Cost-benefit Analysis of Ultra-Low Sulfur Jet Fuel by Stephen Kuhn B.S. in Mechanical Engineering, Carnegie Mellon University, 2008 Submitted to the Department of Aeronautics and Astronautics in Partial Fulfillment of the Requirements for the Degree of Master of Science in Aeronautics and Astronautics at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY ARCHNES 2010 MayMay201 MASSACHUSETTS INSTITUTE © Massachusetts Institute of Technology 2010. All rights reserved. JUN 2 3 2010 OF TECHNOLOGY LIBRARIES .......................... Au th o r........... ..... "i"epart nent of Aeronauhcs and Astronautics lay 24, 2010 Certified by.......... Ian A. Waitz Jerome C. Hunsaker Professor of Aeron Department Head Thesis Supervisor cs and Astronautics Accepted by............ David L. Dar oI Professor of Aeronautics and Astronautics Associate Department Head Chair, Committee on Graduate Students 2 Cost-benefit Analysis of Ultra-Low Sulfur Jet Fuel by Stephen Kuhn B.S. in Mechanical Engineering, Carnegie Mellon University, 2008 Submitted to the Department of Aeronautics and Astronautics in Partial Fulfillment of the Requirements for the Degree of Master of Science in Aeronautics and Astronautics Abstract The growth of aviation has spurred increased study of its environmental impacts and the possible mitigation thereof. One emissions reduction option is the introduction of an Ultra Low Sulfur (ULS) jet fuel standard for global commercial aviation. A full cost-benefit analysis, including impacts on air quality, climate, operations, and lifecycle costs is necessary to justify such a policy. The cost of a ULS jet fuel policy is well-characterized by the adoption of ULS diesel fuel, similar to jet fuel, for ground transportation in the US and elsewhere. The cost of hydrodesulfurization (HDS), the process used to remove sulfur from fuel, is projected to be between 4 and 7 cents per gallon of jet fuel. With 2006 levels of domestic fuel consumption, this translates to a yearly cost of HDS of $540-$940 million within the US. The climate and air quality benefits are characterized by several earth-atmosphere models, which isolate the perturbation of aviation emissions. Comparisons among models, which employ different modeling methods and assumptions as well as different spatial resolution, provide some cross-validation, as well as characterizing the degree of uncertainty in the state of the science. This thesis focuses in detail on the CMAQ (Community Multi-scale Air Quality) model, used by the Environmental Protection Agency (EPA) to support regulatory impact assessment. Other models, their results, and efforts at inter-model comparison are also discussed. Benefits are monetized through valuing the reduction in premature mortality from reduced concentrations of ground-level particulate matter (PM). The central finding from CMAQ is that with nominal health impact parameters, a global ULS jet fuel policy is predicted to save 110 lives per year in the US when considering full flight emissions, a 14% reduction in aviation-attributable mortality resulting in an estimated monetary benefit of $800 million. Acknowledgments During the work entailed in this thesis, I received support and assistance from many people, some of whom I would like to thank here. My advisor Ian Waitz has been instrumental in guiding the course of the ULS project in the face of setbacks and over the course of much spirited debate. He has been a pleasure to work for, and I've learned a great deal from him. His style is well-suited to bringing out the best in the people he works with. Special thanks as well to Steven Barrett, whose knowledge of atmospheric chemistry and tools has been invaluable, and whose assistance and time has been generously gifted. Thanks to my friends and colleagues in the PARTNER lab, with whom I have discussed much. Thanks to Tudor Masek, who helped answer many questions well after he had left us; Elza BrunelleYeung, who brightened the windowless and decrepit expanse of my first office, and helped in many other ways besides; Akshay Ashok, to whom I gladly yield the mantle of pseudo-computer expert; Pearl Donohoo, who suffered mightily, and lived to fight another day; Rhea Liem, whose behaviors can only be described as stochastic; Jim Hileman, a project and lab leader who has helped guide many students' work; Russ Stratton and Matthew Pearlson, who along with Jim and Pearl researched and compiled the cost part of the study; and all the rest I knew and worked with. Thanks as well to the scientists and engineers at UNC, FAA, EPA, Cambridge, Harvard, Stanford, and elsewhere who helped to make, run, and interpret the models used in the study. There were many labors and countless emails; I appreciate the time and effort put in by all. I would also like to thank the MIT Aeronautics and Astronautics Department for helping to make my time at MIT interesting. Table of Contents Abstract..................................................................................................................3 Acknowledgm ents............................................................................................. 4 Table of Contents .................................................................................................. 5 List of Figures .................................................................................................. 8 List of Tables..................................................................................................... 11 List ofSym bols ................................................................................................. 12 Introduction.............................................................................................14 1 1.1 Context ......................................................................................................................... 14 1.2 Motivation ................................................................................................................. 14 1.3 Contributions ............................................................................................................ 15 1.4 Thesis Structure ........................................................................................................ 15 2 2.1 2.2 Setup ....................................................................................................... 15 CMAQ ....................................................................................................................... B ackg ro u n d ............................................................................................................................ 2 .1.1 2 .1.2 M o d el Scen ario s....................................................................................................................16 2 .1.3 M o d el D o m ain ...................................................................................................................... 15 15 Aviation Inventories .................................................................................................. 2 .2 .1 In tro d u ctio n .......................................................................................................................... 2.2.2 Species P opulation .......................................................................................................... 2.2.2.1 AEDT/SAGE, FOA3 and Emissions Indices ...................................................... 2.2.2.2 NOx by Flight Mode...................................................................................................22 2 .2 .2 .3 Lightn ing N O x ................................................................................................................. 2.2.2.4 H ydrocarbon Split........................................................................................................ 2.2.3 Processing Methodology ................................................................................................ 20 20 21 21 17 24 27 28 2.3 28 Background Inventories........................................................................................ 28 2 .3 .1 B ackgro u n d ............................................................................................................................ 2.3.2 Species Population from National Emissions Inventory....................29 .......................................... 29 2.3.2.1 2001 N E I vs. 2005 N E I................................................... . .......................................... 29 2.3.2.2 Surface Concentrations .................................................. 30 2.3.2.3 V ertical profiles ............................................................................................................ 2.4 GEOS-Chem Boundary Conditions ..................................................................... 2 .4 .1 B ackgro u n d ............................................................................................................................ 2.4.2 GEOS2CMAQ and methodology ..................................................................................... 2 .4.3 V erificatio n ............................................................................................................................ 34 34 34 35 2.5 Meteorological Data..................................................................................................38 2 .5 .1 B ackgro u n d ............................................................................................................................ 38 2 .5 .2 2.6 V ertical Pro files.....................................................................................................................40 Particle-Bound Water and Dry vs. W et PM .......................................................... 40 2.7 Population Data and H ealth Impacts....................................................................42 2.7.1 Global Population of the World (GPW) and Global Rural-Urban Mapping Project (GR UM P) ............................................................................................................................................ 42 2.7.2 Concentration Response Functions.................................................... ................... 43 2.7.3 Monetary Value of a Statistical Life (VSL).......... .................................................. 46 2.8 M odeling Context ..................................................................................................... 2.8.1 Earth-Atmosphere Modeling of Aviation.................................................................... 2.8.2 Atmospheric Aerosols .................................................................................................... 2.8.3 Multi-Model Approach.................................................................................................. 46 46 47 50 FullFlightEmissionsAssessment........................................................ 52 3.1 Context ...................................................................................................................... 52 3.2 Changes in Surface Concentration and M ortalities..............................................52 3 4 L TO Emissions Assessm ent.....................................................................57 4.1 Context ...................................................................................................................... 4.2 Changes in Surface Concentration and M ortalities..............................................57 5 Clim ate Feedbacks................................................................................ 57 62 5.1 Background on Climate Feedbacks...................................................................... 62 5.2 Background on GATOR-GCM OM ..................................................................... 62 5.3 Future W ork .............................................................................................................. 64 6 ULS Cost and OperationsAnalysis .......................................................... 64 6.1 Sulfur Content of ULS Jet Fuel.................................................................................64 6.2 Production Costs of ULS Jet Fuel..........................................................................66 6 .2 .1 B ackgro u n d ............................................................................................................................ 6.2.2 Detailed Cost Analysis ..................................................................................................... 6.2.2.1 Normalized Crack Spread of Diesel and Jet Fuels ................................................. 6.2.2.2 Supply and Demand .................................................................................................. 6.2.2.3 The Cost of a Gallon of Diesel.................................................................................. 6.2.2.4 Historic Price Difference in Diesel Fuels ............................................................... 6.2.2.5 Processing Costs.......................................................................................................... 6.2.2.6 Limitations of this Analysis ...................................................................................... 6.2.3 Production Cost Estimate .............................................................................................. 66 67 69 69 69 70 70 70 71 Potential Usability Concerns and Benefits with ULS Jet Fuel.............................. 71 6.3 CMA Q Sensitvity Studies...................................................................... 74 7.1 CM AQ models .......................................................................................................... 74 7.2 Introduction .............................................................................................................. 75 7 7.3 7.4 7.5 8 8.1 Vertical Diffusion and Advection.......................................................................... 7 .3 .1 B ackgro u n d ............................................................................................................................ .............................. 7.3.2 Vertical Diffusion: EDDY vs. ACM 2......................................... 7.3.3 Vertical Advection: PPM vs. YAM O ................................................................................ Static vs. Changing Background Conditions........................................................ 7 .4 .1 C o n tex t...................................................................................................................................7 7.4.2 Dominance of GEOS-Chem Boundary Conditions ....................................................... 76 76 77 78 79 9 79 Lightning N Ox ......................................................................................................... 80 Comparisons to EmpiricalD ata ............................................................ 83 Context ...................................................................................................................... 83 Speciated M odel Attainm ent Test ........................................................................ 8 .2 .1 M eth o d o lo gy ......................................................................................................................... 8 .2 .2 R e su lts .................................................................................................................................... 8.2.3 EPA FRM monitor sites......................................................................................................86 8.2.4 Comparisons of Outputs to M onitor Data.................................................................. 83 83 84 Limitations and Future Work ................................................................... 89 9.1 Scope/Feedback ....................................................................................................... 89 9.2 Uncertainty................................................................................................................90 9.3 Other Differences......................................................................................................90 9.4 UT/LS Transport in CM AQ .................................................................................. 9.5 Proposed Paths Forward........................................................................................92 8.2 9 10 86 91 Conclusion...............................................................................................94 10.1 CTM Findings...........................................................................................................94 10.2 Cost/Benefit Assessm ent...................................................................................... 94 A PPEND ICES................................................................................................. 95 Appendix A: CMA Q Inventory and Model Species ........................................... 95 95 A.1 Aircraft Emissions Inventory ................................................................................ A.2 Background Inventory and Concentrations .......................................................... 102 A.3 CM AQ Outputs...................................................................................................... 106 Appendix B: CMA Q Processes......................................................................... B.1 Carbon-Bond IV Processes and Reactions M odeled ............................................ 108 108 Appendix C : CMA Q R esults ............................................................................ 113 A.1 Additional Aviation Results .................................................................................... 113 A.2 Computational Tim e/Complexity...........................................................................117 Bibliography...................................................................................................... 119 List of Figures Figure 1: CMAQ horizontal domain used in the ULS study (Masek, 2008).......................................18 Figure 2: CMAQ vertical structure (CMAQ 4.6 Operational Guidance Document, 2009) ............. 19 Figure 3: Emission indices of NO, N02, and HONO from a CFM56-3B1 engine (Wood, Herndon, Tim ko, Yelvington, & M iake-Lye, 2008)................................................................................................ 23 Figure 4: Vertical Profile of N from Lightning NO over the CMAQ US domain .............. 25 Figure 5: Vertical columnar sum of N [g N/yr] from lightning emissions of NO for 2006 ............ 26 Figure 6: NASA NSSTC space-based observation of #flashes/km^2/yr., 1995-2000 (Where Ligh tn in g S trik es) ............................................................................................................................................. 26 Figure 7: 2006 background NH3 and PM2.5 concentration showing several co-located hotspots ....30 Figure 8: Comparison of 2001 and 2005 NEI emissions profiles over CMAQ US domain............31 Figure 9: NEI emissions trends of several key species (National Emissions Inventory Air Pollutant T ren d s D ata, 2 0 09) .......................................................................................................................................... 32 Figure 10: Seasonal Variability in the 2001 NEI for several key emissions species ........................... 33 Figure 11: Ground level 03 concentrations [ppbV] from GEOS-Chem (inside the black box) and the CMAQ BCON created from this GEOS-Chem run (outside the black box) ......... .............. 36 Figure 12: Ground level 03 concentrations in a.), and at ~860hPa (layer 10) in b.); ground level HNO3 concentrations in c.), and at ~860hPa in d.) ................................... 37 Figure 13: Background H2SO4 showing higher concentrations at domain boundary at ground level ............................................................................................................................................................................ 37 Figure 14: Full aviation delta in ground-level NH3 and H2SO4, showing clear boundary layer effects ............................................................................................................................................................................ 38 Figure 15: Vertical profiles of US average temperature, cloud water mixing ratio, rain water mixing ratio, and w ater vapor mixing ratio ......................................................................................................... 40 Figure 17: CMA Q dom ain population..................................................................................................... 43 Figure 18: Health impact pathway for air quality assessment (Rojo, 2007)........................................43 Figure 19: Baseline cardiopulmonary and lung cancer mortalities ..................................................... 45 Figure 20: Spatial and temporal scale of atmospheric processes (Seinfeld & Pandis, 2006) ........... 47 Figure 21: Global Radiative Forcing from All Anthropogenic Sources of Emissions (Lee, et al., 2 0 0 9 ) .................................................................................................................................................................. 49 Figure 22: Global Radiative Forcing from Aviation Emissions (Lee, et al., 2009) .............. 50 Figure 23: Full aviation FRM PM2.5 and ULS reduction in same, full aviation (static BCON) FRM PM 2.5, and post-SM AT FRM PM 2.5...........................................................................................................54 Figure 24: Full aviation S04, N03, and NH4, and ULS reduction in same ..................................... 55 Figure 25: Full aviation BC and ORG, and ULS reduction in same ................................................... 56 Figure 26: Speciated components of full aviation-attributable change in FRM PM2.5 and mortalities 56 ............................................................................................................................................................................ Figure 27: Speciated components of ULS aviation-attributable change in FRM PM2.5 and mortalities 57 ............................................................................................................................................................................ Figure 28: LTO aviation FRM PM2.5 and ULS LTO reduction in same, LTO aviation (static BCON) FRM PM2.5, post-SMAT FRM PM2.5........................................................................................58 Figure 29: LTO aviation S04, N03, NH4, and ULS reduction in same....................59 Figure 30: LTO aviation BC and ORG, and ULS reduction in same......................................................60 Figure 31: Speciated components of LTO aviation-attributable change in FRM PM2.5 and 61 m o rtalitie s .......................................................................................................................................................... Figure 32: Speciated components of ULS LTO aviation-attributable change in FRM PM2.5 and 61 mo rtalitie s .......................................................................................................................................................... Figure 33: Jet A average fuel sulfur content in US (Taylor, Feb. 2009), as plotted by (Hileman, et al., 65 20 0 9 ) .................................................................................................................................................................. Figure 34: JP-8 fuel sulfur content as reported by PQI (DESC-BP, 1999-2007), as plotted by 66 (Hilem an , et al., 2 009) ..................................................................................................................................... Figure 35: Jet A-1 average fuel sulfur content in UK (Rickard, June 2008), as plotted by (Hileman, et al., 2 0 09 ) ............................................................................................................................................................ 66 Figure 36: D iesel prices and production................................................................................................... 68 Figure 37: Sensitivity of results to vertical diffusion parameter ........................... 78 Figure 38: Dominance of GEOS-Chem boundary conditions ................................................................ 79 Figure 39: Sensitivity of results to LN O x................................................................................................. 82 Figure 40: Full aviation post-SMAT speciated results in comparison to pre-SMAT model outputs.. 85 Figure 41: Full aviation SMAT constituent contribution to FRM PM2.5 in comparison to pre-SMAT 5 m o d el o u tp u ts...................................................................................................................................................8 Figure 42: EPA FRM monitor sites used in SMAT software (Abt Associates, Inc., 2009)........86 Figure 43: Comparison of CMAQ monthly average ground-level FRM-hydrated PM2.5 to quarterly (native) official FRM monitor data.................................................................... ........ ........ 88 Figure 44: Comparison of monthly average ground-level FRM-hydrated PM across models to quarterly (native) official FRM monitor data.......................................................................................... 89 Figure 45: APMT-AQ projections of changes in FRM PM2.5 (a.) and b.)) in comparison to LTOonly (static B C O N ) results from CM A Q ..................................................................................................... 91 Figure 46: AEDT vertical profiles of US yearly aircraft emissions ..................................................... 97 Figure 47: CMAQ processed input vertical profiles of US yearly total aircraft emissions................99 Figure 48: CMAQ processed input vertical profiles of US yearly total aircraft emissions (continued) ......................................................................................................................................................................... 10 0 Figure 49: CMAQ 2006 yearly average background species concentrations ....................................... 106 Figure 50: Hierarchical relationship between major species in CB-IV (Isukapalli, 1999).................. 109 Figure 51: Full aviation-attributable changes in precursor and oxidizing gases .................................. 115 Figure 52: Aviation's relative contribution to ambient concentrations of several species ................. 116 List of Tables Table 1: Scenarios included in the design of experiment...................................................................... 17 Table 2: CMAQ sigma layers used for ULS, with sea-level referent corresponding pressures/altitudes ............................................................................................................................................................................ 20 T ab le 3: E mission s in dices..............................................................................................................................22 Table 4: Calculation of total N02 and NOx emissions from a CFM56-3B1 during LTO cycle (Wood, Herndon, Timko, Yelvington, & Miake-Lye, 2008) ............................................................... 24 Table 5: Speciated split of NOx into CMAQ species.................................................................................24 Table 6: Hydrocarbon split into CB-IV species .......................................................................................... 28 Table 7: GEOS-Chem to CMAQ species mapping (Moon, Park, Li, & Byun, 2008)...........34 Table 8: CMAQ/MM5 meteorological parameters and mechanisms................................................. Table 9: P coefficient values (O stro, 2004) ............................................................................................. 39 44 Table 10: Summary Model Comparison...................................................................................................51 Table 11: Refining and crude oil costs of a wholesale gallon of diesel, May 2002 - August 2009 (Spot Pric e s, 2 0 10) ...................................................................................................................................................... 69 Table 12: CM A Q schem es/m odels...............................................................................................................74 Table 13: ULS CMAQ "baseline" options (CMAQ 4.6 Operational Guidance Document, 2009).....75 Table 14: Sensitivity studies scenarios proposed.....................................................................................76 T able 15: P roposed future w ork .................................................................................................................... 92 Table 17: Full aviation total aircraft emissions from AEDT.....................................................................95 Table 18: Full aviation total aircraft emissions CMAQ processed input totals ................ 98 Table 19: Corresponding molecular weights for HC species and groupings in Table 18 i.e. (MF/SF) ............................................................................................................................................................................ 98 Table 20: Aircraft emissions pre-processed, speciated hydrocarbon profile (FAA OEE & EPA OT AQ , 2 0 09) ................................................................................................................................................ 100 Table 21: Background inventory species and descriptions ..................................................................... 103 Table 22: CMAQ4.6 concentration output species and descriptions (Isukapalli, 1999) .................... 106 T able 23: CB -IV reactions (Isukapalli, 1999)............................................................................................ 110 List of Symbols AEDT - Aviation Environmental Design Tool APMT - Aviation Portfolio Management Tool AQ - Air Quality AQM - Air Quality Model BC - Black Carbon (soot) BCON - Boundary Conditions CAEP - Committee on Aviation Environmental Protection CBIV - Carbon Bond IV (4) CBV - Carbon Bond V (5) CCN - Cloud Condensation Nuclei CH4 - Methane CI - Confidence Interval CMAQ - Community Multi-scale Air Quality Model CRC - Coordinating Research Council CRF - Concentration Response Function CTM - Chemistry and Transport Model EC - Elemental Carbon (soot, also called BC) El - Emissions Index EIA - Energy Information Agency EPA - Environmental Protection Agency EPAct - Energy Policy Act FAA - Federal Aviation Administration FOA3 - First Order Approximation Version 3.0 FRM - Federal Reference Method FSC - Fuel Sulfur Content GATOR-GCMOM - Gas, Aerosol, TranspOrt, Radiation, General Circulation, Mesoscale, Ocean Model GCM - Global Climate Model GHG - Greenhouse Gas GEOS - Goddard Earth Observing System HC - Hydrocarbon HDS - Hydrodesulfurization HOx - OH and applied radicals HSD - High-Sulfur Diesel ICAO - International Civil Aviation Organization ISORROPIA - CMAQ inorganic gas-aerosol equilibrium model, from the Greek for "equilibrium" IMPROVE - Interagency Monitoring of PROtected Visual Environments IPCC - Intergovernmental Panel on Climate Change LNOx - Lightning-induced Nitrogen Oxides LOSU - Level of Scientific Understanding LS - Lower Stratosphere LSD - Low-Sulfur Diesel LTO - Landing and Takeoff Cycle (<1 000m AGL) MM5 - Mesoscale Model Version 5 NEI - National Emissions Inventory NH3 - Ammonia NH4 - Ammonium ion N03 - Nitrate ion NOx - Nitrogen Oxides (NO + N02) 03 - Ozone OH - Hydroxyl radical ORG - Organics PBL - Planetary Boundary Layer PM - Particulate Matter PM2.5 - Particulate Matter smaller than 2.5 micrometers in diameter (as nearly all aviationattributable aerosols are) PMFO - Primary Fuel Organics PMSO - Primary Sulfur Organics nvPM - non-volatile Particulate Matter PBL - Planetary Boundary Layer ppbV - parts per billion Volumetric ppm - parts per million RF - Radiative Forcing RH - Relative Humidity SAGE - System for assessing Aviation's Global Emissions SANDWICH - Sulfate, Adjusted Nitrate, Derived Water, Inferred Carbonaceous mass Hybrid material balance SMAT - Speciated Model Attainment Test S04 - Sulfate ion SOx - Sulfur Oxides STN - Speciation Trends Network SULF - Sulfuric Acid (H2SO4) THC - Total Hydrocarbon TOG - Total Organic Gases ug/m^3 - micrograms per meter cubed ULS - Ultra Low Sulfur ULSD - Ultra Low Sulfur Diesel ULSJ - Ultra Low Sulfur Jet UT - Upper Troposphere 1 Introduction 1.1 Context Global aviation has historically been a fast-growing industry. Its rate of growth has outstripped the growth of gross domestic product by a factor of 2.4 between 1960 and 1999 (IPCC, 1999), and despite a lull in the past decade due to terrorist attacks and economic downturn, steady growth is projected into the future. Concerns related to the noise, climate, and air quality impacts of aviation are increasingly voiced. On a global scale, aircraft emissions of greenhouse gases (GHGs) and precursors contribute to climate change (IPCC, 1999). On a regional scale, aircraft emissions impact air quality, with resulting health impacts (ICAO, 2007). Poor air quality has been linked to premature mortality (US EPA, 2006), cardiovascular and pulmonary hospital admissions, and asthma aggravation (US EPA, 2004), among other ailments. Aviation is largely unique among the anthropogenic sources of emissions in that its emissions extend well into the atmosphere, to the free troposphere and jet stream, and even into the lower stratosphere, resulting in impacts that can be felt over a larger distance and longer time than groundbased emissions sources. Long-lived species (compounds) and precursors can be transported across oceans and continents. 1.2 Motivation Reductions in aircraft emissions per flight can be effected in one of three ways; changes in operational procedures for greater efficiency; changes in aircraft technology; and changes in fuel composition. Operational changes have been the subject of much research and development, and are embodied in the systems used to route and manage commercial aviation. Such systems are developed at great cost over many years, and often require the development of new technology, including the installation of new avionics equipment in aircraft. Airframe and engine improvements are likewise emphasized in improving fuel efficiency and performance, but a given airframe and engine are typically used in the fleet for 30 or more years (Morrell & Dray, 2009), and new technology takes many years to develop. Furthermore, retrofitting existing aircraft is typically prohibitively expensive. Therefore changes in operational procedures and aircraft technology are felt only gradually, over decades. In contrast, changes in fuel composition result in immediate changes in emissions. In the case of ULS fuels, which are assumed in this study to contain 15ppm sulfur, the technology and infrastructure for HDS exists from the EPA-mandated conversion of diesel to ULS diesel for most types of US ground transportation. The adoption of ULS diesel has phased into effect since 2006 (US EPA, 2006) (US EPA, 2006) and entails an identical chemical process. Projected costs, and especially capital expenditures, are much less speculative for the adoption of ULS jet fuel than for most other emissions mitigation options for aviation. The precursor gases NOx and SOx are the two largest sources of aviation-attributable ground level particulate matter (PM). The species oxidize ammonia (NH3) existing in the atmosphere from nonaviation sources to form sulfates and nitrates. Other sources of aviation-attributable PM include soot (black carbon, or BC), largely dependent on aromatic and paraffmic fuel content, and organic compounds, which typically account for less than 5-10% of aviation-attributable PM in this and prior studies (Rojo, 2007), (Masek, 2008). NOx is created in the aircraft engine combustor as a byproduct of the combustion of air (~78% N2) and fuel at high temperature. Its emission is dictated entirely by the processes within the combustor. As such, NOx reduction is dependent on the implementation of new engine technology in the fleet. SOx is created during combustion as sulfur in the fuel is oxidized. It is therefore possible to completely control SOx emissions by changing fuel composition. 1.3 Contributions This thesis presents some results from the study of a switch to ULS jet fuel. It addresses original work from several parties, which is labeled as such where applicable. Four earth-atmosphere models were used to predict the climate and air quality benefits of a ULS jet fuel policy. These include GATOR-GCMOM (Gas, Aerosol, TranspOrt, Radiation, General Circulation, Mesoscale, Ocean Model), GEOS-Chem (Goddard Earth Observing System Chemistry Model), TOMCAT, and CMAQ (Community Multi-scale Air-Quality Model). This thesis describes some results from several of these models, but focuses primarily on the CMAQ model. The setup, inputs, and outputs of the CMAQ model are described in detail. The mining and integration of data, where results from different models are compared together, are original to this thesis. 1.4 Thesis Structure This thesis is organized into 11 sections. Introduction and setup of the model runs are followed by results from full-flight and landing-takeoff (LTO)-only emissions scenarios. A general discussion of results when climate feedbacks are modeled is included. Afterwards follows analysis of sensitivities to various parameters in CMAQ, as well as comparisons of model data to empirical data. Then limitations, future work, and conclusions are presented. 2 Setup 2.1 CMAQ 2.1.1 Background The EPA's Community Multiscale Air Quality (CMAQ) model is a three-dimensional Eulerian atmospheric chemistry and transport modeling system that simulates ozone, acid deposition, visibility, and fine particulate matter throughout the troposphere (CMAQ 4.6 Operational Guidance Document, 2009). Like all Eulerian computational air quality models, on a fundamental level, CMAQ outputs concentration fields that are the solutions of systems of partial differential equations for the time-rate of change in species concentrations due to a series of individual physical and chemical processes, which are aggregated. Air quality models integrate our understandings of the complex processes that affect the concentrations of pollutants in the atmosphere. Establishing the relationships among meteorology, chemical transformations, emissions, and removal processes in the context of atmospheric pollutants is the fundamental goal of such a model (Seinfeld & Pandis, 2006). CMAQ uses a set of modular process solvers to accomplish this, which incorporate models developed in the literature (Table 12 in Section 7). 2.1.2 Model Scenarios The procedure for studying a particular phenomenon or perturbation, such as the effect of aircraft emissions on ground-level PM and ozone, is as follows. A scenario is run (Run 1) in which every source of emission - biogenic and anthropogenic, industrial, transportation, etc. - is included, except the perturbation. This is the so-called "background" inventory and is compiled from the EPA's National Emissions Inventory (NEI) for the US, and from various sources for the rest of the world. An inventory is a gridded or spatial accounting of emissions rates (i.e. in g/s or mol/s, depending on the phase of the species, using accounting of hourly changing emissions profiles). In this case, the background omits any aviation emissions. The model is run for a given amount of time, processes evolve, and concentrations are saved (i.e. in ug/m^3 or ppbV, depending on the phase of the species). Another scenario is run in which the perturbation inventory, in this case the aircraft emissions inventory, is added to the background inventory (Run 2). The interactions of the aircraft emissions with background species and emissions, including NH3, 03, OH, and S02, among others, are captured. The model is run, and concentrations are saved. Run 1 is subtracted from Run 2, and the impacts of aviation emissions within the wider atmospheric system are isolated. For aviation, this difference, or "delta" case change in concentration, is typically a small fraction of the absolute, or ambient, concentrations (the output of Run 2); depending on location, for all species of PM, aviation-attributable concentrations are at least two orders of magnitude lower than ambient, and often much lower still. Some representative plots of this relative contribution are provided in Appendix C. In addition to the background, there are four aviation scenarios for the ULS project, listed in Table 1. All models used in the study scoped these scenarios at the least, though domains and durations varied, and additional variations were added for different models. The full aviation, or "baseline", scenario represents full flight emissions with current fuel compositions. Fuel compositions are discussed in detail in Section 6.1. The ULS full aviation scenario represents full flight emissions with a ULS fuel composition. Until recently, it was thought that LTO (landing and takeoff, <1km above ground level) emissions contributed to the bulk of ground-level PM. However, that has been called into question by (Barrett, Britter, & Waitz, 2009) and the finding that cruise emissions may contribute between two and 10 times the amount of ground-level PM as LTO emissions, as predicted by GEOS-Chem. A secondary objective of this study is to further examine the relative impact of cruise emissions. As such, scenarios were included in which LTO-only aircraft emissions were modeled, as well as LTO-only emissions with a ULS fuel policy. These also provide valuable comparisons to previous LTO-only studies. A three-month "spin-up" was included for each scenario. This provides sufficient time for concentration gradients and transport processes to develop before the period of study. For CMAQ, that period is the full year of 2006. All aviation inventories for the project, as well as the meteorological data for CMAQ and GEOS-Chem are from 2006. The background inventory is notably based on the 2001 NEI. Over the US, GATOR-GCMOM uses the 2005 NEI, and GEOSChem uses the 1999 NEI. A comparison of the 2001 and 2005 NEIs is provided in Section 2.3.2 for several CMAQ species. It was impractical to harmonize the NEI years across models due to general availability at the time, but the differences between models are far more significant than the differences between the background inventories. The boundary conditions (BCON) for CMAQ (the "picture frame" concentrations at the edge of the domain) were derived from global GEOS-Chem runs in which the scenario was also allowed to change. For instance, the inventory for the global GEOS-Chem run that was used for the background scenario CMAQ BCON also included no aviation emissions. Two additional scenarios beyond the design of experiment were included for CMAQ in which the full aviation and LTO scenarios used static BCON; that is to say, the background BCON were used for these scenarios, eliminating differences in BCON flux in the delta case, and isolating the component of change over the US that is due to US-only changes in aircraft emissions. Table 1: Scenarios included in the design of experiment 2.1.3 Background Spin-up (Oct-Nov 2005), Scenario Gan-Dec 2006) Full aviation, or "baseline" ULS full aviation Spin-up (Oct-Nov 2005), Scenario Gan-Dec 2006) Spin-up (Oct-Nov 2005), Scenario Gan-Dec 2006) LTO aviation ULS LTO aviation Spin-up (Oct-Nov 2005), Scenario (Jan-Dec 2006) Spin-up (Oct-Nov 2005), Scenario Gan-Dec 2006) Model Domain The CMAQ grid is a Lambert conformal conic projection covering the continental United States and parts of Canada, Mexico, and the Caribbean. The 3-D grid is of size 148 x 112 x 35, constituted of 148 cells from western to eastern boundaries, and 112 grid cells from northern to southern, of 36 x 36km within the projection, extending upwards into 35 vertical layers. The domain boundary is not rectilinear in latitude/longitude coordinates, so e.g. the western border spans a range of longitudes. This can be visualized in the CMAQ plots in Section 8. The vertical layers are finer at ground level and become coarser as they ascend, with eight layers in the lowest km and an upper limit of 1OOhPa (-15-17km). The parameters of the Lambert projection, including the standard parallels, central meridian, and latitude of origin, are shown in Figure 1. Lambert conformal projections minimize two-dimensional distortion of the three-dimensional portion of the Earth's surface over the US domain, with zero distortion along the standard parallels, at 33N and 45N. Lambert projections are also favored in general for aviation because a straight line drawn on the projection approximates the great-circle route between endpoints that aircraft generally attempt. Figure 1: CMAQ horizontal domain used in the ULS study (Masek, 2008) The CMAQ vertical domain is a non-hydrostatic sigma coordinate system, described by a set of dimensionless parameters, denoted sigmas by convention, defined in Equation 1. = * P-Prop Pgroundref' Prop Each vertical column has a layer structure that is normalized to the difference between the reference pressure at mean ground-level and 1OOhPa. In essence, this allows the vertical coordinate system to be terrain following. The sigmas (which have values between 1, at ground level, and 0, at 1OOhPa) are then used to determine the pressure at the edges of each vertical layer. As each of the 148 x 112 vertical columns have the same number of layers, and yet different mean ground altitudes (and therefore pressures), each vertical column has a unique grid. This can be visualized in Figure 2. The sigmas themselves are based on the lowest levels of the existing CMAQ 2001 NEI used in prior studies, as well as the reduced vertical domain of GEOS-5 (Goddard Earth Observing System Model-5) for general boundary condition compatibility. For the purposes of displaying the results of this study, the vertical grid normalized to sea level was taken as the referent for plots of domainaverage concentration by pressure. This results in a uniform aggregation of all surface layer concentrations into the same vertical bin, which would otherwise not be the case. A more rigorous method would require paired-in-time-and-space summation, which is prohibitively intensive for the quantity of data generated, and which would conflict with the method of display of other models in the study (which also use sigma, as opposed to eta or hybrid sigma-eta coordinates, in which the ground layer varies). Pr =const =~~~hx p'X( X____ h=/-c )X MSL Figure 2: CMAQ vertical structure (CMAQ 4.6 Operational Guidance Document, 2009) The sigma layers in Table 2 are used to create the pressures at top and bottom of a vertical grid cell. For example, the bottom of layer 1 is described by the sigma 1.000, and the top by 0.9950. The bottom of layer 2 is described by 0.9950, and the top by 0.9900, etc. The pressure and altitude for a vertical column normalized to sea level are also listed, where the pressure and altitude describe the top of the layer. These pressures and altitudes vary, however, with the ground altitude for a given vertical column. Table 2: CMAQ sigma layers used for ULS, with sea-level pressures/altitudes referent corresponding 1 0.6377 682.59 3,260 20 0.5959 644.35 3,652 60.3 21 0.5541 606.17 4,144 995.23 121.0 22 0.5123 567.99 4,662 0.9600 976.96 243.4 23 0.4706 529.89 5,208 6 0.9400 958.69 408.9 24 0.4288 491.71 5,787 7 0.9100 931.29 619.6 25 0.3871 453.62 6,404 8 0.8830 906.62 865.1 26 0.3453 415.53 7,063 9 0.8663 891.37 1,057 27 0.3036 377.34 7,772 10 0.8495 876.02 1,207 28 0.2619 339.25 8,539 11 0.8328 860.76 1,359 29 0.2071 289.19 9,516 12 0.8161 845.50 1,513 30 0.1593 245.52 10,686 13 0.7993 830.16 1,670 31 0.1187 208.43 11,837 14 0.7770 809.79 1,856 32 0.0844 177.10 12,951 15 0.7492 784.39 2,099 33 0.0553 150.52 14,025 16 0.7213 758.91 2,377 34 0.0305 127.86 15,065 17 0.6934 733.42 2,663 35 0.0095 108.68 16,070 18 0.6656 708.03 2,957 0.0000 100.00 16,819 (ground layer) 1.0000 1013.5 2 0.9950 1008.9 20.1 3 0.9900 1004.4 4 0.9800 5 The comprehensive set of CMAQ settings used for such parameters as chemical mechanism, diffusion, advection, aerosol module, etc. are described in more detail in Table 13 in Section 7, the section on CMAQ sensitivity to changes in those parameters. 2.2 Aviation Inventories 2.2.1 Introduction Aviation emissions inventories convert flight data into rates of species emission, using information about the flight mode (i.e. throttle level and altitude), engine and airframe type, fuel type, and temperature, pressure and other environmental conditions. Chord-level flight data is scaled based on relations governed by these parameters, and aggregated to the grid structure. Aviationattributable PM is grouped into two categories: primary PM and secondary PM created from aviation-attributable precursors. Primary PM is created in the combustor and turbine itself; at the plane of the exit nozzle, the species are extant. Secondary emissions are formed in the plume of the engine exhaust and elsewhere once the plume has dissipated, by reactions of the precursor gases NOx and SOx with existing species in the atmosphere (e.g. OH, 03, H20, and later, gaseous ammonia (NH3)) on a scale of meters to many kilometers, from seconds to many hours (IPCC, 1999). The aerodynamic diameter of jet turbine aircraft PM is extremely small in size, with bimodal peaks in the distribution usually occurring at 30 and 100nm (0.3 and lum); as such all aviation emissions are considered PM2.5 (particulate matter with 2.5um or smaller diameter) (Petzold, Dopelheuer, Brock, & Schroder, 1999). The emission of C02 (a greenhouse gas, or GHG), H20 (as contrail), and CO (integral in the CO--OH--CH4 cycle) are significant when considering climate effects. 2.2.2 Species Population 2.2.2.1 AEDT/SAGE, FOA3 and Emissions Indices Aviation emissions are constituted of several species generally grouped into categories. From an air quality perspective, the key species are SOx and NOx, as well as black carbon (BC) and primary fuel organics (PMFO). SOx and NOx are precursor gases that lead to secondary PM (S04, N03, and NH4), and BC and PMFO are primary PM. Hydrocarbons representing unburnt or partially burnt fuel are also included. Total monthly and yearly emissions of all pre- and post-processed species are listed in Appendix A.1. CO is included in the CMAQ emissions, although H20 and C02 are not. For a chemistry and transport model (CTM) not including climate feedbacks or radiative forcing (RF), aircraft emissions of H20 are insignificant (contrails are not treated), and C02 is stable and non-reactive. The creation of these species is governed by emissions indices (Els) based on fuel burn numbers and other relations categorized in the ICAO's (International Civil Aviation Organization) FOA3 (First Order Approximation version 3.0). The System for assessing Aviation's Global Emissions (SAGE) makes use of these relations, in conjunction with global individual flight information, to create speciated inventories of aviation emissions. SAGE is a high fidelity computer model used to predict aircraft fuel burn and emissions for all commercial (civil) flights globally in a given year (DOT/FAA, 2005a). SAGE is part of the Aviation Environmental Design Tool (AEDT), a software system that dynamically models aircraft performance in space and time to produce fuel burn, emissions and noise. The outputs of AEDT/SAGE (referred to hereafter as AEDT) for 2006 over the US CMAQ domain are quantified in Appendix A.1. AEDT has been extensively validated against actual data from US airlines and been found to predict fuel burn with less than 5% error on average (DOT/FAA, 2005b). Standardized techniques for directly measuring volatile and primary aircraft emissions do not exist (Wayson, Fleming, & lovinelli, 2009). As such, a first order approximation was derived. The present iteration of that approximation is version 3.0, denoted FOA3.0, allowing for independent prediction of volatiles and non-volatiles on a more theoretical basis. The more conservative FOA3a approximation was not used for this study. Once an accepted, repeatable method for direct measurement of PM emissions is established, and the fleet is sufficiently represented, the FOA methodology will eventually become obsolete. However, currently, a species-specific emission index is multiplied by the amount of fuel burn per unit time to yield a species emission rate. These relations are quantified in Table 3. Table 3: Emissions indices NOx AEDT(NOx) AEDT(NOx)*0.997 Soot (>1 km) AEDT(fuel burn)*0.035 AEDT(fuel burn)*0.035*0.997 Soot (<1 km) AEDT(PMNV from FOA3) AEDT(PMNV from FOA3)*0.997 SO2 AEDT(fuel burn)*1.176 AEDT(fuel burn)*0.02940*0.997 AEDT(fuel burn)*0.03675 AEDT(fuel burn)*0.0009188*0.997 CO 2 AEDT(fuel burn)*3150 AEDT(fuel burn)*3150*0.997 HC AEDT(HC) AEDT(HC)*0.997 CO AEDT(CO) AEDT(CO)*0.997 PMFO (>1 km) AEDT(fuel burn)*0.015 AEDT(fuel burn)*0.015*0.997 PMFO (<1 kin) AEDT(PMFO from FOA3) AEDT(PMFO from FOA3)*0.997 S(VI) as H 2 SO 4 A baseline fuel sulfur content of 600ppm and ULS fuel sulfur content of 15ppm were assumed. 2% of fuel sulfur is assumed to be emitted as S(VI) (as H2SO4 for CMAQ). The 0.997 ULS emissions correction factor and 3159 g C02/kg-fuel El correspond to an anticipated 0.3% increase in the gravimetric fuel energy density, based on relationships derived from the DESC JP-8 database (DESC-BP, 1999-2007) using data from 2002-2007, and an assumed 1% loss in volumetric energy density. Note that AEDT uses a C02 emissions index of 3155 C02/kg-fuel. 2.2.2.2 NOx by Flight Mode AEDT outputs unspeciated NOx. NOx is comprised of the nitrogen oxides, NO and N02. NO reacts in the presence of 02 to produce N02, the production of which initiates chemical reactions creating 03 (Seinfeld & Pandis, 2006). A dynamic equilibrium exists in which N02 is further photodissociated during the day to NO (Pickering, Huntrieser, & Schumann, 2008). Inter-conversion between NO and NO2 is fairly rapid, with a tropospheric timescale of -5 minutes (Seinfeld & Pandis, 2006). Given this equilibrium, the ratio of NO/NO2 in aircraft emissions is of secondary importance, so long as the number of N molecules are conserved (NOx is expressed in AEDT on an NO2 mass basis). However, for CMAQ, NO and NO2 must be speciated in the aircraft emissions inventory. The fraction of NOx emitted as NO vs. N02 is dependent on the operating condition of the engine, as well as ambient conditions. At lower throttle levels i.e. idling during taxi, N02 dominates; at higher throttle levels, such as takeoff, climb-out, and cruise, NO dominates. This is corroborated through several campaigns in the literature i.e. (Wood, Herndon, Timko, Yelvington, & Miake-Lye, 2008), (Herndon, et al., 2004), (Wormhoudt, Herndon, Paul, Miake-Lye, & Wey, 2007). 0.01 0 W 20 40 60 100 Engine Power (% rated thrust) Figure 3: Emission indices of NO, N02, and HONO from a CFM56-3B1 engine (Wood, Herndon, Timko, Yelvington, & Miake-Lye, 2008) OH-driven oxidation of NO results in a small percentage of aircraft NOx emissions manifested as nitrous acid (HONO) (Miake-Lye, et al., 2002); for this study a uniform value of 0.8% was adopted. Figure 3 shows the split of NOx into NO, N02, and HONO by engine power level for a CFM563B1 engine in (Wood, Herndon, Timko, Yelvington, & Miake-Lye, 2008). This data, as well as the values quantified in Table 4 by flight mode, and similar values in (Wormhoudt, Herndon, Paul, Miake-Lye, & Wey, 2007), resulted in the speciated split by flight mode listed in Table 5. However, there is still a significant variance in values adopted for studies across industry; (Chapter 3 Emission Sources, 2006) (Chapter 3 - Emission Sources, 2006), an Air Quality (AQ) study of Heathrow Airport, used a cruise split with a significantly higher level of N02. Table 4: Calculation of total N02 and NOx emissions from a CFM56-3B1 during LTO cycle (Wood, Herndon, Timko, Yelvington, & Miake-Lye, 2008) LTD phase time in mode (miIn) fuel flow rate {kg/engine/s) NO2 El {g/kg fuel) NOx El (g/kg fuel) total NO2 (kg) total NOx (kg) 0.08 0.63 0.06 0.13 0.80 0.49 approach 4 0.29 1.09 7.04 idle 26 0.114 2.98 3.26 takeoff 0.7 0.946 1.41 17.52 climb-out 2.2 0.792 1.29 14.66 totals per engine/LTO ICAO " NOx emission indices are expressed as NO2 equivalents. The overall NO2/NOx ratio is 0.24. 0.58 0.70 1.53 3.30 3.60 Table 5: Speciated split of NOx into CMAQ species Taxi 7% Takeoff 100% 91.36% / 7.94% / 0.8% Climb-out 85% 90.47% / 8.73% / 0.8% 89.28% / 9.92% / 0.8% Cruise-climb 89.28% / 9.92% / 0.8% Cruise Cruise-descent - Approach 30% Landing Ground Roll with Reverse Thrust 2.2.2.3 8.5% / 90.7% / 0.8% 89.28% / 9.92% / 0.8% 83.8% 83.8% / 15.4% 15.4/ / 0.8% 0.8% Lightning NOx Lightning is characterized by localized heating and ionization of air, resulting in the dissociation of N2 and 02, and the formation of NOx (Pickering, Huntrieser, & Schumann, 2008). As lightning is a significant source of NOx at the same altitudes as aircraft emissions, its inclusion was deemed important to the study. Sensitivity of results to the inclusion of lightning NOx (LNOx) is discussed in Section 7.5. (Schumann & Huntrieser, 2007) have provided a comprehensive summary of the current state of understanding of LNOx, citing a probable range of values in total global emissions from the literature of 2-8 Tg N/yr, with a likely value of 5 Tg N/yr (values listed as equivalent nitrogen mass per year). The IPCC adopted a value of 5 Tg N/yr, with a range of uncertainty of 2-13 Tg N/yr (JPCC: Climate Change 2001, 2001). The GEOS-Chem model, which was used to create a lightning NOx inventory for CMAQ, predicted total global emissions of 6.16 Tg N/yr, with emissions over the CMAQ domain constituting 0.98 Tg N/yr thereof (Figure 4). GEOS-Chem emits 100% of LNOx as NO; the CMAQ LNOx inventories are likewise constituted solely of NO. The value of 0.98 Tg N/yr in LNOx emissions over the CMAQ domain in Figure 4 exceeds that of aviation emissions of NO at 0.633 Tg N/yr (listed in Appendix A.1). Seasonal variation manifests correctly as increased emissions in the summer months, followed by spring, fall and then winter. US Vertical Profile of Nfrom Lightning [g] 2006; Total = 9.83E 11 g 500 600 700 800 900 0 1 2 3 4 5 6 7 8 9 10 10 x 10 Figure 4: Vertical Profile of N from Lightning NO over the CMAQ US domain Figure 5 depicts the spatial distribution of predicted lightning activity in GEOS-Chem by summing all yearly LNOx emissions over the vertical domain of each horizontal grid cell. The 4 degree latitude by 5 degree longitude (4x5) grid cells of the GEOS5 domain is clearly evident in the Lambert conformal conic projection of the CMAQ domain. 2006 yearly sumof vertical columnar LNOx as NO [g] Min = 6.533385e+03 @ (98,4) Max = 2.231915e+08 @ (22,89) Mean = 6.181459e+07 x 10 12.2 0 20 40 , 60 80 Column Number 100 120 140 Figure 5: Vertical columnar sum of N [g N/yr] from lightning emissi ons of NO for 2006 The spatial distribution agrees well with empirical measurement of the frequency of lightning flashes from space-based optical instrumentation, shown in Figure 6. The GEOS-Chem lightning module was indeed written to be brought in line with the Lightning Imaging Sensor/Optical Transient Detector (LIS/OTD) space-based instruments (GEOS-Chem v8-02-01 Online User's Guide). -150 -120 -90 -60 -30 0 30 60 90 120 150 Figure 6: NASA NSSTC space-based observation of #flashes/km^2/yr., 1995-2000 (Where Lightning Strikes) 2.2.2.4 Hydrocarbon Split AEDT outputs a single unspeciated value for aircraft hydrocarbon (HC) emissions, denoted total hydrocarbons (THC). Hydrocarbons are emitted as the product of incomplete combustion processes. They can react photochemically with NOx to form ozone, a key component of smog in urban areas (primarily from ground transportation) (Seinfeld & Pandis, 2006). CMAQ groups hydrocarbons into families, including aldehydes and formaldehydes, simple chains like olefins and paraffins, and aromatics like toluenes and xylenes. A split of these species into CMAQ HC groupings is shown in Table 6, along with their mass fractions, and a full listing of these species by IUPAC nomenclature can be found in Table 20. A description of all CMAQ species by name is provided in Appendix A.1. THC must be converted to total organic gases (TOG) by a scaling factor (1.156 in this case) and multiplied by a split factor (SF) for molar-based emissions (Baek, Arunachalam, Holland, Adelman, & Hanna, 2007). For each species and grouping, the split factor is the product of the mass fraction and a scaling factor divided by the molecular weight, where the scaling factor is used to allocate a species into one or more groupings, based largely on the number and type of carbon bonds (Ryan, 2002). Emission [mol/s] = THC[g/s] * 1.156[TOG/THC] * SF (2) For a given grouping, the sum of all the species split factors for that grouping (many of which are zero) yields a total grouping split factor. For instance, propylene (C3H6) is equally divided between PAR and OLE (paraffins and olefins), and its split factors contribute to the summed split factor of both PAR and OLE (but not any others). All HC emissions are expressed in molar emissions rates i.e. in moles/s. The average molecular weight of all the species in a grouping is the mass fraction (MF) divided by the SF. ETH (ethene or ethylene) and MEOH (methanol) are treated explicitly, not as a grouping of species. The HC profile in Table 20 was based on a measurement campaign different and more recent than that used in the 1099 Civil Aviation profile in the EPA SPECIATE database (FAA OEE & EPA OTAQ, 2009). The SPECIATE database is the EPA repository of volatile organic gas and PM speciation profiles of air pollution sources (Hsu & Divita Jr., 2009). Table 6 compares the resulting formatted CMAQ HC profiles, which still retain broad similarity. Notable differences are increased aromatics in the ULS project profile, as well as slight increases in aldehydes, paraffins, and olefins at the expense of non-reactive compounds. Table 6: Hydrocarbon split into CB-IV species 2.2.3 ALD2 0.123467 0.003378 0.080683 0.002175 ETH 0.154590 0.005511 0.154800 0.005519 FORM 0.137595 0.004651 0.163100 0.005396 NR 0.047957 0.003526 0.168566 0.011104 OLE 0.086442 0.002966 0.059377 0.002584 PAR 0.381871 0.026194 0.356527 0.015279 TOL 0.024358 0.000248 0.009277 0.000139 XYL 0.027511 0.000260 0.007470 0.000120 MEOH 0.018052 0.000563 N/A N/A Processing Methodology Traditionally, CMAQ inventories are compiled from formatted ASCII text files using the SMOKE (Sparse Matrix Operator Kernel Emissions) modeling system (Baek, Arunachalam, Holland, Adelman, & Hanna, 2007). SMOKE is similar to CMAQ in that it is comprised of modules implemented in FORTRAN. For this study, however, greater flexibility was needed in assigning scaling factors, speciated splits by flight mode, custom HC profiles, and lightning NOx emissions. A way to custom-code the processing and injection of the inventories into CMAQ-ready netCDF (network Common Data Format) files was needed. MATLAB 2009a and later contains a netCDF module; this was used to push all processing into a single transparent MATLAB script. The disadvantage of increased processing time was offset by simple parallel implementation of the script, as all CMAQ inventory files contain one day of emissions. A SMOKE utility was still used to merge the background, LNOx, and aviation inventories. 2.3 Background Inventories 2.3.1 Background In this study, the background emissions inventory is the collection of all anthropogenic and biogenic sources of emissions, excluding aviation. Background emissions have a significant effect on the formation of secondary PM. The CMAQ inventory was adapted from that developed for the Clean Air Interstate Rule (CAIR) using the EPA's 2001 NEI database. The NEI is compiled every three years and includes emissions from industry, agriculture, mobile sources (including on-road vehicles, construction equipment, boating and shipping, rail, etc.), and biogenic sources (including VOC emissions, soil uptake and output, dust, wildfire, wildlife, etc.) (EPA, 2009). It is intended to be a comprehensive source of all emissions of criteria air pollutants and hazardous air pollutants (HAPs) within the US. 2.3.2 Species Population from National Emissions Inventory 2.3.2.1 2001 NEI vs. 2005 NEI The 2001 NEI was compiled for use with the Carbon Bond IV (CB-TV) chemical mechanism. The mechanism is discussed further in Appendix B. For a separate project, the 2005 NEI was compiled for use with the Carbon Bond V (CB-V) mechanism, an update of CB-IV with increased treatment of organics and HAPs, as well as updated rate constants. The differences between the mechanism versions are discussed in detail in (Luecken, Phillips, Sarwar, & Jang, 2008) and (Yarwood, Rao, Yocke, & Whtiten, 2008). The reference for results in this paper is CB-IV, as this mechanism was used for all runs. However, a comparison of several key species in the 2001 and 2005 NEIs is provided subsequently. There was insufficient time and space to do runs comparing the CB-IV and CB-V mechanisms and their effect on ULS results for CMAQ. 2.3.2.2 Surface Concentrations CMAQ results for aviation-attributable species concentrations display several "hotspots," or areas of elevated concentrations. The reasons for these hotspots may vary, and those of southern California and New York may be due to locally high air traffic alone; however, a significant cause is elevated concentrations of background gaseous ammonia (NH3), which is oxidized to form sulfates and nitrates. As can be seen in Figure 7, PM spikes in SE North Carolina, SE Pennsylvania, N Georgia, N Mississippi, Louisiana, and California's Central Valley are co-located with increased levels of NH3. Many of these same peaks show up when looking at aviation-attributable PM. Ammonia is the primary basic (i.e. reducing) gas in the atmosphere, and after N2 and N20 is the most abundant nitrogen-containing component in the atmosphere (Seinfeld & Pandis, 2006). Its destruction coincides with nearly all elevated levels of aviation-attributable PM. In this respect, the background inventory exerts perhaps its most potent influence. Animal waste is estimated to be the primary source of ammonia emissions, followed by ammonification of humus, emission from soils, losses of NH3-based fertilizers from soils, and industrial emissions. For example, the NH3 spike in SE North Carolina corresponds to the state's largest and most concentrated hog farming region, in a state that is one of the country's largest producers (Hog & Pig County Estimates, 2008). Further background surface concentration plots are provided in Appendix A.2 Max - 2.125e+01 @(53.21) Ii 2006yearlyaverage FRMPM2.5Iug/m^31 background Min-000@(11) Max- 3.12601 @(14.67) Mean- 5.7825e+00 2006yearlyaverage NH3jppbVJ background Min - 000@(1 1) Mean - 6.9110e-01 70 002S as Ito 20 00 so so 100 120 100 140 120 140 2006yearlyavg (tullaviation-no aviation)FRMPM25[ug/m^31 Min--8.023e-03@(13,69) Max- 1.741e-01@(67.130) Mean-2.3919e-02 2006yearlyavg NH3[ppbVl (tullaviation- no aviation) Min--3.250e-02@(4624) Max-1.976e-03@(13.69) Mean--22032e-03 e&OB 0 40 go so C06LMW*A 100 120 140 0 20 40 W0 W 1W Figure 7: 2006 background NH3 and PM2.5 concentration showing several co-located hotspots 2.3.2.3 Vertical profiles In contrast to aviation, most other sources of emissions (those in the background) are confined to the planetary boundary layer (PBL), a region of the atmosphere variably extending up to about 1km (-900hPa on the plots below), characterized by rapid turbulence and strong vertical mixing (Seinfeld & Pandis, 2006). Figure 8 shows the variability in several key species between the 2001 and 2005 NEIs. S02 and the H2SO4 into which it reacts show peaks above the ground layer from the plumes of coal-fired power plant/industrial flues. Such spikes are visible at the same altitudes for NOx and primary sulfates. All species show strong spikes or maxima at the ground level. US Total S02 Emissions [Tg] US Total N02 Emissions [Tg] US Total H2SO4 Emissions [Tg] 900 900 920 920 940 940 2001 JI -%-- CUI r S960 - 2005 ' 960 960 1000 0 2 6 4 0 US Total NO Emissions [Tg] 0.05 0.1 0.15 0.2 0 920 920 940 940 1 \ 940 960 960 960 960 960 960 1000 1000 15 10 A 1000 0.05 0.1 0.15 0.2 0 900 900 920 920 920 940 940 940 960 960 960 960 960 960 1000 10001 1000 1.5 1 0.5 2 0 0.2 0.4 0.6 0.6 0.005 0.01 0.015 US Total CO Emissions [Tg] US Total PEC Emissions [Tg] 900 0 2 920 \ 0 US Total POA Emissions [Tg] 1.5 900 900 5 1 US Total PNO3 Emissions [Tg] US Total PSO4 Emissions [Tg] 900 0 0.5 50 100 150 US Total NH3 Emissions [Tg] 900 920 940 5 960 960 1000 0 2 4 6 Figure 8: Comparison of 2001 and 2005 NEI emissions profiles over CMAQ US domain The trend in downwards emissions from the 2001 to 2005 NEI reflects increases in efficiency, improved technology, and regulation in the last several decades, which have overcome increases in capacity. Indeed, the EPAs historical data for national emissions trends (National Emissions Inventory Air Pollutant Trends Data, 2009) shows a steady downward trend in S02, NOx, VOCs and CO emissions dating from 1970 (the earliest year for which an NEI was compiled), with a more than 50% reduction over the past four decades in those species (Figure 9). NH3 emissions have remained largely static, while PM10 and PM2.5 have also shown decreases during the years for which data is provided. 40 35 30 25 -2kNH3 -x-NOx 20 -PM10 -e-PM2.5 15 -+-SO2 -- VOC YEAR Figure 9: NEI emissions trends of several key species (National Emissions Inventory Air Pollutant Trends Data, 2009) .... .... ...... US Total S02 Emissions [Tg] 900 US Total H2SO4 Emissions [Tg] US Total NO2 Emissions [Tg] 1 1000 0 0.2 0.6 0.4 0 0.8 US Total NO Emissions [Tg] 0.005 0.01 0 0.015 US Total PSO4 Emissions [Tg] 900, 900. 920 920 920 940 940 940 960 960 960 980 980 980 1000 1000 0 1.5 1 0.5 0 0.005 0.01 0.2 US Total PNO3 Emissions [Tg] 900 1000 0.15 0.1 0.05 1 0.5 0 0.015 x 10 US Total POA Emissions [Tg] 900 900 900 US Total CO Emissions [Tg] US Total PEC Emissions [Tg] 9501 1000 1000 0 0.05 0.1 0.15 0.2 0 0.02 0.04 0.06 0 5 10 15 US Total NH3 Emissions [Tg] 900 920 940 S960 980 1000 0 0.2 0.4 0.6 0.8 Figure 10: Seasonal Variability in the 2001 NEI for several key emissions species As can be seen in Figure 10, there is little seasonal variability in most key species, with the exception of NH3, which shows a more than 200% increase from a winter to summer month. In the absence of other mitigating effects, this would point to higher PM formation in the summer months. Indeed, background concentrations of PM2.5 increase nearly 50% from winter to summer. The extent to which this represents correlation or causation is difficult to quantify, but the role of NH3 in PM formation is well-established. 2.4 GEOS-Chem Boundary Conditions 2.4.1 Background Several species exhibit sufficiently long atmospheric residence times that transport from outside the CMAQ domain, i.e. from Europe or Asia/Pacific, impacts local concentrations. The desire to capture this interaction leads to the use of boundary conditions that are created by a model with global scope. GEOS-Chem was used to create the boundary conditions for CMAQ. The GEOSChem runs were perturbed with the same scenario as the CMAQ runs for which they provided the boundary condition. For example, the global GEOS-Chem inventory for the run used to create the CMAQ full aviation ULS scenario boundary conditions contained full aviation emissions with a ULS fuel policy. Sensitivity to the boundary conditions is documented in Section 7. GEOS-Chem was run on a 4x5 degree global grid. Further details about the model can be found in (GEOS-Chem v802-01 Online User's Guide). 2.4.2 GEOS2CMAQ and methodology GEOS2CMAQ is a modular Fortran utility written to map the GEOS-Chem grid to CMAQ boundary conditions. CMAQ boundary conditions are a simple additional "picture frame" layer of concentrations in the grid around the CMAQ domain. The utility performs horizontal and vertical spatial interpolation of the GEOS-Chem grid to the CMAQ grid, as well as species mapping (listed in Table 7), outputting concentrations every three hours (Moon, Park, Li, & Byun, 2008). Note that CMAQ has no vertical boundary condition, i.e. there is no vertical transport across the 1OOhPa upper layer boundary. Further description of GEOS-Chem species can be found in (GEOS-Chem v8-02-01 Online User's Guide). Description of all CMAQ species can be found in Appendix A. Table 7: GEOS-Chem to CMAQ species mapping (Moon, Park, Li, & Byun, 2008) CMAQ Species Species. ,GEOS-Chem N02 NOx 03 Ox -NOx CO CO N205 N205 HNO3 HNO3 PNA HNO4 H202 H202 UMHP MP NTR R4N2 FORM CH20 ALD2 0.5*ALD2 + RCHO PAR 0.333*PRPE + ALK4 + 0.5*C3H8 + 0.2*C2H6 + ACET + MEK + RCHO OLE 0.333*PRPE PAN PAN + PPN + PMN ISOP 0.2*ISOP ISPD MACR + MVK TERP ALPH + LIMO + ALCO S02 S02 + DMS NH3 NH3 SGTOT TRP 1 0.2*SOG1 + 0.2*SOA1 + 0.2*SOG2 + 0.2*SOA2 + 0.2*SOG3 + 0.2*SOA3 SGTOT TRP_2 0.8*SOG1 + 0.8*SOA1 + 0.8*SOG2 + 0.8*SOA2 + 0.8*SOG3 + 0.8*SOA3 ASO4J S04 + MSA 2.4.3 ANO3J NIT ANH4J NH4 AORGBJ 0.847*SOA1 + 0.904*SOA2 + 1.24*SOA3 AORGPAJ 0.0545*OCPI + 0.0545*OCPO AECJ BCPI + BCPO A25J 0.145*DST1 + 0.145*DST2 ANAJ 1.03*SALA ANAK 1.03*SALC ACLJ 1.03*SALC ACLK 1.03*SALC ASOIL 0.29*DST3 + 0.29*DST4 Verification A simple verification of the GEOS2CMAQ outputs was conducted by plotting the raw GEOSChem outputs for the background case on the same plot as the ring of CMAQ grid cells forming the CMAQ BCON. Yearly averages were used, and two species at two levels were tested: 03 and HNO3, at ground level and ~860hPa (10th layer), plotted in Figure 12. An exhaustive proof was not deemed necessary, as this verifies species and vertical mapping of the implementation of the code, which has itself already been verified at length by (Moon, Park, Li, & Byun, 2008). The species were chosen because they have significant concentrations across nearly the entire CMAQ domain boundary. Note that results need not match exactly, as a less rigorous vertical mapping was performed for this test than that of GEOS2CMAQ (which uses a non-hydrostatic vertical structure created with time-changing pressures in the meteorological data, not from a static definition, as well as other more complex corrections). Nonetheless, results agree very well, both spatially and in magnitude. The efficacy of the comparison diverges the more ground elevation increases from sea level, and the higher the vertical layer; for that reason, two lower layers were used. Figure 11 shows the CMAQ boundary conditions highlighted outside the GEOS-Chem outputs that were used to create them. 0 100 50 150 Column Number Figure 11: Ground level 03 concentrations [ppbV] from GEOS-Chem (inside the black box) and the CMAQ BCON created from this GEOS-Chem run (outside the black box) c.) d.) 100 COWNumbe, ISO Figure 12: Ground level 03 concentrations in a.), and at -860hPa (layer 10) in b.); ground level HNO3 concentrations in c.), and at -860hPa in d.) Of some note, however, is the background concentration of H2SO4, which is the only species that shows a clearly higher concentration at the boundary than within the domain (as shown in Figure 13). The difference also manifests in the full aviation delta scenario (Figure 14b). H2SO4 from the boundaries can be transported to react with NH3 to form (NH4)2SO4, contributing to increased PM within the domain background H2SO4 [ppbV] 2006 yearly average Min = 000 @ (1,1) Max = 5.906e-03 @ (23,66) Mean = 3.4887e-04 0.01 0.009 0.008 0.006 0.005 0.004 0.003 0.002 0.001 0 0 0 20 40 60 80 Column Number 100 120 140 Figure 13: Background H2SO4 showing higher concentrations at domain boundary at ground level Boundary condition gradients are also evidenced in the delta scenarios. This occurs when GEOSChem predicts a larger aviation-attributable perturbation to a species than CMAQ that is able to propagate to the domain boundary. The most marked examples include NH3 (Figure 14a), in which elevated concentrations propagate from the northern boundary, and H2SO4 (Figure 14b), in which elevated concentrations may propagate from the eastern boundary, though weather generally moves west to east over the domain. These ground level plots do not speak to the transport processes in the free troposphere and jet stream, but they do indicate significant boundary effects, which may extend upwards. NH3 in particular drives PM formation in CMAQ. Elevated concentrations of NH3 in GEOS-Chem for full aviation are particularly illustrative. Aviation can only act as a sink for NH3, as it is not produced by aviation, but is destroyed by reaction with SOx and NOx to form NH4, as is shown throughout the rest of the CMAQ domain. Knowing this, it may be possible to ascribe some of the contribution from the boundary conditions to artifacts in the model results. The swath of reduced concentration that breaks the elevated levels, roughly centered at column 50, is in fact due to the reaction of NH3 to form PM, i.e. the swath of elevated difference in emissions from column 20 to 100 is continuous. This can be seen most clearly in the LTO-only plots in Figure 38 in Section 7.4, in comparing the runs with static and changing boundary conditions. The elevated NH3 emissions develop a gradient that extends southeastwards to the main N03 reaction regions of the Midwest. b.) a.) (fullaviation- noaviation)NH3{ppbV) 2006yearlyavg Min- -3.250e-02 @(46.24) Max- 1,976e-03 @(13,69) Mean- -2.2032e-03 100 (full aviation-noaviation)H2S04[ppbV) 2000yearlyavg Mi -- 1.308e-05@(67l0a) Max- 703505@(23,148) Mean-i.9035e-0X -- 0.02 0.015 10 001 0,005 1E 60 40 40 0.01 20 20 -0.015 0 0 0 00 4 40 00 s0 800 Column Number i00 120 14 040 00 20 40 t0 80 100 120 140 Column Number Figure 14: Full aviation delta in ground-level NH3 and H2SO4, showing clear boundary layer effects 2.5 Meteorological Data 2.5.1 Background Meteorological data, or met data, is of integral importance in an atmospheric model. It dictates the advection of species by wind, wet deposition by precipitation, pressure, temperature, relative humidity (and the gas-particle equilibrium that is so dependent upon it), degree of photolytic reaction shielding from clouds, and other parameters that strongly influence the reaction and persistence of species. CMAQ met data is created from the outputs of the Mesoscale Model version 5 (MM5), a model designed to simulate or predict mesoscale or regional scale atmospheric circulation (Chapter 8: MM5). The MM5 parameters used are listed in Table 8. CMAQ met data is not affected by emissions, meaning feedbacks are not included e.g. changes in cloud formation with changes in aerosol cloud condensation nuclei (CCN). The CMAQ met data is derived from the same source as that used for GEOS-Chem; however GATOR and TOMCAT used different data (self-created in the case of GATOR). As an interesting aside, the MM5 data is sufficiently wellresolved in space and time that it accurately modeled the path of Hurricane Ernesto in AugustSeptember of 2006, the only hurricane to make extended landfall along the east coast in the 2006 season. Table 8: CMAQ/MM5 meteorological parameters and mechanisms Meteorology-Chemistry Interface Processor MCIP v3.3 (frozen 08/01/2007) Input Meteorology MM5 v3.6.1 Cumulus Parameterization Grell (Grell & Devenyi, 2002) Microphysics Simple Ice (Dudhia) (Dudhia, 1989) Longwave Radiation RRTM (Mlawer, Taubman, Brown, Iacono, & Clough, 1997) Shortwave Radiation Dudhia (Dudhia, 1989) PBL Scheme MRF (Hong & Pan, 1996) Surface Layer Scheme Standard Land-Surface Scheme OSU or NOAH Land-Surface Model (Mitchell, 2000) Land Use Classification USGS (Andersen, Hardy, Roach, & Witmer, 1976) 3D Analysis Nudging On SFC Analysis Nudging Off OBS Nudging Off 2.5.2 Vertical Profiles US Average Cloud Water Mixing Ratio [g/g] 2006 US Average Temperature [K] 2006 1000 L 200 A 220 240 280 260 300 0 0.5 1 1.5 2.5 2 -s x10 US Average Rain Water Mixing Ratio [g/g] 2006 0 0.5 1 1.5 US Average Water Vapor Mixing Ratio [g/g] 2006 0 0.005 0.01 0.015 x10 x15 Figure 15: Vertical profiles of US average temperature, cloud water mixing ratio, rain water mixing ratio, and water vapor mixing ratio Figure 15 shows several domain- and monthly-averaged vertical profiles of several key parameters. As a domain average, the data is not amenable to comparison with other models, but it shows the extent of intra-annual variability in the met data in CMAQ. For instance, decreased rain water mixing ratio in July below -500hPa may imply higher PM concentrations from reductions in wet deposition, in the absence of other mitigating effects. 2.6 Particle-Bound Water and Dry vs. Wet PM The EPA's federal reference method (FRM), based on the SANDWICH model (Sulfate, Adjusted Nitrate, Derived Water, Inferred Carbonaceous mass Hybrid material balance), dictates that particlebound water (PBW) attached to sulfate, nitrate, and ammonium be included in the accounting of PM2.5 mass. C PM2.5FM = OCM,= + BC + S04+ N0 3 FRm + NH 4 Fm + H20 + crustal + 0.5 (3) Here, OCM,,b is organic carbon mass by difference (i.e. it is derived last), crustal is soil and dust, and 0.5 is a blank mass constant. The blank mass and crustal disappear when the background is subtracted from an aviation scenario. The PM2.5 FRM sampler is specified by design. It contains a sharp-cut particle size separator and a detailed monitoring protocol. PM2.5 mass is determined gravimetrically by weighing the pre- and post-sampling filters that have been equilibrated for a minimum of 24 hours at standardized conditions of 20-23'C and 30-40% relative humidity (RH). Collectively, the FRM is intended to provide reproducible mass concentrations and to determine compliance with standards (Frank, 2006). In this study, dry PM is taken to be the sum of sulfates, nitrates, ammonium, black carbon, and organics. PBW is output by CMAQ, though it is not output in all models. (Frank, 2006) plots PBW as 0.24 * (S04 + NH4), yielding a good first-order approximation derived from measurements at six diverse FRM monitor sites. A more complicated fitted relation varying with S04, NH4, and N03 is used in the EPA Model Attainment Test Software (MATS), which represents the standard technique for allocating PBW in hydrated PM for reporting PM concentrations. The relation itself is comprised of more than twenty terms, conditional on the degree of neutralization (DON), defined as the amount of NH4-associated-with-SO4 divided by the total amount of S04 (Abt Associates, Inc., 2009). However, a comparison of the outputs of the first order approximation described in (Frank, 2006) and that described in the MATS documentation shows good agreement, and as the first order approximation is easier to use, and sufficiently accurate to represent the PBW in the total PM values used to derive the epidemiological Concentration Response Functions (CRFs) upon which health impacts are based, it is used exclusively in this thesis to hydrate PM for all models. A comparison of the CMAQ output PBW (AH2OI + AH2OJ) to the SANDWICH approximation, shows a significant divergence, including a notable bias towards higher water concentrations in nitrate-heavy regions in CMAQ (Figure 16), which are also generally characterized by higher relative humidity. However, the transformation to wet PM should reflect the underlying assumptions in the FRM-monitor-derived epidemiological CRFs. Hence, that method is used to create hydrated PM, with the caveat that it under-predicts model values. b.) a.) yearlyaverage )ug/m'3]2006 approx.) PBW(SANDWICH (fullaviation- background) Mean-21846860-03 Min--,8115650-03@(11,69) Max- G49399-02@(67,130) 2006yearlyavg H20 (PBW)(ug/m^3) (fullaviation-no aviation) Min- -4,98ge-03@(1368) Max- 1.931e-01@(73.85) Man -2.257le-02 .00 .n4000010 0 20 .z so go ColumNurVer we a 4 Colisnn Nunew Figure 16: Comparison of CMAQ output PBW (a) to FRM-derived PBW (b) (note the different scales) 2.7 2.7.1 Population Data and Health Impacts Global Population of the World (GPW) and Global Rural-Urban Mapping Project (GRUMP) Population is used to derive health incidences for a change in concentration within a given grid cell through a relation known as the concentration response function (CRF). The Center for International Earth Science Information Network (CIESIN) of Columbia University hosts data demonstrating the spatial distribution of human populations across the globe (Center for International Earth Science Information Network (CIESIN), 2005). The highest resolution data is available under the Global Rural-Urban Mapping Project (GRUMP) at 30 arc-seconds (-1km). Slightly coarser information is available in the Global Population of the World (GPW) datasets, at 2.5 arc-minutes (-5km). Global baseline health incidences were calculated by (Barrett, Britter, & Waitz, 2009) at this resolution, and so this dataset was adopted exclusively for all models in this study, with population for the year 2000. A comparison of GRUMP and GPW data gridded to the CMAQ domain show very similar results. The data is compiled by CIESIN from censuses and surveys administered for political or administrative units. The approach uses a simple proportional allocation of administrative unit population totals over grid cells (Deichmann, Balk, & Yetman, 2001). Total338 Mlion IoglO0popoiaton) 0AQ doooiln Min.ltt@(1,1) M 6.702692e.00@(67,l30) Moot, lot 338.45M Population [persons)Total: CMAQ domain Min- 0 @(1,1) Max-5.043031e06@ (67,130) Men- 2.041795e+04 100 4 A6 44 3 404 60 Figur 17:CMAQ omai popuatio 80' 2 Ar esons Fnctons(C Conentaton 60 2.7.2 Concntraton Rspons 0 Z84O0 0 20 20 0 MPllio Aidmint c(op~toncenotati33n.4f s)reatechnge 6 zo 40 80 8 Number Column 100 00 20 10 Funtion 40 10 100 80 68 0 100 20 2 120 40 140 80 so 100 120 140 Number C01um Figure 17: CMAQ domain population 2.7.2 Concentration Response Functions Concentration Response Functions (CRFs) relate changes in ambient concentration of PM to changes in health incidences. Figure 18 illustrates the role of CRFs in deriving health cost end points from changes in emissions. Chong"eI aemntr conoentndionsm Ernhionmodeling " changeo In heolah endpoint incdewme rQUN Pokey scengio Aambiet concetration Aheahh costs = cmissions x. ACMssions health mcidence .. x cost aambicn onccntration hcalth incidcne Figure 18: Health impact pathway for air quality assessment (Rojo, 2007) Long-term PM exposure is associated with respiratory illness in children, cardiopulmonary mortality in adults (Pope & Dockery, 2006), premature mortality (US EPA, 2006), lung cancer (Ostro, 2004), cardiovascular and pulmonary hospital admissions, and asthma aggravation (US EPA, 2004). A 2004 Journal of American Medicine study estimated total yearly PM-attributable mortalities in the US at between 22,000 and 52,000 (Mokdad, Marks, Stroup, & Gerberding, 2004). The shape of the CRFs and the existence of a no-effects threshold level have been the subject of much study in the evolution and evaluation of air quality standards (Pope & Dockery, 2006). Exponential and piecewise functions have been used, but (Pope & Dockery, 2006) state that the shapes are generally linear or weakly log-linear with no well-defined threshold. A World Health Organization (WHO) report recommends the use of the following log-linear CRF (Ostro, 2004): RR = [(X + 1)/(Xo + 1)]f (4) Here, RR is the relative risk, X is the annual mean concentration of PM2.5 under the aviation P is an empirical scenario, X 0 is the annual mean background concentration of PM2.5, and coefficient for each health condition. Low, nominal and high values are cited in Table 9. For this study, health incidences were derived only for cardiopulmonary mortality and lung cancer in individuals over the age of 30, based on the methodology in (Barrett, Britter, & Waitz, 2009). However, (Sequiera, 2008) has shown that 97% of annual monetized health costs from aviation are due to premature mortality, and of those more than 99% are due to premature mortality in adults over 30 (as opposed to infants < 1 year old). Therefore, the overwhelming majority of aviationattributable health costs are included. Table 9: P coefficient values (Ostro, 2004) Low Nominal High Cardiopulmonary Mortality 0.0562 0.15515 0.2541 Lung Cancer 0.08563 0.23218 0.37873 1 Aviation-attributable mortalities are derived by scaling existing incidences by the reduction in relative risk: -mortalities= baseline * [1 - [(X + 1)/ (X + 1)] fl] (5) Here, baseline is the total number of deaths by a given cause (here, lung cancer or cardiopulmonary mortality), compiled from WHO Global Burden of Disease (GBD) reports (Barrett, Britter, & Waitz, 2009). Global and US-only lung cancer and cardiopulmonary mortalities are plotted in Figure 19 on a logarithmic scale, showing highest global incidences in India, China and Europe. In the US, the WHO yearly lung cancer mortality figure of ~176,000 closely matches data from the US Center for Disease Control (CDC) for 2006 at 158,599 (Lung Cancer, 2006). The difference may be explained by the inclusion of mortalities from parts of Canada, Mexico, and the Caribbean in the CMAQ domain, as well as different sample years. Baselinecardiopulmonary Incidences from PM2.5[ogl0(mortalities)] Total- 14.135Million Baselinelung cancer incidences [og10(mortalities)] Total - 1.262Million 6 1N N NV N N 0: S E 0 'E 9000E 120'E 150 E101 20', I Beg o 90', 60 30', 0 30 E 60' E 90' E 120 E 150*E 180*E Totl 76 s Baseline ncden 941120a d l0(muities] Total [og 678 s)] a dacancer mon deces L,0 O 0'N-0 60 -6 2a m mema 66 -2L 1.5 z6 40 40 GOMH4ind VOf(,s byx OH14in2the00presence ag 0.5 of20 N--x. 20 0 , 20 40 60 s0 Column Number 100 120 140 a 2D MortII@(~ aieA0.5 from3e.0@6,10 aain-atriutbl 40 so 80 Column Number 100 120 140 Figure 19: Baseline cardiopulmonary and lung cancer mortalities In addition to PM, O3 in the troposphere is a pollutant with adverse health impacts; only in the stratosphere does it play a positive role, where much higher concentrations contribute to the absorption of UV radiation. Ozone is produced in the troposphere by photochemical oxidation of CO CH4, and VOCs by OH in the presence of NOx. Mortalities from aviation-attributable changes in 03 at the surface are not within the scope of this thesis. However, (Ratliff, et al., 2009) found changes in aviation-attributable mortalities and other health incidences from 03 to be neutral or even slightly beneficial on balance, reflecting both increases and decreases in ground concentrations depending on conditions. Potential differential toxicity between the species of PM is not treated, which may be important to the extent that aviation-attributable PM differs in chemical composition and size with respect to the polluted urban air upon which epidemiological CRFs are derived. PM precursor pollutants emitted from aviation are not different than those emitted from other combustion sources, implying that the impacts of secondary aerosols are comparable to non-aviation sources. However, aviation BC emissions are typically finer than non-aviation sources, and recent studies indicate that ultra-fine BC may have particularly high toxicity (Liao & Chio, 2008); as such, aviation BC impacts may be more important than results from the present analysis indicate (Barrett, Britter, & Waitz, 2009). 2.7.3 Monetary Value of a Statistical Life (VSL) Once mortalities have been derived from the CRFs, it is necessary to assign a monetary cost per death. The EPA uses a hedonic value of a statistical life (VSL). Hedonic valuations measure the willingness of an individual to pay to avoid increased risk. For example, if an individual is willing to pay $600 to avoid a 1 in 10,000 chance of dying, that individual's life is statistically valued at $6 million. In its 2000 Guidelines, the EPA advocates a VSL of $7.4 million in 2006 dollars (EPA, 2000); however, there is quite a bit of variability in the literature. The Aviation Portfolio Management Tool (APMT) uses a mean value of $6.3 million in 2000 dollars, which translates again to $7.4 million in 2006 dollars. That value is used exclusively in this study for valuing US mortalities. Elsewhere around the world, such a valuation is inflated, and project results are provided only in non-monetized mortalities. 2.8 Modeling Context 2.8.1 Earth-Atmosphere Modeling of Aviation The atmosphere itself presents a range of spatial scales in its motions that spans eight orders of magnitude, from eddies on the scale of centimeters, to huge air mass movements of continental size. (Seinfeld & Pandis, 2006). CMAQ is a regional, or mesoscale model. Mesoscale denotes phenomena occurring on scales of tens to hundreds of kilometers, such as land-sea breezes, mountain-valley winds, and migratory high- and low-pressure fronts (Seinfeld & Pandis, 2006). Several species are important in analyzing the production of aviation-attributable PM. NOx, S02, tropospheric 03, and aerosols all have lifetimes of days to several weeks. Strongly oxidizing species such as H02, H202, and OH (the most important oxidizing species for tropospheric compounds) are short-lived but integral to the production of sulfates and nitrates from SOx and NOx. For example, HNO3 reacts with NH3 to form NH4NO3, or ammonium nitrate; HNO3 is produced by the reaction of N02 and OH. OH is itself formed during the day by the reaction of H20 and 01D, created from photolysis of 03. In the analysis of CMAQ results, the following species were examined as significant in the study of aviation-attributable PM: S02, NO, N02, 03, HNO3, H2SO4, aerosols (S04, N03, NH4, BC, ORG), OH, H202, and NH3. The temporal and spatial scale of reaction of several of these species are plotted in Figure 17. Urban or Regional or Synoptic to Local scale Mesoscale| Global scale Microscale ng-lived . 100 yr species CFCs) NPI CH4 CH 1CCl1, 10 yr I ~ kOCH3Br -- Lived Species L0*Aerosols Trop O 1Iday .@s02 H2O2 day* -I NI x t Mixing Tim Intra-hemisphenc Mixing Time Bounday Layer Mixing Time -- ./~ ~ Shr-lived ioo [ DMS Ih Inter-hemispheric * Co ' Modertely Long u 100 g *CH3Oj O gCH H02 ON03 Im l10m 100m 1km 10km 1(K0kmlOOWkrnIO,OOkm Spatial Scale Figure 20: Spatial and temporal scale of atmospheric processes (Seinfeld & Pandis, 2006) One of the most important atmospheric cycles in the study of aviation's impacts is the creation and destruction of 03, which is integral to the oxidation of precursors to PM. Ozone in the troposphere is generated from two major types of precursors: VOCs and NOx (Seinfeld & Pandis, 2006). 03 formation is initiated by the reaction of OH with organics, and a subsequent interwoven network of free-radical reactions is catalyzed by NOx. Aviation is the sole significant source of VOC emissions above the PBL, and one of the two major sources of NOx emissions, along with lightning. The significance of the role NOx plays as a catalyst can be seen in Section 7 describing the sensitivity of results to lightning NOx, in which the removal of lightning NOx results in a significant increase in oxidizing capacity from the NOx created by aviation, especially in the summer months. 2.8.2 Atmospheric Aerosols Particulate matter is any substance, except pure water, that exists as a liquid or solid in the atmosphere under normal conditions and is of microscopic or submicroscopic size but larger than molecular dimensions. In studying aviation, we usually refer to "dry" and "wet," or hydrated PM. Wet PM contains water that has condensed onto a particle, inflating its mass. In analyzing the health impacts of PM the particle-bound water (PBW) contributes mass, but the health effects are correlated with the amount of hydration assumed in the FRM as described earlier (30%-40% RH and 20-23'C). A full description of atmospheric particles requires specification of not only their concentration but also their size, chemical composition, phase (liquid or solid), and morphology (Seinfeld & Pandis, 2006). In spite of the specific processes that affect particulate aging, the usual residence time of particles in the lower atmosphere does not exceed several weeks. Very close to the ground, removal is effected by dry deposition and settling on surfaces. Above 100m, precipitation scavenging (incorporation into cloud droplets during formation of precipitation) is the predominant removal mechanism. Differences in meteorology between models have significant effects on aerosol concentration. Particles that can become activated to grow to cloud droplets in the presence of a super-saturation of water vapor are termed cloud condensation nuclei (CCN) (Seinfeld & Pandis, For sulfur compounds with an estimated average residency of one week, the troposphere is 2006). not even well-mixed vertically. Particles less than 2.5um in diameter are generally referred to as "fine" and those greater than 2.5um diameter as "coarse." The fine and coarse particle modes originate separately, are transformed separately, are removed from the atmosphere by different mechanisms, require different techniques for their removal from sources, have different chemical composition, have different optical properties, and differ significantly in their deposition patterns in the respiratory tract. Therefore the distinction between fine and coarse particles is a fundamental one in the physics, chemistry, measurement and health effects of aerosols (Seinfeld & Pandis, 2006). That being said, the CRFs used in this study do not distinguish between type of PM, which is as of yet an unquantified uncertainty. CMAQ contains three size bins, or modes, of aerosol. The nucleation (or nuclei) mode comprises particles with diameters up to 10nm. The Aitken mode spans particles from 10nm to 100nm (0.1um) in diameter. These modes account for the large majority of particles by number, but only a small percentage of particles by mass and are lumped together in CMAQ. The accumulation mode, spanning particles from 0.1um to 2.5um in diameter, usually accounts for most of the aerosol surface area and a substantial part of the mass. Particles in the accumulation mode are formed from coagulation of smaller particles; at this size, particle removal mechanisms are least efficient, hence particles tend to accumulate (Seinfeld & Pandis, 2006). The coarse mode is comprised of particles larger than 2.5um in diameter; however, essentially all aviation emissions are smaller than 2.5um. As discussed in the section on climate feedbacks, atmospheric aerosols can have either a warming or a cooling effect on the atmosphere, depending on cloud interactions, altitude and optical properties. Several past studies have found aerosols to have a net cooling effect. In particular, in the 2009 authors update of the IPCC reports, (Lee, et al., 2009) predict that total global radiative forcing (RF) from aerosols exert a net cooling effect from direct backscattering and some types of cloud formation that's nearly 75% that of the warming effect from elevated C02, notably with uncertainty bounds falling entirely on the side of cooling (Figure 21). However, the level of scientific understanding (LOSU) and characterization of these processes is cited as medium to low. In general, however, there is thought to be a tradeoff between the air quality and climate effect of aerosols, as increased PM (and the health incidences it brings) nonetheless combat global warming through negative RF. Any such effect, however, is far from symmetric, as the spatial patterns of warming from GHGs and cooling from aerosols are quite different. Global Radiative Forcing Components in 2005 RF values RF Terms Long-lived greenhouse gases 1 LHalocarbons SOzone-005 CD OzntaopeicToopei Stratospheric water vapour from CH 4 Surface albedocarbon SraeabdLadueon < Total aerosol 1.. effect Solar irradiance Global High Global High [-0.15 to 0.051 0.35 [0.25 to 0.65] Continental to global Med 1 -0.56 [-0.9 to -0. 1] Contgnenal Medo 1 -0.7 [-1.8 to -0.3] Continental Low 0.01 [0.003 to 0.03] Continental Low 0.12 [0.06 to 0.30] Global LOW 1 1 1 -1 0 Med Low - Total net anthropogenic. -2 1.66 [1.49 to 1.831 0.48 [0.43 to 0.53] 0.16 [0.14 to 0. 18] Local to continental I Linear contrails 1t Spatial scale LOSU -0.2 -0.4 to 0.0 0.1 [0.0 to 0.2] snow Direct effect Directblffect Cloud albedo 2 (W M ) 1 Radiative Forcing (W rn- 2 ) [0. to global to .] 2 Figure 21: Global Radiative Forcing ftom All Anthropogenic Sources of Emissions (Lee, et al., 2009) Aviation-attributable aerosols are emitted over a larger range of altitudes than other anthropogenic sources and contribute to induced upper-altitude cirrus cloudiness. (Lee, et al., 2009) predict sulfate aerosols from aviation have a cooling effect, but the LOSU is low. Sulfate aerosols in the upper troposphere and stratosphere from volcanic emissions have had profound climate effects. In 1991, the eruption of Mt. Pinatubo in the Philippines and corresponding injection of -i0Tg i02 into the tropical stratosphere resulted in an average cooling of the earth by adiC the following year (Crutzen, 2006). Stratospheric aerosols have significant longer residence times due to low cross-tropopause transport. Indeed, a leading strategy for geoengineering, or the deliberate manipulation of the climate to combat the effects of global warming, is the injection of sulfate precursors into the upper stratosphere. The Nobel Prize winning chemist Paul Crutzen has posited that a (sizeable) continuous stratospheric sulfate loading of 5.3Tg S would be sufficient to counteract the 4 W/m^2 RF from a doubling of atmpospheric C02 concentrations (Crutzen, 2006). By comparison, total global aviation output of SOx is on the order of 0.1Tg S. Aviation Radiative Forcing Components in 2005 [(Wm) RF Terms 00280 Carbon dioxide produiction Methane Ozoe0263 I pro Neduction I SU Globa igh 9hemisphei jj -LT9 L Global (0.0115) Kemnspheric -0.0028 Water vapour Sulphate aerosol LO scale Conflnenital tonphn (0.219) -0.0125Md obal (4,0104) 0 Total NOx Spatial LOW Best estimate Sulhae arool I ~-0.0048 L2Estimate values) o -- 90% confidence Local gloalto Low Local Low '(IPCCAR4 Soot 0aeroso 001 0.0034 0-0025) Linear contrails Induce cirrus0033 cloudiness 0.055 (0.0478) Total aviation (Excl. induced cirrus) Total aviation -0.08 -0.04 0.08 0 0.04 Radiative Forcing (W rn- 2 ) Low Local to hemispheic Very Low ll 0.078_Global 7l (Inct induced cIrrus) to global Low continental L Low 0.12 Figure 22: Global Radiative Forcing from Aviation Emissions (Lee, et al., 2009) 2.8.3 Multi-Model Approach Given the uncertainties of some of the processes that affect aviation's impact on air quality and climate, it is desirable to introduce a number of independently-developed models for some measure of cross-comparison. The Intergovernmental Panel on Climate Change (IPCC) used at least 18 distinct models, some with one or more versions. The models varied significantly in resolution, upper-most altitude, coupling, and other key parameters (Randall, et al.). Considerable confidence can be developed as model predictions can be verified by empirical data. Studying an individual perturbation to the atmosphere, such as aviation, defies easy comparison to empirical data, though comparisons of the baseline case for each model to FRM monitor data is presented in Section 8 as a way of aggregating the differences in model processes and inventories. Table 10 summarizes differences in model resolution, data, duration, and scope. Table 10: Summary Model Comparison GATOR global 4x5 degree GATOR GATOR 1x1 degree US-nested GEOSChem global 58 layers, 8 Coupled layers in meteorology bottom km, created in ~19km model ceiling 45 layers ' 14 layers in bottom km Coupled Culd meteorology created in model GEOS-5 reduced, 47 layers, 8 2006 GEOS5 4x5 degree layers in meteorology, bottom km, static 80km 12 years, looped 2006 aircraft emissions 1 +1 CMAQ 3x36 km CMAQ 36x36 k 2006 GEOS5 bottom km, oo meteorology, static .. (~15km) TOMCAT 2.8x2.8 degree 35 layers, 10hPa .. ceiling (-32km) Static boundary conditions, includes nearly all of the Pacific month (Jan, ' and Atlantic Oceans above 15N Jl 06 in domain Tropospheric + stratospheric chemistry transport model (CTM), no climate feedbacks, 1 year199NJbcgon (2006) 1999 NEI background inventory over US, default GEOS-Chem inventories ceiling 35 layers, 18 layers in Coupled meteorology/ chemistry, climate feedbacks, more detailed/nearexplicit physical and chemical processes than other models, computationally expensive in comparison, 2005 NEI background over US elsewhere 1 year + 3 Tropospheric CTM, no climate feedbacks, nested with varying month spin-up GEOS-Chem boundary conditions, 2001 NEI background inventory No treatment of BC, NH4 Full Flight Emissions Assessment 3 Context 3.1 The inclusion of full flight emissions in AQ studies has been a recent development. The APMT Air Quality (AQ) module (hereinafter APMT-AQ), which used CMAQ to create a response surface to the perturbation of aviation emissions, and EPAct studies used aviation inventories with LTO-only emissions. However, (Barrett, Britter, & Waitz, 2009) have used GEOS-Chem to show that cruise level emissions can have significant effects on ground concentrations of PM in CTMs, a finding borne out in the CMAQ results comparing full flight and LTO-only emissions. The addition of cruise emissions with varying boundary conditions (to reflect the scenario) results in increased surface concentrations of PM by a factor of -5.4. The addition of cruise emissions in comparison to LTO-only emissions both with static BCON (background BCON) results in increased surface concentrations of PM by a factor of -2.3. Therefore, there are both significant CMAQ modelspecific reasons for increased PM, as well as boundary condition transport/effects. Changes in Surface Concentration and Mortalities 3.2 When including full flight emissions, the main findings from CMAQ are as follows: e When considering US + non-US, full flight emissions, a ULS fuel policy results in a 20% reduction of domain-average ground-level PM2.5, with 110 lives saved (a 14% reduction). * Reducing direct aircraft sulfate emissions (as oxidized aviation-attributable S02) to essentially zero results in a reduction in aviation-attributable S04 of ~50%. The rest is a result of aviation-attributable NOx reacting with VOCs to form 03, which feeds OH, which oxidizes background S02 to S04. The chain of reaction is thus NOx + VOCs -* {03} + {OH} + S02 -+ {S03} + H20 -* {H2SO4} + NH3 -> (NH4)2SO4. This effect is discussed further in Section 8 describing the sensitivity of results to LNOx, as well as in Section 2. ho -+ * {02 + [O} + H20] -+ This limit on sulfate reduction is also a key finding of the corresponding GEOS-Chem runs, which set the upper limit at ~40%. * Background ammonia (NH3) drives the formation of N03 and NH4 in the US, which constitute almost 50% of aviation-attributable PM. * As a percentage of all sources of particulate matter, global full aviation-attributable concentrations average about 0.4% in the US, with a maximum of 1.3% (Appendix C). * The application of SMAT yields very similar results, with a marginal reduction in mortalities and marginal increase in concentration, indicating a good general agreement in model baseline and FRM monitor concentrations (Section 8.2). There has been some contention that these results overestimate the significance of cruise emissions due to an overly diffusive vertical transport scheme, for both CMAQ and GEOS-Chem. CMAQ used a (comparatively) simple first order local closure K-theory eddy diffusion scheme. Sensitivity tests (described in further detail in Section 7) in which the value of K was set to zero above the PBL showed negligible differences in domain average vertical profiles, and minor differences in surface concentrations, including with respect to a more sophisticated local + non-local closure scheme (ACM2). Indeed, changes in concentration and mortalities between full aviation and LTO from BC, a non-reactive aviation emission that can be treated as a tracer, are relatively low, with a 30% reduction in ground-level concentration and less than 10% reduction in mortalities. As emissions of BC are much higher at cruise altitude (see Appendix A), this implies modest vertical transport downwards, with quick removal of BC at altitude through wet deposition. Nonetheless, significant aviation-attributable concentrations exist between the top of the PBL and cruise altitudes, where emissions are sparse. Some initial comparisons have shown subsidence rates (W-component wind) in CMAQ (and by extension GEOS-Chem, given the provenance of the met data), may be higher than that in GATOR. Subsequent phases of the study will seek to investigate this and other effects further. It is fair to say that given these results, and a comparison of the runs in which the boundary conditions were varied and static, GEOS-Chem is driving the results of CMAQ when cruise emissions are considered. Indeed, domain average concentration of PM2.5 is approximately five times higher with varying BCON than with static BCON. Note that all domain average concentrations were taken using a subset of the total CMAQ grid displayed, windowed tightly on the CONUS landmass and bounded by 125W, 67W, 25N, and 51N. Min- -8.023e-03@(13,69) (ugim^31 - 797(290to1304) 2006yearlyavg.mortalities Max- 1 741e-01@ (67.130) Mean-2 3919e-02 (fullaviation- noaviation)FRMPM25 to 160) 2006avg,mortalities - 110(40 fullaviationULSreductionin FRMPM25ug/m^31 Mn - 78e-02@(1468) Mao 4195e-02@(4622 0,0 00 01 Z 20 40 0 0 s o 004 003 008 20 20 0 Mean 47415e03 0 01 so s0o ComnNumber 100 120 140 0 0 20 40 so so Columnuber 1 = 106(7210325) avg.mortalities FRMPM25)ug/m31 2006yeariy - no aviation) (fullaviation(sBCON) (full aviation (sBCON) - no aviation) FRM PM25 [ug/m^3] 2006 yearly avg, mortalities - 198 (72 to 325) Min-000@(1,1) Max- 9.577e-02@(67,130) Mean- 5.1547e-03 (fullaviation- noaviation,SMAT)FRMPM2.5[ug/m^3] 2006avg.morts.- 783(285to1281) Min- -8.300e-03 @(13,69) Max- 3.542e-01 @(53,21) Mean- 2.4608e-02 1so 0 20 40 so so 100 120 140 Column Number Column Number Figure 23: Full aviation FRM PM2.5 and ULS reduction in same, full aviation (static BCON) FRM PM2.5, and post-SMAT FRM PM2.5 By component, the largest reductions with ULS are seen in S04, as was expected. However, significant aviation-attributable S04 concentrations remained, almost exclusively in the East and Southeast, where background S02 and VOCs are highest, and OH concentrations are depressed from more rapid oxidation. Background S04 adheres to the same pattern of elevated concentration. A smaller but non-trivial reduction in mortalities was seen from decreased NH4, primarily in the West, where a greater percentage of S04 was removed with a ULS fuel policy. BC and ORG are negligibly affected, confined to boundary condition effects. 4 (fullaviation- noaviation)S04[ug/m^3] 2006yearlyavg,mortalities - 174(63to 284) @(13,69) Max- 3.073e-02 @(46,22) Mean- 7.4230e-03 Min- -4.079e-03 0 20 40 0 80 Column Number 100 120 140 - 85 (31to140) full aviationULSreductionin S04 [ug/m312006avg,mortalities Min- 4923e-03 @(13,69) Max- 2.865e-02 @ (46.22) Mean-3.6228e-03 0 20 40 60 0 ColumnNumber 100 120 140 (fullaviation- noaviation)N03 [ug/m^3]2006yearlyavg.mortalities - 342(124to560) Min- -1.144e-03@(43.47) Max- 8.211e-02 0(60.18) Mean- 7.9871e-03 full aviationULSreductionin N03 lug/m^312006avg.mortalities - -18(-7to-30) Min- 9.789"43 0(72.116) Max-2 716e-030(54.22) Mean- -5.6634e-04 CohannNumber ColumNumber (full aviation -no aviation) NH4 (ug/m^3) 2006yearly avg mortalities - 142 (52 to 233) Min- -1.902e-03@(13,69) Max- 30114e-02@(67.130) Mean-39471e-03 full aviation ULS reduction in NH4 (ug/m^3] 2006avg. mortalities - 19 (7 to 31) Min. -2.283e-03@(14,68) Max - 7803e-03 @(46.22) Mean - 7.101le-04 0035 '002 0 02 01015 10005 101 0 20 40 60 s0o Column Number 100 120 20 140 40 80 Coklum Nurrbe 60 100 120 140 Figure 24: Full aviation S04, N03, and NH4, and ULS reduction in same (fullaviation-no aviation) BC [ug/nm3] 2006yearlyavg.mortalities - 14(5 to22) Min - 1.092e-04 @(13,68) Max - 1.797e-02@(67,130) fullaviationULSreductionin BC[ug/m^3]2006avg,mortalities - 0 (0 to 0) Mean - 1.5474e-04 Min - -1.298e-04@(13,69) Max- 1.797e-04@(13,62) Mean- 57789e-06 100 20 40 60 so Column Number 100 120 140 0 20 40 60 80 Cokonn Number 100 120 140 - -1 (0to-2) in ORGIppbV]2006avg,mortalities fullaviationULSreduction @(13,62) Mean- -. 7975a-05 Min- 3,126e-03@(94,103) Max- 1.030e-03 - 55(20to 90) ORG[ppbV)2006yearlyavg.mortalities (fullaviation-no aviation) @(40,111) Mean- 1.9604e-03 @(88,5) Max- 1.800e-02 Min- -6.621e-04 0.01 0.00" 0-002 0.001 0.01 140 20 40 s0 s0 100 120 140 Column Number Column Numbern Figure 25: Full aviation BC and ORG, and ULS reduction in same The contribution of each constituent species of FRM PM2.5 to total mortalities and concentration is shown in Figure 26. N03 shows higher concentrations in population centers in the Midwest and East, dominating total mortalities. Full Aviation Mortalities ORG 7% BC 1% Full Aviation dConc ORG 8% BC 1%U Figure 26: Speciated components of full aviation-attributable change in FRM PM2.5 and mortalities The constituents of ULS aviation-attributable FRM PM2.5 are shown in Figure 27, in which N03 exerts a larger influence at the expense of S04. Other species contributions remain largely unchanged from those of the full aviation scenario. tTLS Aviation dConc ULS Aviation Mortalities ORG BC 8% 2% BC 1% Figure 27: Speciated components of ULS aviation-attributable change in FRM PM2.5 and mortalities 4 LTO Emissions Assessment Context 4.1 LTO-only emissions are defined as those occurring below 1km above ground level. They have a more localized effect given their emission within the PBL, where strong vertical mixing reduces residence times. Prior aviation AQ studies have almost exclusively included LTO-only emissions. Providing a comparison of full aviation results to LTO-only results provides some continuity in our understanding of aviation's impact. Changes in Surface Concentration and Mortalities 4.2 The main findings from CMAQ when considering LTO-only emissions are as follows: e When considering US + non-US, LTO-only flight emissions, a ULS fuel policy results in a 28% reduction in domain-average ground-level PM2.5, with 60 lives saved (a 32% reduction). * Reducing direct aircraft sulfate emissions (as oxidized aviation-attributable S02) to essentially zero results in a reduction in aviation-attributable sulfates of almost 90%, including more than 90% of mortalities that would be attributable to aviation-related sulfate PM. This is a significantly larger relative reduction in S04 with ULS than that of the full aviation scenario. * Background ammonia (NH3) drives the formation of N03 and NH4 in the US, which constitute almost 60% of LTO-aviation-attributable PM and more than 70% of mortalities. * The application of SMAT yields very similar results, with marginal increases in domainaverage concentration and mortalities, indicating a good general agreement in baseline and FRM monitor concentrations (Section 8.2). (LTOaviation- noaviation.SMAT)FRMPM2.5 lug/m^3]2006avg,morts.- 188(68to308) Min- -7.808e-03 @(13,68) Max- 1.080e-01@(46,22) Mean- 4.6618e-03 Figure 28: LTO aviation FRM PM2.5 and ULS LTO reduction in same, LTO aviation (static BCON) FRM PM2.5, post-SMAT FRM PM2.5 By component, the largest reductions with ULS are seen in S04, as was expected. Again, however, significant aviation-attributable S04 concentrations remained, almost exclusively in the East and Southeast (due to the role of NOx in increasing the oxidative capacity of the atmosphere through ozone and OH, leading to greater conversion of S04 from non aviation sources). In contrast to the full aviation scenario, in which ULS reduced S04-attributable mortalities by only ~50%, LTO aviation S04-attributable mortalities were cut by more than 90%. A smaller but significant reduction in mortalities was seen from decreased NH4, primarily in the West, where a greater percentage of S04 (as (NH4)2SO4) was removed with a ULS fuel policy. negligibly affected, confined primarily to boundary condition effects. BC and ORG are LTO aviation ULSreduction in SO4 [ug/m^3] 2006avg, mortalities - 36 (13to 60) @ (46.22) Mean-8 6166e-04 Min- -4.810e-03 @(13.74) Max-2 536e-02 (LTO aviation - no aviation) S04 [ug/m^31 2006yearly avg. mortalities - 39 (14 to 64) Min - 3215e-03 @(13,68) Max - 2.702e-02 @(46,22) Mean - 98271e-04 too 40 20 20 i0 ColumNurme - 74 (27to121) (LTOaviation- noaviation)N03|ugm^312006 yearlyavg.mortalities Max- 2.87le-02 @(58.19) Mean- 1.9207e-03 Min- -1.877e-03@(36.28) 20 40 00 SO Coiumn Nuer Oe 100 120 20 40 st Column Number 60 100 120 140 LTO aviation ULSreduction in N03 [ug/m^31 2006avg. mortalities - 1 (0 to 1) Min - -2.452e-03 6M46.22) Max - 6.342e-03 @ (94.24) Mean - -5 6563e-05 ColumNumber + Max- 1.402e-02@(67.130) 60 0 140 - 32(12to52) NH4[ug/m^3]2006yearlyavg,mortalities (LTOaviation- no aviation) Min - -1.179e-03@(13,68) 40 i 100 Mean- 7.3203e-04 I2N I4" - 11 (4to18) 2006avg.mortalities LTOaviationULSreductionin NH4(ug/m^3] Min - -2.119e-03@(13,74) 20 40 Max- 6810e-03@(46.22) 60 90 Column Number Figure 29: LTO aviation S04, N03, NH4, and ULS reduction in same 10o Mean- 1.4938e-04 120 140 LTOaviationULSreduction in BC)ug/m^3] 2006avg,mortalities - 0 (0to0) Min- -1,647e-04@(13.74) Max- 1.330e-04 @(13,62) Mean- 3.8011e-06 (LTOaviation- no aviation) BC [ug/m^3] 2006yearlyavg.mortalities - 13(5 to21) @(13,68) Max- 1.794e-02 Min- -7.554e-05 @(67,130) Mean- 1.080le-04 100 D 20 40 s o 100 120 0 140 20 40 Column Number 20 40 60 s0 Column Number 100 120 140 Numbu Column (LTOaviation- no aviation) ORG[ppbV]2006yearlyavg,mortalities - 12(4to19) Min- 1.126e-03 @(41.109) Max- 1,081s-02 @(67.1301 Mean- 3.0160e-04 0 00 60 100 LTOaviationULSreduction in ORG(ppbV]2006avg.mortalities - 1(0 to 2) @(87.881 Max- 1.388s-03 0013.681 Mean- 2834a-as Min- -9.715e-04 120 0 20 Column Number Figure 30: LTO aviation BC and ORG, and ULS reduction in same The contribution of each constituent species of FRM PM2.5 to total mortalities and concentration is shown in Figure 31. N03 again shows higher concentrations in population centers in the Midwest and East, dominating total mortalities. By domain-average concentration, S04 contributes a significantly larger fraction to PM2.5 than in the full aviation scenario. 60 LTO Aviation dConc LTO Aviation Mortalities ORG ORG 6%1m BC ,%a Figure 31: Speciated components of LTO aviation-attributable change in FRM PM2.5 and mortalities The constituents of ULS LTO-only aviation-attributable FRM PM2.5 are shown in Figure 32, in which N03 again exerts a larger influence at the expense of S04, which is almost entirely eliminated. PBW 4% PBW Aviation Mortalities ULS LTO Aviation dConc 4% BC 3% 4% 2% Figure 32: Speciated components of ULS LTO aviation-attributable change in FRM PM2.5 and mortalities 5 Climate Feedbacks 5.1 Background on Climate Feedbacks Climate effects from aviation (IPCC, 1999) include: * Increased radiative forcing (RF) from C02, an established GHG with a high level of understanding of its creation and effects e Increased RF from contrails, linear ice clouds formed in the wake of aircraft in regions with sufficient humidity. Contrails reflect more space-bound infrared radiation back to earth than earth-bound radiation back to space and exert a wholly warming effect at night. In 1992, it was estimated that by annual average, contrails covered 0.10% of the earth's surface. e Increased RF from induced cirrus cloudiness, or clouds formed of ice crystals above -6km and estimated to cover nearly 30% of the earth. Cirrus clouds can condense around persistent contrails. * Increased RF from aircraft emissions of water vapor, a greenhouse gas e Increased RF from NOx-catalyzed ozone production e Decreased RF from methane reduction by NOx-catalyzed ozone production, which also produces the hydroxyl radical OH, which breaks down CH4 into C02 and H20, which are weaker GHGs than CH4. * It is thought that sulfate aerosols result in decreased RF through direct backscattering and reduction in optical aerosol depth, as well as indirectly by acting as cloud condensation nuclei (CCN), though the sign of this indirect effect depends on the type of cloud formation. There was a low LOSU of this result in the IPCC report on global aviation, and the finding of negative total RF is questioned by the results from GATOR in this study. * Increased RF from soot aerosols, since the (black) carbon of soot absorbs solar radiation and decreases albedo when deposited, especially over snow. It can be seen that some clouds have a net cooling effect by reflecting more solar radiation back to space than they trap heat from the earth; others have a net warming effect by trapping energy near the surface. On balance, clouds exert a significant cooling effect, although this varies by region. Low-level, white clouds reflect sunlight, thereby preventing sunlight from reaching the Earth, which exerts a cooling effect, whereas high, convective clouds absorb energy from below at higher temperatures than they radiate energy into space from their tops and thereby exert a warming effect (Seinfeld & Pandis, 2006). Cloud formation as a function of the density of aerosols, which function as CCN, can therefore either increase or decrease RF depending on conditions. 5.2 Background on GA TOR-GCMOM GATOR-GCMOM is described and evaluated for the application to aircraft climate and air quality effects in (acobson et al, 2010a) and (Jacobson et al, 2010b). Briefly, it is a one-way-nested (feeding information from coarser to finer domains) global-regional Gas, Aerosol, Transport, Radiation, General Circulation, Mesoscale, and Ocean Model that simulates climate, weather, and air pollution in all domains. It treats emissions, urban, tropospheric, and stratospheric gas photochemistry, sizeand composition-resolved aerosol microphysics, size- and composition-resolved cloud microphysics, size- and composition-resolved cloud-aerosol interactions, sub-grid cumulus cloud thermodynamics, grid-scale stratiform thermodynamics accounting for sub-grid variations in energy and moisture, spectral UV, visible, near-IR, and thermal-IR radiative transfer for heating rates and photolysis, dynamical meteorology, 2-D ocean dynamics, 3-D ocean diffusion, 3-D ocean chemistry, oceanatmosphere exchange, soil, vegetation, road, rooftop, snow, and sea-ice energy transfer, and soil and vegetation moisture transfer, among other processes (PARTNER Project 27 Internal Interim Report, 2010). With respect to aircraft, the model accounts for the microphysical evolution of contrails from all commercial aircraft worldwide at the sub-grid scale (at their actual size). Specific sub-grid processes treated include: * spreading and shearing of individual aircraft exhaust plumes (Naiman, Lele, Wilkerson, & Jacobson, 2010) * the calculation of a unique supersaturation for each plume based on emitted and ambient water vapor, ambient temperature, and emitted and ambient particle size distributions * time-dependent, discrete size-resolved ice deposition and condensation/freezing onto sizeresolved aerosols * size-resolved coagulation to create and grow linear contrails at the sub-grid scale e * discrete size-resolved evolution of grid-scale cirrus and other clouds from the remnants of sub-grid linear contrails radiative effects of sub-grid linear contrails as a function of ice crystal size and composition and contrail plume shape. Two discrete (multiple size bins) aerosol distributions, each with multiple components, were treated here on the grid scale - an emitted fossil-fuel soot (EFFS) distribution and an internally-mixed (IM) distribution. EFFS particles could coagulate with other EFFS particles, IM particles, or discrete sizeand composition-resolved liquid, ice, or graupel hydrometeor particles. EFFS or IM particles could also serve as nuclei for new hydrometeor particles. Once sub-grid contrails dissipated due to subgrid plume expansion, dilution, and water sublimation, their aerosol cores were released to the EFFS size distribution on the grid scale. Aerosol cores released from sub-grid contrails to the grid scale eventually either coagulated with aerosol particles from other sources or hydrometeor particles or served as cloud nuclei. Grid-scale aerosol particles and their components were tracked within hydrometeor particles, both through cloud formation and precipitation (PARTNER Project 27 Internal Interim Report, 2010). 5.3 Future Work At this time, no conclusions can yet be made from GATOR as to the monetized health impacts from ULS fuel that result from also including climate feedbacks. There is significant intra-annual variability in the nested US-Ocean runs, and it is thought that the global runs are insufficiently resolved to carry through mortalities in the cost-benefit analysis. Further runs and tests have been proposed for GATOR (PARTNER Project 27 Internal Interim Report, 2010). 6 6.1 ULS Cost and Operations Analysis Sulfur Content of ULS Jet Fuel This section presents work cited from (PARTNER Project 27 Internal Interim Report, 2010). It was adapted from work conducted and reported by Dr. James Hileman and others and is not an original contribution of this thesis. It is included here since it will be drawn upon in deriving conclusions. The current specification for Jet A (and Jet A-1) fuel sulfur content (FSC) is 0.30% or 3,000 ppm; however, the actual fuel sulfur levels typically fall below this value, with both spatial and temporal variation. The Coordinating Research Council (CRC) sponsored a voluntary survey of the fuel sulfur content of jet fuel leaving refineries to determine if the incorporation of the US ULS diesel standard would impact jet fuel sulfur content (Taylor, Feb. 2009). The results from the survey, presented in Figure 33, show that the eastern US refineries produced jet fuel with reduced fuel sulfur content concurrently with the incorporation of the ULS diesel standard. The data also shows the regional variation in the fuel sulfur content of jet fuel leaving US refineries. The highest sulfur contents were recorded in the Gulf coast while the lowest values were recorded on the west coast. The sulfur content values for the west coast showed considerable variability with a minimum weighted monthly mean value of 116 ppm and a maximum value of 573 ppm. The U.S. Military's annual jet fuel quality survey (Defense Energy Support Center (DESC), Petroleum Quality Information System (PQIS) and (DESC-BP, 1999-2007)) presented in Figure 34, shows that sulfur content within jet fuel varies by location and has been changing with time. Individual samples of jet fuel have sulfur levels ranging from under 10 ppm to over 2000 ppm. The DESC survey collects data from U.S. military bases around the world including Europe and the Pacific. In comparison, in the United Kingdom, as can be seen in Figure 35, the mean jet fuel sulfur content has fluctuated between 360 ppm and 640 ppm over the past 20 years (Rickard, June 2008). In 2007, the mean for the United Kingdom was 500 ppm with a standard deviation of 600 ppm. This demonstrates the international variability of jet fuel sulfur content. ................ ...... x Nationwide aGulf 0 East A West 1000 0 800 600 0 - 0 0 XD x a D 0 0 x ex x - a x _ xx Ax X A 0 XX xx x xxa X X X A A * 400 - 0 010 0 x @0 AAA A AS @0.00 200 dA Sep-05 Jan-06 Jun-06 Nov-06 Apr-07 Sep-07 Feb-08 US East 446 321 US Gulf 858 800 US West 240 395 Nationwide 709 677 Figure 33: jet A average fuel sulfur content in US (Taylor, Feb. 2009), as plotted by (Hileman, et al., 2009) 60 50l 1 40 30 - 20 - 10 - 0 1800 2400 Fuel Sulfur Content (ppm) 3000 2002 530 514 2003 580 525 2004 666 504 2005 668 581 2006 754 648 2007 762 590 Figure 34: JP-8 fuel sulfur content as reported by PQI (DESC-BP, 1999-2007), as plotted by (Hileman, et al., 2009) 7001 0 E 600 500 0 .0 0 S400 0 0@ * @0e 0 S S 0 -* .0 300 200 LL 100 0i 1985 ___ - -------------- 1990 1995 2000 2005 Year Figure 35: Jet A-1 average fuel sulfur content in UK (Rickard, June 2008), as plotted by (Hileman, et al., 2009) 6.2 Production Costs of ULS Jet Fuel 6.2.1 Background Desulfurization processing units are well established in the petroleum industry to remove sulfur containing chemical species from middle distillate products. Ultra-low sulfur middle distillates, such as ULS gasoline, diesel and jet, require harsher processing conditions than low-sulfur products and increase refining costs. Additional hydrogen gas, process energy, and capital expenses for new or upgraded reactors are necessary to operate at more severe processing conditions. During the transition to ULS diesel (ULSD), some refineries left the diesel market due to their processing capabilities and the economics of ULS production. Additional capital and operating costs for producing diesel shifted production to other products, such as gasoline, jet, and heating fuels. However, by the time the proposed ULS jet (ULSJ) specification rolls out, nearly all products leaving the refinery gate must be ULS. In an effort to estimate the costs of ULSJ production, diesel prices were examined to determine the production costs during the transition to ULSD. Diesel and jet fuels exhibit nearly identical crack spread profiles. As a result, refining costs for diesel are believed to be a good surrogate for jet fuel. Prior to the ULSD transition the Energy Information Agency (EIA) conducted a comprehensive study to evaluate economic impacts and create estimates for both the refinery and end-user costs (EIA, 2001) (EIA, 2001). The EIA report explored several ULSD production strategies and assumptions taking into account detailed refinery supply and demand information. The additional cost of ULSD was estimated to be between 3.1 and 8.3 cents per gallon depending on refiner capacity, existing on-road diesel production, feed stock quality, and existing hydro-treating capacity. These models assumed a 5.2% after-tax return on investment, production of -7ppm ULSD at the refinery gate, and world crude oil price of no more than 58.57 cents per gallon ($24.60 per barrel) in 2010. In reality the world price of crude oil increased well beyond predicted prices during the phase-in period of ULSD production. For example, the spot price of West Texas Intermediate at Cushing, Oklahoma increased from 46 cents per gallon in December of 2001 to 318.76 cents per gallon in March 2008 (Spot Prices, 2010). The price then fell to 97 cents per gallon in January 2009, 65% higher than what the EIA predicted. Additionally the U.S. price for natural gas, used as a process fuel and in the desulfurization process, followed a similar price trend and reached historic high prices. The increased cost of process fuel increased the cost of refining diesel during the phase in period and is discussed below. Additionally, the transition to ULSD fuel occurred during some of the most dramatic events in recent history. The terrorist attacks of September 11 and the wars in Iraq and Afghanistan sent crude oil prices to unprecedented high levels. Hurricane Katrina's devastating effects on the Gulf Coast took several refineries off line resulting in diminished supply levels. These market-influencing events affected energy and fuel prices significantly. 6.2.2 Detailed Cost Analysis Figure 36 was created with historical fuel prices, and refinery production data obtained from the EIA. The figure shows crack spread, diesel fuel production levels, the component costs of diesel, and the price differential of diesel fuels. ................... 0.6 0.5 0.4 L1.0 -jet Crack Spread 0.3 Diesel Crack Spread 0.2 U.S --- Product Supphed Of DIsonlate Fuel 0, 0 to 15 ppm Sulfur (Thousand GaHion) 450 350 0i -0- 250 IL 150 U S. Product Supplied of Distwilate fuel Od. Greater than 15 to 500 ppm Sulfur (Thousand Gallon) U.S Product Supplied of Ddnllate fuel Oil, Greater Than SOO ppm Sulfur (thousand Gallon) 50 .50 400 8 350 300' 41 S 'Jo. Crude 20 Ott (cents/gal) lRehmnt cents/gal) Wholiesale Pnce (cents/gal) I 150 100 so 0 25 20 I 15 wholesae DetatLS-HS) ($0.01/ga) 10 wholesale Delta(tuL-tts) (So01/gal) Figure 36: Diesel prices and production - - Wholesale Delta(ULS-L5s ($0 01/gal) 6.2.2.1 Normalized Crack Spread of Diesel and Jet Fuels. The normalized crack spread of diesel and jet fuels was calculated by subtracting the monthly wholesale price from the Cushing, OK spot price of crude oil, and diving by the spot price of crude (Diesel Fuel Explained, 2009) Crack Spread= Pricee - Pricec'u (6) It can be seen from Figure 36a that the crack spreads are very similar. The average crack spread is 0.24 percent for both fuels. The average cost of crude was 91.31 cents/gal, and 112.8 cents/gal and 113.23 cents/gal for jet and diesel respectively. The refining costs of ULSD are therefore believed to be a good basis for estimating the costs of ULSJ production. 6.2.2.2 Supply and Demand High Sulfur Diesel (HSD) is a substitute for heating oil, and production fluctuates with seasonal variations (higher in the winter). HSD is also used in off-road applications such as locomotives and agricultural machinery. Low Sulfur Diesel (LSD) is used for on-road trucks and buses. It has been replaced by ULSD for on-road use. Figure 36b shows that the production of ULSD was reported beginning in January 2004. It ramps up quickly in June 2006, and over took LSD production in September 2007. All diesels will be ULSD in 2011. 6.2.2.3 The Cost of a Gallon of Diesel Wholesale prices at the refinery gate include the cost of crude and refining. Figure 36c plots the monthly US average cost the reseller pays for refining and crude from May 2002 through August 2009. These prices do not include distribution costs, taxes, or retail mark-up which are typically included in retail prices. Figure 36c also shows that the costs of refining increased through May 2008. The cost of crude also increases through the same time period. In addition to the capital costs and additional process energy, the cost of all refining processes increases because of the rising costs of crude and natural gas. Table 11 reports the average, maximum, and minimum costs of producing a gallon of diesel fuel. Table 11: Refining and crude oil costs of a wholesale gallon of diesel, May 2002 - August 2009 (Spot Prices, 2010) Average [percentage] 100.0 23.89 76.11 Average [cents/gal] 167.84 40.1 127.74 Maximum [cents/gal] 383.05 92.48 307.58 Minimum [cents/gal] 65.97 6.65 57.48 6.2.2.4 Historic Price Difference in Diesel Fuels It can be seen from Figure 36d that the difference between the cost of LSD and HSD was relatively stable from May 2002 through June 2004, exhibiting typical seasonal variations. The difference then began to increase significantly because of the high price of crude oil, natural gas, and amortization of capital investments made for ULSD production. The high price difference of ULSD compared to HSD in early 2007 ranges between approximately 13 and 25 cents per gallon. After the summer of 2007 prices fell and stabilized with some fluctuations in the winter of 2008. 6.2.2.5 Processing Costs US monthly wholesale diesel prices from the EIA were used to estimate the cost of processing ULSD. The US average wholesale monthly diesel price is available from January 1994 for high (>500ppm) and low (15-500ppm) sulfur diesel fuels, and from January 2007 through July 2009 for ultra-low sulfur (<15ppm). The cost of ULSD production was found by averaging the difference between the monthly wholesale prices of ULSD and high-sulfur diesel (HSD). Overall, the average price difference of ULSD from HSD is 6.7 cents per gallon and from LSD is 3.3 cents per gallon. In spite of the unforeseen high costs of feedstocks, these production costs are within the range given by EIA estimates and all sensitivity analysis. 6.2.2.6 Limitations of this Analysis There are several limitations of this analysis. First, ULSD production prices are reported beginning in January 2007, though small scale ULSD production is reported to begin in January 2004 with approximately 40.5 million gallons produced. The production then ramped up significantly to three billion gallons in August 2006. Capital costs for new equipment and increased labor costs were amortized during this phase in period and were difficult to observe directly with the cost of ULSD. Second, this is an incomplete data set of national averages. ULSD has only been produced for a couple of years, and the full costs of ULSD transition will not be fully realized until 2011 when 100% production of ULSD for on- and off-road applications is achieved. Additionally, the data obfuscates regional differences in costs, sales, and production disruptions such as scheduled downtime caused by machinery malfunctions, accidents, or severe weather. Moreover, every refinery is different, and some refiners enter the market, or leave maintenance, unexpected depending on the economics and capabilities of their equipment. Therefore, this analysis assumes that the average cost of ULSD is spread among all refineries, not just ULSD producers. Third, this does not take into account imports and exports of diesel and jet fuels. International market for diesel can affect domestic prices and production. Similarly, jet fuel is traded on the open international market. Lastly, HSD, LSD, and ULSD supply and demand changes throughout the phase in period as production levels change. Some production changes are attributed to refineries leaving the diesel market due to the increased operating and capital costs for harsh ULS processing conditions. This is complicated by the fact that prices may shift due to seasonal demand changes, and because these products are substitutes for heating oil and non-road applications. 6.2.3 Production Cost Estimate Based on a review of the literature supporting the EPA ULS diesel rulemaking, (Hileman, et al., 2009) found, "the additional marginal cost of producing and distributing ULS diesel fuel from thenstandard diesel fuel was estimated during rule-making at an additional $0.054 per gallon. This estimated cost is comprised of a refinery desulfurization cost of $0.041 per gallon, a distribution cost of $0.011 per gallon, and lubricity additives with costs of $0.002 per gallon (US EPA, 2000a) (US EPA, 2000a) and (EIA, 2001) (EIA, 2001). Recently, (US EPA, 2006) (US EPA, 2006) reiterated that '[ULS] diesel fuel costs an additional $0.04 to $0.05 per gallon to produce and distribute.' Given the similarities between jet fuel and diesel fuel, the experience gained in producing ULS diesel fuel is relevant." Based on the analysis of historical data, the cost of ULSD production was estimated to range between three and 25 cents per gallon. The average cost was 6.7 cents per gallon and 3.3 cents per gallon for desulfurization to ULSD from HSD and LSD, respectively. The cost of desulfurization process changes over time due to the amortization of equipment and labor costs, as well as fluctuations in the cost of process energy. Even with unprecedented high costs of crude oil and natural gas, the average production costs are within the ranges given by the EIA report estimates and all sensitivity analysis. Based on the literature review and data analysis, the expected increase in cost of ULS jet fuel is between 4 and 7 cents per gallon. Statistics from the Bureau of Transportation Statistics' National Transportation Statistics show that in 2006, the most recent year for which data is provided (and the nominal study year for this modeling effort), the domestic commercial aviation fleet consumed 13.458 billion gallons of fuel (Fuel Consumption by Mode of Transportation, 2007). The HDS of that amount of fuel translates to a yearly marginal cost of between $540 million and $940 million. 6.3 Potential Usability Concerns and Benefits with ULS Jet Fuel This section presents an excerpt on ULS jet fuels from (Hileman, et al., 2009). The main compatibility and operability concerns in switching to ULS fuels are energy density, lubricity, fuel leakage due to reduced elastomeric swelling, and pipeline compatibility. Within diesel fuels, the HDS process results in a small loss in volumetric fuel economy because of the approximately 1% loss in energy density (US EPA, 2000b) (US EPA, 2000b), (US EPA, 2006) (US EPA, 2006), (BP, 2006) (BP, 2006), (Exxon, Undated) and (Chevron, 2007). This decrease was observed for ULS diesel fuels, and it is expected that a ULS jet fuel would have a similar loss. In (Hileman, et al., 2009), relationships of fuel energy density as a function of hydrogen content were used to estimate the impact of deep HDS on the specific energy of jet fuel. A 1% loss in energy density corresponds to a change from 34.8 MJ/L to 34.4 MJ/L. It was assumed that a desulfurized jet fuel would follow the trends for hydrocarbons derived from the PQIS database; thus, a 1% change in energy density should be accompanied by an increase in mean hydrogen content from 13.8 percent to 14.0 percent and a 0.3% increase in mean specific energy. The net result is that 1 percent more fuel volume, but 0.3% less fuel weight, is required to fly a typical distance. These estimates take into account the reduction in fuel consumption that would be associated with reduced aircraft weight. HDS can result in a reduction in fuel lubricity, as it results in the removal of the compounds that provide jet fuel its natural lubricity. However, as noted by (Chevron, 2004), low sulfur or aromatics levels in jet fuel are not necessarily signs of inadequate lubricity. The boundary lubricity of jet fuel cannot be predicted from bulk physical or chemical properties, it can only be measured in a specially designed test apparatus. Fuels with similar sulfur and aromatics content can have different lubricity. Further, a near-zero-sulfur, synthetic fuel has been made with similar lubricity properties to conventional jet fuel without the use of fuel additives (Moses & Wilson III, Evaluation of Sasol Synthetic Kerosene for Suitability as Jet Fuel, Dec. 2003). If the lubricity of a ULS jet fuel were found to be less than desired, lubricity additives may be added to the fuel such that the end user does not notice a difference, such as has been done with ULS diesel fuel (BP, 2006) (BP, 2006), (Exxon, Undated) and (Chevron, 2007). The additives contain esters (10-50 ppm) or fatty acids (20-250 ppm) (Chevron, 2007). For example, all of Exxon's diesel fuels have incorporated lubricity additives since 2005 (Exxon, Undated). These fuels all meet the diesel-fuel standard ASTM D975 (ASTM Standard Specification for Aviation Turbine Fuels, 2007a). A framework for lubricity additives already exists in aviation; US military standards for jet propellants 4, 5, and 8 (JP-4, JP-5, and JP-8) require a corrosion inhibitor/lubricity improver. Lubricity and corrosion inhibitors can also be added to Jet A-1 (Chevron, 2004). For this purpose, the US military has approved RPS 613, produced by Champion Technologies, and DCI-4A, by Octel Starreon, which are also used in civil aviation (Moses & Wilson III, Evaluation of Sasol Synthetic Kerosene for Suitability as Jet Fuel, Dec. 2003). However, additives usually have not been used in Jet A (Chevron, 2004). The use of ULS fuel may lead to a reduction in certain maintenance costs (Edwards, et al., July 2004). A low-sulfur fuel may have increased thermal stability, which would reduce the formation of corrosive deposits, possibly leading to increased component life (US EPA, 2000a) (US EPA, 2000a). The reduced aromatic content of a ULS fuel could lead to reduced combustor-liner temperatures and increased combustor life. (Moses & IKarpovich, Fuel Effects on Flame Radiation and HotSection Durability, 1988) conducted a set of combustor experiments with fuels of varying hydrogen contents, and they found that a 1% decrease in fuel-hydrogen content resulted in a combustor life reduction of between 20% and 80% (ULS fuels have slightly higher hydrogen contents). These results are based on a reduced-fidelity model for older combustors, but the trends should extend to newer designs. The reason for this increase in life is that the wall temperature of the combustor is influenced by the emitted radiant energy and the formation of soot in the primary zone of the combustor. Both of these are related to hydrogen content (Martel & Angello, 1973), (Shirmer & Quigg, 1965), (Lefebvre, 1984) and (Moses & Karpovich, Fuel Effects on Flame Radiation and HotSection Durability, 1988). It thus appears that a higher-hydrogen fuel such as that resulting from HDS may lead to increased combustor life. At present, available testing data are insufficient to provide a foundation for reliable quantitative estimates of the potential impact of ULS Jet A fuel on maintenance costs. Specifically, further study is required to determine whether there is an appreciable positive or negative impact. Current Jet A, with its relatively high sulfur content, causes problems for pipeline operators. They must use extra steps to flush out sulfur contamination after shipping high-sulfur fuels, such as Jet A, to avoid contamination of low-sulfur fuels, such as ULS diesel e.g. (EIA, 2001). A transfer to ULS Jet A would ease this problem and enable more-efficient pipeline system use, providing a benefit that might be passed on to aviation consumers. ULS jet fuel may also benefit the U.S. Department of Defense (DoD) in its goal of reducing the number of different fuels required during military operations, and it could potentially lead to lower overall procurement and maintenance costs. The U.S. Army uses JP-8 in its diesel equipment, but new diesel engines are designed for ULS diesel fuels. These engines will encounter problems if operated on higher-sulfur fuels. To use current JP-8, which has ~700 ppm sulfur content, the U.S. Army may need to special order diesel engines that can tolerate higher sulfur levels. However, if Jet A were ULS, then JP-8 would likely become ULS as well, and then off-the-shelf diesel engines could be used. The specification for Jet A, ASTM D1655 (ASTM Stamdard Specification for Diesel Fuel Oils, 2007b) has a maximum limit for fuel sulfur content, but not a minimum limit. As such, a ULS jet fuel falls within the current specification outlined by ASTM D1655. As previously noted, the degree of hydrotreatment that is necessary to achieve a ULS fuel can affect lubricity. The D1655 standard does not have a lubricity requirement, and lubricity issues can be overcome through the use of appropriate fuel additives. 7 CMAQ Sensitivity Studies 7.1 CMAQ models CMAQ uses a number of models to describe atmospheric processes, developed by academic and government organizations and cited in the literature, summarized in Table 12 with references. These include: " Carbon Bond IV gas-phase equilibrium model, developed by the EPA for use in urban and regional photochemical modeling e ISORROPIA II gas-aerosol equilibrium model for inorganic PM, developed by the Georgia Institute of Technology with assistance from the National Oceanic and Atmospheric Agency (NOAA); ISORROPIA is also an option in GEOS-Chem in replacement of the default RPMARES scheme (GEOS-Chem v8-02-01 Online User's Guide) e Dry deposition Regional Acid Deposition Model * MM5 Meteorological Model, developed by the National Center for Atmospheric Research (NCAR), using NASA Global Modeling and Assimilation Office (GMAO) GEOS-5 data, which is also used in GEOS-Chem. Table 12: CMAQ schemes/models Carbon Bond IV (CB-IV) Gas-phase equilibrium model (Gery, Whitten, Killus, & Dodge, 1989) ISORROPIA II Gas-aerosol equilibrium model for inorganic PM (Nenes, Christodoulos, & Pandis, 1998), (ISORROPIA v1.7 Reference Manual, 2006) SOA Organic PM model (Odum, Hoffmann, Bowman, Collins, Flagan, & Seinfeld, 1996) RADM (Regional Acid Deposition Model) Dry-deposition model (Stockwell, Middleton, & Chang, 1990), (Wesely, 1989) RADM Bulk Aqueous Chemistry model MM5 Meteorological model (Grell, Dudhia, & Stauffer, 1995) K-theory Eddy Vertical Diffusion method Piecewise Parabolic Method (PPM) Vertical/Horizontal Advection method (Colella & Woodward, 1984) 7.2 Introduction In line with its modular nature, active development community, and serial updates of modules, CMAQ has several different options or versions for diffusion, advection, and chemistry. Table 13 shows the options used by default in all the results presented. Table 13: ULS CMAQ "baseline" options (CMAQ 4.6 Operational Guidance Document, 2009) Lightning NOx from 4x5 global GEOSChem added into background inventory as 100% NO Lightning NOx included Background inventory year and source 2001 EPAct ModInit - chemistry/ transport timestep init - air density-based scheme for mass-conserving initialization advection adjustmmsscse denrate - adjust the vertical advection error term from the meteorology model by air density from the chemistry/transport module ero ModHadv - horizontal advection hppm - Piecewise Parabolic method ModVadv - vertical advection vppm - Piecewise Parabolic method multiscale - diffusion coefficient based on local wind ModHdiff - horizontal diffusiondeomtn deformation ModVdiff - vertical diffusion eddy - calculate vertical diffusion using eddy dfuiiyter diffusivity theory ModPhot - photolysis phot - photolysis included ModPing - plume-in-grid ping-noop - plume-in-grid not included ebi cb4 - use the Euler Backward Iterative solver optimized for the carbon bond-IV mechanism ModAero - aerosol chemistry aero4 - 4th generation modal CMAQ aerosol model with extensions for sea salt emissions and thermodynamics ModAdepv - aerosol deposition velocity aero-depv2 - second generation CMAQ aerosol deposition velocity routine ModCloud - cloud module for modeling cloudacm - RADM-based cloud processor using impacts of clouds on deposition, mixing, asymmetric convective model to compute photolysis, and aqueous chemistry convective mixing Mechanism - specifies the combination of gas-phase, aerosol, and aqueous phase cb4_ae4_aq chemical mechanisms Tracer tracO - no tracer specified In order to characterize the sensitivity of CMAQ results to options that may have a significant effect on vertical transport, several additional test cases were run (and proposed). These test cases entail changing a single mechanism from the "baseline" values listed in Table 13. For each test, the full aviation scenario with static BCON only was considered to better isolate the importance of the mechanism within the model. Four months only were considered due to time and space constraints. Likewise, not all proposed sensitivity tests were carried out. In particular, a full aviation and background run with the CB-V mechanism necessitated the reprocessing of nearly all data involved, including met data, aircraft, and background inventories. The other scenarios were limited by lack of storage space and time. In subsequent phases of the project, when time and further computing resources will exist, it may be useful to revisit these. There is also direct applicability to future aviation studies using CMAQ. Table 14: Sensitivity studies scenarios proposed (full aviation (static BCON) no aviation) (completed) Lightning NOx not included Jan, Apr, Jul, Nov 2006 Vertical Diffusion - acm2 ) . (fl noaviation aviation)(tB(limited) (asymmetric convective model, vrin2 version 2) Jan, Apr, Jul, Nov 2006 Vertical and Horizontal (full aviation (static BCON) - Advection - YAMO no aviation) (proposed) (Yamartino-Blackman cubic Jan, Apr, Jul, Nov 2006 scheme) (full aviation (static BCON) - 2005 EPAct inventory, CB-V no aviation) (proposed) mechanism, CMAQ4.7 7.3 Vertical Diffusion and Advection 7.3.1 Background Oct 2005-Dec 2006 The equations and assumptions for the diffusion and transport schemes in CMAQ are described in detail in (Byun, Young, Pleim, Odman, & Alapaty). Briefly, diffusion is here defined as eddy or turbulent mixing on a sub-grid scale. Advection is defined (somewhat differently than it is sometimes taken to be in the literature) as both horizontal and vertical transport of the matter in one grid cell to another, through wind or convection. Dry deposition is another transport process that results in the settling of aerosols to the ground over time, under gravitational force alone. Wet deposition occurs when aerosols wash out in precipitation. Vertical Diffusion: EDDY vs. ACM2 7.3.2 On the advice of Dr. Daewon Byun of NOAA, K-theory eddy diffusivity was used for all final results in lieu of the default asymmetric convective model, version 2 (ACM2). By K-theory eddy diffusivity, it is meant that a first order approximation for turbulent diffusion is taken, in which it is assumed that turbulent fluxes of a quantity flow down the mean gradient of that quantity at a rate of flow proportional to a local eddy diffusivity coefficient K. Such schemes assume that turbulent eddies are sub-grid scale and hence can be described by the local K coefficient. Second or higher order relations (variances, covariances, etc.) are subsumed by the first order approximation. ACM2 combines local eddy diffusivity with a non-local scheme that is better able to represent the supergrid-scale components of diffusion (Pleim, 2007). It was initially proposed that the vertical diffusion scheme in CMAQ was excessively diffusive given the high relative contribution of cruise emissions to surface PM. Therefore a sensitivity test was conducted in which three schemes were compared. In addition to local eddy diffusivity and ACM2, eddy diffusivity with all turbulence coefficients K set to zero above the PBL was also custom-coded for comparison. This artificially induces zero vertical diffusion in the free troposphere, which is a not unreasonable approximation in any case, as the free troposphere is usually non-turbulent, or only intermittently turbulent, with subsidence rates on the order of 0.1-1 cm/s (Seinfeld & Pandis, 2006). (U 3 [51 AI TOTALPL4O, ,00(&-02 vdi)f 00000600000 00ta)Zv~ avbo 0006 5I.."A.90) .g OCT DECN 06.kWO.4 TOTAtPIO$ 00 00 000 000 0050000 004 003 000 000 002 20 40 60 .so ColumNumbe. 100 120 0 140 003 02 001 001 20 *0025*0 009 000 0 200 .o.V t OCT OEC 00 aviatio(K2 PL .0) ba ound(K2Pal.. 0))TOTAL.PM15 [ogik931 Ava forOCT DEC2006 MA..011000o-30200112) Max.1.53335-03 @29.53) Mo3,,.11.6110203*-02 100 (full aviation - background) domain averageSC, varyingyd1 schemes(average OCT-DEC 2005) 0.1 20 0.09 100 00 0.03 0.02 Z 300 - 0 500' Kr>0BL~ 700 Kz ALPB3 0 0.32 0 0 20 43 0 00 so 0 1OU 10 0.2 0.4 140 0.0 038 1 concentration ug/m3 1.2 14A 10 1.8 X10' Figure 37: Sensitivity of results to vertical diffusion parameter The results are shown in Figure 35. Note that these results used slightly different aviation emissions processing than that used for the final results. There was no relative difference, however, and so the results should hold in general for any study of aviation with CMAQ. A period of three months was chosen as sufficient to allow downward gradients to develop to steady state. The result is that all three schemes showed quite similar behavior, resulting in changes in ground level concentrations of only a few percent. The artificially depressed K and eddy diffusivity schemes resulted in the lowest concentrations (diverging by less than a percent) and ACM2 in the highest. It is interesting that the zero diffusion scheme resulted in slightly higher ground concentrations, implying as much or more upwards transport as downwards, something that's borne out in time-lapse vertical transport movies of non-reactive BC. The noticeable divergence of all three schemes occurred only roughly within the PBL. The existence of concentrations between cruise altitude and ground level, where emissions are sparse, implies subsidence in lieu of an "overly diffusive" numerical method. 7.3.3 Vertical Advection: PPM vs. YAMO On the advice of Dr. Daewon Byun of the NOAA, a piecewise parabolic method (PPM) was adopted as a vertical advection scheme, in lieu of the default Yamartino-Blackman cubic scheme. The Piecewise Parabolic Method is a numerical technique using discrete, monotonic parabolas fit to the zonal average of a variable using information from neighboring averages, providing for accurate representation, particularly of sharp gradients. The Yamartino-Blackman cubic scheme (YAMO) replaces a parabola with a cubic spline as the grid-cell interpolating polynomial for vertical advection but uses PPM for horizontal advection. There was insufficient time and space to conduct a sensitivity test of the differences between these methods, though early phases of the study used YAMO, and the final results were run with PPM; however, other parameters differed between runs. Qualitatively, given the magnitude of these other differences and the evolution of CMAQ model results throughout the study, the advection scheme may have a more significant effect than the diffusion scheme. Additional sensitivity studies isolating the effects of advection from cruise may be profitable. Analysis of hourly data, showing surface concentrations, as well as vertical profiles averaged by latitude and longitude, in time-lapse movie format, has been of some value in understanding these effects. 7.4 Static vs. Changing Background Conditions 7.4.1 Context Most aviation modeling work done to date with CMAQ has used constant boundary conditions. However, as the scope of this work has been to study a global change in policy, it was necessary to create a differing set of boundary conditions for each aviation scenario, by perturbing the global GEOS-Chem runs used to create the BCON with the corresponding scenario. 7.4.2 Dominance of GEOS-Chem Boundary Conditions talities - 812(295to 1328) avg mor (tull aviation- noaviation)FHMPM25[ugm 3| 2006yearlIy Mean- 2 3919e-02 @(1369) Mao-i1 741e-0Ot@0(67.130) Min. -8.023e-03 (full aviation (sBCON) - no aviation) FRM PM25 (ug/m^312006 yearly avg. mortalities - 213 (77 to 348) Min.0000@1,1) Max- 9.577e-02@(67,130) Mean- 5.1547e-03 100 10000 1 s0 000 o24 2 4 vg otlte 06yal M5)gm3 oaito)02 sCN (LOaito 40 6o 00 s Number Column 100 120 140 - 201(73to 329) (LTOaviation- noaviation)FRMPM25[ug/m^32006yearlyavg,mortalities @(67,130) Mean-4.4404e-03 @(13,68) Max- 9.473e-02 Min- -5.766e03 s 0 20 40 60 00 Column Number 100 120 140 (L TO aviation (sBCON) -no aviation) FRM PM25 [ug/m^3] 2006 yearly avg, mortalities - 125 (45 to 204) Min- 000@(11) @ (67,130) Mean- 2.0521e-03 Max- 1.064e-01 100. 0 20 40 60 00 40 Column Numberx 60 0 120 140 Coiumn Number Figure 38: Dominance of GEOS-Chem boundary conditions Figure 38 shows that using changing boundary conditions results in ~4.6 times higher surface PM in the full aviation scenario and ~2.2 times higher surface PM in the LTO-only aviation scenario. The dominance of the GEOS-Chem boundary conditions can manifest in one of two ways. Firstly, PM can be transported directly into the domain from the ring of boundary cells. Secondly, precursors can be transported into the domain and react to form PM. In the boundary condition analysis described in Figure 7, NH3 and H2SO4 developed significant gradients along the boundaries that could result in elevated concentrations of PM as the species react into NH4 and S04, whereas PM species did not develop these gradients (see Appendix C). In particular, the northern swath of elevated PM extending from north of Montana to the lower Midwest was traced to significantly elevated NH3 (even in the aviation-impact-only delta plot, sometbing that is difficult to explain) that was advected southeastwards into the main aviation-attributable PM zones in the Midwest and Southeast. In the LTO-only scenario in Figure 38, if one were to discount this vector of southeasterly elevated concentrations, due wholly or in part to the elevated NH3, the static and changing BCON runs may be quite similar. 7.5 Lightning NOx Lightning creates NOx emissions in the troposphere of the same order of magnitude at cruise altitude as aviation (see Section 2 and Appendix A). Given this significant source of emissions at upper altitudes, which would otherwise be lacking in CMAQ, an LNOx emissions inventory was created from GEOS-Chem. A comparison of full aviation (static BCON) delta results (for runs in which both the full aviation and background cases contained no LNOx, and boundary conditions were kept constant) with full aviation delta results (in which both the full aviation and background cases contained LNOx) is presented in Figure 39. It was thought desirable to limit the scenario to static boundary conditions to isolate model-specific effects (from boundary condition effects). Note that these results used slightly different aviation emissions processing than that used for the final results. (fullaviation(sBCON) - noaviation)FRMPM25[Ug/m^3] JANaverage Min- 8.322e-03@(74,121) Max- 8.641e-02 @(46,22) Mean- 1.0104e-03 (full aviation(staticBCON,noLNOx)- no aviation(noLNOx))FRMPM25[ug/m^3]JANaverage Min. -6.657e-03 @(37,44) Max- 8.669e-02 @(46,22) Mean- 8.7584e-04 0.1 100 mm0.1 0.09 0.09 100 00 400.04 0-0. am Column 0 40 as so 80 Column Number 100 Nubo 100. ol03Nnio 140 0 0 0 20 40 so 00 Column Number 100 ooo ian (fullaviation(sBCON) - noaviation)FRMPM25fug/m^31 JULaverage Min - -2195e-03 @ (33.140) Max- 2.303e1 @(67.130) Mean- 5.8182e-03 (lull aviation(staticBCON,noLNOxo-no aviation(noLNOx))FRMPM25(og/m^3] JULaverage Mmn-t-166e-03 @(34,142) Max- 5.672e-01 @(67.130) Mean.- 2 7389e-02 009 100 0,08 80 00 J I 000 aor 00% 6 -005 004 003 0.01 20 40 so 0 100 120 140 0 0 002 0 01 40 s0 00 100 120 140 Column Nomber ColumNumber (fullaviation(sBCON) - noaviation)SO4[ugim^31 JULaverage @ (67.130)Mean- 2.8512e-03 Min- -1.281e-03@(92,24) Max- 6.479e-02 (full aviation (static BCON.no LNOx)- noaviation(noLNOx)) S04(ug/m^3JULaverage Min- -8.391e-04 @ (92.24) Max- 9611e-02 @ (58,113) Mean- 1 1270e-02 40 00 20 40 100 140 120 20 40 60 Column Number ^ (fullaviation(sBCON) - noaviation)N03 lugim31JULaverage k un. 04A0-non [AA al A.- - S97Ao-na s0o Colum Number 100 120 140 N03lug/m^3]JULaverage (lullaviation(staticBCONnoLNO) - noaviation(noLNOx)) Min- -2030e-03@(51.33)Max- 3.378e-01 @(29.93) Mean-50073e03 RAI - .1 D01.-m G1111 411 co1 0.00 1 100 006 0006 0 20 40 so 00 Colum Number 100 120 m. 140 0 a 20 I 40 60 so Column Number 100 120 140 JULaverage - noaviation)OHIppbV] (fullaviation(sBCON) 0 (59.59) Mean- 1.2599e-07 Min- 4.555e-06 0 (46.221 Max- 6.227e-06 Column Number Figure 39: Sensitivity of results to LNOx Results of the sensitivity study indicate that LNOx may have a significantly greater impact in the study of aviation than was previously expected. Although there was a modest (though nonnegligible) decrease in aviation-attributable PM2.5 concentrations by domain average when no LNOx emissions were included in a winter month, there was a very large increase in S04, 03, and PM2.5 for a summer month. The small impact in the winter months can be explained given the much smaller amount of lightning emissions in the winter, coupled with lower photo-dissociation rates, which create and destroy 03, feeding OH and the oxidation of S02. In the summer months, lightning emissions and photolysis rates are much higher, and the inclusion of NO from lightning at altitude apparently leads to a nearly saturated oxidation pathway in which the NOx and VOCs from aviation VOCs react with NO to create N02, which bypasses the normal mechanism in which 03 is destroyed to maintain this equilibrium. 03 is both created and destroyed by photolysis, in a way that results in relatively low 03 levels; however, when VOCs block a path to 03 destruction, levels of 03 rise. As aviation is the only source of VOCs above the PBL, but not have a much smaller effect. the only source of NOx, its effect on increased 03 is dampened; however, when it becomes the only source of both, its impact is magnified. This can be seen in the huge, 16-fold increase in domain average aviation-attributable surface 03 in July when no LNOx is included. Increased 03 levels result in increased OH (again, by nearly 20-fold) and oxidation of precursor gases to PM. Interestingly, this elevated value for 03 is quite similar to yearly average aviation-attributable 03 (though not quite so prominently of OH) when boundary conditions are not held constant (Appendix C). 8 8.1 Comparisons to Empirical Data Context In contrast to studies of cumulative effects on the atmosphere and climate that can use monitor data to gauge the efficacy of a model, studying the small perturbation of aviation defies easy verification of how well the models are doing in their predictions. In the aftermath of the September 11 terrorist attacks, when planes were grounded for more than three days, scientists looked at the effect of a lack of aviation on contrail formation and the resulting changes in temperature this brought (Travis, Carleton, & Lauritsen, 2004), but it was too short a period to isolate air quality effects, especially from cruise emissions, given the transport times, and it is thought that the effect may not have been significant relative to natural meteorological variations (Kalkstein & Balling Jr., 2004). In general, models are evaluated by paired-in-time-and-space comparisons of model predictions to monitors (ground based stations, radiosondes, aircraft measurement, space-based optical instruments, etc.). This cannot isolate the impact of aviation, but it does provide a means of comparing the broader differences in model processes and inventories. Aggregating these measurements to temporal averages is not rigorous, as it doesn't measure the standard deviation of instantaneous model outputs to monitor measurements; however, as the CRFs and mortality calculations assume lengthy exposure, they are useful in gauging the efficacy of model predictions of totalmortalities from PM, and are technically interestingin any case. 8.2 Speciated Model Attainment Test 8.2.1 Methodology The Speciated Model Attainment Test (SMAT) is used as EPA best practice to bring model results in line with the Federal Reference Method (FRM) for computing PM (which method is used in any official EPA figures for PM concentration), as well as shifting results to reflect differences between model outputs and measured monitor concentrations. SMAT maintains the PM species fractions in the ambient speciated data, interpolating monitor data, of irregular distribution across the US, to each model data grid cell on a quarterly basis. Put more simply, one has to trust the model in its prediction of the total relative magnitude of a perturbation, since monitor data only determines the baseline, i.e. there is no isolating the contribution of aviation in monitor data. However, the species fraction split of the monitor data can be mapped to the baseline case of the model data, and the extent of the difference between model and monitor baselines assessed. In this way, the SMAT process can alter both the total magnitude and species fractions of an aviation scenario. Results 8.2.2 Full aviation results retain good agreement pre- and post-SMAT as shown in Figure 40, both spatially and in magnitude. The main shift is an increase in ORG at the expense of N03 as shown in Figure 41. This may be attributed to a good general agreement between CMAQ baseline and FRM monitor concentrations. LTO-only results show similar good agreement. Note that monitor data was only used from 2006, and all other settings were kept at the software defaults. - 797(290to1304) 2006yearlyavg.mortalities (fullaviation-no aviation)FRMPM25[ug/m^3] @(67.130) Mean-2.3919e-02 Max. 1741)-01 Min- -8.023e-03@(13.89) SMAT)FRMPM2.5[ug/m'3) 2006avg,morts.- 783(285to 1281) (full aviation-no aviation. @ (13.69) Max- 3.542e-01@ (53,21) Mean- 2.4608e-02 Min - 8.300e-03 0 07 0406 004 0,03 & 02 001 0 20 4 W we0 O colum Number zo 40 Go 60 W ColumnX 04 - 342(124to560) 2006yearlyavg.mortalities (full aviation- noaviation)N03 [ug/m^3] Mn-1.144e-03@(43,47) Max- 8,211e-02@(60,18) Mean-7.9871e-03 2006avg,morts.- 263(96to 430) SMAT)N03[ug/m^3] (full aviation-no aviation, Min--1300e-03@(36.28) Max-2.373e-01 @(53,21) Mean-6.1534e-03 0 in0 Camrn Nuneer I Numer 0 100 12a 140 I (full aviation - no aviation) S04 [ug/m^312006 yearly avg. mortalities - 174(63 to 284) Min - -4.079e-03 @ (13,69) Max - 3.073e-02 @ (46.22) Mean - 7.4230e-03 (full aviation -no aviation SMAT) S04 )ug/m^3] 2006avg. morts.- 188(68 to 307) Mon- -5,900e-03@(13,69) Max- 5.320e-02@(46.22) Mean - 74226e-03 0035 oO 0025 so 4002 -0015 0005 1001 0 20 40 0 0 Cok.irwiri NL"Ilw 100 12o 140 (full aviation - no aviation. SMAT) NH4 lugim^3| 2006avg morts - 136 (49to 222) Min - -2.900e-03@(78.20) Max - 7.740e-02@(53,21) Mean - 4 2242e-03 o 0 2 i 0 -go 5e o Colim Nuff6w (full aviation -no aviation) NH4 lugJm^312006 yearly avg. mortalities - 142(52 to 233) Min - -1.902e-03@(13.69) Max - 3.011e-02@(67,130) Mean - 39471e-03 003 *005 100 0025 0025 0 02 0 001 001 0 1 20 0 20 40 s0 10 Ca0m Nurmbe 100 0 0 005 20 120 40 s0 0 Colurm Number 100 120 140 Figure 40: Full aviation post-SMAT speciated results in comparison to pre-SMAT model outputs SMAT Full Aviation dConc Full Aviation dCone ORG 8% BC 1% , BC 1% Figure 41: Full aviation SMAT constituent contribution to FRM PM2.5 in comparison to pre-SMAT model outputs 8.2.3 EPA FRM monitor sites There are ~1200 FRM monitor sites across the country, of which -250 are in the STN (Speciation Trends Network) and ~165 in the IMPROVE (Interagency Monitoring of PROtected Visual Environments) network, which provide speciation of PM (Abt Associates, Inc., 2009). Data from locations with species fractions information is interpolated to the larger number of sites which can only measure total PM. Figure 42: EPA FRM monitor sites used in SMAT software (Abt Associates, Inc., 2009) 8.2.4 Comparisons of Outputs to Monitor Data The comparisons between quarterly average monitor data and pre-SMAT quarterly average model data for the absolute baseline is shown in Figure 43. Note that because the monitor sites are irregularly spaced, a natural neighbor interpolation based on Voronoi tessellation was chosen to display concentrations from monitor sites across the domain. Note that FRM PM here denotes dry PM that has been (lightly) hydrated using the (Frank, 2006) first order PBW approximation. This under-predicts model apportionment of PBW by as much as an order of magnitude or more in CMAQ, but is nonetheless the assumption used in the monitor data from which the epidemiological CRFs are derived. It can be seen that CMAQ successfully picks up seasonal changes in PM and matches several regional and urban hotspots, with generally good agreement with monitors by magnitude across the domain. OfficialFF monitorPM2.51stquarter2006[ug/m^3] 7 Min 2 412 e+00@(35.2,-113,5) Max = 2.0846e+01@(30.2,-113.2) Mean NaN (fullaviation+ background) FRMPM2.5|ugim^3 1 s quarler2006 Mm- 4 8915e-01@(184,121 7) Max- 3 6097e+O01 @)(256. 100.3)Mean- 46120e.00 do 10 0 40 W 130W 120'W 110W lo0W 90'W 80W 70'W 60*W OfficialFIRMmonitorPM2.52ndquarter2006[ug/^3] Mn - 3,3995e+00 @(37.8,107,2) Max = 2 4915e+01@(38.0,117,0) Mean (fullaviation+ background) FRIMPM25 NaN lug/m^312nd quarter2006 Min - 1 0318o+00@(18 4 -121 7) Max - 4 4406e+01@(490 1230) Mean - 5 4841e+00 so I0 140W 130'W 120'W 110* W 100*W 90'W 80W 70'W 60'W OfficialFRMmonitorPM2.53rd quarter2006[ug/^3] @(33.5,-84.5) Mean = NaN Min 4.3519e+00@(43.0,-112.8) Max - 2.1160e+01 (full aviation + background)FRM PM2.5 |ug/m^3 3rd quarter 2006 Min - 7 2195e-01@(18.4-121 7) Max - 36369e+01 @(45 5,73.4) Mean - 6.9539e+00 10 5 140W 130'W 120*W 110'W 100'W 90'W 80' W 70'W 60 WY OfficialFRMmonitorPM2.54thquarter2006[ug/^3] Max= 2.7943e+01@(38.0,-117.0) Mean Min 2.7558e+00@(44.5,-106.5) (fullaviation+ background) FRMPM2.5[ug/m^3) 4thquarter2006 Min- 4.3789e-01@(18.4,-121,7) Max- 4.0290e+01 @(25.6,-100.3) Mean- 5.5036e+00 NaN Fj 14....*. W 1'W.. 140W 130*W 2' 120*W 11W 10W 1lltW 100,W 0W.'70W go'W 80*W 70'W w rt~r;w 0W 60 W Figure 43: Comparison of CMAQ monthly average ground-level FRM-hydrated PM2.5 to quarterly (native) official FRM monitor data The aggregation of the differences in all model processes and inventories is reflected in the baseline concentrations predicted by each model (i.e. full aviation + background). A comparison of the results of each model to actual FRM monitor data is possible. Although this does not speak to what would happen if some emission is perturbed, or the feedbacks it might create, it is nonetheless a valuable comparison when fundamental differences between models are examined. The three leading proposed tasks for future work have been comparison of the background inventories, comparisons of model processes, and comparison of numerical methods. The aggregate of these three things is the model baseline results. Note that GEOS-Chem results were interpolated from their native 4x5 resolution. GATOR and CMAQ are displayed by native grid cell size. The months of January and July alone were chosen given the unavailability of other months in the US-ocean nested GATOR runs. OfficialFRMmonitorPM2.51st quarter 2006[ugl3] Max = 2.0846e+01@(30.2-113.2) Mean = NaN Min= 2.4127e+00@(35.2,-113.5) 14W...... ........ ........ ....... OfficialFRMmonitorPM2.53rd quarter 2006[ug/^3] Mean = NaN Max= 2.1160e+01@(33.5,-84.5) Min= 4.3519e+00 @(43.0,-112.0) ..... ....... .......... I 40'W 130'W 120'W 1 10'W 10'W go*W 80'W 70'W 60'W GATOR(full aviation+ background) JUL2006wet PM25[ug/m^3] + background) JAN2006wetPM25[ug/m^3] GATOR(fullaviation Mean- 8,0535e+00 @(44.2-79.8) Max- 3.6301e+01 Min- 8.7864e-01@(16.2-208.8) ""i -A*IC* 112-n A 11 kA-^ " .. - 0 Aac-Ain LWA o_700 kA- . -An 4111- 30 25 20 15 10 5 0 GEOS-Chem (full aviation + background) JUL2006wetPM2.5[ug/m^31 Min- 1.7195e+00Max- 1.5579e+01 Mean- 4.84950+00 GEOS-Chem (full aviation+ background) JAN2006wetPM2.5[ug/m^3] Min- 1.0481e+00 Max- 1.7943*+01Mean- 4.3443e+00 JAN2006 FRMPM2.5(ugim^31 (fullaviation+ background) @(18.4,121.7) Max- 4.5148e+0l @(25.6-100.3)Mean- 4.1668e+00 Min- 1 93700-01 (fullaviation+background) FRMPM2.5Iugm^3] JUL2006 Mn - 6.8681e-01@(18,4.-121.7) Max- 3,9229e+01@(40.5,-73.8) Mean- 6.5300e+00 30 * i lie I: *in 10 *r Figure 44: Comparison of monthly average ground-level FRM-hydrated PM across models to quarterly (native) official FRM monitor data 9 . Limitations and Future Work 9.1 Scope/Feedback There is inherent difficulty in establishing areas of agreement and disagreement when each model has differing scope, resolution, variable duration, and degree of independence. In establishing the relative importance of climate feedbacks, for example, it is necessary to separate the CTM-only and climate-perturbed components of model results, a capability that is currently lacking. Indeed, this is one of many proposed tasks for future work. In addition, CMAQ results are not independent when 80% or more of the full aviation results are attributable to boundary condition effects from GEOSChem. As TOMCAT does not fully treat PM, in assessing mortalities for monetized benefits as of this moment, CTM-only results reduce largely to that of 4x5 GEOS-Chem runs, given their dominance in CMAQ. High-resolution US-nested GEOS-Chem runs are also being conducted and when completed will provide a truer comparison to CMAQ results, providing a solid foundation for the estimation of CTM-only effects. Assuming the nested GEOS-Chem results hold with CMAQ to first order, and that these combined results might be taken as an approximation for the CTM-only portion of the GATOR results (which they may not), preliminary results suggest that the climate effects from GATOR are significantly larger than the chemistry/transport effects. GEOS-Chem and CMAQ use static meteorological data based on the NASA GMAO (Global Modeling and Assimilation Office) GEOS-5 model. As many of the climate feedback effects from GATOR are due to aviation-attributable changes in precipitation and clouds that are created within GATOR, this is a significant difference between models. It is known that aviation has a significant effect on cloud formation, as particulate emissions function as cloud condensation nuclei (CCN). 9.2 Uncertainty The most fundamental source of uncertainty in the study is the extent to which CMAQ accurately describes the atmospheric processes that result from aviation. It is known that CMAQ does not treat feedbacks or changes to weather and cloud formation, which GATOR has shown may be significant. There is no way of quantifying this uncertainty, and so it remains a qualitative caveat in any model results. The aviation inventories themselves have a small level of uncertainty on the order of 5%, which is not carried through to the CIs (Confidence Intervals) in the final results. Likewise, the background inventories derived from EPA's 2001 NEI are assumed to be representative, and the quantification of any uncertainty in these inventories is beyond the scope of this study. It is known that using a 2001 NEI to model a 2006 study year results in higher domainaverage background concentrations, with somewhat lower peaks in some species; however, given the range of inventory years in other models and data sets, this is of lesser relative significance. For instance, population totals were only available for the year 2000. The health impacts CRFs result in a value for nominal mortalities, which is bounded by 95% confidence low and high values. This uncertainty is carried through in the conclusions as a range of calculated benefits. 9.3 Other Differences Comparisons to APMT-AQ outputs show significantly higher concentrations for full flight APMT-AQ is currently derived from training runs that include LTO-only emissions; however, even the LTO-only results show (less) significantly higher aviation-attributable concentrations than APMT-AQ. This can be explained given that APMT-AQ used exclusively static emissions results. boundary conditions for all the CMAQ training runs used to create the response surface. In comparing the LTO-only static BCON case to APMT-AQ, agreement is quite good (with slightly higher domain average but moderately lower mortalities), as can be seen in Figure 45. The ............... ................ . ... differences can likely be explained by different aircraft inventories and diffusion/transport settings used within CMAQ, as well as the domain used for averaging (see below). (full aviation- noaviation)FRMPM2.5[ug/m^32006yearly avg.mortalities - 172 Min- -1288e-02 @(64.39) Max- 8.590e-02 @(46.22) Mean- 1.0902e-03 ULSreduction InFRMPM2.5[ug/m^3|2006 yearly avg mortalities - 72 Min - -3.921e-03@ (63,46) Max- 3.768e-02 @(42.24) Mean- 6.08540-04 0.1 0.00 0.07 0U 0,01 0 20 40 60 0 100 120 140 20 CO~nNumr W 40 00 00 100 120 140 Cou1mnNumber c.) - 125(45to204) -no aviation)FRMPM25[ug/m^3|2006yearly avg.mortalities (ILTO aviation(sBCON) Min-000@(1.1) Max- 1.064e-01 @(67,130) Mean-2.0521e-03 t044 0.0 001 ao alummer 20 44 00 so Cooo umbe.r 100 120 140 Figure 45: APMT-AQ projections of changes in FRM PM2.5 (a.) and b.)) in comparison to LTO-only (static BCON) results from CMAQ In order to apply a consistent means of averaging over dissimilar domains across models, a tightly bound box on the CONUS landmass described by 125W, 67W, 25N, and 51N was used in this project. This must be a caveat in comparison of domain averages to other projects for which this was not done, such as all APMT-AQ work. 9.4 UT/LS Transportin CMA( CMAQ is a nominally tropospheric-only CTM that nonetheless extends to 1OOhPa, which in northern latitudes lies well above the tropopause. The characterization of upper troposphere/lower stratosphere (UT/LS) transport may be poorly defined, and indeed is accounted for in some corrections i.e. setting all concentrations above the tropopause in CMAQ to the GEOS-Chem concentrations at the tropopause (GEOS-Chem v8-02-01 Online User's Guide). 9.5 Proposed Paths Forward When it became apparent that significant model differences necessitated further work to understand the reasons for these differences, a list of proposed tasks was created. These tasks were further weighted across the working group to isolate those that would be most beneficial, given constraints on time and manpower. These are briefly summarized in Table 15, listed from the top in order of the degree to which each project member thought them useful, where each member received one vote. Table 15: Proposed future work ,Argnments Test/T ask Comparison of background Background emissions have a strong effect through emissions interactions with aviation emissions Run GATOR with and without This has been central to the argument of the relative passive tracer to separate climate importance of climate feedbacks, also tests cross-tropopause feedbacks from transport transport; is it possible to run in CTM-only mode, given that this may not necessarily isolate climate effects? Expand scope of high-resolution In progress GATOR runs for further scenarios/durations Characterize cross-tropopause Not likely to be main source of difference, but useful given transport in each model 30% of emissions are in LS, may be possible to roll in with passive tracer test Run CTMs with perturbed wet removal parameters Useful in understanding cloud processes across models, assessing sensitivity of results to wet removal process, would need to define uncertainty range, where/when it rains - not easy to fix Create inventory of processes modeled Information in literature, likely highly qualitative; however, may be a good investment for future work using these models Create inventory of numerical Likely to provide insight, given the importance of numerical methods used in each model for each process methods in a model, but information alone is still qualitative, may be a good investment for future work using these models Comparison of 7-Be observations The only proposed empirical test of aviation with model performance emissions/transport processes, may be a good proxy, but depends on wet removal Comparison of 03, CO, and PAN May be a useful proxy for general model performance for vertical profiles with data vertical mixing, but we are most interested in modeling perturbation, not baseline; however, this is a good sanity check as well, been done before Evaluation of met model outputs Perform constant mixing ratio test Compare against satellite data products Do vertical grid resolution sensitivity study for CTMs Assess inter-annual variability Perform rotating plume test Introduce a Lagrangian transport test Proceed without CMAQ Introduce additional coupled chemistry-climate model Introduce CMAQ hemispheric model Do RF calculations for GEOSChem results Important to have confidence in met models, given their significance, but such evaluation has been done before; both CMAQ and GEOS-Chem use NASA GEOS-5 data derived in part from satellite observations, likely about as good as it gets for static weather Already known for GEOS-Chem; routine, easy, could be used to compare vertical transport CALIPSO and MODIS AOD for aerosols aloft, OMI for N02 and HCHO aloft, essentially model validation Unlikely to be significant driver, all models have roughly similar high resolution in PBL, lower in free trop; may be useful for sensitivity Aviation-attributable changes usually smaller than inter-annual changes, but would entail much rework to do an entirely different year Not much bang for the buck, may need to be custom-coded, some references in literature already, direct test of numerical method used, some think "simpler is better" in trying to do a comparison, Rastigejev (citation) paper would seem to obviate importance of test Would identify transport trajectories, but a lot of work to implement (and take away from other work), can't do nonlinear chemistry, difficult for vertical transport CMAQ currently dominated by GEOS-Chem boundary condtions, but CMAQ needed for urban resolution and fidelity to population over US domain; CMAQ is precisely designed for regional AQ, the central focus of valuation; would focus science primarily on GEOS-Chem and GATOR and add CMAQ tests only where time/utility allows Depends on availability, potentially a lot of work to implement, but currently GATOR has no comparison for climate feedbacks, will be done via ACCRI, but may add to general confusion currently Would be a better comparison for the GATOR US-Oceannested runs, but code is not yet mature, may be issues even if it were, based on comparison to an intercontinental version of CMAQ Simplified stratospheric chemistry is not really a problem since you want delta 03, would provide a comparison to TOMCAT, doesn't really address any central issue 10 Conclusion 10.1 CTM Findings Both GEOS-Chem and CMAQ show that when a global full flight emissions ULS policy is studied, the benefits, as monetized reduction in premature mortality from reduced PM2.5 in the US, are of the same order of magnitude as the costs in the US. Reducing the fuel sulfur content by nearly 98% results in a significant reduction in aviation-attributable S04; however, there is an upper bound on this reduction due to aviation NOx emissions and resulting oxidation of background S02, predicted to be ~40-50%. As other PM species are only marginally affected by a ULS fuel policy, this effectively limits the total reduction in aviation-attributable PM and mortalities to ~10-20%. Runs in which boundary condition effects were isolated suggest that transport of above-LTO emissions to ground occurs both with and without changing boundary conditions. Given the vertical transport and circulation times, it is thought that cruise-level emissions from Europe and Asia may contribute to elevated concentrations over North America. Indeed, using changing boundary conditions to model global full aviation results in a higher ratio of full aviation to LTO aviation domain average surface PM concentration, with a value of -5.4. However, runs in which both full and LTO aviation scenario boundary conditions were held static still show a ratio of ~2.3, indicating significant transport of US emissions to ground-level in the US. LNOx emissions were found to have a significant dampening effect on the contribution aviation emissions are able to exert by partially saturating oxidative capacity at altitude. The climate and weather feedback effects of a ULS fuel policy, as well as the significance of their omission in the CTMs, remains an interesting subject of ongoing study. 10.2 Cost/Benefit Assessment The cost of HDS is projected to be between 4 and 7 cents per gallon of jet fuel. With 2006 levels of domestic jet fuel consumption, this translates to a yearly cost of $540-$940 million. CMAQ predicts that a global ULS jet fuel policy saves 110 lives per year in the US (with a 95% CI of 40-180) when considering full flight emissions, a 14% reduction in aviation-attributable mortality resulting in a nominal monetary benefit of $800 million (corresponding to a 95% CI of $300-$1300 million). This valuation does not monetize other operational benefits of such a switch, such as pipeline transport, military diesel operations, maintenance, combustor lifetime, and the introduction of technology relevant to the use of biofuels for aviation. Even so, the monetized health impact benefits compare favorably to the cost of HDS. APPENDICES Appendix A: CMAQ Inventory and Model Species A.1 Aircraft Emissions Inventory Table 16 shows the full list of CMAQ aircraft inventory species and a brief description thereof, followed by monthly totals of the pre-processed AEDT outputs showing variation in emissions throughout the year. This is followed by vertical profiles of the AEDT outputs, a table of postprocessed CMAQ species throughout the year, and vertical profiles of the same. Finally, the preprocessed, speciated hydrocarbon emissions inventory (showing species splits as a fraction of THC) is shown. Table 16: Aircraft inventory species CO Carbon Monoxide NO Nitrogen Oxide N02 Nitrogen Dioxide ALD2 Grouping of Acetaldehyde and higher aldehydes ETH Ethene (CH2=CH2) FORM Formaldehyde and related MEOH Methanol NR Non-Reactive Hydrocarbon OLE Grouping of Olefin Carbon Bond (C=C) PAR Grouping of Paraffin Carbon Bond (C-C) TOL Toluene (C6H4-CH3) and related XYL Xylene (C6H5-(CH3)2) and related S02 Sulfur Dioxide POA Primary Organic Aerosols PEC Primary Elemental Carbon PSO4 Primary Sulfates SULF Sulfuric Acid (H2SO4) HONO Nitrous Acid Table 17: Full aviation total aircraft emissions from AEDT FEB 4.19 2.12E-2 3.21E-3 5.36E-2 1.53E-4 2.74E-4 2.50E-4 3.36E-3 13.2 5.18 4.93E-3 MAR APR 4.81 4.64 2.46E-2 3.65E-3 2.36E-2 3.44E-3 6.14E-2 5.97E-2 1.76E-4 3.14E-4 2.88E-4 3.86E-3 15.2 5.95 5.66E-3 1.69E-4 3.03E-4 2.79E-4 3.72E-3 14.7 5.74 5.46E-3 MAY 4.69 2.37E-2 3.49E-3 6.03E-2 1.71E-4 3.06E-4 2.85E-4 3.75E-3 14.8 5.80 5.52E-3 JUN 4.66 2.37E-2 3.44E-3 6.OOE-2 1.70E-4 3.04E-4 2.83E-4 3.73E-3 14.7 5.76 5.48E-3 JUL 4.86 2.40E-2 3.46E-3 6.31E-2 1.77E-4 3.18E-4 2.94E-4 3.90E-3 15.4 6.02 5.72E-3 AUG 4.94 2.49E-2 3.61E-3 6.41E-2 1.80E-4 3.23E-4 3.OOE-4 3.95E-3 15.6 6.11 5.81E-3 SEP 4.54 2.33E-2 3.37E-3 5.87E-2 1.65E-4 2.96E-4 2.76E-4 3.62E-3 14.3 5.61 5.34E-3 OCT 4.71 2.45E-2 3.56E-3 6.10E-2 1.71E-4 3.07E-4 2.86E-4 3.76E-3 14.9 5.83 5.54E-3 NOV 4.52 2.33E-2 3.40E-3 5.81E-2 1.64E-4 2.95E-4 2.73E-4 3.61E-3 14.3 5.59 5.31E-3 DEC 4.77 2.36E-2 3.48E-3 6.14E-2 1.73E-4 3.12E-4 2.82E-4 3.82E-3 15.1 5.90 5.61E-3 2006 55.9 0.720 0.00204 0.00365 0.00337 0.0447 17.7 69.1 0.0657 0.283 0.0414 ............ ...... !t ....... ... ....... 2006 US Aircraft FB Emissions [Tg] 2006 US Aircraft HC Emissions [Tg] 2006 US Aircraft CO Emissions [Tg] a- 400 t 600 1000 2006 US Aircraft NOx Emissions [Tg] 0.1 0.05 0.015 0.01 0.005 0 0.015 0.01 0.005 0 2006 US Aircraft PMNV Emissions [Tg] 2006 US Aircraft PMFO Emissions [Tg 200 200 400 600 "NOx 800 0.01 1000 0 1000 "w 0 0.03 0.02 - -- 2 4 - 4 a 6 2 0 4 x10 x10 2006 US Aircraft PM Emissions [Tg] 200 """ ----- ~ 1000 I- - 0 2 ...-.. M 4 6 200 200 -0 1 1000 -1 2 3 4 -4 x10 2006 US Aircraft SOx Emissions [Tg] 200 n sox --- 1000 0 1 2 3 -6 2006 US Aircraft H20 Emissions [Tg] 2006 US Aircraft C02 Emissions 1000 ft0 6 6 4 Figure 46: AEDT vertical profiles of US yearly aircraft emissions - M0 - -- 2 4 6 4 8 Table 18: Full aviation total aircraft emissions CMAQ processed input totals Outputs CO [Tg] NO [Tg] NO2 [Tg2 5.14E-2 4.71E-2 6.58E-3 6.04E-3 4.78E-4 5.98E-4 5.33E-4 7.00E-5 2.85E-4 OLE [Tg] 3.35E-4 FEB 2.24E-2 2.12E-2 4.60E-4 5.72E-4 5.09E-4 6.71E-5 2.73E-4 3.20E-4 MAR 2.46E-2 5.40E-2 6.93E-3 5.19E-4 6.50E-4 5.78E-4 7.63E-5 3.10E-4 3.65E-4 JAN 4 [P3 Tg] ETH T4 FO RM[g 4 E T] 4 APR 2.36E-2 5.25E-2 6.73E-3 4.92E-4 6.16E-4 5.46E-4 7.17E-5 2.92E-4 3.43E-4 MAY 2.37E-2 5.30E-2 6.81E-3 4.96E-4 6.24E-4 5.54E-4 7.29E-5 2.96E-4 3.48E-4 JUN JUL 2.37E-2 5.27E-2 6.79E-3 4.92E-4 6.16E-4 5.46E-4 7.17E-5 2.92E-4 3.43E-4 5.55E-2 5.63E-2 7.13E-3 7.25E-3 4.92E-4 6.20E-4 5.50E-4 7.23E-5 2.94E-4 3.45E-4 AUG 2.40E-2 2.49E-2 5.14E-4 6.46E-4 5.74E-4 7.52E-5 3.07E-4 3.60E-4 SEP 2.33E-2 5.16E-2 6.65E-3 4.83E-4 6.03E-4 5.37E-4 7.06E-5 2.86E-4 3.38E-4 OCT 2.45E-2 5.36E-2 6.91E-3 5.05E-4 6.37E-4 5.66E-4 7.40E-5 3.02E-4 3.55E-4 NOV 2.33E-2 5.10E-2 6.58E-3 4.87E-4 6.07E4 5.42E-4 7.11E-5 2.89E-4 3.40E-4 DEC 2.36E-2 5.40E-2 6.94E-3 4.96E-4 6.20E-4 5.54E-4 7.29E-5 2.96E-4 3.22E-4 2006 0.283 0.633 0.0814 0.00592 0.00741 0.00659 0.000866 0.00352 0.00411 POA [Tg]f PEC-[Tg Outputs PAR [Tg] TOL [Tg-] XYL [Tg]-SO2 [Tg]1 PSO04 [,Tg]. SULF [Tg,] HONO-[g 1.48E-3 1.42E-3 9.35E-5 8.97E-5 1.06E-4 1.02E-4 5.36E-3 2.72E-4 1.66E-4 2.98E-4 1.67E-4 4.67E-4 FEB 4.93E-3 2.50E-4 1.53E-4 2.74E-4 1.54E-4 4.28E-4 MAR 1.61E-3 1.02E-4 1.16E-4 5.66E-3 2.88E-4 1.76E-4 3.14E-4 1.77E-4 4.92E-4 APR 1.52E-3 9.61E-5 1.09E-4 5.46E-3 2.79E-4 1.69E-4 3.03E-4 1.71E-4 4.77E-4 MAY 1.54E-3 9.76E-5 1.11E-4 5.52E-3 2.85E-4 1.71E-4 3.06E-4 1.72E-4 4.82E-4 4.8E-4 JAN JUN JUL 1.51E-3 9.61E-5 1.09E-4 5.48E-3 2.83E-4 1.70E-4 3.04E-4 1.71E-4 1.53E-3 9.66E-5 1.10E-4 5.72E-3 2.94E-4 1.77E-4 3.18E-4 1.79E-4 5.05E-4 AUG 1.59E-3 1.01E-4 1.15E-4 5.81E-3 3.00E-4 1.80E-4 3.23E-4 1.82E-4 5.13E-4 SEP 1.49E-3 9.42E-5 1.07E-4 5.34E-3 2.76E-4 1.65E-4 2.96E-4 1.67E-4 4.70E-4 OCT 1.57E-3 9.95E-5 1.13E-4 5.54E-3 2.86E-4 1.71E-4 3.07E-4 1.73E-4 4.88E-4 NOV 1.50E-3 9.52E-5 1.08E-4 5.31E-3 2.73E-4 1.64E-4 2.95E-4 1.66E-4 4.65E-4 DEC 1.54E-3 9.73E-5 1.10E-4 5.61E-3 2.82E-4 1.73E-4 3.12E-4 1.75E-4 4.91E-4 2006 0.0183 0.00116 0.00132 0.0657 0.00337 0.00204 0.00365 0.00205 0.00576 Table 19: Corresponding molecular weights for HC species and groupings in Table 18 i.e. (MF/SF) 105.8 97.44 14.58 29.14 13.60 32.04 29.59 28.05 36.55 [g/mol] grouping explicit grouping explicit grouping grouping grouping grouping grouping 2006 US Aircraft CO Emissions [Tg] 0.1 0.05 0 2006 US Aircraft ALD2 Emissions [Tg] 2006 US Aircraft NO Emissions [Tg] 0 0.05 0.1 0.15 2006 US Aircraft N02 Emissions [Tg] 0 0.2 0.01 0.005 0.015 0.02 2006 US Aircraft ETH Emissions [Tg] 2006 US Aircraft FORM Emissions [Tg] FORM 2 - - -- Jm - 1000 0 -- 6 4 .4 2006 US Aircraft MEOH Emissions [Tg] 2006 US Aircraft NR Emissions [Tg] 1000 0 xl10 x10 x 10 x105 2 4 6 6 2006 US Aircraft OLE Emissions [Tg] 0 2 x10 Figure 47: CMAQ processed input vertical profiles of US yearly total aircraft emissions 2005 US Aircraft PAR Emissions [Tg] 2005 US Aircraft TOL Emissions [Tg] 2006 US Aircraft XYL Emissions [Tg] 200 200 200 ', "-XYL )eIo . - 400 400 600 600 600 500 innn - I- 1000 1 -----------------------------------0 2 4 6 1000 6 0 2 4 6 .6 -4 x 10 x106 2006 US Aircraft S02 Emissions [Tg] x10 2006 US Aircraft POA Emissions [Tg] 2006 US Aircraft PEC Emissions [Tg] POA 200 400 600 X"I"*.*~ c 1 600 1000 0 1 2 3 4 0 1 2 3 4 -4 x10 x10 0 2 4 6 -4 -4 2006 US Aircraft SULF Emissions [Tg] 2006 US Aircraft HONO Emissions [Tg] 200 "" " -200 -- 0 0 2 2 4 4 6 6 000 0 is.,--e .5 0.5 1 1 .-5 1.5 Figure 48: CMAQ processed input vertical profiles of US yearly total aircraft emissions (continued) Table 20: Aircraft emissions pre-processed, speciated hydrocarbon profile (FAA OEE & EPA OTAQ, 2009) Ethylene 0.15458986 28.05316 Acetylene 0.039385952 26.03728 2 C2H2 Ethane 0.005214505 30.06904 2 C2H6 Propylene 0.045336437 42.07974 3 C3H6 Propane 0.000780871 44.09562 3 C3H8 Isobutene/1-Butene 0.017538274 56.10632 4 C4H8 1,3-Butadiene 0.016869627 54.09044 4 C4H6 cis-2-Butene 0.002104593 56.10632 4 C4H8 3-Methyl-1-butene 0.001403439 70.1329 5 C5H10 100 C2H4 1-Pentene 0.007760686 70.1329 5 C5H10 2-Methyl-1-butene 0.001744648 70.1329 5 C5H10 n-Pentane 0.00198433 72.14878 5 C5H12 trans-2-Pentene 0.003593968 70.1329 5 C5H1O cis-2-Pentene 0.002757017 70.1329 5 C5H1O 2-Methyl-2-butene 0.001846216 70.1329 5 C5H1O 4-Methyl-1-pentene 0.000858179 84.15948 6 C6H12 2-Methylpentane 0.004084956 86.17536 6 C6H14 2-Methyl-1-pentene 0.000428122 84.15948 6 C6H12 1-Hexene 0.00736025 84.15948 6 C6H12 trans-2-Hexene 0.000371141 84.15948 6 C6H12 Benzene 0.01681482 78.11184 6 C6H6 1-Heptene 0.004384568 98.18606 7 C7H14 n-Heptane 0.000638894 100.20194 7 C7H16 Toluene 0.006421156 92.13842 7 C7H8 1-Octene 0.002757017 112.21264 8 C8H16 n-Octane 0.000624801 114.22852 8 C8H18 Ethylbenzene 0.001742866 106.165 8 C8H1O m-Xylene/p-Xylene 0.002821783 106.165 8 C8H1O Styrene 0.003094253 104.14912 8 C8H8 o-Xylene 0.001659872 106.165 8 C8H1O 1-Nonene 0.002455358 126.23922 9 C9H118 n-Nonane 0.000623583 128.2551 9 C9H20 Isopropylbenzene 3.96117E-05 120.19158 9 C9H12 n-Propylbenzene 0.00066586 120.19158 9 C9H12 m-Ethyltoluene 0.001926704 120.19158 9 C9H12 p-Ethyltoluene 0.000802048 120.19158 9 C9H12 1,3,5-Trimethylbenzene 0.000675821 120.19158 9 C9H12 o-Ethyltoluene 0.000817972 120.19158 9 C9H12 1,2,4-Trimethylbenzene 0.004377359 120.19158 9 C9H12 1-Decene 0.001846216 137.1921245 10 C1OH20 n-Decane 0.003201988 142.28168 10 C10H22 1,2,3-Trimethylbenzene 0.001327673 120.19158 9 C9H12 n-Undecane 0.004441511 156.30826 11 C11H24 n-Dodecane 0.004615541 170.33484 12 C12H26 n-Tridecane 0.005354028 184.36142 13 C13H28 C14-alkane 0.00186031 198.388 14 C14H30 C15-alkane 0.00177053 212.41458 15 C15H32 n-tetradecane 0.00416355 198.388 14 C14H30 101 C16-alkane 0.001459826 226.44116 16 C16H34 n-pentadecane 0.001726267 212.41458 15 C15H32 n-hexadecane 0.000486609 226.44116 16 C16H34 C18-alkane 1.76775E-05 254.49432 18 C18H38 n-heptadecane 8.84283E-05 240.46774 17 C17H36 phenol 0.007261785 94.11124 6 C6H50H naphthalene 0.00541181 128.17052 10 C1OH8 2-methyl naphthalene 0.002061886 142.1971 11 C11H10 1-methyl naphthalene 0.002466177 142.1971 11 C11H1O dimethylnapthalenes 0.000898492 156.22368 12 C12H12 C4-Benzene + C3-aroald 0.006564325 134.21816 10 C1OH14 C5-Benzene+C4-aroald 0.003241136 148.24474 11 C11H16 Methanol 0.018051895 32.04186 1 CH30H Formaldehyde (FAD) 0.123081099 30.02598 1 CH20 Acetaldehyde (AAD) 0.042718224 44.05256 2 C2H40 Acetone 0.003693477 58.07914 3 C3H60 Propionaldehyde 0.007265856 58.07914 3 C3H60 Crotonaldehyde 0.012909514 70.08984 4 C4H60 Butyraldehyde 0.001481767 72.10572 4 C4H80 Benzaldehyde 0.004695067 106.12194 7 C7H60 Isovaleraldehyde 0.000406083 86.1323 5 C5H100 Valeraldehyde 0.003064793 86.1323 5 C5H100 o-Tolualdehyde 0.002872119 120.14852 8 C8H80 m-Tolualdehyde 0.003471951 120.14852 8 C8180 p-Tolualdehyde 0.000602286 120.14852 8 C8H80 Methacrolein 0.005362609 70.09 4 C4H60 Glyoxal 0.018164641 58.03608 2 C2H202 Methylglyoxal 0.015032806 72.06266 3 C3H402 acrolein 0.024493139 56.06326 3 C3H40 C-10 paraffins 0.141565 142.28168 10 C10H22 C-10 oleffins 0.056626 140.2658 10 C1OH20 Decanal 0.056626 156.2652 10 C1OH200 dodecenal 0.028313 184.31836 12 C12H240 A.2 Background Inventory and Concentrations It has been established that the background inventory has a significant effect on the reaction of aviation-attributable species. The list of post-processed 2001 NEI CMAQ background species is shown in Table 21, along with a brief description of the species. This is followed by yearly average 102 surface plots of background concentrations of several significant species and groupings of species, which are driven by the background inventory. Table 21: Background inventory species and descriptions Grouping of Acetaldehyde and higher aldehydes Carbon Monoxide Ethene (CH2=CH2) Formaldehyde (CH20) and related ALD2 CO ETH FORM ISOP NH3 NO Isoprene (C5H8) Ammonia Nitrogen Oxide Nitrogen Dioxide Non-Reactive Hydrocarbon Grouping of Olefin Carbon Bond (C=C) Grouping of Paraffin Carbon Bond (C-C) N02 NR OLE PAR Primary Elemental Carbon Coarse Particulate Matter Fine Particulate Matter Primary Nitrates Primary Organic Aerosols Primary Sulfates Sulfur Dioxide Sulfuric Acid (H2SO4) Grouping of Terpenes Toluene (C6H4-CH3) and related Xylene (C6H5-(CH3)2) and related PEC PMC PMFINE PNO3 POA PSO4 S02 SULF TERPB TOL XYL lug/m^3) backgroundNH3 [ppbV]2006 yearlyaverage Max - 2.125e+01 @(53.21) Mean- 6.9110e-01 Mn -000@(1,1) 2006yearly average backgroundFRM PM2.5 Min - 000@(1 1) Max -3126e+01 @(1467) Mean - 5,7825e+00 404 u0 Com CokorNorreor 0 O C N0 ICok 103 N0 o N.,,t&o 00 4 background S02IppbV] 2006yearlyaverage Min - 000@ (1.1) Max- 1.365e+01 @ (30,92) Mean- 8.1943e-01 background N02 (ppbV 2006yearlyaverage Min- 000@(1,1) Max- 3.786e+01 @(67,130) Mean- 2.1725e+00 60 40 40 20 20 --~~~~ ~~ 0 20 - -~--~ - --- 40 -~ 60 00 100 120 0 140 m o Column Numbor ColumnNumber background NO[ppbV]2006yearlyaverage Min- 000@(1,1) Max- 2.598e+01 @ (67,130) Mean- 2.0451e-01 background N03 lug/m^3]2006yearlyaverage Min-000 @(1,1) Max- 3.609e+00 @(64,125) Mean- 3.2596e-01 5 4.5 100 4 so -5 60 20 0 20 40 s0 00 Column Number 100 120 140 0 background S04 [ug/m^3] 2006yearlyaverage Min-0000@(1,1) Max- 7.837e+00 @(43,106) Mean- 1.7349e+00 0 40 00 00 ClumnNumber 100 120 140 background H2SO4[ppbV]2006yearlyaverage Min- 000@(1,1) Max- 5.906e-03 @(23,66) Mean- 3.4887e-04 -r 20 0 0 40 50 C 0 000 000 140 1 00 a ColumnNumber 20 100 40 Column Number 104 120 140 background NH4[ugm^3) 2006yearlyaverage Mean- 5.4595e-01 Min- 000@(1,1) Max- 2.975e+00@(64.125) background HNO3[ppbVj2006yearlyaverage Max- 3.098e+00@(29,95) Mean-5.9693e-01 Min-000@(1.1) 60 2.s 0: 00___ 20 40 10g 120 Cokmn Numbner background OH[ppbV] 2006yearlyaverage, Min-000@(1.1) Max- 1.375e-04@(36.60) Mean-5.1958e-05 u uy a n NIr Colukm e 140 I, 0 20 100 40 120 ComnNumber background H202 [ppbVj 2006yearlyaverage Min-000@(1.1) Max- 2.889e+00@(16.43) Mean- 1.3072e+00 40 v 60 00 100 120 140 CokumNumber background COIppbV2006yearlyaverage 2006yearlyaverage background 03 [ppbv) @(30,101) Mean-3.7612e+01 Min- 000@(1.1) Max- 4.859e+01 Ain- n Wn 1i 11 Ma.. 7071.4006i7 1 Um Moan.1 017a4n 250 150 too 0 Coumn Numbr 105 20 40 60 so 0 CoklunNumber 100 120 140 background BC[ug/m^3]200Byearly average Min- 000@(1,1) Max- 4.098e+00 @(94,24) Mean- 1.9054e-01 0 20 40 u uu Column Number 100 background H20(PBW)[ug/m^3]2006yearlyaverage Min- 000@(1,1) Max- 1.488e+01 @(67,130) Mean- 3.2276e+00 0 0 120 background ORG[ppbV]2006yearlyaverage Min- 000@ (1,1) Max- 1.S04e+01 @(94,24) Mean- 1.2037e+00 0 0 20 xe 40 00 100 120 - - - - 20 40 bu n ColumnNumber background VOC(ppbVl 2006yearly average Min- 000@(1,1) Max- 1.828e+02 @(67,130) Mean- 2.6936e+01 140 0 20 40 ColumnNumber so 0 100 120 140 Column Number Figure 49: CMAQ 2006 yearly average background species concentrations Note that in Figure 49, PBW is directly output from the model, not derived using the lightly hydrated FRM method. A.3 CMAQ Outputs CMAQ particulate outputs are divided into three modes: Aitken, accumulation, and coarse (described in Section 2). Aitken mode particulates are denoted by "J", accumulation mode by "I", and coarse mode by "K". A brief description of all CMAQ species and groupings treated in the model is listed in Table 22. Table 22: CMAQ4.6 concentration output species and descriptions (Isukapalli, 1999) SpeciesNierid NO2 NO Nitrogen Oxide Nitrogen Dioxide O 03 Ground Potential Atomic Oxygen (singlet) Ozone N03 01D Energized Atomic Oxygen (singlet) Nitrate 106 OH H02 N205 HNO3 HONO PNA Hydroxide Hydroperoxyl radical Dinitrogen Pentoxide Nitric Acid Nitrous Acid Peroxynitric Acid H202 CO FORM ALD2 Hydrogen Peroxide C203 X02 PAN PACD PAR XO2N ROR NTR OLE FACD AACD ETH TOL CRES T02 OPEN CRO XYL MGLY ISOP ISPD S02 SULF UMIHP TERP AS04J ASO4I ANH4J ANH4I ANO3J ANO3I AORGAJ AORGAI Carbon Monoxide Formaldehyde (CH2O) and related Acetaldehyde and higher aldehydes Peroxyacyl radical (CH3C(O)OO) NO-to-NO2 Operation Peroxyacyl Nitrate Higher Peroxycarboxylic Acid Grouping of Paraffin Carbon Bond (C-C) NO-to-Nitrate Operation Secondary Organic Oxy radical Alkylnitrate (>C3) Grouping of Olefin Carbon Bond (C=C) Formic Acid Higher Carboxylic Acid Ethene (CH2=CH2) Toluene (C6H4-CH3) and related Cresol and higher molecular weight Phenols Toluene-hydroxyl radical adduct High molecular weight aromatic oxidation ring fragment Methylphenoxy radical Xylene (C6H5-(CH3)2) and related Methylglyoxal (CH3C(O)C(O)H) Isoprene (C5H8) Products of Isoprene reactions Sulfur Dioxide Sulfuric Acid (H2SO4) Methyl Hydroperoxide Terpenes Accumulation mode Sulfates Aitken mode Sulfates Accumulation mode Ammonium Aitken mode Ammonium Accumulation mode Nitrates Aitken mode Nitrates Accumulation mode Secondary Anthropogenic Organics Aitken mode Secondary Anthropogenic Organics AECJ AECI Accumulation mode Secondary Biogenic Organics Accumulation mode Secondary Biogenic Organics Accumulation mode Elemental Carbon Aitken mode Elemental Carbon A25J Aitken mode unspecified anthropogenic mass AORGBJ AORGBI 107 ACORS ASOIL NUMATKN NUMACC NUMCOR SRFATKN SRFACC AH2OJ AH2OI ANAJ ACLJ ANAK ACLK ASO4K NH3 SGTOT_ALK SGTOTOLI_1 SGTOTOLI_2 SGTOTXYL_1 SGTOTXYL_2 SGTOTCSL SGTOTTOL_1 SGTOT TOL_2 SGTOT TRP 1 SGTOTTRP_2 HCL Other coarse mass Soil (dust) Aitken mode number Accumulation mode number Coarse mode number Aitken mode surface area Accumulation mode surface area Accumulation mode Water Aitken mode Water Accumulation mode Sodium Accumulation mode Chloride Coarse mode Sodium Coarse mode Chloride Coarse mode Sulfate Ammonia Combined Gas + Particle Alkanes Combined Gas + Particle SOA precursor species produced by alkene reactions 1 Combined Gas + SOA precursor species produced by alkene reactions 2 Combined Gas + Particle Xylenes 1 Combined Gas + Particle Xylenes 2 Combined Gas + Particle Cresol Combined Gas + Particle Toluenes 1 Combined Gas + Particle Toluenes 2 Combined Gas + Particle Terpenes 1 Combined Gas + Particle Terpenes 2 Hydrochloric Acid Appendix B: CMAQ Processes B.1 Carbon-Bond IVProcesses and Reactions Modeled The default chemical mechanism used for all ULS runs is Carbon Bond IV (CB-IV), developed mainly for urban smog and regional air quality mode (Isukapalli, 1999), (Morris & Myers, 1990). The mechanism is a hybrid of explicit chemistry, surrogate approximations, and lumped or generalized chemistry designed to simulate the features of urban smog chemistry. Explicit chemistry is used for the inorganic and carbonyl species and the chemistries of ethene, isoprene, and formaldehyde. The ethene chemistry, for example, is treated explicitly because it reacts much slower than other alkenes, it is a large fraction of common hydrocarbon emissions, and that it yields a high fraction of formaldehyde. Isoprene is an alkene but it is treated explicitly also because its reactions with free radicals are much faster, and it forms a prominent part of biogenic emissions in rural areas. Many peroxy radicals have been lumped into a single X02 universal peroxy radical. The lumping method of carbon-bonds is used mainly for paraffins and olefins. Molecular surrogates toluene and xylene are used for higher aromatic compounds. For instance, a complex molecule with both 108 aromatic and alkene structures might be represented with a combination of TOL, OLE, and PAR surrogates. Also, some of the rates and the stoichiometries of some incorporated reactions depend upon the atmospheric composition of reacting hydrocarbons (Isukapalli, 1999). The hierarchical relationship between the major species in the carbon bond mechanism is depicted in Figure 50. The most complex species, in terms of their oxidation products (paraffins, olefins, isoprene and aromatic hydrocarbons), are placed at the highest levels. The simplest species, in terms of molecular complexity (NOx, HOx, CO and formaldehyde), are placed at the lowest levels (Isukapalli,1999). Table 23 presents the chemical reactions that constitute the CB-IV mechanism. Note that in Table 23, h V denotes the addition of a photon of light of frequency V in a photolytic reaction, and M denotes an additional reagent. Figure 50: Hierarchical relationship between major species in CB-IV (Isukapalli, 1999) 109 Table 23: CB-IV reactions (Isukapalli, 1999) Reacti No. R1 ]NO2 +h R2 O Reaction .. NO + O 0 3 021 R3 3 + NO R4 O+NO2 NO R5 O+ N02 N03 R6 O+NO R7 NO2 + O3 N02 [ N02 NO3 R8 03 + h .. 0 R9 03 + h y -. 01D R10 01D R1 1 01D + H2O R12 03 + OH R13 ] 03 + HO2 --- OH FN3 + hu .-- 0.89 N02 + 0.89 0 + 0.11 NO R15 ]INO3 + NO -. 2 N03 R16 ]NO3 .-- NO + N02 R14 + NO2 _0-+_O .-. 2 OH HO2 R17 N03 + N02 R18 N2O5 + H20 R19 N205 N03 + N02 R20 NO + NO 2 N02 R21 NO + N2 H20 2 HNO2 R22 NO + OH HNO2 R23 HNO2 +h R24 OH + HNO2 N205 -- 2 HNO3 NO + OH .....+ NO2 110 HNO2 + HNO2 NO + N02 N02 + OH HN03 OH + HNO3 NO3 H02 + NO OH + N02 H02 + N02 PNA PNA H02 + N02 OH + PNA N02 H02 + H02 H202 H02 + H02 H20 H202 H202 + hy 2 OH OH + H202 H02 OH + CO H02 FORM + OH H02 + CO FORM + hy 2 HO2 + CO FORM + hV CO FORM + 0 OH + HO2 + CO FORM + N03 HNO3 + HO2 + CO ALD2 + O C303 + OH ALD2 + OH C2O3 ALD2 + N03 C2O3 + HNO3 ALD2 + hv FORM + 2 H02 + CO + X02 C203 + NO FORM + NO2 + HO2 + XO2 C203 + N02 PAN PAN C2O3 + NO2 C203 + C203 2 FORM + 2 XO2 + 2 H02 C203 + H02 0.79 FORM + 0.79 X02 + 0.79 H02 + 0.79 OH 111 OH FORM + X02 + H02 PAR + OH 0.87 X02 + 0.13 XO2N + 0.11 H02 + 0.11 ALD2 + 0.76 ROR - 0.11 PAR ROR 0.96 X02 + 1.1 ALD2 + 0.94 H02 + 0.04 XO2N + 0.02 ROR - 2.1 PAR ROR ROR + NO H02 2 0 + OLE 0.63 ALD2 + 0.38 H02 + 0.28 X02 + 0.3 CO + 0.2 FORM + 0.02 XO2N + 0.22 PAR + 0.2 OH OH + OLE FORM + ALD2 - PAR + X02 + H02 02 + OLE 0.5 ALD2 + 0.74 FORM + 0.22 X02 + 0.1 OH + 0.33 CO + 0.44 H02 - PAR N02 + OLE 0.91 X02 + FORM + 0.09 XO2N + ALD2 + N02 - PAR O + ETH FORM + 1.7 H02 + CO + 0.7 X02 + 0.3 OH OH + ETH X02 + 1.56 FORM + 0.22 ALD2 + H02 03 + ETH FORM + 0.42 CO + 0.12 H02 TOL + OH 0.44 H02 + 0.08 X02 + 0.36 CRES + 0.56 T02 T02 + NO 0.9 N02 + 0.9 H02 + 0.9 OPEN T02 CRES + H02 OH + CRES 0.4 CRO + 0.6 X02 + 0.6 H02 + 0.3 OPEN CRES + N03 CRO + HNO3 CRO + N02 OH + XYL 0.7 H02 + X02 + 0.2 CRES + 0.8 MGLY + 1.1 PAR + 0.3 T02 OPEN + OH X02 + 2 CO + 2 H02 + C203 + FORM 112 R71 OPEN + h v _Ly1 R72 OPEN + O3 _ C2O3 + HO2 + CO 110.03 ALD2 + 0.62 C203 + 0.7 FORM + 0.03 X02 + 0.69 CO + 0.08 OH + 0.76 H02 + 0.2 MGLY R73 OH + MGLY -y R74 MGLY + hv _y R75 OII +ISOP .. XO2 + C2O3 C2O3 + HO2 + CO - 1 0.6 H02 + 0.8 ALD2 + 0.55 OLE + X02 + 0.5 CO + 0.45 ETH + 0.9 PAR R76 X02 + FORM + 0.67 H02 + 0.13 XO2N OH + ISOP + ETH + 0.4 MGLY + 0.2 C2O3 + 0.2 ALD2 R77 FORM + 0.4 ALD2 + 0.55 ETH + 0.2 MGLY 03 + ISOP +0.1 PAR +0.06 CO +0.44 HO2 +0.1 OH R78 NO3 + ISOP XO2N R79 XO2 + NO NO2 R80 XO2 + XO2 R81 XO2N + NO R82 XO2 + HO2]I ..... Appendix C : CMAQ Results A.1 AdditionalAviation Results Note that similar sets of plots for all scenarios, in addition to monthly averages, were made and can be accessed on request, but cannot be shown here given their number. Likewise, movies of several species and meteorological parameters were made on hourly time steps for the full 15 months of study, which help to illustrate boundary condition effects, the importance of meteorological data, the extent of transport and residence times, transient initial condition times, plume transport, vertical mixing, and other key effects. 113 I 2006yearlyavg -no aviation)N2 JppbVl (full aviation Min - -7.612e-02@(59.111) Max-1446e+00@(67.130) Mean- -1,5102e-02 (full aviation -no aviation) S02 (ppbV] 2006yearly avg Min - -2.908e-03 @(58,112) Max - 1,780e-01 @ (67.130) Mean - -2.1936e-04 0.0 0 40 0 Uin - 60 to ColumNumber 100 0 0 140 120 40 20 40 Cakmm Nmb+ (full aviation - no aviation) HNO3 [ppbV)2006yearly avg Min - -5.680e-04@(98.126) Max - 4.991e-02 @(41,109) Mean - 1.0647e-02 9 100 zu Column_ Numer 120 _ so G0 CoblmNuiber to0 10 14 (full aviation - no aviation) H2SO4[ppbV 2006yearly avg Min- -1,308-05@(67,108) Max- 7.035e-05@(23,148) Mean-1.9035046 (full aviation - no aviation) NO [ppbV] 2006yearly avg Max - 4.760e-01@ (67.130) Mean - -1.22160-02 6(2A.941 M7W-1 100 20 140 120 Uin. 0 140 _ 114 G0 my Numlber Column (full aviation - no aviation) OH [ppbV] 2006yearly avg Mean - 6.0023e-07 Max- 4.18e-06 @(18.132) -1 AirA.M006221 20 40 s0 0 I (fullaviation-no aviation)H202[ppbV]2006yearlyavg Mean- 8.5872e-03 Min- 6.761e-04@(41.109) Max- 2.464e-02@(16,121) 03 [ppbv]2006yearlyavg (fullaviation- no aviation) Mean-10576e+00 Min- 000@(1,1) Max- 1.716e+00@(57.25) 0 20 40 59 0 P 20 40 sx so 100 120 140 Column Nume& 2006yearlyavg (full aviation- noaviation)VOC[ppbV] @ (20.45) Max- 1.861e+00@ (67.130) Mean- -8.8552e-02 Min- -1.989e-01 (tull aviation- noaviation)CO[ppbV)2006yearlyavg Min- -2.188e+00 @(53,52) Maxu-5501e+00@(67,130) Mean- -18567e+00 20 40 s0 x0 -2 100 0 120 20 40 s0 so 100 120 140 CoumNumere Figure 51: Full aviation-attributable changes in precursor and oxidizing gases 2006yearlyaverage / fullaviation)S04 1%contribution) ((fullaviation-no aviation) @0(1369) Max- 1.970e+000@(46,22)Mean-5.0803e-01 Min- -2.034e-01 2006yearlyaverage ((fullaviation- noaviation)/ fullaviation)FRMPM251%contribution - 1239e+000(4524) Mean- 41363e-01 Uin - 1 qO7-Al1 i I Ro Mulax 0i4 02Z 0 02 120 20 140 40 60 so ColumNumbe 115 100 120 140 {(fullaviation- noaviation) / fullaviation)N03 [%contribution 2006yearlyaverage Min- -4.670e+00 @(30,37) Max- 1.044e+01 @(17,36) Mean- 2.0845e+00 0 20 40 en so 100 120 {(fullaviation- noaviation)/ fullaviation)VOC[%contribution| 2006yearlyaverage Min- -1.613e+00@(22,6) Max- 1.256e+00 @(4622) Mean- -4.8183e-01 140 20 Column Number 40 en n0 1-lu -- nb ((fullaviation-no aviation)/full aviation)BC[%contribution) 2006yearly average Min- -6.360e42 @(13,68) Max- 1.131e+00 @(18,132) Mean- 9.4334e-02 {(fullaviation- noaviation)/ fullaviation)NH4[%contribution) 2006yearlyaverage Min- -2.802e-01 @(13,69) Max- 3.309e+00 0(60,15) Mean- 8.5163e-01 0 go 20 40 60 80 100 120 140 umnb erN Columum ((full aviation- noaviation)/ full aviation)03 [%contribution] 2006yearlyaverage Min- 000@ (11) Max- 4.220e+00 @ (63,42) Mean- 2.7368e+00 {(fullaviation- no aviation) / full aviation)OH[%contribution) 2006yearlyaverage Min- -6.476e+00 @(46,22) Max- 7.093e+000 (18,132) Mean- 1.1040e+00 0 0 20 40 ColumnNumber Figure 52: Aviation's relative contribution to ambient concentrations of several species 116 A.2 Computational Time/Complexity The resolution, time step, number of reactions treated (and degree of explicit treatment), the numerical method used and the manner in which it is implemented all affect computational complexity. For this work, each CMAQ scenario was run on an 8-core E5530 using 1066 DDR3 RAM and Ethernet-attached PCI-e 16-bay RAID external storage formatted to XFS. Parallelized CMAQ is implemented with MPI (Message Passing Interface). With 148 x 112 x 35 = 580,160 grid cells, and using the CB-IV mechanism, each day of CCTM output takes about 25 minutes (20-30 minutes) to compute (after several optimizations of the storage arrays for better concurrent use I/O performance). In comparison, some similarly resolved CB-V runs appear to take about twice as long. 117 Bibliography Abt Associates, Inc. (2009). ModeledAttainment Test Software User's Manual Research Triangle Park, NC: Office of Air Quality Planning and Standards, US EPA. Andersen, J., Hardy, E., Roach, J., & Witmer, R. (1976). A Land Use and Land Cover Classification Systemfor Use with Remote Sensor Data.Washington: US Government Printing Office. (2007b). ASTM StamdardSpecificationforDieselFuel Oils. American Society for Testing Materials. (2007a). ASTM StandardSpecificationforAviation Turbine Fuels. American Society for Testing Materials. Baek, B., Arunachalam, S., Holland, A., Adelman, Z., & Hanna, A. Development of an Interfacefor the Emissions and DispersionModeling System (EDMS) with the SMOKE Modeling System. Chapel Hill, NC: Institute for the Environment, UNC. Barrett, S., Britter, R., & Waitz, I. (2009). Global Mortality Attributable to Aircraft Cruise Emissions. Submitted to EnvironmentalScience & Technology . BP. (2006, June). Retrieved from Clearing the Air About Ultra-Low Sulfur Diesel: http://www.bp.com/sectiongenericarticle.do?categoryID=9009862&contentID=7018720 Byun, D., Young, J., Pleim, J., Odman, T., & Alapaty, K. Chapter 7: NumercalTransportAlgorthmsfor the Community Multiscale AirQuality (CMAQ) Chemical TransportModel in Generalized Coordinates.EPA. Center for International Earth Science Information Network (CIESIN). (2005). GriddedPopulation of the World and the GlobalRural-UrbanMapping Project.Palisades, NY: CIESIN, Columbia University. Chapter3 - Emission Sources. (2006, July 19). Retrieved Mar 11, 2010, from Department for Transport - Project for the Sustainable Development of Heathrow: http://www.dft.gov.uk/pgr/aviation/environmentalissues/heathrowsustain/chapter3emissionsourc es Chapter8: MM5. (n.d.). From http://www.mmm.ucar.edu/mm5/documents/tutorial-v3-notespdf/mm5.pdf Chevron. (2004). From Aviation Fuels: A Technical Review: http://www.chevronglobalaviation.com/ga/gacommercialaviation Chevron. (2007). From Diesel Fuels Technical Review: http://www.chevron.com/products/PRODSERV/fuels/diesel/documents/DieselFuelTechRe view.pdf (2009). CMAQ 4.6 OperationalGuidance Document. Chapel Hill, NC: University of North Carolina Chapel Hill. Colella, P., & Woodward, P. (1984). The Piecewise Parabolic Method (PPM) for gas-dynamical simulations. Journalof CompuationalPhysics, 54 , 174-201. Crutzen, P. (2006). Albedo Enhancement by Stratospheric Sulfur Injections: a Contribution to Resolve a Policy Dilemma? Climatic Change , Volume 77, Numbers 3-4. Deichmann, U., Balk, D., & Yetman, G. (2001). Transforming PopulationDataforInterdisczglinay Usages: From Census to Grid.Palisdades, NY: CIESEN, Columbia University. DESC-BP. (1999-2007). From Defense Energy Support Center, Petroleum Quality Information System Reports: http://www.desc.dla.mil/DCM/DCMPage.asp?pageid=99 DieselFuelExplained. (2009). From U.S. Energy Information Administration, Independent Statistics and Analysis: http://tonto.eia.doe.gov/energyexplained/index.cfm?page=dieseLprices DieselNet. (n.d.). From Fuel Regulations Japan: Diesel Fuel: http://www.dieselnet.com/standards/jp/fuel.php Dockins, C., Maguire, K., Simon, N., & Sullivan, M. (2004). Value ofStatisticalLife Analysis and EnvironmentalPolig:A White Paper.US EPA: National Center for Environmental Economics. DOT/FAA. (2005a). SAGE Version 1.5 TechnicalManual DOT-VNTSC-FAA-05-14. 118 DOT/FAA. (2005b). SAGE Version 1.5 Validation Assessment, ModelAssumptions and Uncertainties. FAA-EE-2005-03. Dudhia, J. (1989). Numerical Study of Convection Observed during the Winter Monsoon Experiment Using a Mesoscale Two-Dimensional Model. Journalof the Atmospheric S*ences, Vol. 46, No. 20, 3077-3177. Edwards, T., Minus, D., Harrison, W., E., C., DeWitt, M., S., Z., et al. (July 2004). Fischer-Tropsch Jet Fuels: Characterization for Advanced Aerospace Applications. Proceedings ofthe 40th AIA A/ ASM E/SAD/ASEE Joint PropulsionConference andExhibit. Ft. Lauderdale, FL: American Institute of Aeronautics and Astronautics AIAA-2004-3885. EIA. (2001, May). Retrieved from U.S. Energy Information Administration, Independent Statistics and Analysis; The Transition to Ultra-Low-Sulfur Diesel Fuel: Effects on Prices and Supply (SR/OIAF/2001-01): http://www.eia.doe.gov/oiaf/servicerpt/ulsd/uls.html (2000). EPA. Guidelines for Preparing Economic Analyses, Office of the Administrator. EPA-240R-00-003. EPA. (2009). About the NationalEmission Inventory Database. From EPA: http://www.epa.gov/oar/data/neidb.html Europa. (2007). From Europa, Activities of the European Union, Summaries of EU Legislation, Quality of petrol and diesel fuels: sulfur and lead. Exxon. (Undated). From Diesel FAQ: http://www.exxon.com/USAEnglish/GFM/ProductsServices/Fuels/Diesel FuelsFAQ.asp FAA OEE & EPA OTAQ. (2009). Recommended Best PracticeforQuantifying SpecieatedOrganic Gas Emissions FromAircraft Equipped with Turbofan, Turbojet, and Turboprop Engines. Frank, N. (2006). Retained Nitrate, Hydrated Sulfates, and Carbonaceous Mass in Federal Reference Method Fine Particulate Matter for Six Eastern U.S. Cities. J.Air & Waste Manage. Assoc., 56, 500511. FuelConsumption by Mode of Transportation.(2007). From Bureaau of Transportation Statistics: http://www.bts.gov/publications/national transportation statistics/html/table_04_05.html GEOS-Chem v8-02-01 Online User's Guide. (n.d.). From http://acmg.seas.harvard.edu/geos/doc/man/ Gery, M., Whitten, G., Killus, J., & Dodge, M. (1989). A Photochemical Kinetics Mechanism for Urban and Regional Scale Computer Modeling. Journalof GeophysicalResearch, 12925-12956. Grell, G. A., & Devenyi, D. (2002). A Generalized Approach to Parameterizing Convection Combining Ensemble and Data Assimilation Techniques. GeophysicalResearch Ietters, Vol. 29, No. 14, 1693. Grell, G., Dudhia, J., & Stauffer, D. (1995). A Descrption of the Fifth-GenerationPenn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR. Herndon, S., Shorter, J., Zahniser, M., Nelson Jr., M., Layne, J., Brown, R., et al. (2004). NO and N02 Emission Ratios Measured from In-Use Commercial Aircraft during Taxi and Takeoff. Environ. Sci. Technol., 38, 6078-6084. Hileman, J., Ortiz, D., Bartis, J., Wong, H. M., Donohoo, P., Weiss, M., et al. (2009). Near-Term Feasibility ofAlternative Jet Fuels. (TR-554-FAA). Hog & Pig County Estimates. (2008). From North Carolina Dept. of Agriculture: http://www.ncagr.gov/stats/coest/HogsCoest.pdf Hong, S.-Y., & Pan, H.-L. (1996). Nonlocal Boundary Layer Vertical Diffusion in a Medium-Range Forecast Model. Monthly Weather Review, Volume 124 , 2322-2339. Hsu, Y., & Divita Jr., F. (2009). SPECIATE 4.2 Speciation DatabaseDevelopment Documentation. Research Triangle Park, NC: US EPA Office of Research and Development. ICAO. (2007). EnvironmentalReport 2007. International Civil Aviation Organization. 119 IPCC. (1999). Aviation and the GlobalAtmosphere: A Special Report of IPCC Working Groups I and II. Costa Rica: Intergovernmental Panel on Climate Change. (2001). IPCC: Climate Change 2001. New York: Cambridge University Press. (2006). ISORROPIA v1.7 Reference Manual. Isukapalli, S. (1999). UncertaintyAnalysis of Transport-TransformationModels. New Brunswick, NJ: Rutgers University. Kalkstein, A., & Balling Jr., R. (2004). Impact of unusually clear weather on United States daily temperature range following 9/11/2001. Climate Research, Vol. 26:1-4. Kuhn, S. (2010). Cost-BenefitAnaysis of Ultra-Low Sulfur Jet Fuels. Cambridge, MA: Massachusetts Institute of Technology, S.M. Thesis. Lee, D., Fahey, D., Forster, P., Newton, P., Wit, R., Lim, L., et al. (2009). Aviation and Global Climate Change in the 21st Century. Atmospheric Environment, 1-18. Lefebvre, A. H. (1984). Fuel Effects on Gas Turbine Combustion-Liner Temperature, Pattern Factor, Pollutant Emissions. JournalAircraft,Vol. 21, No. 11 , 887-898. Liao, C.-M., & Chio, C.-P. (2008). Assessment of atmospheric ultrafine carbon particle-induced human health reisk based on surface area dosimetry. Atmos. Environ., 42 (37) , 8575-8584. Luecken, D., Phillips, S., Sarwar, G., &Jang, C. (2008). Effects of using the CB05 vs. SAPRC99 vs. CB4 Chemical Mechanism on Model Predictions: Ozone and Gas-Phase Photochemical Precursor Concentrations. Atmospheric Environment42, 5805-5820. Lung Cancer. (2006). From Center for Disease Control: http://www.cdc.gov/cancer/lung/ Martel, C., & Angello, L. (1973). Hydrogen Content as a Measure of the Combustion Performance of HydrocarbonFuels, Tech. Rep. No. AFAPL-TR-72-103. Wright Paterson AFB, OH: Air Force APL. Masek, T. (2008). A Response Surface Model of the AirQualiy Impacts ofAviation. S.M., Massachusetts Institute of Technology. Miake-Lye, R., Yam, C., Han, S., Lukachko, S., Waitz, I., Wormhoudt, J., et al. (2002). The Evolution of HONO Emissions in an Aircrafte Engine Turbine. EGS XXVII GeneralAssemby. Nice. Mitchell, K. (2000). The Community NOAH Land-Surface Model (LSM). NOAA. Mlawer, E., Taubman, S., Brown, P., lacono, M., & Clough, S. (1997). Radiative Transfer for Inhomogeneous Atmospheres: RRTM, a Validated Correlated-k Model for the Longwave. Journalof GeophysicalResearch, Vol. 102, No. D14, 16,663-16,682. Mokdad, A., Marks, J., Stroup, D., & Gerberding, J. (2004). Actual Causes of Death in the United States, 2000. Journalof the American MedicalAssociation , 291:1238-1245. Moon, N., Park, R., Li, X., & Byun, D. (2008). A Simple User's Guideforgeos2cmaq: Linking CMAQ with GEOS-Chem. Houston, TX: Institute for Multidemensional Air Quality Studies (IMAQS), University of Houston. Morrell, P., & Dray, L. (2009). EnvironmentalAspects of Fleet Turnover, Retirement, and Life Cycle Final Report. Morris, E., & Myers, T. (1990). User's Guidefor the UrbanAirshed Model Volume 1: User's Guidefor the UAM (CB-IV). U.S. EPA (EPA-450/4-90-007A). Moses, C., & Karpovich, P. (1988). Fuel Effects on Flame Radiation and Hot-Section Durability. Advisory GroupforAerospace Research and Development (AGARD) Conference ProceedingsNo. 422, AGARDCP-422, (pp. 15.1-15.15). Moses, C., & Wilson III, C. (Dec. 2003). Evaluation of Sasol Synthetic Kerosenefor Suitability as Jet Fuel. Naiman, A., Lele, S., Wilkerson, J., & Jacobson, M. (2010). Parameterization of subgrid aircraft emission plumes for use in large-scale atmospheric simulations. Atmos. Chem. Phys Disc., 9, 2475524781. 120 NationalEmissions Inventory Air Pollutant Trends Data. (2009). From US EPA Technology Transfer Network Clearinghouse for Inventories & Emissions Factors: http://www.epa.gov/ttnchiel/trends/ Nenes, A., Christodoulos, P., & Pandis, S. (1998). ISORROPIA: A New Thermodynamic Equilibrium Model for Multiphase Multicomponent Inorganic Aerosols. Aquat. Geoch. Odum, J., Hoffmann, T., Bowman, F., Collins, D., Flagan, R., & Seinfeld, J. (1996). Gas/Particle Partitioning in Secondary Organic Aerosol Yields. Environ. Sci. Technolog., 30 , 2580-2585. Ostro, B. (2004). OutdoorAirPollution:Assessing the EnvironmentalBurden of Desease at Nationaland Local Levels. World Health Organization, Environmental Burden of Disease Series, No. 5. PARTNER Project 27 Internal Interim Report. (2010). Cost-BenefitAnalysis of Ultra-Low Sulfur Jet Fuels. Petzold, A., Dopelheuer, A., Brock, C., & Schroder, R. (1999). In Situ Observations and Model Calculations of Black Carbon Emission by Aircraft at Cruise Altitude. J. Geophys. Res., 104 , 2217122181. Pickering, K., Huntrieser, H., & Schumann, U. (2008). Laghtning: Pinciples,Instruments andApplications, Chapter26: Lightning and NOx Productionin GlobalModels. Springer. Pleim, J. (2007). A Combined Local and Nonlocal Closure Model for the Atmospheric Boundary Layer. Part I: Model Description and Testing. JournalofApplied Meteorology and Climatology, 46, 13831395. Pope, C. A., & Dockery, D. (2006). Health Effects of Fine Particulate Air Pollution: Lines That Connect. J.Air & Waste Manage. Assoc., 56, 709-742. Queena, A., Zhanga, Y., Gilliam, R., & Pleim, J. (2007). Examining the sensitivity of MM5-CMAQ predictions to explicit microphysics schemes and horizontal grid resolutions, Part I-Database, evaluation protocol, and precipitation predictions. Atmosphenc Environment, Volume 42, Issue 16, 38423855. Randall, D., Wood, R., Bony, S., Colman, R., Fichefet, T., Fyfe, J., et al. 2007: Climate Models and Their Evaluation. Climate Change 2007: The PhysicalScience Basis. Contributionof Working Group I to the FourthAssessment Report of the IntergovermentalPanelon Climate Change. Cambridge, UK and New York, NY: Cambridge University Press. Ratliff, G., Sequeira, C., Waitz, I., Ohsfeldt, M., Thrasher, T., Graham, M., et al. (2009). Aircraft Impacts on Local and RegionalAirQualityin the United States. PARTNER-COE-2009-002. Rickard, G. Gune 2008). The Quality of Aviation FuelAvailable in the United Kingdom, Annual Survey 2007. QinetiQ FLC Report QINETIQ/08/01656. Rojo, J. (2007). Future Trends in LocalAirQualityImpacts ofAviation. S.M., Massachusetts Institute of Technology. Ryan, R. (2002). Memorandum: Speciation Profiles andAssignment FilesLocated on EMCH. US EPA, Emissions Factor and Inventory Group. Schumann, U., & Huntrieser, H. (2007). The Global Lightning-Induced Nitrogen Oxides Source. Atmos. Chem. Phys., 7, 3823-3907. Seinfeld, J., & Pandis, S. (2006). Atmospheric Chemisty and Physics: FromAir Pollution to Climate Change. Hoboken, NJ: John Wiley & Sons, Inc. Sequiera, C. (2008). RelationshipsBetween Emissions-RelatedAviationRegulations and Human Health. Cambridge, MA: Massachusetts Institute of Technology, S.M. Thesis. Shirmer, R., & Quigg, H. (1965). High Pressure CombustorStudies of Flame Radiation as Related to Hydrocarbon Structure R&D Report No. 5450-69. Phillips Petroleum Company. Spot Prices. (2010). From U.S. Energy Information Administration, Independent Statistics and Analysis: http://tonto.eia.doe.gov/dnav/pet/petLpri-spt s1_m.htm 121 Stockwell, W., Middleton, P., & Chang, J. (1990). The Second Generation Regional Acid Deposition Model Chemical Mechanism for Regional Air Quality. Journalof GeophysicalResearch, Volume 95, 16343-16367. Taylor, W. (Feb. 2009). Survey ofSulfur Levels in CommercialJetFuel.Alpharetta, GA: Coordinating Research Council Report AV-1-04. Travis, D., Carleton, A., & Lauritsen, R. (2004). Regional Variations in US Diurnal Temperature Range for the 11-14 September 2001 Aircraft Groundings: Evidence of Jet Contrail Influence on Climate. J. Clim. 17, 1123-1134. US EPA. (2000a, December). Retrieved from US EPA; Regulatory Impact Analysis: Heavy-duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements (EPA420-R00-026): http://www.epa.gov/otaq/highway-diesel/regs/2007-heavy-duty-highway.htm US EPA. (2000b, December). Retrieved from US EPA; Heavy-duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements: Response to Comments (EPA420-R-00027): http://www.epa.gov/otaq/highway-diesel/regs/r00027.zip US EPA. (2006, October). Retrieved from US EPA, Office of Transportation and Air Quality; Introduction of Cleaner-Burning Diesel Fuel Enables Advanced Pollution Control for Cars Trucks and Buses (EPA420-F-06-064): http://www.epa.gov/otaq/highway-diesel/regs/420f06064.htm US EPA. (2004). Air Qualiy CriteriaforParticulateMatter.Research Triangle Park, NC: US EPA, National Center for Environmental Assessment. US EPA. (2006). E xpanded ExpertJudgmentAssessment of the Concentration-ResponseRelationship Between PM2.5 Exposure andMortaliy. Research Triangle Park, NC: US EPA Office of Air Quality and Planning Standards. Wayson, R., Fleming, G., & lovinelli, R. (2009). Methodology to Estimate Particulate Matter Emissions from Certified Commercial Aircraft Emissions. J.Air and Waste Manag.Assoc. 59, 91-100. Wesely, M. (1989). Parameterization of Surface Resistances to Gaseous Dry Deposition in RegionalScale Numerical Models. Atmospheric Environment, Vol. 23, 1293-1304. Where Lightning Strikes. (n.d.). From Science@NASA: http://science.nasa.gov/headlines/y2001/ast05dec_1.htm Wood, E. C., Herndon, S. C., Timko, M. T., Yelvington, P. E., & Miake-Lye, R. C. (2008). Speciation and Chemical Evolution of Nitrogen Oxides in Aircraft Exhaust near Airports. Environ. Sci. Technol. , 1884-1891. Wormhoudt, J., Herndon, S., Paul, Y., Miake-Lye, R., & Wey, C. (2007). Nitrogen Oxide (NO/NO2/HONO) Emissions Measurement in Aircraft Exhausts. JournalofPropulsion andPower , Vol. 23, No. 5. Yarwood, G., Rao, S., Yocke, M., & Whtiten, G. (2008). FinalReport:Updates to the Carbon Bond ChemicalMechanism: CB05. RT-04-00675. 122