Atmospheric Environment 139 (2016) 20e29
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Ultrafine particle size as a tracer for aircraft turbine emissions Erin A. Riley a, Timothy Gould b, Kris Hartin a, Scott A. Fruin c, Christopher D. Simpson a, Michael G. Yost a, Timothy Larson a, b, * a
University of Washington, Department of Environmental and Occupational Health Sciences, Box 357234, Seattle, WA, 98198, United States University of Washington, Department of Civil & Environmental Engineering, Box 352700, Seattle, WA, 98198, United States c University of Southern California, Keck School of Medicine, SSB 225F, MC 9237, 2001 N Soto Street, Los Angeles, CA, 90032, United States b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Spatial measurements of ultrafine particle number (PN) and diameter in two cities. PN concentrations more than 5 km from airport are similar to those on freeways. Spatial distribution of mean ultrafine particle diameter is distinct near airport. Ratio of PN to black carbon is higher beneath approach path than elsewhere.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 13 January 2016 Received in revised form 6 May 2016 Accepted 8 May 2016 Available online 10 May 2016
Ultrafine particle number (UFPN) and size distributions, black carbon, and nitrogen dioxide concentrations were measured downwind of two of the busiest airports in the world, Los Angeles International Airport (LAX) and Hartsfield-Jackson International Airport (ATL e Atlanta, GA) using a mobile monitoring platform. Transects were located between 5 km and 10 km from the ATL and LAX airports. In addition, measurements were taken at 43 additional urban neighborhood locations in each city and on freeways. We found a 3e5 fold increase in UFPN concentrations in transects under the landing approach path to both airports relative to surrounding urban areas with similar ground traffic characteristics. The latter UFPN concentrations measured were distinct in size distributional properties from both freeways and across urban neighborhoods, clearly indicating different sources. Elevated concentrations of Black Carbon (BC) and NO2 were also observed on airport transects, and the corresponding pattern of elevated BC was consistent with the observed excess UFPN concentrations relative to other urban locations. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Ultrafine particles Airport emissions Mobile monitoring Ultrafine particle size Black carbon Nitrogen dioxide
1. Introduction The largest impacts of jet engine emissions on neighboring residential areas has until recently been thought to be primarily
* Corresponding author. University of Washington, Department of Civil & Environmental Engineering, Box 352700, Seattle, WA, 98198, United States. E-mail address: [emailprotected] (T. Larson). http://dx.doi.org/10.1016/j.atmosenv.2016.05.016 1352-2310/© 2016 Elsevier Ltd. All rights reserved.
limited to areas within about a kilometer of edge of the runway. Characterization of air pollutants from aircraft traffic and airport activity has been primarily conducted at fixed sites located within 2 km of the airport runways (Carslaw et al., 2006; Hsu et al., 2013; Hu et al., 2009; Lai et al., 2013; Lobo et al., 2012; Masiol and Harrison, 2014, 2015; Westerdahl et al., 2008; Zhu et al., 2011) but mobile monitoring has also been used to confirm the area-wide impact of emissions from airport activities on near-airport neighborhoods within a few kilometers (Choi et al., 2013; Hsu et al.,
E.A. Riley et al. / Atmospheric Environment 139 (2016) 20e29
2014). Recently, a mobile monitoring campaign was conducted in the approach path of LAX where a spatially dense sampling scheme uncovered a much larger area of impact covering 60 km2 and extending 20 km downwind from the airport, the largest monitoring area effort to date (Hudda et al., 2014). The authors reported increased ultrafine particle number concentrations (UFPN) of 4e5 times normal at distances of 8e10 km from the airport on multiple independent days of sampling. Subsequently, fixed-site downwind measurements conducted at Schiphol airport in the Netherlands and supported by Gaussian plume models have shown substantial impacts of airport emissions on UFPN concentration extending >8 km from the airport (Keuken et al., 2015). With its ability to cover large areas, mobile monitoring may be the most effective means to refine the current understanding of the impact that aircraft (and airport) emissions have on air quality in a variety of urban settings. Here we present an analysis that uses mobile monitoring data from Atlanta, GA and Los Angeles, CA to compare the spatial distributions of ultrafine particle metrics derived from three different mobile sampling strategies: downwind airport transects perpendicular to flight paths, on-road highway routes, and repeated measures at fuzzy points (neighborhood sampling centered on an intersection and utilizing a clover leaf approach to include all upwind and downwind conditions). This wide area approach provided a novel verification of the wide area impacts found by Hudda et al. (2014), and demonstrated the relatively high UFPN impacts of airports compared to freeways, which are usually considered to dominate UFPN concentrations in most urban areas in the U.S.
2. Methods Ultrafine particle metrics included: PN10 (in Los Angeles only; UFPN, diameter > 10 nm), PN25 (UFPN, diameter > 25 nm), PN50 (UFPN, diameter > 50 nm), DpN (mean number diameter from PN25 measurement), and the ratio of PN25/PN50. We also simultaneously measured nitrogen dioxide (NO2) and black carbon (BC). Instruments and their reporting limits are listed shown in Table 1. Instruments were checked or calibrated for zero and span (NO2 analyzer) in the field periodically. Particle number instruments were previously evaluated by traceable methods at their respective manufacturers: the P-Trak was certified in May 2011 measuring a particle count that averaged 100.4% of test standard particle counts determined by a TSI 3080 Electrostatic Classifier and two TSI 3010 Condensation Particle Counters, and the NanoCheck 1.320 was equipped with a new aerosol detector by the manufacturer Grimm Technologies which confirmed the device to be working within specifications on July 30, 2012. More detailed discussion of the mobile platform can be found in the supplemental material. Downwind airport transects were selected along residential neighborhood streets to minimize the impact of roadway traffic on the observations. The transects are expected to have distributions of traffic related air pollutant (TRAP) concentrations similar to the
21
distributions captured by the fuzzy point (FP) monitoring, which were located on both high and low traffic-impacted roads. In LA, the transect locations were within the sampling area previously reported by Hudda et al. (2014), where a raster-pattern sampling scheme was used to detect a gradient in the 5th percentile of rolling 1 s UFPN measurements (over 30 s intervals) beneath the approach path of aircraft compared to adjacent neighborhoods. Finally, highway monitoring was meant to capture TRAP concentrations and characteristics associated with the most trafficked roadways in these cities. The comparison between the results of these three sampling strategies allowed us to investigate whether neighborhoods downwind of airports experience different air pollutant impacts compared to other urban locations that encompass a distribution of TRAP impacts.
2.1. Sampling methodology Sampling consisted of highway, airport transect, and FP monitoring. FP locations were clusters of data ~600 m across, composed of a clover leaf pattern that produces more robust average and distributional concentrations of a given location by including all directions and orientations to surrounding traffic (Larson et al., 2009). We therefore define the term ’fuzzy point‘ or “FP,” to include data points encompassed within a circle 300 m in radius centered on a given intersection. For each FP we obtained mobile data at 10 s resolution for a total of ~6e10 min during each transit, thus, the mobile monitor captures a distribution of values assigned to this location. FPs are selected in residential urban neighborhoods encompassing a range of traffic characteristics from low traffic residential streets, to arterials with mixed traffic composition, to near highway locations (see Figs. 1 and 2). The 43 FPs were divided among three routes, with the airport transects and highway sampling comprising the fourth route. Routes were driven 3e4 times, with the order of sites visited reversed in direction for successive transits through each route. Sampling in Los Angeles took place between June 14th, 2013 and July 1st, 2013 and was divided into four routes with one route driven per day between the hours of 12:00e19:00 local time. Airport transects and freeways were sampled on June 22nd, 27th, and July 1st and an additional day of highway-only driving was performed June 14th. Sampling in Atlanta took place during a shorter time period between the 8th and 17th of September 2013 with airport transects sampled on September 9th, 11th, 13th, 14th, and 17th. The complete sampling space for the LAX and Atlanta campaigns is depicted in Figs. 1 and 2. Highway sampling in Atlanta took place on September 10th and 15th; highway sampling data were more limited in spatial extent in Atlanta than in LA because it was not a design priority for the former campaign. In Atlanta additional highway data were selected manually by defining a polygon around the urban-center highways in Google Earth; points within the polygon were added to the highway sampling data set in Atlanta from the remaining sampling days. FP sampling in Atlanta
Table 1 Mobile platform instrumentation. Parameter
Instrument
Manufacturer
Measurement range
PN25 concentration and mean diameter (25e400 nm) PN50 number concentration (50e1000 nm)a PN10 number concentration (10e1000 nm) (Los Angeles only) Black carbon (BC) NO2 Positioning & real-time tracking
NanoCheck 1.320 P-Trak 8525, with particle diffusion screens CPC model 3007 (ethanol based) Micro-Aethalometer AE52b Cavity Attenuated Phase Shift (CAPS) monitor GPS Receiver BU-353
GRIMM TSI TSI AethLabs Aerodyne Research US GlobalSat
0e4 106 particles/cm3 0e5 105 particles/cm3 0e1 105 particles/cm3 0e1 mg/m3 0.1e3000 ppb 10 m accuracy
a b
Normal P-Trak lower size limit without screen is 20 nm, the screen captures 20e50 nm particles by diffusion to provide additional independent size resolved count data. MicroAeth® Model AE51, custom modified by AethLabs for dual wavelength acquisition.
22
E.A. Riley et al. / Atmospheric Environment 139 (2016) 20e29
Fig. 1. Map of Los Angeles mobile monitoring data for three pollutant metrics. Data were spatially binned into a grid with 300 m 300 m cell size, and the median was calculated for each bin. Color scale represents quantiles of the resultant spatial bins. Original PN25 concentrations have been adjusted by aligning data collected on separate days using the daily 5th percentile. PN25 concentrations presented are the incremental concentration change from the 5th percentile; there are no negative measurements in the original data. Particle size and PN25/PN50 are calculated with the original (unaligned) data. Figures a,c, and e: The solid blue box encloses the airport transects, solid blue circles indicate airport FPs. Dashed blue lines leading to LAX airport indicate approximate approach paths to the two landing strips (see supplemental Figure S2). Figures b,d, and f: The black box indicates the zoomed-in spatial region displayed in Figures a,c, and e.
E.A. Riley et al. / Atmospheric Environment 139 (2016) 20e29
23
Fig. 2. Map of Atlanta mobile monitoring data for three pollutant metrics. Data were spatially binned into a grid with 300 m 300 m cell size, and the median was calculated for each bin. Color scale represents quantiles of the resultant spatial bins. Original PN25 concentrations have been adjusted by aligning data collected on separate days using the daily 5th percentile. PN25 concentrations presented are the incremental concentration change from the 5th percentile; there are no negative measurements in the original data. Particle size and PN25/PN50 are calculated with the original (unaligned) data Figures a,c, and e: The solid blue box encloses the airport transects and the dashed blue lines leading to ATL airport indicate approximate approach paths to the two landing strips (see supplemental Figure S3). Figures b,d, and f: The black box indicates the zoomed-in spatial region displayed in Figures a,c, and e.
24
E.A. Riley et al. / Atmospheric Environment 139 (2016) 20e29
consisted of overlapping clusters of FPs arrayed in a gradient at increasing distance from freeways along with 20 dispersed FPs. The spatial extent of the FP, highway, and airport transect data can been seen in Fig. 2. For both cities the airport transects consisted of low traffic residential roadways that were selected to minimize the impact from mobile sources. Maps of the airport transect data are provided in the supplemental material (Fig. S1). A detailed description of the methods for sample collection, sample analysis and statistical treatment of the data can be found in the supplemental material. 2.2. Mobile platform temporal adjustments A central challenge to the use of mobile monitoring data for spatial analysis is adjusting for (or removing) temporally-varying background contributions to the measurements, such as day to day differences in meteorology that affect the entire region. For this analysis we derived distributional properties (e.g. 5th, 25th, 50th, 75th and 95th percentiles) of the pollutant concentrations measured by the mobile platform. To allow more direct comparisons across locations, between-day temporal trends were accounted for by aligning the data to the daily 5th percentile concentrations. The 5th percentiles tend to reflect regionally consistent concentrations when no localized source contributions are present. We use this value as an “estimated urban background”. This adjustment was particularly important since the locations were sampled on different days (refer to supplementary material for additional details and sensitivity tests). The daily 5th percentile was first calculated from all available mobile platform data, including time periods when the platform was in transit between selected sites and then subtracted from all the measurements. The two exceptions involved variables other than concentration where the 5th percentile does not necessarily represent an urban background, specifically the measured mean particle diameter and PN25/PN50.
mean diameter observed in Los Angeles occurs in neighborhoods beneath the aircraft approach path. The ratio of PN25/PN50 shows a similar spatial pattern, where the airport jet pathway and airport FPs (enclosed by 6 blue circles in eFig. 1a and c) all have percentiles above 75%, corresponding to a prevalence of smaller diameter particles in this location compared to both the freeway and FP locations. The spatial extent of the Atlanta campaign is shown in Fig. 2aef. As in Los Angeles, elevated concentrations of PN25 were measured along the near airport transects relative to the FP sampling (Fig. 2a) but the highest PN25 concentrations were observed on the freeways (Fig. 2b). Particle diameters in the airport transect were in the smallest 25th percentile (90 nm). In Atlanta the smallest diameter particles (