Figure 7 shows the BGI samplers deployed at the APA Central monitoring site with runway 17L/35R in the background. Figure 7 BGI PQ100 PM samplers for PM2.5 (PM10 inlet followed by a PM2.5 cyclone) (left) and TSP inlet (right) -48-

Twelve-hour integrated PM samples were collected each day. These sampling events were conducted during the 12-hour period of highest piston-engine aircraft activity based on discussions with the airport authorities. This approach was preferred over 24-hour integrated sampling for several reasons. Piston-engine aircraft activity is very low at night and thus the additional 12 hours of sampling would increase the relative contribution from background Pb to the time average concentration. The 24-hour time window for sampling also increases the likelihood of wind direction variability. This is not a hard constraint for the modeling, but persistent winds do simplify the data interpretation. Finally, calm winds are more frequently observed at night and these periods are more difficult to model. PM sampling and chemical analysis protocols are described in detail in Appendix B and are summarized as follows. PM samplers were mounted on wood platforms. Filter holders containing the Teflon® filter media were installed in the samplers each morning immediately prior to the start of sampling and retrieved each evening immediately following the end of sampling. While Pb is nonvolatile, bromine (Br) is also of interest and it is relatively volatile so cold transport and storage was adopted. Samples were transported to and from the field sites in coolers with ice packs and were stored in a freezer after sampling. For each airport study, a subset of samples was analyzed by X-Ray Fluorescence (XRF) at Cooper Environmental Services (CES, Beaverton, OR) to obtain data for a range of elements. XRF data were reported as areal densities (e.g., ng/cm2 filter) and converted to ambient concentrations using the filter effective cross- sectional area and the ambient air volume sampled. All samples—including those analyzed by XRF, which is a non-destructive method—were digested and analyzed for Pb at WUSTL using Inductively Coupled Plasma – Mass Spectrometry (ICP-MS). Two sequential digestions were performed using a hot-block at 90 °C with nitric acid and hydrofluoric acid for the first digestion and boric acid for the second digestion. Digestion solutions were diluted to a known volume and filtered to remove any remaining particulate matter. ICP-MS data were reported as concentrations in the diluted digestion solutions and converted to ambient concentrations using the diluted digestion solution volume and the ambient air volume sampled. All samples were analyzed for Pb isotopes as well as total Pb. ICP-MS analysis for Pb isotopes was not included in the original study design but was added to strengthen the connection between airborne Pb and piston-engine aircraft emissions. The isotopic composition of Pb used to make the avgas additive tetraethlylead (TEL) is distinct from the isotopic composition of native soils at these airports. Thus, isotopic composition can be used to discriminate the origins of Pb in the airborne PM samples. Pb isotopes are stable and therefore cannot be used to distinguish PM-Pb in freshly emitted exhaust from exhaust PM-Pb that has locally deposited over the years and is resuspended by wind or aircraft-induced turbulence during the PM sampling events. The isotopes data were also used to screen PM-Pb samples for contamination. Appendix B presents the analytical protocol and use of the isotopic composition for data validation. Data collection objectives included 20 sampling events at each airport with 85% data completeness (17 events) for valid PM2.5-Pb data at both of the primary sites. Additional data collection objectives were nine events with PM2.5 and TSP sampling at the primary sites and nine events with PM2.5 sampling at the primary and secondary sites. Table 21 -49-

Table 21 PM Data Collection Summary Parameter Objective RVS APA SMO PM Sampling Events1 20 31 25 25 PM2.5-Pb (two primary sites) 17 31 25 24 PM2.5-Pb and TSP-Pb (two primary sites) 9 9 9 9 PM2.5-Pb (two primary and two secondary sites) 9 8 9 4 Note: PM-Pb measures include both valid data collection and valid chemical analysis for Pb content. 1 Excludes collocated TSP sampling events which, by design, do not include PM2.5 sample collection. summarizes the data completeness from the perspective of not only valid sample collection but also valid chemical analysis for Pb. Additional details are provided in the airport-specific summaries. For each airport, the number of attempted PM sampling events and valid PM2.5-Pb data collection at the primary sites far exceeded the objectives. PM2.5-Pb and TSP-Pb data collection at the primary sites met the objective of nine events per airport. The four-site PM2.5-Pb data collection objective of nine events was met at APA; however, RVS included only eight events and SMO only four. Low capture at SMO was from a failed sampler that needed to be returned for repair, which resulted in nine valid samples at one secondary site and five valid samples at the other secondary site. As discussed in the SMO case study summary, this data collection shortcoming did not compromise the data analysis and interpretation. PM-Pb quality assurance data collection included field blanks and collocated sampling for PM2.5 and TSP. Eight PM2.5 and four TSP field blanks were collected at each airport by placing filters in the samplers overnight (nominally 12 hours) between scheduled sampling events. A one-way nonparametric analysis of variance test demonstrated the field blanks distributions for each airport were statistically indistinguishable (95% confidence level) and thus the field blanks data were pooled across the airports. Effective ambient concentrations were calculated using the target air volume of 12 m3 drawn through a 16.7 LPM sampler during a 12-hour sampling event. PM2.5-Pb median and 90th percentile field blank concentrations were 0.1 ng/m3and 0.5 ng/m3, respectively (N = 48). For TSP-Pb the median field blank concentration was 0.4 ng/m3 (N = 12). The 90th percentile TSP-Pb field blank was 1.8 ng/m3 but extremes of a distribution, such as the 90th percentile, might be non-representative for small sample sizes such as the 12 samples in this case. The higher field blanks value for TSP compared to PM2.5 is consistent with windblown dust intrusion into the sampler in the absence of air sampling. Median field blank Pb levels are similar to the analytical MDL of 0.2 ng/m3 and thus airborne PM-Pb concentration values were not corrected using the field blanks data. PM-Pb measurement precision was evaluated by collocating samplers with matched inlets (i.e., operating two matched samplers side-by-side). These measurements were conducted at both of the primary sites at each airport. PM-Pb collocated data are presented in the data tables for each airport summary (see Sections 5.2, 5.3, and 5.4). -50-

Thirty-two PM2.5 sample pairs were collected: 16 at RVS, 4 at APA, and 12 at SMO. Ten TSP sample pairs were collected: four at APA and six at SMO. These data were pooled across the airports and the collocated precision was calculated as the root mean square difference over all sample pairs divided by √2. Precision estimates are presented in Table 22. Measurement precision for ambient PM sampling typically has two components—an additive (absolute) contribution that dominates at low concentrations approaching the detection limit, and a proportional (relative) contribution that dominates at high concentrations. The concentration dependence of precision was evaluated by splitting the PM2.5 data into three groups using tertile concentrations and splitting the TSP data into two groups using the median concentration. Table 22 shows that at low PM-Pb concentrations the absolute precision is 0.35 ng/m3 for PM2.5 and 0.59 ng/m3 for TSP; at high concentrations, the relative precision is 12% for PM2.5 and 5% for TSP (bolded values in Table 22). Precision estimates for TSP might be influenced by the small sample sizes after stratifying the data into two groups (N = 5 for each group). Table 22 PM-Pb Measurement Precision from Collocated Sampling Parameter Sample Pairs Mean Pb (ng/m3) Collocated Precision Absolute (ng/m3) Relative1 (%) PM2.5-Pb - All Data - Bottom 1/3 Concentrations - Middle 1/3 Concentrations - Top 1/3 Concentrations 29a 9 10 10 10.7 0.83 4.8 25.7 1.8 0.35 0.85 3.0 17 43 18 12 TSP-Pb - All Data - Bottom 1/2 Concentrations - Top 1/2 Concentrations 10 5 5 15.6 1.8 29.5 1.2 0.59 1.6 8 37 5 1 Relative precision is the absolute collocated precision divided by the pooled mean concentration. a The 7/27/13 SMO Northeast site sample pair was excluded because the concentration difference is an extreme value that exerts high influence on the precision estimates. Additional analysis, presented in Appendix B, demonstrates the PM2.5-Pb relative precision of 12% at high concentrations is a stable estimate. Assuming the additive and proportional contributions to measurement error are independent, sample-specific uncertainties can be estimated by adding the precision contributions in quadrature, i.e. 2 5.25.2 015.012.0 PbPM CPbPM −×+=−σ -51-

where C is the measured PM2.5-Pb concentration and σ is its uncertainty, both in units of ng/m3. PM2.5-Pb measurement precision of ~0.35 ng/m3 at low concentrations and 12% at high concentrations demonstrate high data quality for use in the model evaluation. 5.1.3 Avgas Data Collection Avgas dispensed by all FBOs at three airports is 100LL grade, which has a maximum Pb content of 2.12 g/gal (0.56 g/L). The actual Pb content in 100LL can be considerably lower, however, and thus avgas samples were collected at each airport and analyzed for Pb content. Avgas samples were collected from FBOs at RVS and APA. At SMO, however, the FBOs were unwilling to provide avgas samples for this study; therefore, samples were collected from two privately owned, SMO-based piston-engine aircraft. In general, avgas samples were collected from FBOs within days after new fuel deliveries; however, some samples were obtained from FBOs with low avgas sales volumes, resulting in samples drawn as long as 10 months after the most recent delivery. A total of 15 avgas samples were collected and shipped to Intertek Caleb Brett for Pb content analysis using test method ASTM D5059. Airport-specific results are presented in Sections 5.2, 5.3, and 5.4. Pooling over the three airports, mean and median Pb concentrations in avgas were 1.56 g/gal and 1.33 g/gal, respectively, with a maximum Pb content of 2.12 g/gal. The avgas samples were analyzed by ICP-MS for Pb isotopes for comparison to the airborne PM-Pb data. 5.1.4 Soil Data Collection Bulk soil samples were collected near each of the four PM sampling sites at each airport. A portion of each sample was resuspended in a chamber with PM2.5 collected onto filters. These filter samples were digested and analyzed by ICP-MS for Pb mass fraction in the resuspended soil and also Pb isotopes. The remaining portion of each bulk soil sample will be archived at WUSTL until at least January 1, 2016. 5.1.5 Meteorological Data Collection Automated Surface Observing System (ASOS) data and Integrated Surface Hourly (ISH) data for each airport have been obtained from the National Climatic Data Center. In addition, a 3D sonic anemometer (Model 81000, RM Young Co., Traverse City, MI) was deployed starting midway through the first field study (RVS) and throughout the remaining two studies. 3D wind speeds and temperature at 3 m height were logged at 10 Hz. These data can be used to calculate horizontal wind direction and speed in addition to the vertical wind speed. To further enhance the ASOS routine meteorology data collection, a portable weather station (PortLog, Rainwise Inc., Bar Harbor, ME) was deployed for each study period. The following parameters were recorded as one-minute averages: wind speed and direction at nominally 2 m height; temperature; relative humidity; barometric pressure; rainfall; and solar radiation. While the ASOS 10 m surface winds data are used for all modeling, the additional meteorology data are used to further characterize environmental conditions during sampling. -52-

5.2 RVS Field Study Figure 8 shows the RVS airport layout. There are three runways: the two north/south runways are used predominantly, while the east/west runway is rarely used. Prevailing winds are from the south. Figure 8 Airport Diagram and PM Sampling and Activity Data Collection Locations Deployed at RVS Data Collection Locations Note: PM sampling was conducted at the North (N), East (E), South (S), and West (W) sites; video cameras were deployed at the VC1 and VC2 sites; and other activity data were manually collected at the P1, P2, and TCC sites. The field study was conducted from March 27 to April 28, 2013. March 27 was a shakedown day for PM measurements and these data are valid for precision estimates but, for reasons described in Section 5.2.2, should not be used for the model-to-monitor comparison. A video camera to continuously record LTOs was deployed starting on March 27. However, for reasons described in Section 5.2.1, this camera did not adequately capture LTO operations. A second camera was deployed starting on April 4 and the data collected thereafter are most suitable for the model-to-monitor comparison, although the data collected prior to April 4 can be used with additional assumptions about -53-

LTO activity. PM sampling and aircraft activity data were collected from 8 AM to 8 PM CDT. 5.2.1 Aircraft Activity Data Collection Aircraft activity data collection at RVS is summarized in Table 23; data collection locations are shown in Figure 8. Video cameras were continuously operated during each 12-hour PM sampling event to record LTOs. Initially one camera was deployed at the VC1 location (Figure 8) to capture LTOs for aircraft takeoffs and landings at the north end of both north-south runways, which are the operating conditions for prevailing southerly winds. However, VC1 could not capture the takeoff portion of touch-and-go operations and only partially captured landings that originated at the south end of the north-south runways, which are the operating conditions for northerly winds. Periods with northerly winds were more frequent than anticipated. Starting on April 4, a video camera was also deployed at the VC2 location (Figure 8) to resolve these data collection issues and LTO operations are reported in the database only for the period from April 4 through April 28. Figure 9 shows the hourly distribution of total operations for all aircraft (not just piston- engine aircraft) as determined from the video camera data. Touch-and-go activities are counted as two operations each and are distinguished from normal takeoffs and landings. Over the study period there were, on average, 19 operations per hour. Total operations peaked between 10 AM and 12 PM, with the lowest levels of activity in the early morning. Operations were nearly evenly distributed between the two north/south oriented runways with 46% on 1R/19L and 49% on 1L/19R. East/west runway (13/31) and helicopter activity were 3.6% and 1.5% of documented operations, respectively. However, the video cameras are not ideal for capturing helicopter activity since they have a different spatial extent of operation. Operations on the north/south runways were evenly split between those originated at the north end (19L/19R) and south end (1L/1R) at 50% each. Thirty hours of LTOs were photographed from the locations marked as P1 and P2 in Figure 8. The photographs were reviewed to identify tail numbers, which were matched to aircraft and engine specifications in the FAA Registry. The resulting fleet inventory database includes a record for each operation but with the tail numbers removed. Over the 30 hours of observation, 171 unique aircraft were identified. Nine aircraft (5%) accounted for one-third of the operations and 20 aircraft (12%) accounted for half of the operations (19 fixed-wing single-engine aircraft, and one fixed-wing multi-engine aircraft). Table 24 summarizes the distribution of LTOs by aircraft type; more than three-fourths of the operations were single-engine piston aircraft. TIM data were manually collected. Piston-engine aircraft run-up activities were observed for 15 hours and included 109 run-up operations, with magneto test duration recorded for 76 of these operations. Missing magneto test data primarily resulted from confounding sources of noise. Tail numbers were recorded for 95% of the run-up operations. Twenty- eight planes bypassed the run-up area and did not perform any observed run-ups. -54-

Table 23 RVS Aircraft Activity Data Collection at RVS Date Activity Data Collection VC1 VC2 Tail ID TIM Run-up Comments 03/27/2013 P - 2 0 0 VC1 ran 1200-2000 CDT 03/28/2013 Y - 2 0 0 03/29/2013 Y - 1 1 0 03/30/2013 Y - 2 0 1 03/31/2013 Y - 1 2 0 04/01/2013 Y - 2 0 1 04/02/2013 N - 0 0 0 04/03/2013 Y - 0 2 0 04/04/2013 Y Y 1 0 2 VC2 deployed starting 4/4/2013 04/05/2013 Y P 0 0 0 VC2 ran 0800-1338 CDT 04/06/2013 Y N 0 0 0 VC2 hardware failure 04/07/2013 Y Y 0 0 0 04/08/2013 Y Y 2 0 1 04/09/2013 Y Y 1 0.5 0 04/10/2013 N N 0 0 0 VCs not deployed - severe weather 04/11/2013 Y Y 0 0 1 04/12/2013 Y Y 1 0 0 04/13/2013 Y Y 2 2 0 04/14/2013 Y Y 2 1 1 04/15/2013 Y Y 2 1 0 04/16/2013 Y Y 1 2.5 0 04/17/2013 Y Y 0 1 0 04/18/2013 Y Y 1 0 0 04/19/2013 Y Y 1 0 0 04/20/2013 Y Y 1 0 2 04/21/2013 Y Y 3 1 0 04/22/2013 Y Y 1 1 1 04/23/2013 Y Y 0 0 0 04/24/2013 Y Y 0 0 2 04/25/2013 Y Y 0 0 2 04/26/2013 Y Y 0 0 0 04/27/2013 Y Y 0 0 0 04/28/2013 Y P 1 0 1 VC2 ran 0800-1631 CDT Total Hours 30 15 15 Notes: VC = video camera for time-resolved takeoffs and landings (Y = yes, N = no, P = portion of the 12 hour period); Tail ID = still photographs of planes for tail number identification; TIM = time-in-mode data collection (e.g., taxiing, takeoff, climb-out); and Run-up = run-up area activity including TIM for magneto testing. -55-

Figure 9 Hourly Average Operations at RVS – All Aircraft PM sampling was conducted 8 AM to 8 PM CDT Table 24 Distribution of Aircraft Types Identified by Tail ID at RVS Plane Type Count % of Total Piston Single Engine 437 79% Multi Engine 59 11% Turboprop 24 4% Jet 30 5% Note: Based on 30 hours of still photography. Table 25 and Figure 10 summarize the run-up results. Mean TIM values were 69 seconds for the magneto test and 296 seconds for the total time in the run-up area. There was large variation in these times, with standard deviations of about 50% and 80% of the means for total run-up and magneto testing, respectively. Total run-up and magneto test -56-

Table 25 Time in Mode Data Collected for Run-Up Operations at RVS Total Run-Up Magneto Testing Number of Aircraft 109 76 Mean ± Std Dev (sec) 296 ± 150 69 ± 56 Median (sec) 284 50 Notes: Based on 15 hours of data collection. Means are reported with 1σ standard deviation values. Figure 10 Time-in-Mode Data for Total Time in the Run-Up Area and Duration of Magneto Testing at RVS (a) (b) Notes: (a) box plots (interior solid line is the median, interior dashed line is the arithmetic mean; box boundaries are 25th and 75th percentiles, whiskers are 10th and 90th percentiles, and circles are all records below the 10th percentile and above the 90th percentile); and (b) cumulative distributions as a log-probability plot. TIM data are shown as box plots in Figure 10(a) and cumulative distributions in Figure 10(b). Both total run-up time and magneto test duration data are approximated relatively well by a lognormal distribution as evidenced by the nearly linear trend for the log-probability plot. This means that a few aircraft have much longer TIM than would be expected from the standard deviations about the mean times. -57-

TIM data were also manually collected for piston-engine aircraft taxiing, idling, landings, and takeoffs. Fifteen hours of operations were viewed from an observation tower. Table 26 shows summary statistics for landing, takeoff, and touch-and-go times, as well as average locations for wheels-up and wheels-down. TIM for touch-and-go operations represents the time between wheels-down on landing and the subsequent wheels-up on takeoff. Wheels-up and wheels-down locations are measured as the distance from the start of the runway. There is less variation in TIM for landing and takeoff activities than for run-up activities. Activities were logged by aircraft so trip-based times can be constructed. Similar TIM data collection and processing has been performed for other aircraft activities, such as taxiing and idling, and the data are included in the database. Table 26 Summary of Time-in-Mode and Location of Aircraft Landing and Takeoff Operations at RVS Activity/Location Mean Time (s) Std. Dev (s) Mean Wheels-Up (ft) Mean Wheels-Down (ft) Landing Runway 1L 25 8 -a 562 Runway 1R 22 4 - 1036 Runway 19L 39 29 - 1117 Runway 19R 39 13 - 1016 Takeoff Runway 1L 18 9 1064 - Runway 1R 13 7 727 - Runway 19L 14 5 1324 - Runway 19R 20 8 1595 - Touch-and-Go Runway 1L 23 8 3370 976 Runway 1R 17 6 1324 - Runway 19L 16 8 1371 729 Runway 19R - - - - Notes: Based on 15 hours of data collection. TIM means are reported with 1σ standard deviation values. a Dashes indicate no data are available. 5.2.2 Airborne PM-Pb Data Collection PM sampling locations are shown in Figure 8 with key characteristics summarized in Table 27. The North site is the downwind primary site with presumably high impacts from run-up, taxiing and idling, and takeoff activities on runway 19R for prevailing southerly winds. The East site is the upwind primary site and should capture background Pb concentrations regardless of wind direction with the exception of westerly winds, which were rare during the study. The South site is impacted by climb-out from runway 19R for southerly winds and run-up, taxiing and idling, and takeoffs from runway 1L for -58-

Table 27 Airborne PM Sampling Locations for the RVS Study Site Location with Respect to Nearest Runway Comments North Downwind Primary ~125m NW of 19R For prevailing southerly winds, this site was impacted from runway 19R run-ups and takeoffs, as well as idling and taxiing. (Lat: 36.047435° Long: - 95.984719°) East Upwind Primary ~500m SE of 31 For winds from the south, east, and north, this site is upwind of all ground-based activities. It is ~700m east of runway 1R and may be modestly impacted by aircraft operations for winds from the west. (Lat: 36.033631° Long: -95.976139°) West Downwind Secondary ~250m NW of 13 For prevailing southerly winds, this site was impacted by the southern half of runways 19L and 19R and ground-based activities on the west side of the airport. (Lat: 36.042370° Long: -95.989708°) South Upwind Secondary ~200m SW of 1L For winds from the south, east, and west, this site is upwind of all ground-based activities. For northerly winds, it was impacted by ground-based activities on the west site of the airport including run-ups and takeoffs for runway 1L, as well as idling. (Lat: 36.032130° Long: -95.989700°) northerly winds. Emissions from ground-based operations west of the runways might impact this site for northerly winds. The West site is potentially impacted by ground- based operations on the west side of the airport for southerly winds and runway operations for easterly winds, which were rare during the study. Twelve-hour integrated PM samples were collected each day using up to four PQ100 samplers. Sampling was conducted from 8 AM to 8 PM CDT. Table 28 shows the PM samples collected each day. Although the goal was to operate four samplers during each event, one of the PQ100 samplers failed early in the study and eight runs were conducted with three samplers until a rental unit could be obtained while the original sampler was being repaired. Due to this and also the relatively high frequency of northerly winds, the campaign was extended an additional week. Samplers were operated on 32 of the 33 days—no sample collection was attempted on 4/2 because of heavy rain. The first day of sampling, conducted on 3/27, was not considered to be an attempted sampling event because it was a hardware shakedown day; the North site -59-

Table 28 Airborne PM Sampling Configurations and Wind Direction Characteristics at RVS PM Sampling Configuration Wind Direction Frequency and Classification Date N E S W Calm N E S W Winds 03/27/2013 {F,F} F,F 0 2 0 98 0 S 03/28/2013 F,F F,F 0 2 0 98 0 S 03/29/2013 F,F F,F 0 13 13 56 19 var 03/30/2013 F,F F,[F] 0 6 10 83 0 S 03/31/2013 F,F F,F 0 94 6 0 0 N 04/01/2013 F,F F 0 100 0 0 0 N 04/02/2013 no sample collection 0 46 54 0 0 var 04/03/2013 F,F F 0 69 25 4 2 N 04/04/2013 F,F F 0 88 0 0 13 N 04/05/2013 F,F F 4 0 0 85 10 S 04/06/2013 F,[F] F 0 2 0 94 4 S 04/07/2013 F,F F 2 0 4 94 0 S 04/08/2013 F F [F] 0 2 0 98 0 S 04/09/2013 F F F 0 2 0 98 0 S 04/10/2013 F F F 0 52 0 0 48 var 04/11/2013 F F F F 0 21 0 0 79 W 04/12/2013 F F F F 4 71 6 0 19 N 04/13/2013 F F F [F] 0 2 8 90 0 S 04/14/2013 F F F F 0 2 0 98 0 S 04/15/2013 F,T F,T 0 94 6 0 0 N 04/16/2013 F,T F,T 0 100 0 0 0 N 04/17/2013 F,T F,T 2 0 6 92 0 S 04/18/2013 F,T F,T 0 29 0 0 71 W 04/19/2013 F F F F 0 4 0 0 96 W 04/20/2013 F F F F 0 2 6 92 0 S 04/21/2013 F F F F 0 2 0 98 0 S 04/22/2013 F F F F 0 2 0 98 0 S 04/23/2013 F F F F 0 100 0 0 0 N 04/24/2013 F,T F,T 2 88 0 0 10 N 04/25/2013 F,T F,T 4 0 23 73 0 S 04/26/2013 F,T F,T 0 6 81 13 0 E 04/27/2013 F,T F,T 0 96 0 0 4 N 04/28/2013 F,T F,T 0 2 2 96 0 S Notes: The row following each interior horizontal line is a Monday. PM Sampling: F = PM2.5, T = TSP, [ ] = invalid sample; { } = samplers moved during the sampling period; and empty cells indicate no sample collection. Wind Direction: percentage frequency of 15-minute average ASOS 10m wind direction; and wind direction classification for events with at least two-thirds of the 15-minute winds from a given quadrant (var = variable winds). -60-

samplers were moved ~75 m during the sampling period from a temporary location to their permanent location and the South site samplers did not start until 11 AM. Thus, there are 31 events with valid PM2.5 sample collection at the North and East sites, which is nearly twice the objective of 17 events. The additional measurement objective of nine events with PM2.5 and TSP sampling at the North and East sites was met, while the eight events with PM2.5 sampling at four sites was one short of the objective because of a sampler failure. All samples were analyzed for Pb with no identified issues; thus, the sample collection completeness carried through to the overall PM-Pb data completeness. Figure 11 shows the wind rose for NWS ASOS (10 meter) 15-minute average data over all sampling periods (8 AM to 8 PM CDT). Daytime wind speeds were relatively high throughout the study with only 1.2% calms (operationally defined as wind speeds ≤ 1 m/s) and only 13% of 15-minute periods with wind speeds less than 4 m/s. While prevailing winds were from the south quadrant (46%), winds from the north quadrant were also common (32%). The model-to-monitor comparison does not require a consistent wind direction throughout the day, but instead relies upon the background site not being significantly impacted by emissions from aircraft operations; however, for many reasons, interpretation of the PM data is simplified when the wind direction is consistent throughout the sampling period. Wind direction persistence was examined by assigning each 15-minute average wind direction to a quadrant centered on the cardinal wind directions. A wind direction was assigned to a sampling event if at least two-thirds of the 15-minute average winds were from a given quadrant; otherwise the winds were deemed to be variable. Table 28 summarizes the daily wind direction patterns. Excluding the first sampling day (shakedown) and the day with no sample collection attempted, the 31 sampling events had the following distribution of winds: southerly, 15 events (48%), including 13 events after April 4 when the second video camera was deployed to improve the LTO data collection; northerly, 10 events (32%); westerly, 3 events (10%); easterly, 1 event (3%); and variable, 3 events (6%). While the criterion of winds being from a given quadrant for two-thirds of the time might seem relatively lax, for 19 of the 29 sampling events with a classified wind direction the winds were from the designated quadrant for at least 90% of the sampling period. Table 29 shows the PM-Pb data for all ambient PM samples as measured by ICP-MS. Four samples were invalidated, as explained in Appendix B. Of the 99 PM2.5 samples, 93 (94%) have a Pb concentration at least three times the PM2.5 median field blank level. Of the 18 TSP samples, 16 (89%) have Pb concentrations at least three times higher than the median TSP field blank. These results demonstrate that acceptable detectability was achieved for the airborne Pb data. -61-

Figure 11 Daytime (0800-1959 CDT) 15-minute Average 10m ASOS Winds at RVS, March 25 – April 29, 2013 -62-

Table 29 Airborne Pb Concentrations Observed at RVS Pb Concentration, ng/m3 North East Date Primary Collocate Primary Collocate South West 03/27/2013 {49.6} {50.1} 4.3 2.6 03/28/2013 37.7 32.9 2.5 2.4 03/29/2013 19.1 18.3 7.2 8.0 03/30/2013 8.6 8.0 1.2 -- 03/31/2013 1.2 1.1 1.2 0.9 04/01/2013 3.3 2.2 3.6 04/02/2013 04/03/2013 2.2 2.3 2.9 04/04/2013 6.2 2.8 3.3 04/05/2013 46.9 50.7 3.0 04/06/2013 17.2 -- 3.5 04/07/2013 7.3 7.0 1.4 04/08/2013 14.0 0.8 -- 04/09/2013 18.4 1.4 2.4 04/10/2013 0.5 0.4 0.3 04/11/2013 0.7 1.9 7.6 0.8 04/12/2013 1.9 2.6 29.2 1.8 04/13/2013 33.2 2.3 2.2 -- 04/14/2013 11.0 2.6 1.2 2.1 04/15/2013 1.6 [3.4] 1.6 [3.4] 04/16/2013 0.9 [3.3] 1.0 [2.2] 04/17/2013 15.5 [18.0] 1.1 [2.0] 04/18/2013 0.3 [0.4] 2.4 [2.2] 04/19/2013 1.2 6.7 9.9 1.2 04/20/2013 31.2 1.6 1.4 7.1 04/21/2013 7.7 1.5 1.4 2.9 04/22/2013 19.4 1.7 2.1 4.9 04/23/2013 1.0 0.4 12.3 0.6 04/24/2013 2.2 [4.2] 2.4 [3.3] 04/25/2013 26.8 [37.1] 3.2 [3.1] 04/26/2013 3.9 [4.6] 2.4 [3.2] 04/27/2013 1.5 [1.9] 0.3 [0.6] 04/28/2013 14.7 [22.5] 2.7 [2.0] Notes: The row following each interior horizontal line is a Monday. Data are PM2.5-Pb unless otherwise noted. [ ] = TSP;{ } = samplers moved during the sampling period; “--“ = invalid sample; and empty cells indicate no sample collection. -63-

Trends for PM-Pb are presented within the context of wind patterns during sample collection. Figure 12 compares PM2.5-Pb at the North and East sites with the data stratified by prevailing wind direction. For southerly winds, PM2.5-Pb at the North site is at least five times higher than PM2.5-Pb at the East (background) site. This pattern clearly indicates the downwind site is impacted by aircraft operations. In contrast, PM2.5-Pb concentrations at the North and East sites are quite similar for northerly winds. There are three events with PM2.5-Pb at the East site significantly higher than at the North site (i.e., the inverted triangles along the x-axis in Figure 12). These three events had winds from the west, which led to aircraft emissions impacts at the East site. In summary, the East site appears to represent background Pb levels for all conditions except when the winds are from the west. PM2.5-Pb for the North and South sites are compared in Figure 13. Southerly winds yield elevated Pb at the North site while northerly and westerly winds lead to elevated Pb at the South site, which is near a run-up and takeoff area for operations when winds are from the north. Figure 14 shows TSP-Pb versus PM2.5-Pb for the North and East sites. The mean concentration for Pb in the TSP-minus-PM2.5 size range (i.e., coarse particles) was 1.8 ng/m3. For the three samples with elevated Pb levels, PM2.5-Pb was 65-86% of TSP- Pb. The low concentrations for Pb in PMTSP-2.5 compared to Pb in PM2.5 at sites impacted by the piston-engine aircraft strongly suggest that the observed PM2.5-Pb in these samples is from direct aircraft exhaust emissions and not the tail of coarse mode PM-Pb extending below 2.5 µm. Figure 12 PM2.5-Pb at the North and East Sites Stratified by the Prevailing Wind Direction During Sample Collection at RVS -64-

Figure 13 PM2.5-Pb at the North and South Sites Stratified by the Prevailing Wind Direction During Sample Collection at RVS Figure 14 TSP-Pb versus PM2.5-Pb at the North and East Sites -65-

5.2.3 Avgas Data Collection Four avgas samples were collected from the FBOs. Christiansen Aviation Inc. is the largest FBO at the airport with activities including, but not limited to, a flight school. Avgas samples were obtained from Christiansen after three fuel deliveries. One sample was obtained from Spartan College of Aeronautics and Technology, which operates a flight school. A third flight school—Riverside—obtains avgas from Christiansen. The mean lead content of the seven avgas samples provided by Christiansen and Spartan college was 0.34 ± 0.008 g/L. 5.3 APA Field Study Figure 15 shows the airport layout. There are three runways. Prevailing winds are from the south but tend to shift throughout the day; however, the east/west runway is still used except at high wind speeds. Figure 15 Airport Diagram and PM Sampling and Activity Data Collection Locations Deployed at APA Data Collection Locations Note: PM sampling was conducted at the North (N), East (E), South (S), Center (C), and Center Secondary (CS) sites; video cameras were deployed at the VC1 and VC2 sites; and other activity data was manually collected at the C1, NER and FAA sites. -66-

The field study was conducted from May 15 to June 10, 2013. Video cameras were deployed to record LTOs starting on April 16. PM sampling and aircraft activity data were collected from 7 AM to 7 PM MDT. 5.3.1 Aircraft Activity Data Collection Aircraft activity data collection activities at APA are summarized in Table 30 and data collection locations are shown in Figure 15. Video cameras were continuously operated at the VC1 and VC2 locations during each 12-hour sampling event to record LTOs. Given the runway layouts at APA, it was necessary to review the video for both cameras: VC1 was used to capture LTOs on runway 10/28, and VC2 was used to capture LTOs and touch-and-go operations (TGOs) on runways 17L/35R and 17R/35L. Figure 16 shows average hourly distribution of total operations for all aircraft (not just piston-engine aircraft) as determined from the video camera data. Touch-and-go activities accounted for 25-50% of the total operations depending on the hour with higher proportions of such operations in the mornings. Total operations peaked around 11 AM and the lowest levels were towards the end of the 7 AM to 7 PM MDT sampling periods. About half (51%) of the operations were on runway 17R/35L, which is normally used only by piston-engine aircraft and has high touch-and-go activity. Forty percent (40%) of operations were on runway 17L/35R, which has all of the jet activity and some of the piston-engine activity. Only 9% of operations were on runway 10/28, which is used exclusively by piston-engine aircraft. For the north/south-oriented runways, 60% of operations originated at the north end (17L/17R) and 40% originated at the south end (25L/35R). For the east/west-oriented runway, 97% of operations originated at the west end (runway 10) and 3% originated at the east end (runway 28). Thirty hours of LTOs were photographed to develop the active fleet inventory. The photographs were reviewed to identify tail numbers, which were matched to aircraft and engine specifications in the FAA registry. The resulting fleet inventory database includes a record for each operation but with the tail numbers removed. Over the 30 hours of observation, there were 365 unique aircraft identified. Eighteen aircraft (5%) accounted for one-third of the operations and 39 aircraft (11%) accounted for half of the operations (35 fixed-wing single-engine aircraft and four fixed-wing, multi-engine aircraft). Table 31 summarizes the distribution of LTOs by aircraft type; about two-thirds of the operations were single-engine piston aircraft. -67-

Table 30 Aircraft Activity Data Collection at APA Date Activity Data Collection VC1 VC2 Tail ID TIM Run- up Comments 05/15/2013 N N 0 0 0 VCs not deployed during PM sampler shakedown 05/16/2013 Y Y 0 0 0 05/17/2013 Y Y 1 0 1 05/18/2013 Y Y 1 2 1 05/19/2013 Y Y 1 1 1 05/20/2013 Y Y 1 0 0 05/21/2013 Y Y 1 1 1 05/22/2013 Y Y 2 1 0 05/23/2013 Y Y 2 0 1 05/24/2013 Y Y 1 1 0 05/25/2013 Y Y 2 0 2 05/26/2013 Y Y 3 1 0 05/27/2013 Y Y 1 1 1 05/28/2013 Y Y 2 0 1 05/29/2013 Y Y 0 1 1 05/30/2013 Y Y 0 0 0 05/31/2013 Y Y 0 0 0 06/01/2013 Y Y 0 0 0 06/02/2013 Y Y 0 0 0 06/03/2013 Y Y 2 0 0 06/04/2013 Y Y 2 1 0 06/05/2013 Y Y 1 1 1 06/06/2013 Y N 1 1 1 VC2 hardware failure 06/07/2013 Y N 0 1 0 VC2 hardware failure 06/08/2013 Y Y 2 0 2 06/09/2013 Y Y 3 0 1 06/10/2013 Y Y 1 2 0 Total Hours 30 15 15 Note: VC = video camera for time-resolved takeoffs and landings (Y = yes, N = no); Tail ID = still photographs of planes for tail number identification; TIM = time-in-mode data collection (e.g., taxiing, takeoff, climb-out); and Run-up = run-up area activity including time in mode for magneto testing. -68-

Figure 16 Fixed-wing Aircraft Average Hourly Operations at APA PM sampling was conducted 7 AM to 7 PM MDT Table 31 Distribution of Aircraft Types Identified by Tail ID APA Plane Type Count % of Total Piston Single Engine 598 64% Multi Engine 64 7% Turboprop 93 10% Jet 182 19% Note: Based on 30 hours of still photography. Time-in-mode data were manually collected. Piston-engine aircraft run-up activities were observed for 15 hours and included 53 run-up operations, with magneto test duration recorded for 42 of these operations. Missing magneto test data primarily resulted from confounding sources of noise. Tail numbers were recorded for 89% of the run-up operations. Table 32 and Figure 17 summarize the run-up results. Mean times in mode were 97 seconds for the magneto test and 327 seconds for the total time in the run-up area. There was large variation in these times, with standard deviations of about 60% and 100% of the means for total run-up and magneto testing, respectively. Figure 17(a) shows box plots of the total run-up and magneto test TIM data; Figure 17(b) shows cumulative distributions for the TIM data. The magneto test data are relatively -69-

Table 32 Time in Mode Data Collected for Run-Up Operations at APA Total Run-Up Magneto Testing Number of Aircraft 53 42 Mean ± Std Dev (sec)1 327 ± 189 97 ± 102 Median (sec) 287 71 1 Means are reported with 1σ standard deviation values. Figure 17 Time-in-Mode Data for Total Time in the Run-Up Area and Duration of Magneto Testing at APA (a) (b) Notes: (a) box plots (interior solid line is the median, interior dashed line is the arithmetic mean; box boundaries are 25th and 75th percentiles, whiskers are 10th and 90th percentiles, and circles are all records below the 10th percentile and above the 90th percentile); and (b) cumulative distributions as a log- probability plot. well approximated by a lognormal distribution as evidenced by the nearly linear trend for the log-probability plot. The total run-up time data are not well represented by normal or lognormal distributions. Compared to RVS, the mean total run-up time and mean magneto test time were longer at APA. Both run-up time and magneto test time also showed higher variability at APA compared to RVS. -70-

TIM data were also manually collected for piston-engine aircraft taxiing, landings, and takeoffs. Fifteen hours of operations were viewed from an observation tower. Table 33 shows summary statistics for landing, takeoff, and touch-and go times as well as average locations for wheels-up and wheels down. TIM for touch-and-go operations represents the time between wheels down on landing and the subsequent wheels-up on takeoff. Wheels-up and wheels-down locations are measured as distance from the start of the runway. There is less variation in TIM for landing and takeoff activities than for run-up activities. Wheels-up for touch-and-go operations on Runway 17R were a shorter distance down the runway than wheels-up for takeoffs on the runway; this unexpected pattern likely results from only three regular takeoffs recorded on that runway during the TIM data collection periods. Taxiing activities were logged by aircraft so trip-based taxiing times can be constructed. Similar TIM data collection and processing have been performed for other aircraft activities such as taxiing and idling, and the data are included in the database. Table 33 Summary of Time-in-Mode and Location of Aircraft Landing and Takeoff Operations at APA Activity/Location Mean Time (s) Std Dev (s) Avg Wheels-Up (ft) Avg Wheels-Down (ft) Landing Runway 10 27 2 -a 700 Runway 28 37 10 - 600 Runway 17L 45 23 - 1600 Runway 17R 39 11 - 1200 Runway 35L 40 16 - 1500 Runway 35R 57 14 - 2100 Takeoff Runway 10 20 7 1700 - Runway 28 20 4 2700 - Runway 17L 27 6 2800 - Runway 17R - - 3300 - Runway 35L 19 - 1800 - Runway 35R 37 13 3400 - Touch-and-Go Runway 17L 23 - 3200 1700 Runway 17R 23 5 3000 1100 Runway 35L 23 6 2500 1400 Runway 35R 32 - - 2100 a Dashes indicate no data are available. -71-

5.3.2 Airborne PM-Pb Data Collection PM sampling locations are shown in Figure 15, with key characteristics summarized in Table 34. The Central site is the downwind primary site, with presumably high impacts from run-up and takeoff activities on runway 10 for prevailing southerly winds as well as taxiing on taxiway A. Although information provided prior to the field campaign suggested there was little piston-engine aircraft activity on runway 17L/35R, we were subsequently informed that there is considerable piston-engine activity on this runway that may lead to significant impacts at the Central site. The East site is the upwind primary site and should capture background Pb concentrations because of its distance from airport activity. The South site is impacted by climb-out from runway 17L for southerly winds and run-up, taxiing and idling, and takeoffs from runway 35R for northerly winds. Emissions from ground-based operations east of the runways might impact this site for northerly winds. The North site should be impacted primarily by climb-out from runway 35R for northerly winds and run-up, taxiing and idling, and takeoffs from runway 17L for southerly winds. Sampling was conducted from 7 AM to 7 PM MDT. Table 35 shows the PM samples collected each day. The goal was to operate four samplers during each event. Valid PM2.5 samples were collected at the Central and East sites (i.e., the primary sites) for each event. Thus, there are 25 events with valid PM2.5 data collection at the Central and East sites, which is 40% greater the objective of 17 events. The additional measurement objective of nine events with PM2.5 and TSP sampling at the Central and East sites was met; the objective of nine events with PM2.5 sampling at four sites was exceeded, with three additional valid samples collected at the Central Secondary site instead of the South site. All samples were analyzed for Pb with no identified issues; thus, the sample collection completeness carried through to the overall PM-Pb data completeness. Figure 18 shows the wind rose for NWS ASOS (10 meter) 15-minute average data over all sampling periods (7 AM to 7 PM MDT). Daytime wind speeds were relatively high throughout the study, with only 1.1% calms (operationally defined as wind speeds ≤ 1 m/s) and 76% of 15-minute periods with wind speeds greater than 4 m/s. While prevailing winds were from the south quadrant (36%), winds from the west quadrant (22.7%) and north quadrant (21.3%) were also common. The wind direction was more variable than expected from the climatology for this period. Wind direction persistence was examined by assigning each 15-minute average wind direction to a quadrant centered on the cardinal wind directions. A wind direction was assigned to a sampling event if at least two-thirds of the 15-minute average winds were from a given quadrant; otherwise the winds were deemed to be variable. Table 35 summarizes the daily wind direction patterns. Excluding the two days with mostly missing wind data, the remaining 25 sampling events had the following distribution of winds: southerly, 2 events (8%); northerly, 4 events (16%); westerly, 1 event (4%); easterly, 1 event (4%); and variable, 17 events (68%). In contrast to RVS, there were a high proportion of events with variable winds at APA. However, the variations within sampling periods tended to be distinct shifts followed by persistent winds rather than light and variable winds. Thus, these sampling events are still conducive to dispersion modeling. -72-

Table 34 Airborne PM Sampling Locations for the APA Study Site Location with Respect to Nearest Runway Comments Central Downwind Primary ~250m NW of 10 For prevailing southerly winds, this site was impacted by runway 10 run-ups and takeoffs. It was also impacted by taxi and idle activities around the center of the airport. (Lat: 39.574860° Long: -104.849210°) East Upwind Primary ~1km SE of 10 For winds from the south, east, and north, this site is upwind of all ground-based activities. It is ~850m east of runway 35R and was modestly impacted by aircraft operations for winds from the west. (Lat: 39.566290° Long: -104.840830 °) North Downwind Secondary ~300m NW of 17L For prevailing southerly winds, this site was impacted by the northern portions of runways 35L and 35R and ground-based activities on the east side of the airport. (Lat: 39.586810° Long: -104.850550°) South Upwind Secondary ~250m SE of 35R For winds from the south, east, and west, this site is upwind of all ground-based activities. For northerly winds, it was impacted by ground-based activities on the east side of the airport, including run-ups and takeoffs for runway 35R. (Lat: 39.555010° Long: -104.847950°) Central Secondary Upwind Tertiary ~200m SE of 10 For winds from the south and east, this site is upwind of all landing and takeoff operations but was impacted by taxiing on the ramp to the south of the site. For northerly winds, it was impacted by activities on runway 10. (Lat: 39.571300° Long: -104.846660°) -73-

Table 35 Airborne PM Sampling Configurations and Wind Direction Characteristics at APA Date PM Sampling Configuration Wind Direction Frequency and Classification C E N S CS Calm N E S W Winds 05/15/2013 F F F [F] 2 17 8 19 54 var 05/16/2013 F F F F 0 17 17 33 33 var 05/17/2013 F F F F 6 0 8 46 40 var 05/18/2013 F F F F 2 15 31 6 46 var 05/19/2013 F F F F 0 46 4 8 42 var 05/20/2013 F F F F 4 42 23 23 8 var 05/21/2013 F F F F 0 67 23 10 0 N 05/22/2013 F F F F 4 10 65 6 15 var 05/23/2013 F F F F 0 2 2 96 0 S 05/24/2013 F,T F,T 77% of records missing 05/25/2013 F,T F,T 0 8 4 40 48 var 05/26/2013 F,T F,T 0 29 29 10 31 var 05/27/2013 F,T F,T 0 17 8 33 42 var 05/28/2013 F,T F,T 2 15 4 79 0 S 05/29/2013 F F F F 0 63 31 0 6 var 05/30/2013 F F F F 0 10 0 2 88 W 05/31/2013 F F F F 77% of records missing 06/01/2013 F F F F 0 77 19 4 0 N 06/02/2013 F,T F,T 8 29 42 19 2 var 06/03/2013 F,T F,T 0 15 0 25 60 var 06/04/2013 F,T F,T 0 58 40 0 2 var 06/05/2013 F,T F,T 0 15 85 0 0 E 06/06/2013 F,F F,F 0 67 8 4 21 N 06/07/2013 F,F F,[F] 0 8 0 31 60 var 06/08/2013 F,F F,F 0 90 2 0 8 N 06/09/2013 T,T T,T 0 19 27 23 31 var 06/10/2013 T,T T,T 0 27 17 38 19 var Note: The row following each interior horizontal line is a Monday. PM Sampling: F = PM2.5, T = TSP; [ ] = invalid sample; and empty cells indicate no sample collection. Wind Direction: percentage frequency of 15-minute average ASOS 10m wind direction; and wind direction classification for events with at least two-thirds of the 15-minute winds from a given quadrant (var = variable winds). -74-

Figure 18 Daytime (0700-1859 MDT) 15-minute Average 10m ASOS winds at APA, May 15 – June 10, 2013 -75-

Table 36 shows the PM-Pb data for all PM samples as measured by ICP-MS. Two samples were invalidated and explanations are provided in Appendix B. Of the 80 analyzed PM2.5 samples, 79 (99%) have a Pb concentration at least three times the PM2.5 median field blank; of the 26 TSP samples, 24 (92%) have Pb concentrations at least three times higher than the median TSP field blank. These results demonstrate that acceptable detectability was achieved for the airborne Pb data. Table 36 Airborne Pb Concentrations Observed at APA Date Pb Concentration (ng/m3) Central East North South Central Secondary Primary Collocate Primary Collocate 05/15/2013 16.2 1.4 2.0 -- 05/16/2013 22.7 1.5 5.4 4.3 05/17/2013 27.2 1.3 12.0 4.0 05/18/2013 20.5 3.5 2.3 9.3 05/19/2013 8.1 1.2 1.3 3.5 05/20/2013 7.7 2.2 1.5 1.9 05/21/2013 35.4 1.8 1.2 3.4 05/22/2013 8.4 3.0 2.3 2.0 05/23/2013 47.1 0.4 6.6 0.2 05/24/2013 18.3 [25.5] 1.6 [2.2] 05/25/2013 16.9 [21.7] 1.9 [2.6] 05/26/2013 12.4 [22.0] 2.8 [2.1] 05/27/2013 12.8 [12.5] 1.8 [1.8] 05/28/2013 26.1 [27.7] 1.8 [2.1] 05/29/2013 3.1 1.0 1.0 0.9 05/30/2013 4.7 1.9 1.6 1.1 05/31/2013 4.3 1.2 0.3 2.0 06/01/2013 4.1 1.1 4.9 1.3 06/02/2013 7.8 [10.1] 2.0 [1.7] 06/03/2013 8.8 [1.2] 1.4 [2.6] 06/04/2013 3.3 [4.0] 1.2 [1.8] 06/05/2013 11.1 [9.8] 0.4 [0.0] 06/06/2013 17.3 15.4 5.6 4.9 06/07/2013 19.6 -- 1.2 06/08/2013 5.7 5.5 2.6 2.4 06/09/2013 [13.2] [11.7] [2.3] [3.8] 06/10/2013 [13.9] [15.5] [1.9] [0.9] Note: The row following each interior horizontal line is a Monday. Data are PM2.5-Pb unless otherwise noted. [ ] = TSP; “--“ = invalid sample; and empty cells indicate no sample collection. -76-

Figure 19 compares PM2.5-Pb at the Central and East sites. The minimum ratio of Central site concentration to East site concentration was 2.2. While it is difficult to stratify the data based on wind direction because of the within-sample wind variations, the highest ratios usually occurred for higher wind frequencies from the south or west. Figure 19 PM2.5-Pb at the Central and East Sites at APA This pattern clearly indicates the downwind Central site is impacted by aircraft operations. In contrast, for sampling events with a higher frequency of northerly or easterly winds the ratios of PM2.5-Pb at the Central and East sites are usually lower. In summary, the East site appears to represent background Pb levels for all conditions, with the Central site more strongly impacted for winds from the south or west. The PM2.5-Pb levels at the North and South sites were above or within the measurement error, as defined by the collocated precision, of the PM2.5-Pb at the East site. This supports the selection of the East site as a proper background site for the airport and also demonstrates the localized nature of concentration hot spots near zones such as the Central site. Additionally, the Central Secondary site was used for three days of sample collection. The PM2.5-Pb measured at the Central Secondary site was about three times higher than the East site for each day and the difference between the two sites exceeded the measurement error. -77-

Figure 20 shows TSP-Pb versus PM2.5-Pb for the Central and East sites. The high- concentration samples show enrichment of Pb in TSP but most of the Pb is in the PM2.5 size range. One sample has a higher Pb concentration in PM2.5 than in TSP, which is physically unrealistic; however, this sample may have been contaminated as there were large specks and dirt on the PM2.5 filter after sampling (such cases were rare and these observations were flagged in the data set). The mean concentration for Pb in the TSP- minus-PM2.5 size range (i.e., coarse particles) was 2.6 ng/m3, excluding the sample with higher PM2.5-Pb than TSP-Pb. For the four samples with elevated Pb levels, PM2.5-Pb was 56-94% of TSP-Pb. The relatively low concentrations for Pb in PMTSP-2.5 compared to Pb in PM2.5 at the piston-engine aircraft-impacted sites strongly suggest that the observed PM2.5-Pb in these samples is from direct aircraft exhaust emissions and not the tail of coarse mode PM-Pb extending below 2.5 µm. Figure 20 TSP-Pb versus PM2.5-Pb at the Central and East Sites at APA 5.3.3 Avgas Data Collection Nine aviation gasoline (avgas) samples were collected from the FBOs. TAC Air and Denver Jet Center are the largest FBOs at the airport in terms of gasoline distribution. Three avgas samples were obtained from Denver Jet Center and four were obtained from TAC Air. One sample was obtained from Signature Flight Support and another sample was obtained from XOJet. The mean lead content of the seven avgas samples provided by Denver Jet Center and TAC Air was 1.57 ± 0.3 g/gal. The samples from XOJet (1.32 g/gal) and Signature Flight Support (2.08 g/gal) were excluded from this calculation because they are jet-focused FBOs and distribute limited amounts of avgas. -78-

5.4 SMO Field Study Figure 21 shows the airport layout. There is only one runway at SMO. Prevailing winds were from the southwest throughout the entire field study, which was conducted from July 3 to July 30, 2013. Figure 21 Airport Diagram and PM Sampling and Activity Data Collection Locations Deployed at SMO Note: PM sampling was conducted at the North (N), Northeast (NE), West (W), and Southwest (SW) sites; video cameras were deployed at the North site; and other activity data was manually collected at the Skydeck and NER sites. 5.4.1 Aircraft Activity Data Collection Aircraft activity data collection activities at SMO are summarized in Table 37 and data collection locations shown in Figure 21. Video cameras were continuously operated at -79-

Table 37 Aircraft Activity Data Collection at SMO Date Activity Data Collection VC1 VC2 Tail ID TIM Run- up Comments 07/03/2013 Y Y 0 0 0 07/04/2013 Y Y 1 1 0 07/05/2013 Y Y 0 0 0 07/06/2013 Y Y 2 2 0 07/07/2013 Y Y 2 1 0 07/08/2013 Y Y 2 0 1 07/09/2013 Y Y 0 1 2 07/10/2013 Y Y 2 2 0 07/11/2013 Y Y 0 1 2 07/12/2013 Y Y 1 1 0 07/13/2013 Y Y 0 0 0 07/14/2013 Y Y 0 0 0 07/15/2013 Y Y 0 0 0 07/16/2013 Y Y 0 0 0 07/17/2013 Y Y 1 0 0 07/18/2013 Y Y 2 0 0 07/19/2013 Y Y 2 0 1 07/20/2013 Y Y 2 0 2 07/21/2013 Y Y 2 1 1 07/22/2013 Y Y 1 2 0 07/23/2013 Y Y 2 1 0 07/24/2013 Y Y 0 0 1 No run-up activity during collection period 07/25/2013 Y Y 2 0 0 07/26/2013 Y Y 2 0 0 07/27/2013 Y Y 2 0 2 07/28/2013 Y Y 2 1 3 07/29/2013 Y Y 1 1 0 07/30/2013 Y Y 0 0 1 Total Hours 30 15 15 Note: VC = video camera for time-resolved takeoffs and landings (Y = yes, N = no); Tail ID = still photographs of planes for tail number identification; TIM = time-in-mode data collection (e.g., taxiing, takeoff, climb-out); and Run-up = run-up area activity including time in mode for magneto testing. -80-

the North site during each 12-hour sampling event to record LTOs. One camera was set up to determine the number of LTOs, while a second camera was set up to enhance run- up time characterization. In contrast to RVS and APA, where the runways covered large footprints, the activities at SMO are more concentrated and the fractions of jets, turboprops, and piston-engine aircraft could be determined from the video camera data. Given the positioning of the video camera, however, it was sometimes difficult to distinguish touch-and-go operations from normal landings; therefore, touch-and-go operations are underrepresented and normal landings are overrepresented in this data set. The TIM data collected at SMO may be able to close this data gap. Figure 22 shows average hourly piston-engine operations for the entire study period (in contrast, Figure 9 for RVS and Figure 16 for APA include all fixed-wing aircraft). Touch-and-go activities are counted as two operations each. From 11 AM to 4 PM, the total hourly operations were relatively high and consistent from hour to hour. Thirty hours of LTOs were photographed from the Skydeck to develop the active fleet inventory. The photographs were reviewed to identify tail numbers, which were matched to aircraft and engine specifications in the FAA registry. The resulting fleet inventory database includes a record for each operation but with the tail numbers removed. Over the 30 hours of observation, 247 unique aircraft were identified. Twelve aircraft (5%) accounted for one-third of the operations and 25 aircraft (10%) accounted for half of the operations (24 fixed-wing, single-engine aircraft and one fixed-wing, multi-engine aircraft). Table 38 summarizes the distribution of LTOs by aircraft type; more than three-fourths of the operations were single-engine piston aircraft. TIM data were manually collected. Piston-engine aircraft run-up activities were observed for 15 hours and included 41 run-up operations, with magneto test duration recorded for 36 of these operations. Missing magneto test data primarily resulted from confounding sources of noise. Tail numbers were recorded for 95% of the run-up operations. Twenty- three planes bypassed the run-up area and did not perform run-ups that were observed. Table 39 and Figure 23 summarize the run-up results. Mean TIM was 61 seconds for the magneto test and 328 seconds for the total time in the run-up area. There was a large variation in these times, with standard deviations of about 70% and 80% of the means for total run-up and magneto testing, respectively. Figure 23(a) shows box plots of the total run-up and magneto test TIM data, and Figure 23(b) shows cumulative distributions for the TIM data. Both total run-up time and magneto test data are relatively well approximated by a lognormal distribution as evidenced by the nearly linear trend for the log-probability plot. Mean total run-up times at SMO and APA were similar, with shorter total run-up times at RVS. The highest variability in total run-up time was observed at SMO. Mean magneto test times at SMO were shorter than at RVS and APA. The magneto test times at SMA and RVS had similar variability, with higher variability observed at APA. -81-

Figure 22 Piston-Engine Aircraft Average Hourly Operations at SMO Note: PM sampling was conducted from 8 AM to 8 PM PDT. Table 38 Distribution of Aircraft Types Identified by Tail ID at SMO Plane Type Count % of Total Piston Single Engine 447 78% Multi Engine 11 2% Turboprop 32 6% Jet 81 14% Note: Based on 30 hours of still photography. Table 39 Time-in-Mode Data Collected for Run-Up Operations Including Magneto Testing at SMO Total Run-Up Magneto Testing Number of Aircraft 41 36 Mean ± Std Dev (sec)a 328 ± 215 61 ± 52 Median (sec) 244 42 Notes: Means are reported with 1σ standard deviation values. -82-

Figure 23 Time-in-Mode Data for Total Time in the Run-Up Area and Duration of Magneto Testing at SMO (a) (b) Notes: (a) box plots (interior solid line is the median, interior dashed line is the arithmetic mean; box boundaries are 25th and 75th percentiles, whiskers are 10th and 90th percentiles, and circles are all records below the 10th percentile and above the 90th percentile); and (b) cumulative distributions as a log- probability plot. TIM data were also manually collected for piston-engine aircraft taxiing, landings, and takeoffs. Fifteen hours of operations were viewed from an observation deck. Table 40 shows summary statistics for landing, takeoff, and touch-and-go times, as well as average locations for wheels-up and wheels-down. TIM for touch-and-go operations represents the time between wheels down on landing and the subsequent wheels up on takeoff. Wheels-up and wheels-down locations are measured as the distance from the start of the runway. There is less variation in the time-in-mode for landing and takeoff activities than for run-up activities. Taxiing activities were logged by aircraft so trip-based taxiing times can be constructed. Similar time-in-mode data collection and processing has been performed for other aircraft activities such as taxiing and idling and the data are included in the database. -83-

Table 40 Summary of Time-in-Mode and Location of Aircraft Landing and Takeoff Operations at SMO Activity/Location Mean Time (s) Std Dev (s) Avg Wheels-Up (ft) Avg Wheels-Down (ft) Landing Runway 21 37 14 -a 1200 Takeoff Runway 3 21 2 2100 - Runway 21 15 3 1200 - Touch-and-Go Runway 21 10 4 1600 930 a Dashes indicate no data. 5.4.2 Airborne PM-Pb Data Collection Figure 21 depicts PM sampling locations, and Table 41 summarizes the key characteristics. The Northeast site is the downwind primary site, with presumably high impacts from run-up, taxiing and idling, and takeoff activities on runway 21 for prevailing southwesterly winds. In contrast to RVS and APA, it was not possible to locate the upwind primary site on the airport footprint far removed from aircraft activities. The Southwest site is the upwind primary site and should, to a large extent, capture background Pb concentrations; however, it might be influenced by climb-out, for prevailing southwesterly winds. The North site is impacted by taxiing and climb-out from runways 3 and 21 for southwesterly winds. Emissions from ground-based operations north of the runway might impact this site for prevailing winds. The West site should be impacted primarily by climb-out from runway 3 for southwesterly winds and ground-based activities for southerly and southeasterly winds. Sampling was conducted from 8 AM to 8 PM PDT. Table 42 shows the PM samples collected each day. Although the goal was to operate four samplers during each event, one of the PQ100 samplers failed on the first day of the study and only three samplers were available until a rental unit could be obtained while the original was being repaired. Valid PM2.5 samples were collected at the Northeast and Southwest sites (i.e., the primary sites) for each event. Thus, there are 25 events with valid PM2.5 sample collection at the Northeast and Southwest sites, which is nearly 50% greater than the objective of 17 events. The additional measurement objectives of nine events with PM2.5 and TSP sampling at the Northeast and Southwest sites was met; however, the objective of nine events with PM2.5 sampling at four sites was not met, with only six valid samples collected at the West site because of the sampler failure that required a replacement unit. All samples were analyzed for Pb and three samples were invalidated because of contamination issues (Northeast and Southwest on July 13, and West on July 17). The justification for invalidating samples is provided in Appendix B. This reduced the PM-Pb data completeness to 24 events with PM2.5 data at the Northeast and Southwest sites and four events with PM2.5 sampling at the two primary sites and two secondary sites. There were eight events with PM2.5 sampling at the two primary sites and the North secondary site. -84-

Table 41 Airborne PM Sampling Locations for the SMO Study Site Location with Respect to Nearest Runway Comments Northeast Downwind Primary ~100m N of 21 For prevailing southwesterly winds, this site was impacted from runway 21 run-ups and takeoffs. It was also significantly impacted by taxiing and idling on Taxiways A and B. (Lat: 34.021490° Long: -118.445531°) Southwest Upwind Primary ~150m NW of 3 For winds from the southwest, this site is upwind of all ground-based activities but may be impacted by climb- out. It may be impacted by aircraft operations for winds from the east. (Lat: 34.011560° Long: -118.458439°) North Downwind Secondary ~150m W of 21 For prevailing southwesterly winds, this site was impacted by takeoffs and run-ups on runway 3, as well as most ground-based activities on Taxiways A and B. (Lat: 34.021030° Long: -118.447011°) West Upwind Secondary ~400m NE of 3 For winds from the southwest, this site is upwind of all activities on runway 21 and the northeast side of the airport. It may be impacted by activities at the southwest end of the airport. For southeasterly winds, it is potentially impacted by ground-based activities at the southern end of the airport. In contrast to the other three sites that were sited in open fetch, the West site was near obstructions (buildings, trees). (Lat: 34.014300° Long: -118.456050°) Figure 24 shows the wind rose for NWS ASOS (10 meter) 15-minute average data over all sampling periods (8 AM to 8 PM PDT). Daytime wind speeds were relatively high throughout the study with only 7.1% calms (operationally defined as wind speeds ≤ 1 m/s) and 54% of 15-minute periods with wind speeds less than 4 m/s. Prevailing winds were from the southwest quadrant for the vast majority of the study period (94%). -85-

Table 42 Airborne PM Sampling Configurations and Wind Direction Characteristics at SMO PM Sampling Configuration Wind Direction Frequency and Classification Date NE SW N W Calm NW NE SE SW Winds 07/03/2013 F F 0 0 0 2 98 SW 07/04/2013 F F F 0 0 0 4 96 SW 07/05/2013 F F F 4 0 6 10 79 SW 07/06/2013 F F F 0 2 0 4 94 SW 07/07/2013 F,T F,T 0 0 0 0 100 SW 07/08/2013 F,T F,T 0 0 0 0 100 SW 07/09/2013 F,T F,T 0 0 0 2 98 SW 07/10/2013 F,T F,T 0 6 0 4 90 SW 07/11/2013 F,T F,T 2 0 0 21 77 SW 07/12/2013 F F F F 0 0 0 19 81 SW 07/13/2013 [F] [F] F F 0 2 0 2 96 SW 07/14/2013 F F F F 2 0 0 6 92 SW 07/15/2013 F F F F 0 0 0 0 100 SW 07/16/2013 F F F F 58% of records missing 07/17/2013 F F F [F] 0 0 0 0 100 SW 07/18/2013 F,T F,T 0 0 0 10 90 SW 07/19/2013 F,T F,T 0 0 0 0 100 SW 07/20/2013 F,T F,T 0 0 0 17 83 SW 07/21/2013 F,T F,T 0 2 0 10 88 SW 07/22/2013 F,F F,F 0 0 0 0 100 SW 07/23/2013 F,F F,F 0 0 0 0 100 SW 07/24/2013 F,F F,F 0 0 0 2 98 SW 07/25/2013 F,F F,F 0 0 0 0 100 SW 07/26/2013 F,F F,F 0 0 0 6 94 SW 07/27/2013 F,F F,F 0 0 0 0 100 SW 07/28/2013 T,T T,T 0 0 0 0 100 SW 07/29/2013 T,T T,T 0 0 0 0 100 SW 07/30/2013 T,T T,T 0 6 0 0 94 SW Notes: The row following each interior horizontal line is a Monday. PM Sampling: F = PM2.5, T = TSP, [ ] = invalid sample, and empty cells = no sample collection. Wind Direction: percentage frequency of 15- minute average ASOS 10m wind direction; and wind direction classification for events with at least two- thirds of the 15-minute winds from a given quadrant. -86-

Figure 24 Daytime (0800-1959 PDT) 15-minute Average 10m ASOS Winds at SMO, July 3 – 30, 2013 -87-

Wind direction persistence was examined by assigning each 15-minute average wind direction to a quadrant centered on the wind directions midway between the cardinal directions. A wind direction was assigned to a sampling event if at least two-thirds of the 15-minute average winds were from a given quadrant; otherwise, the winds were deemed to be variable. Table 42 above summarizes the daily wind direction patterns. Excluding the day with mostly missing wind data, all sampling events were classified as southwesterly winds. Winds were much more consistent at SMO than they were at either APA or RVS. While the criterion of winds being from a given quadrant for two-thirds of the time might seem relatively lax, for 22 of the 27 events with available ASOS wind direction the winds were from the southwest for at least 90% of the sampling period. A laboratory contamination issue was discovered that affected samples from July 3 through July 21, with the exception of those samples sent for XRF analysis. The source of this contamination was determined to be the acid bath for cleaning the glassware used in sample preparation. To quantify the contamination, “glassware blanks” were collected prior to and following replacement of the acid bath and these samples were analyzed using ICP-MS. The median contamination level corresponded to 2.8 ng/m3, and this value was subtracted from all samples digested in the contaminated batches. The overall impact of the contamination is quite modest for samples collected at the Northeast (downwind primary) site, with 21 of the 23 samples having a corrected concentration more than five times the correction value and 50% of the Northeast samples having corrected concentrations more than ten times the correction value. Corrected field blank concentrations ranged from -2 to +2 ng/m3; for the Northeast site, six of the seven TSP- Pb values were greater than the corresponding PM2.5-Pb values, with the one exception having a difference within the measurement error. The contamination had a greater impact on data collected at the remaining sites (Southwest, North, and West) where the PM-Pb concentrations are much lower. Appendix B discusses the rationale for voiding the Northeast and Southwest samples on July13 and the West sample on July 17. Table 43 shows the PM-Pb data for all PM samples as measured by ICP-MS and including the correction for contamination. Of the 57 analyzed PM2.5 samples, 45 (79%) have a Pb concentration at least three times the PM2.5 median field blank. Additionally, 29 (97%) of the 30 analyzed PM2.5 samples collected at the North and Northeast sites have a Pb concentration at least three times the PM2.5 median field blank. Of the 30 TSP samples, 26 (87%) have Pb concentrations at least three times higher than the median TSP field blank. These results demonstrate that acceptable detectability was achieved for the airborne Pb data. Figure 25 compares PM2.5-Pb at the Northeast and Southwest sites. Of the 30 data pairs, 29 (97%) have downwind to upwind ratios of greater than five, and 20 (67%) have a ratio of greater than 20. This pattern clearly indicates the downwind Northeast site is impacted by aircraft operations. PM2.5-Pb levels at the Southwest site were below or within the measurement error, as defined by the collocated precision, of the PM2.5-Pb at the North site and West site for days without contaminated samples. These patterns support the assignment of the Southwest site as a background site for the airport. For 6 of the 9 days with sampling at the North site, PM2.5-Pb at that site was at least double the Southwest site. This shows that the North site is likely impacted from aircraft climb-out and possibly ground operations. -88-

Table 43 Airborne Pb Concentrations Observed at SMO Pb Concentration, n/m3 Northeast Southwest Date Primary Collocate Primary Collocate North West 07/03/2013 29.0 1.7 07/04/2013 16.9 < 0 3.3 07/05/2013 < 0 3.3 6.8 07/06/2013 36.8 3.7 8.6 07/07/2013 27.5 [35.3] 0.3 [4.5] 07/08/2013 20.5 [50.5] 1.0 [2.5] 07/09/2013 25.5 [39.6] 0.4 [2.9] 07/10/2013 37.7 [46.6] 3.2 [3.9] 07/11/2013 27.9 [37.8] 4.7 [2.9] 07/12/2013 30.7 1.6 7.3 0.0 07/13/2013 -- -- 10.7 7.6 07/14/2013 71.8 1.0 5.4 1.7 07/15/2013 6.8 < 0 5.5 < 0 07/16/2013 29.9 1.0 5.6 1.7 07/17/2013 30.5 < 0 3.4 -- 07/18/2013 32.4 [42.4] 2.0 [0.8] 07/19/2013 44.9 [61.4] 4.2 [2.3] 07/20/2013 27.6 [26.8] < 0 [1.7] 07/21/2013 34.2 [47.1] 0.1 [0.9] 07/22/2013 20.6 21.6 0.9 0.6 07/23/2013 19.9 20.8 0.2 0.3 07/24/2013 18.9 25.4 0.2 1.3 07/25/2013 15.0 17.1 0.0 0.9 07/26/2013 36.1 31.9 1.4 1.2 07/27/2013 67.1 32.1 0.4 0.5 07/28/2013 [33.2] [35.1] [1.0] [1.4] 07/29/2013 [26.0] [29.3] [1.7] [1.8] 07/30/2013 [59.6] [57.2] [2.0] [1.0] Notes: The row following each interior horizontal line is a Monday. Data are PM2.5-Pb unless otherwise noted. [ ] = TSP; and empty cells indicate no sample collection. 2.8 ng/m3 was subtracted from the underlined samples through to correct for laboratory contamination; negative concentration values after applying this correction are denoted by “< 0”. Samples denoted with a dash (--) were invalidated because of contamination. -89-

Figure 25 PM2.5-Pb at the Northeast and Southeast Sites at SMO Note that the y-axis range is 20 times the x-axis range. Figure 26 shows TSP-Pb versus PM2.5-Pb for the Northeast and Southwest sites. One sample has a higher Pb concentration in PM2.5 than in TSP, which is physically unrealistic; however, the difference is within the measurement error. The average PMTSP-2.5 concentration at the Northeast site was 12 ng/m3, which is 40% of the average PM2.5-Pb for the same sample days. PM2.5-Pb was 41-100% with only one sample pair below 64%. Thus, TSP-Pb at the Northeast site is dominated by PM2.5-Pb, which supports the Pb at this site being predominantly from direct piston-engine aircraft exhaust, but with significant contributions from the tail of coarse mode PM-Pb extending below 2.5 µm. -90-

Figure 26 TSP-Pb versus PM2.5-Pb at the Northeast and Southwest Sites at SMO 5.4.3 Avgas Data Collection Because the SMO FBOs that sell avgas declined to provide gasoline samples, two avgas samples were collected from SMO-based planes. One sample was collected from an aircraft that had recently refueled at the American Flyers FBO and another sample was collected from an aircraft that was under maintenance (at Bill’s Air Center). Avgas samples were analyzed by Intertek Caleb Brett for Pb content analysis using test method ASTM D5059. The lead content was 0.489 g/L (1.85 g/gal) from the plane fueling at American Flyers and 0.522 g/L (1.98 g/gal) from the plane under maintenance. The relative volumes of avgas in the aircraft tanks from fueling at SMO versus other airports are not known. 5.5 LTOs from On-Site Observations and ATADS As described in Section 5.1, LTOs were captured using video cameras deployed during each 12-hour PM-Pb sampling period. ATADS data were obtained for each day of on- site data collection and are compared to the LTOs recorded by video camera. Figure 27 shows the distributions of the ratio of 12-hour (daytime) video recorded operations to corresponding daily ATADS-reported operations. For all three airports, LTOs from the video data are lower than ATADS data with mean ratios of ~60% at RVS and APA, and ~85% at SMO. Part of the video data deficit is from LTOs prior to or following the 12- hour video data collection period. However, at APA the FAA also provided hourly LTO counts used to generate the ATADS-reported data and only ~10% of all FAA counted -91-

observations occurred outside the 12-hour video data collection period. For APA, this still leaves a difference of ~30% between the video data and ATADS data. The relatively better agreement at SMO might arise in part from the restrictions on nighttime activity, which reduce the LTOs outside the 12-hour daytime video data collection period. There are several factors that could contribute to the discrepancies, such as overflights that are counted by the FAA and requested operations that are cleared by the FAA but subsequently aborted. More work is needed to explain these differences, which do impact the emissions estimates. Figure 27 Distributions of Daily Video-Recorded and ATADS-Reported LTOs Notes: Interior solid lines are medians and interior dashed lines are arithmetic means, whiskers are 10th and 90th percentiles. RVS, APA, and SMO are 12-hour video data and daily ATADS data; APA-Hourly corresponds to 12-hour video data and FAA data, used to generate ATADS values, for the same 12-hour period as the video data. 5.6 Additional Lines of Evidence for the Origins of Airborne Pb At each airport, the highest ambient PM-Pb concentrations were measured at the sites downwind of aircraft operations. PM2.5-Pb at the aircraft-impacted sites accounted for most of the TSP-Pb, with ranges of 65-86% at RVS, 71-100% at APA, and 41-100% at SMO. The enrichment of Pb in the fine particulate matter size range is consistent with fresh exhaust emissions rather than exhaust emissions that previously deposited and are resuspended by wind or aircraft-induced turbulence during the sampling events. Additional lines of evidence for the contribution of aircraft exhaust—whether fresh or resuspended—are provided by Br/Pb ratios from XRF analysis and Pb isotope ratios from ICP-MS analysis. -92-

5.6.1 PM2.5 Br/Pb Ratios Halide compounds such as ethylene dibromide and ethylene dichloride are part of the tetraethyl lead (TEL) additive blended into avgas. These compounds scavenge Pb in the engine and the resulting exhaust emissions are bromolead compounds such lead bromide (PbBr2) and lead bromochloride (PbBrCl). There is only one supplier of the TEL additive (Innospec) and it supplies TEL-B, which contains only ethylene bromide (http://www.innospecinc.com/market/octane-additives). The Br/Pb ratio in ambient particulate matter has long been used as an indicator for combustion of leaded fuels and, in particular, motor gasoline (Harrison and Sturges 1998). A mixture of brominated and chlorinated additives was most commonly used in motor gasoline and the Br/Pb ratio of 0.386—which corresponds to the compound PbBrCl and is commonly termed the “ethyl ratio”—was used as a reference ratio. If brominated but not chlorinated additives are used, then a Br/Pb ratio of 0.772, corresponding to the compound PbBr2, would be the reference. There are several caveats to use of the Br/Pb ratio as an indicator for combustion of leaded fuels. Most importantly, bromine is relatively volatile and volatility losses can lead to a decrease in the Br/Pb ratio. The loss can occur during transport in the atmosphere, during sample collection and storage, or during analysis using high energy beams such as XRF. For this study, Br/Pb ratios were examined for evidence of Pb from piston-engine aircraft exhaust emissions. Br losses during atmospheric transport of fresh exhaust emissions are likely small because the time between emission and sample collection is on the order of seconds to minutes. Br losses from resuspended exhaust PM could have substantial losses. Precautions were taken to minimize losses after sample collection by transporting in coolers and storing in a freezer. A subset of the airborne PM2.5 samples collected at each airport was sent to Cooper Environmental Services for elemental analysis by XRF. Details are provided in Appendix B, including a comparison of PM-Pb by XRF and ICP- MS. Figure 28 shows the relationship between PM2.5-Br and PM2.5-Pb as measured by XRF. All but one sample are above a distinct edge with Br/Pb ~ 1/3. At low Pb concentrations, the samples can be enriched in Br. Figure 29 shows the same data after classifying each sample as having expected high or low aircraft exhaust impacts. First, concentrations less than three times the XRF MDL values were screened out for this analysis. Next, expected high-impact samples were identified on an airport-by-airport basis. At RVS, expected high-impact samples were those collected at either the North or South sites, depending on the wind pattern. Samples from the East site or upwind of the airport based on the daily winds were categorized as low impact. Samples collected at RVS on days with variable winds are not included in Figure 29. At APA, all samples from the Central site were categorized as high impact while samples from the East site were considered low impact. At SMO, all samples from the Northeast site were categorized as high impact and the samples from the Southwest site were considered low impact. Br and Pb are highly correlated for the high-impact samples (r = 0.92). In contrast, Pb was weakly correlated with markers for resuspended soil, especially for the high-impact samples (r = 0.00 and 0.01 for Si-Pb and -93-

Figure 28 PM2.5 Br and Pb Measured by XRF and Stratified by Airport Figure 29 PM2.5 Br and Pb Measured by XRF and Stratified as Samples with High or Low Expected Impacts from Aircraft Exhaust -94-

Ca-Pb, respectively). The strong correlation of Pb with Br and weak correlation of Pb with soil markers such as Ca and Si provide support that the PM2.5-Pb originates from combustion of leaded fuel. The downwind Br/Pb ratio was then adjusted to correct for Br and Pb background contributions. The adjusted ratios were calculated as: The median adjusted downwind Br/Pb ratio is 0.30 with 25th and 75th percentile ratios of 0.24 and 0.39, respectively. These ratios are much smaller than the expected ratio of 0.772 if all the lead was present as PbBr2. It is also less than the ethyl ratio. This discrepancy may be caused by volatilization of Br during the XRF process. Chlorine was also examined to determine if Pb was in the form of PbBrCl, and it was poorly correlated with Pb and Br; however, volatilization of Cl could also be an issue. 5.6.2 PM-Pb Isotopic Composition Pb isotopes have long been used to identify sources of PM-Pb and in some cases to quantify the source contributions. Pb isotopes are stable, but fractionation that has occurred over geologic time leads to distinct isotopic signatures for Pb of different origins. In this study, if the Pb isotope ratios for both the native soil and other “background” Pb sources are different from the isotope ratios for Pb used in leaded fuels, these ratios can be used to evaluate the prevalence of PM-Pb from leaded fuel combustion. Given that lead was phased out of motor gasoline several decades ago, a leaded fuel signature at general aviation airports should be a good indicator for piston- engine exhaust emissions. While this approach cannot distinguish between direct exhaust emissions and exhaust emissions that have locally deposited and are subsequently resuspended, it may be able to discriminate Pb from avgas compared to other sources. A common approach to examining Pb isotopes data is to make a scattergram of the 207Pb/206Pb versus 208Pb/206Pb ratios. The isotopic composition of airborne PM-Pb is a simple mixture of Pb from different geologic sources (whether native soil, or mined and refined such as the Pb in avgas). Emission source compositions are “end members” and the source contributions to Pb in airborne PM are estimated by the PM-Pb sample location along the mixing line between the end members. This construct is sound if there are only two end members, which is often the case, but becomes more complicated in the presence of three or more end members. All PM-Pb samples were analyzed for Pb isotopes using the protocol described in Appendix B. While precise quantification of Pb isotopes is best performed using a high resolution ICP-MS, previous work conducted by the WUSTL team has determined that the ICP-MS available to this project was adequate for at least semi-quantitative analyses and the measurement precision has been quantified. upwinddownwind upwinddownwind Pb-Pb Br-Br = Pb Br -95-

Figure 30 shows the 207Pb/206Pb versus 208Pb/206Pb ratios for all airborne PM samples collected in this study with the data stratified by airport. All of the samples lie along a line, albeit it with some scatter, which generally supports the notion of a two-source model for Pb. Four resuspended soil samples from each airport and 15 total avgas samples were also analyzed. Figure 30(b) shows the 207/206Pb versus 208/206Pb for all airborne PM samples and the soil and avgas samples collected for each airport. The avgas ratios are similar to measured ratios of lead mined in Australia, which is reported to be the source of the lead used in avgas (Townsend et al., 1998). Isotope ratios for resuspended soil are generally consistent with crustal material in the continental United States and are more similar to each other than to Australian lead. Therefore, samples with high Pb ratios likely contain more avgas contribution while samples with low ratios are likely indicative of background ambient Pb and in particular resuspended soil. It is not clear whether the soil samples include significant Pb originating from the use of avgas. However, isotopic compositions of soil samples collected at different locations within and between the three airports are indistinguishable. This suggests that avgas Pb does not dominate the Pb in these soils. A potential confounder to the soils analysis is that the airport topsoil samples can have dramatically different histories, and in some cases there was evidence of soil being moved. At high 208Pb/206Pb ratios, the APA data tend to fall below the mixing lines for the RVS and SMO data. Figure 30 Pb Isotope Ratios for Airborne PM-Pb, Soil, and Avgas Samples Collected at the Three Airports Figure 31 shows the same data but now stratified into samples expected to have low or high impacts from piston-engine aircraft emissions. Concentrations less than three times the ICP-MS MDL for Pb were screened out for this analysis, and the classification -96-

scheme used for the Br/Pb ratios analysis was used for this analysis. Most of the high- impact samples are clustered towards the avgas end member, whereas most low-impact samples are at greater distances from the avgas end member. This pattern is consistent with Pb in high-impact samples being dominated by avgas combustion. The threshold composition between high- and low-impact samples is 208Pb/206Pb ~ 2.15. All of the low- impact samples with 208Pb/206Pb above this threshold are from APA. Figure 31 Pb Isotope Ratios for Airborne PM-Pb with Samples Stratified as High or Low Expected Impacts from Aircraft Exhaust Data from collocated PM sampling can be used to gauge measurement precision. Shown in Figure 32 are the 208Pb/206Pb ratios and 207Pb/206Pb ratios for the collocated airborne PM samples collected at the three airports. Collocated sample isotope ratios are highly correlated, with r2 = 0.90 for 208Pb/206Pb and r2 = 0.94 for 207Pb/206Pb. While the data are highly correlated, the high scatter suggests that even if the end member compositions are appropriately identified, caution should be used when quantitatively apportioning airborne PM-Pb to the end members. Again, this scatter is likely a limitation of not using a high resolution ICP-MS instrument. -97-

Figure 32 Pb Isotope Ratios Collocated PM Samples Collected at the Three Airports (a) (b) Correlation was also observed between Pb isotope ratios and total lead concentration. Figure 33 shows total Pb concentration versus the 208Pb/206Pb ratio for airborne PM samples collected at the three airports. High Pb concentrations correspond to high 208Pb/206Pb ratios, suggesting that lead from avgas combustion is the primary driver of the high PM-Pb measured at each airport. Some of these data represent paired PM2.5 and TSP samples. For the high-impact data, there was no consistent trend in the directional difference of the PM2.5-Pb and TSP-Pb isotope ratios and the differences were generally small. This pattern suggests that significant Pb in coarse particles at the high-impact sites originates from previously deposited TEL-Pb. 5.7 Comparison of SMO Data to Previous Studies In addition to this ACRP study, there have been two other recent studies of PM-Pb at SMO. The South Coast Air Quality Management District (SCAQMD) conducted a study in 2006-2007 to assess the impact of airport operations for a suite of air pollutants, including PM-Pb (SCAQMD 2010). A 2009 study performed by ICF International for EPA (Carr et al. 2011) collected airborne PM-Pb samples towards evaluating an air quality modeling approach for local-scale impacts from general aviation activity. In this section, the results from the SCAQMD and ICF studies are compared to the results from this ACRP study. -98-

Figure 33 Pb Total Concentration versus the 208Pb/206Pb Ratio for PM Samples Collected at the Three Airports Table 44 summarizes the key attributes for each of the three SMO studies. Both the SCAQMD and ICF studies performed wintertime and summertime campaigns, and in both cases PM-Pb concentrations were higher in the winter than the summer. Comparisons to the ACRP study use only the summertime data because the ACRP study was conducted in July 2013. Both the SCAQMD and ICF studies primarily focused on TSP-Pb sample collection using FRM High-Volume (Hi-Vol) samplers. This ACRP study focused primarily on PM2.5-Pb measurements and secondarily on TSP-Pb measurements because the objective is to refine the inventory methodology for aircraft exhaust emissions, which are fine particles. BGI PQ100 samplers were used for both size ranges because they are portable and battery operated, which provided flexibility for siting and moving the samplers. ICF also used Mini-Vol samplers during the first, wintertime campaign. The Mini-Vol is attractive because it is a portable, battery-operated sampler. However, its performance for TSP-Pb measurements was deemed unacceptable and therefore only Hi-Vol samplers were used in the second, summertime campaign. During the SCAQMD summertime campaign, a BGI PQ100 sampler was used to collect TSP-Pb at a site near the runway to overcome the need for electrical service. MetOne SASS samplers were also deployed at multiple sampling locations to collect PM2.5 samples. -99-

Table 44 Measurement Attributes for the Three Airborne PM-Pb studies at SMO Parameter Study SCAQMD1 EPA/ICF2 ACRP Sampling Period April-July 2006 (12 wks) July 2009 (1 wk) July 2013 (4 wks) Sampling Frequency 1-in-3 day Daily Daily Sampling Duration 24-hour 24- and 16-hour 12-hour Sampling Locations 7 4 4 PM Size Sampled (Primary Size/Other Size) TSP / PM2.5 TSP / PM2.5 PM2.5 / TSP Primary PM Samplers3 Other PM Samplers Hi-Vol FRM SASS, PQ100 Hi-Vol FRM PQ100 Total Samples for Primary Size 191 28 77 Primary Size Samples Collected at Site Northeast of Runway 214 12 7 27 Analytical Method (Primary Method /Other Method) ICP-MS / XRF XRF ICP-MS / XRF ATADS-reported daily mean aircraft activity LTOs) 385 (April-July 2006) 290 (July 25-31, 2009) 293 (July 3-31, 2013) 1 Measurement attributes for the SCAQMD summertime campaign only. 2 Measurements attributes for the ICF summertime campaign only. 3 PM sampler manufacturers: Anderson Hi-Vol, MetOne SASS, BGI PQ100. 4 Does not include collocated sampler data. Finally, both the SCAQMD and ICF studies primarily collected 24-hour samples while this ACRP study collected 12-hour samples to focus on periods with highest aircraft activity. The SCAQMD and ICF studies included sampling locations outside the airport footprint while this ACRP study included locations only within the footprint. Figure 34 shows the site locations at or immediately adjacent to the airport and including only the summer period sites from the ICF study. The SCAQMD and ICF studies had additional sampling locations in residential areas around the airport, some of which are not shown on the map. All three studies had sampling locations to the northeast of runway 21 (designated NE in ACRP, and East Tarmac in SCAQMD and ICF) as well as near the maintenance shed (designated West in ACRP). The ACRP and SCAQMD studies both had sampling locations west of runway 3 (designated SW in ACRP, West Tarmac in SCAQMD). For clarity, the remainder of this section uses only the ACRP site designations. The NE site zone is downwind of the activity on runway 21 for prevailing -100-

Figure 34 Sampling Locations for the SCAQMD, ICF, and ACRP PM-Pb Studies winds during the summertime. However, the precise locations are not identical across the three studies and this could lead to significant differences in measured concentrations because this is a zone of steep concentration gradients. PM-Pb concentration data are summarized in Table 45. Since samples from the ACRP study were primarily PM2.5, the measured PM2.5-Pb concentration values were multiplied by a factor of 1.3, which was the mean ratio of TSP-Pb to PM2.5-Pb for simultaneously collected TSP and PM2.5 samples. For the SCAQMD study, the TSP-Pb to PM2.5-Pb ratio ranged from 0.68 to 1.20; however, different sampler types, and possibly different analytical methods, were used for the TSP and PM2.5 samples. For all three studies, PM-Pb concentrations measured on the airport footprint were highest at the Northeast site, with lower and similar concentrations at the West and Southwest sites. For the SCAQMD and ICF studies, PM-Pb concentrations at monitoring locations off the airport footprint were similar or less than the concentrations at the Northeast site. Mean and Median TSP-Pb at the Northeast site were highest for the SCAQMD study, intermediate for the ICF study, and lowest for the ACRP study. Differences in aircraft activity may explain some of the differences. Overall airport activity as measured by ATADS was 11% higher for the SCAQMD study than the ICF and ACRP studies, which had similar levels of activity. As summarized below, there are several factors that potentially confound the comparison across studies. -101-

Table 45 TSP-Pb (ng/m3) Summary Statistics for the SCAQMD, ICF, and ACRP Summertime Studies Site Northeast West Southwest Study SCAQMD1 ICF2 ACRP3 SCAQMD ICF ACRP SCAQMD ICF ACRP Number of Samples 12 7 27 32 7 5 31 - 28 Mean 85 49 39 4 3 3 5 - 2 Median 84 53 36 3 4 2 4 - 1 Maximum 135 62 93 14 6 10 18 - 6 Minimum 27 34 <MDL 0 <MDL <MDL 0 - <MDL Note: ACRP TSP-Pb estimated as 1.3 times the measured PM2.5-Pb; <MDL = concentration below the minimum detection limit. 1 Source: South Coast Air Quality Management District 2010 2 Source: Carr et al. 2011 3 Current study • First, ACRP sampling was conducted during the daytime 12 hours to collect PM only during periods with high aircraft activity. SCAQMD and ICF sampling durations were primarily 24 hours; the inclusion of nighttime hours would mix in periods with lower activity but perhaps also lower dispersion; therefore, the overall impact on the reported average concentrations is not clear. • Second, based on the ICF and Sierra modeling, the sampling zone northeast of runway 21 is an area with steep PM-Pb concentration gradients. Thus, even modest differences in sampling location can have a large impact on observed PM- Pb levels. The ACRP northeast site was ~20 m north of the SCAQMD and ICF sites to be farther away from the blast wall. • Third, there are likely differences in the actual TSP collection efficiencies between the different samplers. For example, the Hi-Vol TSP sampler has an inlet collection efficiency that depends on the sampler orientation with respect to the wind. • Finally, there can be analytical biases. For example, the SCAQMD study included PM sample collection at the Northeast site using a BGI PQ100 with a TSP inlet and operated at 12 LPM. Pb as measured by both XRF and ICP-MS were highly correlated but with an absolute bias of ~25 ng/m3 with XRF higher than ICP-MS. A similar comparison was made with 19 PM2.5-Pb samples collected during the ACRP study. In this case, the data were highly correlated but with a relative bias of ~28% with the XRF higher than the ICP-MS. -102-

Despite these potential confounders, the PM-Pb concentration data are generally consistent across the three studies. While the measurements are not directly comparable to the Pb NAAQS because of differences in data collection averaging times, it is noted that none of the summertime 12- or 24-hour samples had a TSP-Pb concentration above the three-month average NAAQS of 150 ng/m3 and study mean concentrations were 25- 57% of that value. 5.8 Key Observations Field studies were conducted at three general aviation airports that each had distinguishing features in terms of airport layout and meteorology. Airborne PM-Pb samples were collected with high completeness, detectability, and precision. Overall, aircraft activity patterns and PM-Pb patterns were similar across the three airports. The patterns outlined below were observed for all three airports. • Aircraft LTOs measured by the video cameras were lower than the FAA ATADS counts. There may be several reasons for this discrepancy, including differences in counting methods and helicopter activity that is not necessarily well-captured by the video data. • A small number of aircraft disproportionately contributed to total operations with 5% of the observed aircraft conducting one-third of the operations and 10-12% of the observed fleet conducting half of the operations. • Run-up operation TIM was more variable than landing-and-takeoff TIM. PM-Pb hot spots tend to be downwind of run-up areas, so the run-up TIM variability will lead to variability in the hot spot intensity. • PM-Pb concentrations were downwind of aircraft ground operations and especially downwind of run-up areas. • Most of the PM-Pb at the high-impact sites is in the PM2.5 size range, consistent with direct exhaust emissions from piston-engine aircraft, but there is considerable Pb in the PMTSP-2.5 size range. The median TSP-Pb/PM2.5-Pb ratio across all three airports was 1.3 with a narrow interquartile range of 1.2 and 1.4 for the 25th and 75th percentiles, respectively. While coarse mode particles are not the dominant contributor to PM-Pb, they cannot be neglected. • PM-Br and PM-Pb are highly correlated at the high-impact sites, which is consistent with TEL-Pb origins. The Br/Pb ratio is much lower than predicted by the presumed form of PbBr2. There are several possible reasons for the lower than expected ratio, which was also commonly observed in studies during the era of leaded automobile gasoline. • The Pb isotopic compositions for PM samples collected at sites with expected high impact from TEL-Pb are distinct from those for samples collected at sites -103-

with expected low impacts. Furthermore, PM-Pb isotopic compositions for the high-impact sites are consistent with avgas samples collected at the airports, while the isotopic compositions for the low-impact sites are generally consistent with resuspended soil samples collected at the airports. • TSP-Pb at the high-impact sites does not systematically exhibit a shift in isotopic compositions towards background (e.g., resuspended soil) Pb compositions compared to the corresponding PM2.5-Pb samples. This pattern suggests that the lead in the coarse particle size mode is strongly influenced by TEL-Pb. PM data collection focused on size ranges and averaging times that do not support direct comparisons to the Pb NAAQS. Nonetheless, none of the individual 12-hour PM-Pb values exceeded the three-month average NAAQS of 150 ng/m3, and the highest observed 12-hour concentration was a TSP-Pb value of 72 ng/m3 at SMO. Study-average PM2.5-Pb values at the highest concentration sites were 15 ng/m3 at APA, 21 ng/m3 at RVS, and 30 ng/m3 at SMO. ### -104-