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.

25 Our model evaluation and other analyses have identified several areas of improvement for dispersion models applied to aircraft and airport sources. They include: 1. Incorporation of background concentrations, 2. Representation of aircraft sources at the airport, 3. Inventory of ultra-fine particles (UFPs), 4. Plume rise from aircraft emissions, 5. Aircraft downwash effects on plume rise and dispersion, 6. Aircraft dispersion based on instantaneous “line puffs,” 7. Effects of light winds and atmospheric stability on dispersion, and 8. Other limitations, such as lack of chemical treatment in some models. This chapter presents a short description of each of these problem areas and offers some recommendations. Given that AEDT (coupled with AERMOD) is the current regulatory model for modeling aircraft emissions at airports in the United States, all the recommendations in this guidebook focus on this model. 6.1 Incorporation of Background Pollutant Concentrations The models used in this study capture the contributions to local air quality from airport sources; however, the regional background also contributes to air quality. To account for background contributions, estimates can be incorporated from a regional-scale model like the Community Multi-scale Air Quality (CMAQ) model, or ambient monitoring data from background sites can be used through a geostatistical technique, such as Space-Time Ordinary Kriging (STOK), which was developed to support exposure assessment studies (Arunachalam et al. 2014). The CMAQ model is a three-dimensional Eulerian (i.e., gridded) atmospheric chemistry and transport modeling system developed and maintained by EPA. (The term Eulerian refers to a fixed spatial coordinate system, which means the model is not Lagrangian and does not fol- low the plume or flow; the term is also used to refer to the solution scheme, which is based on solving equations that govern variables such as the wind, temperature, and concentration, on a fixed-grid system.) CMAQ was designed as a modular system, able to incorporate data from other models that have alternate mathematical processes, so it works well in concert with EDMS. CMAQ has multi-pollutant capabilities to address air quality issues such as O3, PM, and air toxics, all of which are known to have adverse health impacts. It is non-proprietary and includes emission, meteorology, and chemical modeling components. The chemistry-transport component can simulate chemical transformation and fate. CMAQ is a multi-scale system that C h a p t e r 6 Future Research Needs

26 Dispersion Modeling Guidance for airports addressing Local air Quality health Concerns can be applied over local, regional and continental areas with progressively finer resolution in a series of nested grids. With the temporal flexibility of the model, simulations can evaluate short-term (weeks to months) transport from local sources as well as longer-term (annual to multi-year) pollutant evolution. CMAQ’s atmospheric chemistry modeling capability and its ability to model large geographical areas are important aspects for using the model with EDMS to analyze airport air quality. NWS uses CMAQ to produce daily US forecasts for O3 air quality, and it is also used by states to assess implementation actions needed to attain EPA’s air qual- ity regulations. CMAQ can be obtained at no cost from the Center for Community Modeling and Analyses System (CMAS), which is hosted at the University of North Carolina at Chapel Hill. CMAQ has a large user community across the globe, and user support is offered through the CMAS Center. Although the research team did not use CMAQ in ACRP Project 02-58, it is discussed here as an alternate regional-scale model that can be used to estimate background concentrations. CMAQ was used in conjunction with AERMOD using a hybrid approach to predict total concentrations at an airport in a previous ACRP-funded study published as ACRP Report 71, and explained later in this chapter (Kim et al. 2012). CMAQ Details on CMAQ resources and products can be found at the following website: • https://www.cmascenter.org/cmaq/ 6.2 Source Characterization The evaluation of AERMOD and other models at LAX required the use of about 5,000 area sources to describe the aircraft plumes and trajectories. This was by far the largest source cat- egory used in the airport modeling, and it required the most computational resources and time. Each area source was modeled as a continuous release. The SCICHEM model had the second- order closure treatment and offered the most detailed chemical treatment for gas-phase and aerosol species that are of interest, but even without turning on the chemical mechanism and running SCICHEM in tracer mode, the computational burden was prohibitively large. Additional work is suggested to perform a range of sensitivity simulations to determine the level of detail adequate for accurately characterizing local-scale air quality and health. The researchers suggest exploration of these options: 1. Revisit the current approach for EDMS/AERMOD to create a uniform set of rectangular area sources to represent aircraft activity during landing and take-offs; 2. Consider aggregating these sources to have fewer numbers, and evaluate potential loss of information in characterizing local air quality and health; and 3. Consider alternate treatments. (Additional discussion of alternate treatments is provided in this chapter in the section on “line puffs.”) 6.3 Inventory of UFPs Concern about contributions to adverse health effects from UFPs (< 100 nm in size) from airport emissions is increasing. Currently, EDMS (and AEDT) inventories do not report UFP as a separate category. As described in the methodology for estimating PM emissions from aircraft, almost all particles emitted by aircraft are < 100 nanometers, with a small fraction ranging in size

Future research Needs 27 between 100–250 nanometers (Petzold et al. 1999; Wayson et al. 2008; Lobo et al. 2015; Brem et al. 2016; Ortega et al. 2016). However, current official guidance for the AEDT model refers to these emissions as PM2.5. It is suggested that future updates to AEDT explicitly include an inven- tory of ultra-fine particles from aircraft engines as a primary pollutant. 6.4 Plume Rise from Aircraft Emissions A major limitation in applying AERMOD to aircraft emissions is not accounting for plume rise from the engine exhaust. This leads to overestimation of ground-level concentrations (GLCs) for on-airport and near-airport measurements, as found in the June 2013 LAX AQSAS. The results showed that the modeled NOx GLCs at four monitor sites around LAX were roughly an order of magnitude higher than observations at the high end of the quantile-quantile (Q-Q) plots. The researchers found similar overpredictions of 1-hour NOx concentrations in ACRP Project 02-58, not only for AERMOD but also for the other models. A correction for plume rise would be needed to improve the GLC predictions. In AERMOD, the aircraft plume is modeled using a series of area sources that are treated as neutrally buoyant releases from a fixed height (the engine height zs above the surface). No option exists to treat distance-dependent plume rise, but the source height could be greater than zs. In view of this limitation, the researchers suggest that the area-source height be equal to the engine height plus the “final rise” of the aircraft plume or the “effective source height” as used in the Gaussian plume model. The final rise typically depends on the wind speed, stability, and engine heat or buoyancy flux, and separate final rise models can be used to address neutral, stable, and unstable atmospheric conditions. The researchers suggest consideration of three general plume rise models for aircraft emissions: 1. The existing model with a zero plume rise; 2. An empirical model for plume rise and initial spread based on the LIDAR measurements of Wayson et al. (2004 and 2008); and 3. A fluid-mechanical entrainment model (FEM). The FEM suggested in this guidebook is a more general and fundamental model for aircraft plume rise. From the FEM (presented in greater detail in the Contractors’ Final Report), sev- eral choices or options are available for using the plume rise in AERMOD and other dispersion models. The choices include: 1. An average rise (over time) at the aircraft starting position akin to the average rise obtained from LIDAR by Wayson et al. (2008); 2. An average rise over the ground roll; 3. The final rise based on atmospheric conditions that limit plume rise—stable stratification and atmospheric turbulence; and 4. A local or separate plume rise for each elemental area source along the aircraft trajectory. Further research and comparisons between model-predicted concentrations and observations are required to help determine the most appropriate plume rise prediction. 6.5 Aircraft Downwash Effects on Plume Rise and Dispersion Wing-tip (WT) vortices and downwash are generated by the pressure difference across the bottom and top wing surfaces, which leads to the rotational flow about the wings. (For more information, see Appendix G in the contractor’s final report). Once generated, the WT vortices

28 Dispersion Modeling Guidance for airports addressing Local air Quality health Concerns leave the wing as trailing vortices, which are long lived and can cause safety issues for nearby aircraft. Extensive literature examines WT vortices, or wakes, and their effects on engine exhaust dispersion and concentrations (Garnier, Brunet, and Jacquin 1997; Schumann et al. 1995; Unterstrasser et al. 2014). However, most studies address dispersion at high altitude, where the vortices are far removed from the effects of solid boundaries. Based on their review, the research team suggests further work/research be performed on the effects of aircraft wake downwash, specifically to: 1. Develop a new, coupled plume rise-wake model for predicting and assessing the effects of wake vortices on plume rise, dispersion, and GLCs. This model would be directed mostly at the take- off phase, but also should be made applicable to plume behavior during the ground roll. 2. Conduct further analysis of observed near-runway surface concentrations of aircraft pol- lutants (at any airport) to determine if high concentrations occur near the runway take-off end and how they compare with those near the runway starting position. Results from ACRP Project 02-58 show that high concentrations do indeed occur near the starting end. 3. Consider a potential field experiment deploying a dense array of near-runway surface moni- tors of an aircraft pollutant that can be measured in real time (e.g., NOx). The experiment could again determine if high concentrations occur near the take-off end of the runway and how these concentrations compare with concentrations near the aircraft starting position. Such an experiment could be considered if: i. Any existing near-runway monitors suggest that concentrations near the take-off end are significant, and/or ii. Sufficiently high surface concentrations are found using the new coupled plume rise-wake model. 4. Analyze existing large eddy simulation results of wake vortices for aircraft on or near the ground and use these to guide the development and testing of the coupled plume rise-wake model. 6.6 Aircraft Dispersion Based on Instantaneous “Line Puffs” The evaluation of AERMOD and other models at LAX required the use of approximately 5,000 area sources to describe the aircraft plumes and trajectories. This was by far the largest source category in the airport modeling and required the most computational resources and time; each area source was modeled as a continuous release. To improve the calculation effi- ciency, the researchers believe that it would be useful to consider instantaneous “line puffs” or “segmented line puffs” as the basic source configuration for the aircraft plume. Essentially, one would model the ground-roll plume as one line puff and the initial climb-out plume as a second line puff—albeit, as a tilted or slanted line puff. Two key reasons for considering a line puff as the main source geometry for the aircraft plume are: 1. The line puff is more physically representative of the plume than a number of continuous area sources, each with its own plume, and 2. The concentration and dose at some location and time due to a single aircraft (or line puff) can be determined and then the concentration and dose from many aircraft can be super- imposed to obtain appropriate values over an averaged time (e.g., 1 hour). An example of this approach is the dose or exposure due to an instantaneous crosswind line source (in the y direction, normal to the mean wind) that is transported by the local wind; the dose or exposure is the integral of the concentration over time. Near the ground, a crosswind line source could be created by a mobile vehicle releasing tracer along a road normal to the mean

Future research Needs 29 wind; similarly, an elevated crosswind line source could be generated by an aircraft releasing tracer along a path normal to the mean wind. The contractor’s final report includes an analyti- cal dose expression for a crosswind line source from which the concentration can be found by dividing the dose by the average time. The researchers suggest that it would be useful to conduct a preliminary feasibility study to develop and implement the line-puff model for aircraft plumes and compare its concentra- tion predictions and performance, especially run time, with those of the existing area-source approach. The results of such a study would provide an assessment of the potential benefits of the approach for further consideration. 6.7 Effects of Light Winds and Atmospheric Stability on Dispersion Analysis of the model performance results shows clear trends of predicted and observed con- centrations with time of day, wind speed, and atmospheric stability, especially at the four LAWA core sites. Several processes or causes are either not included or incompletely addressed in the models evaluated, leading to the overestimation of maximum concentrations in light winds. They are: 1. Along-wind dispersion (Sharan and Yadav 1998), which is typically not considered in a Gaussian plume model and leads to a reduction in concentration, but which can be accom- modated more easily in a puff model; 2. Plume or puff meandering in light winds, which leads to enhanced lateral dispersion and lower concentrations; 3. Insufficient dispersion and turbulence parameterization (e.g., of the lateral root mean square [RMS] turbulence velocity, sv, or the RMS angular wind deviation, sq) when using the fric- tion velocity (up) to parameterize sv and the dispersion (sy) due to the difficulty in uniquely characterizing up in very low winds, creating a very large scatter and uncertainty in up(Qian and Venkatram 2011); and 4. Adequate accounting for the increase of wind speed with height near the ground using the log law or similarity theory for the wind speed increase, which is important for near-surface sources (Eckman 1994; Qian and Venkatram 2011). As noted by Qian and Venkatram, use of direct measurements of the lateral turbulence, sv or sq, leads to improved estimates of lateral dispersion and concentration in light winds (Qian and Venkatram 2011). One suggestion would be to take one model (e.g., AERMOD) or a surrogate (AERMOD simu- lator) and address/improve all or as many of these four processes as possible and re-evaluate the model with the LAWA data set. This approach would permit an assessment of proper treatment of the processes and the resulting change in the model performance. 6.8 Other Limitations The range of local-scale models considered in this study had a wide range in their treatment of chemical processes. From a health impact point of view, the six criteria pollutants of interest in the NAAQS are CO, Pb, NO2, SO2, O3, and PM2.5. Given the large amount of NOx emissions from aircraft, any O3 that is formed is immediately titrated; hence, O3 concentrations in the vicinity of the airport are usually of less interest. At a minimum, the local-scale models used for airport dispersion modeling should have the ability to predict CO, NO2, SO2, and directly emitted PM2.5.

30 Dispersion Modeling Guidance for airports addressing Local air Quality health Concerns AERMOD has a detailed three-tiered approach to predict NO2 given NOx concentrations: (1) the Plume Volume Molar Ratio Method (PVMRM), (2) the Ozone Limiting Method (OLM), and (3) the Ambient Ratio Method (ARM). These approaches are designed primarily for emis- sions from tall stacks such as power plants and other large stationary sources. ACRP has spon- sored a separate study to develop a NOx-to-NO2 chemistry module specifically designed for aircraft sources in EDMS-AEDT. At publication of ACRP Research Report 179, the final report from that study was not yet available. It is recommended that future dispersion studies consider use of the new module and the associated guidance, when they become available, to predict NO2. Given that secondary PM2.5 has a more homogenous characteristic spatially, estimates for secondary PM2.5 can be obtained from a more comprehensive model like CMAQ. Hybrid approaches described in other studies can be used to obtain total PM2.5 concentrations (Isakov et al. 2009; Davis and Arunachalam 2009; Kim et al. 2012; Yim et al. 2015; Chang et al. 2017). In these hybrid approaches, the local-scale contribution for PM2.5 is assumed to be primary (directly emitted), and the secondary component is assumed to be formed from atmospheric chemical reactions by interactions with other emitted species from non-aircraft sources. The secondary component is estimated from the regional-scale CMAQ model, and then the two estimates (primary and secondary) are simply added to obtain the hybrid estimate. From these prior studies, the hybrid approach has been shown to give much better model performance than single models when compared to observations at local scales. The advantages of the hybrid approach are that it incorporates complex chemistry as well as accounts for regional-scale trans- port of the PM2.5 component from upwind regions of the airport region. Given that this hybrid approach requires additional expertise in a more complex model such as CMAQ, however, one can also obtain a spatial field of these concentrations using statistical approaches such as STOK, as was illustrated by Arunachalam et al. in support of another environmental exposure study (Arunachalam et al. 2014). Given that aircraft-related air quality impacts are primarily due to PM2.5 (primary in the near field and secondary in the far field), use of a hybrid approach that estimates both near- and far-field concentrations will enhance this characterization to a large degree and improve model performance. 6.9 Interim Guidance Given that several of the recommendations put forth in this report are for the longer-term, until dispersion models are improved, the interim recommendation is for airport air quality practitioners to adopt the following approaches: 1. Assume that most or all of the PM2.5 emitted from aircraft during LTO cycles is of a size less than 100 nanometers, and hence the AEDT-based PM2.5 inventory for aircraft emissions is essentially the same as aircraft inventories of UFP. 2. When it becomes available, use the new NOx-to-NO2 module being developed as part of ACRP Project 02-43 for estimating NO2 concentrations. An alternative is to use the three- tiered approach from AERMOD in sensitivity mode and assess NO2 model performance. 3. Use a hybrid modeling approach to estimate PM2.5 (using CMAQ or STOK, as described in this chapter) for improved model performance. 4. Explore the treatment of aircraft sources as volume sources instead of the current default approach, which treats them as area sources in AERMOD, and then assess model performance.