Dispersion Modeling Guidance for Airports Addressing Local Air Quality Health Concerns (2017)

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.

13 Some of the dispersion models described in this chapter have been used in modeling studies of local airport air quality, particularly for NOx, NO2 and CO, in addition to the intercompari- son study for this project. This chapter summarizes key findings from individual studies that highlight some differences in model performance. More information about these studies can be found in the contractor’s final report for ACRP Project 02-58, which can be accessed using a link on the project webpage. ADMS-Airport, EDMS-AERMOD, and LASPORT (Lagrangian Simulation of Aerosol-Transport for Airports), which were not specifically used in ACRP Proj- ect 02-58, were all designed to be used for airports (Janicke, Fleuti, and Fuller 2007). The Inter- national Civil Aviation Organization Committee on Aviation Environmental Protection (ICAO CAEP) approved all three models for assessing air quality impacts of airport emission sources (International Civil Aviation Organization 2011). Generally, applications of these three models are country-specific: ADMS-Airport is the preferred model at London Heathrow Airport and London Gatwick Airport in the United Kingdom; LASPORT is generally used at airports in Switzerland and Germany; and AEDT is designated as the required model by the FAA for assessing air quality impacts of airports within the United States. With the exception of some analyses at Heathrow Airport (Carruthers et al. 2007), no com- prehensive model evaluations have used the same airport and sources, meteorology, and moni- toring data until this ACRP-sponsored model intercomparison at LAX. Most evaluations have been of individual models at a single airport. The nature or quality of a model evaluation depends strongly on the concentration averaging time, given that the effects of turbulence and wind direction variability can fluctuate significantly across time periods of different lengths. Large turbulence intensities and variability lead to large concentration fluctuations and statistical variability for short (e.g., 1-hour) averaging periods in contrast to the smaller fluctuations associated with longer (e.g., annual) averaging periods. Evaluations for short averaging times are negatively affected by the large variability. Finally, the airport location, surrounding terrain, and local meteorology and climatology also affect the quality of a model evaluation. More complex terrain and meteorology lead to more uncertainty in the meteorological inputs to models and a larger variance between model predictions and observations. In a model intercomparison study at Heathrow Airport that included ADMS-Airport, EDMS and LASPORT, ADMS-Airport’s annual average NOx predictions were closest to actual mea- surements (Carruthers et al. 2007). The good performance for ADMS-Airport was attributed to the model’s more robust treatment of plume buoyancy and dispersion for jet aircraft exhaust plumes compared to that used by the other models. Also, the annual mean NOx concentra- tions for these three models varied by a factor of 2.1 despite their use of the same emissions and meteorological data. These results were based on model comparisons with observations at nine C h a p t e r 3 Airport Modeling Studies

14 Dispersion Modeling Guidance for airports addressing Local air Quality health Concerns NOx monitoring stations around Heathrow over a year. In addition, modeled airport NOx con- centration contributions were approximately 1.7 times too high at a single monitoring station north of the northern runway. It is unclear how the apportionment analysis was conducted and if it applied to the annual average concentrations. In a more recent study, EDMS-AERMOD performed equally as well as ADMS-Airport in matching the observed annual-averaged NOx and NO2 concentrations around Heathrow Airport (Sabatino et al. 2011). Other studies have applied ADMS-Airport further at Heathrow Airport. One study examined ADMS-Airport predictions of NOx, NO2, and PM10 around Heathrow and found good model performance for each of the three pollutants (Carruthers et al. 2008). Another study found that ADMS-Airport accurately represented the overall pattern of NOx concentrations around Heath- row with the treatment of aircraft exhaust as buoyant jets (as opposed to volume sources), which resulted in significantly different results at receptors nearest the airport (Carruthers et al. 2011). At low wind speeds, however, predicted concentrations were too high, suggesting that the model overestimated the impact of nearby passive sources, such as those along the airport perimeter road. Such overpredictions also may have been caused by underestimates of turbulence fluctuations and wind variability under low wind conditions. In a fourth study using ADMS-Airport over several days at a busy regional airport, predicted NOx concentrations for complete aircraft traffic including aircraft trajectories were found to be satisfactory (Sarrat, Aubry, and Chaboud 2012). Prior to the development of ADMS-Airport, ADMS-Urban was applied at the Manchester and Heathrow airports. One study found that aircraft flying at altitudes between 200–1,000 m had minimum impact on ground-level concentrations (Peace et al. 2006). This finding high- lighted the importance of investigating the total contribution from many distributed sources to local air quality at an airport versus considering the airport as just one source. In the Heathrow Airport study, ADMS-Urban tended to overestimate the concentrations of NOx at the one moni- tor site considered (Farias and ApSimon 2006). Another study using LASPORT reported airport-attributable NO2 concentrations were typi- cally below 1 µg/m3 at locations 3 km or more away from the airport, and that major highways dominated the regional air pollution (Fleuti and Hofmann 2005). However, comparisons of mea- sured and modeled NO2 concentrations were mixed with model predictions at some locations that were well correlated with measurements, while others were not. For example, LASPORT underpredicted NOx concentrations at monitors near roadways, which were dominated by road traffic, but overestimated NOx concentrations near runways, which were dominated by aircraft activities. The overestimation may have been due to insufficient plume rise or issues with low winds and wind and turbulence variability. The study by Fleuti and Hofmann suggested that the emission factors for aircraft in actual operation were lower than the results of ICAO certification tests, a result that other studies have corroborated. Because it is a proprietary model, however, LASPORT was not used in the ACRP Project 02-58 study. Other work with EDMS-AERMOD has found a range of performance results, usually for short-term (1-hour) averaged concentrations. For example, one study examined lead concen- trations from aircraft piston engines at a general aviation airport near Santa Monica, California (Carr et al. 2011). This study showed that the model-to-monitor performance at two sites was good to excellent (within a factor of 2), particularly on 4 of the 6 modeled days. EDMS-AERMOD also was used for the LAX AQSAS. In an extensive evaluation of AERMOD using four measurement sites around LAX, model performance was reported as “generally fair to poor” (Tetra Tech 2013). In particular, AERMOD showed that greater than 50% of modeled values of NOx differed from observations by at least a factor of 2. This poorer performance rela- tive to the results at Heathrow was likely attributed to: (1) the shorter averaging time used at LAX (1 hour); (2) the more complex terrain at this coastline site, with generally more complicated

airport Modeling Studies 15 meteorology including land- and sea-breezes; and (3) a greater frequency of unstable conditions, with light and variable winds that would lead to higher concentration variability. Moreover, the neglect of plume rise for the aircraft sources probably led to high overpredictions of concentra- tions (> factor of 2). In a 3-day study comprising only 18 hours, a comparison of EDMS-AERMOD predictions of 1-hour. CO concentrations with observations around Washington Dulles International Air- port showed that the model frequently captured the hourly trend in the data but, overall, it underpredicted the measurements (Martin 2006). There was evidence that mobile sources (i.e., automobiles) were the largest CO contributor, but low measurements reported for many hours suggested either an underestimate of airport traffic emissions or possibly background and traffic sources were not included in the study. Another study found that EDMS-AERMOD performed quite well for annual average pre- dictions around Heathrow Airport (Sabatino et al. 2011). The EDMS-AERMOD results were determined to be within 20% for NOx and NO2 concentrations. Trends of hourly EDMS NOx predictions at Budapest Ferenc Liszt International Airport (Ferihegy Airport) also agreed well with these measurements (Steib et al. 2008). However, CO predictions in the 2008 Ferihegy Airport study were mixed and generally underestimated the peak CO observations, which was possibly due to CO transport from the nearby Budapest urban area, not included in the emis- sions data. Recent work reported in ACRP Report 135: Understanding Airport Air Quality and Public Health Studies Related to Airports found that PM2.5 dominated the overall health risk posed by airport emissions (Kim et al. 2012). Especially for aviation-attributable PM, considerations of chemistry could have significant implications on both the total PM mass and composition. The three models evaluated in these studies—LASPORT, EDMS, and ADMS-Airport—are limited in their consideration of atmospheric chemistry, as well as their treatment of fine particulate matter (PM2.5) by either including only primary PM2.5 (LASPORT, EDMS) or by including only some secondary PM2.5 formation pathways (ADMS-Airport). Recent efforts to quantify secondary PM formation from aircraft found that secondary organic aerosols (SOA) make up a significant amount of aircraft PM after a few hours of chemical pro- cessing (Miracolo et al. 2011; Woody et al. 2015). Furthermore, an additional study found that aircraft-attributable PM (which already accounted for secondary inorganic PM) was enhanced by up to 10% near the airport and 20% downwind (Riley et al. 2016). Recent studies, including one by Peters et al. (2016), have measured ultra-fine PM emissions on and near airports; however, these data have not been incorporated into or evaluated using airport air quality dispersion models. Findings from these studies indicate that aircraft produce particles—predominantly in the 10–20 nm size range—that are smaller than the particles pro- duced by other sources commonly found at airports. On-airport measurements found peak concentrations of these particles under arriving aircraft and behind aircraft taking off. Peak concentrations also were found off-airport downwind of the runways. A clear relationship exists between LTO operations, wind direction, and distance to the airport and the ultra-fine particle (UFP) concentration that is observed at monitoring sites around the airport. The contribution decreases with increasing distances, but effects were measurable at a distance of 7 km from the airport. In other recent studies, a 4- to 5-fold increase in particle number concentrations (PNC) was observed 8–10 km downwind of LAX, and a doubling of PNC was observed at a site 4 km down- wind of Boston’s Logan International Airport (Hudda et al. 2014; Hudda and Fruin 2016). In contrast, airport activity does not contribute more to black carbon, NOx and PM10 concentra- tions at monitoring sites than does traffic at nearby roadways.