2010 Aviation Emission Inventory Development and Analysis

lable at ScienceDirect

ARTICLE IN PRESS

Environmental Modelling & Software xxx (2010) 1e16

Contents lists avai

Environmental Modelling & Software

journal homepage: www.elsevier .com/locate/envsoft

Aviation emission inventory development and analysis

Viet Van Pham*, Jiangjun Tang, Sameer Alam, Chris Lokan, Hussein A. AbbassDefence and Security Applications Research Center, University of New South Wales, the Australian Defence Force Academy, Canberra, Australia

a r t i c l e i n f o

Article history:Received 5 August 2009Received in revised form31 March 2010Accepted 6 April 2010Available online xxx

Keywords:Aviation emissionAir trafficEnvironmental modellingEmission inventory

* Corresponding author.E-mail address: [email protected] (V.V.

1364-8152/$ e see front matter Crown Copyright � 2doi:10.1016/j.envsoft.2010.04.004

Please cite this article in press as: Pham, Vdoi:10.1016/j.envsoft.2010.04.004

a b s t r a c t

An up to date and accurate aviation emission inventory is a prerequisite for any detailed analysis ofaviation emission impact on greenhouse gases and local air quality around airports. In this paper wepresent an aviation emission inventory using real time air traffic trajectory data. The reported inventoryis in the form of a 4D database which provides resolution of 1� � 1� � 1000 ft for temporal and spatialemission analysis. The inventory is for an ongoing period of six months starting from October 2008 forAustralian Airspace.

In this study we show 6 months of data, with 492,936 flights (inbound, outbound and over flying).These flights used about 2515.83 kt of fuel and emitted 114.59 kt of HC, 200.95 kt of CO, 45.92 kt of NOx,7929.89 kt of CO2, and 2.11 kt of SOx. From the spatial analysis of emissions data, we found that the CO2

concentration in some parts of Australia is much higher than other parts, especially in some major cities.The emission results also show that NOx emission of aviation may have a significant impact on the ozonelayer in the upper troposphere, but not in the stratosphere.

It is expected that with the availability of this real time aviation emission database, environmentalanalysts and aviation experts will have an indispensable source of information for making timely deci-sions regarding expansion of runways, building new airports, applying route charges based on envi-ronmentally congested airways, and restructuring air traffic flow to achieve sustainable air traffic growth.

Crown Copyright � 2009 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Aviation has grown dramatically worldwide (Michot et al., 2003).Within Australia, air traffic has grown by more than 50% over thelast decade. Australia now has approximately 1 million domesticand international flights every year, with up to 4000 flights on thebusiest days.Worldwide air traffic is expected to continue to grow atrates of 3e5% per year (Penner et al., 1999). In Australia, it is fore-casted to grow by four percent per annum on average, doubling by2025. This forecast is based on growth in traveling passengers andthe increase in demand for air freight (Shepherd et al., 2007).

These levels of growth will mean a considerable increase inaviation traffic and emissions over the next 10 years. Whilst aviationbrings considerable economic benefits, this growth is also associ-ated with increased environmental pressures, and consequentlya growing need to improve the environmental performance of theindustry. Aviation has considerable environmental impacts both ata local airport level and at a regional and global level. Local atmo-spheric issues are related to airport contributions to local air qualityand the potential for health impacts on residential populations in

Pham).

009 Published by Elsevier Ltd. All

.V., et al., Aviation emission

surrounding areas. At a global level, aviation emission has thepotential of atmospheric impacts to affect climate. Aviation emis-sions are increasingly recognized as a significant contributor toglobal climate impacts through global warming (Penner et al., 1999).Models of global warming associated with aviation emissionssuggest that aviation contributes approximately 3.5% of the totalanthropogenic forcing, and this may increase to between 7 and 12%by 2050 (Penner et al., 1999).

Understanding the impact of aviation emissions is importantdue to the characteristics of aviation emissions occurring in highaltitude, and the growth of air traffic. However, the current state ofknowledge of aviation impacts on the atmosphere is severelylimited due to unavailability of timely and accurate aviation emis-sion data. New approaches and models are required in order toreduce the considerable uncertainties in emission calculations.Aviation emission inventories are a critical resource because of thisrequirement.

Aviation emission modelling is an important component in thedomain of environmental modelling. There is a wide interest inemission inventories worldwide (Samaali et al., 2007; Winiwarterand Schimak, 2005; Keogh et al., 2009; Monforti and Pederzoli,2005; Borge et al., 2008). Also, various aviation emission invento-ries have been developed worldwide by the U.S. National Aero-nautics and Space Administration (NASA) (Sutkus et al., 2001), the

rights reserved.

inventory development and analysis, Environ. Model. Softw. (2010),

mailto:[email protected]

www.sciencedirect.com/science/journal/13648152

http://www.elsevier.com/locate/envsoft

V.V. Pham et al. / Environmental Modelling & Software xxx (2010) 1e162

ARTICLE IN PRESS

U.S. Federal Aviation Administration (FAA) (FAA, 2005a) and theEuropean Organisation for the Safety of Air Navigation (EURO-CONTROL) (Eyers et al., 2004). Most of these inventories use historicair traffic data, not real time air traffic data. They use severalassumptions to simplify the complex flight trajectories, and dealwith few representative aircraft types. Moreover the emissioninventories are not comprehensive i.e. they lack details regardingemission in different flight phases (such as taxi-out and taxi-in).Therefore the results obtained are not very accurate, and detailed,and by the time they are compiled they become outdated.

In this paper we present an aviation emission inventory meth-odology that is generic in nature. It is applied here to the Australianairspace, due to data availability. The methodology uses real timeair traffic trajectory data and real time computation of aviationemissions. We have developed software called TOP-LAT (TrajectoryOptimization and Prediction of Live Air Traffic), that constructs thetrajectory of a flight based on real time flight plans, routes and radarsignatures of the flight. The fuel flow and emissions of the flight arecalculated every time step using the speed, altitude, and phase ofthe flight, along with aircraft-specific aerodynamic models. Theseresults are combined with aircraft position and time to generatea 4D (latitude, longitude, altitude and time) emission database.With these characteristics, the model provides more accurate fuelflow and emissions computation than the other models.

We also present a generic temporal and spatial analysis meth-odology of the impact of air traffic on the environment, with a casestudy on the Australian airspace. Furthermore, spatial distributionand analysis of aviation emissions including HC, CO, NOx, CO2 andSOx are presented.

2. Environmental impact of aircraft emissions

Typically, aviation emission consists of 71% Carbon dioxide (CO2)and 28% water vapour (H2O). In the remaining 1%, NOx is the mostimportant emission (Penner et al., 1999, p. 28). Four main factors(Janic, 1999) influencing aviation emissions include

� the intensity and volume of aircraft movements� the type and spatial concentration and distribution of theparticular pollutants

� fuel consumption and energy efficiency� and the rate of renewing of the aircraft fleet by introducing“cleaner” aircraft.

The effects of aviation emission are classified in the literature aslong-range air pollution, effect on the ozone layer, and greenhouseeffects (Janic, 1999; Brasseur et al., 1998). They are presented inmore detail in the following subsections.

2.1. Long-range air pollution

Long-range air pollution refers to the effects of air pollution faraway from the polluting source. One such effect is “acid rain”caused partially by SOx and mainly by NOx produced by air traffic(Brasseur et al., 1998). Acid rain has harmful effects on plants,aquatic animals, and infrastructure. Acid rain is mostly caused byhuman emissions of sulphur and nitrogen compounds which reactin the atmosphere to produce acids.

2.2. Effect on the ozone layer

When aircraft-related pollution appeared, complex chemicalreactions were initiated. The NOx from aviation emission in thetroposphere (where subsonic flights take place) has enhanced theozone concentration (Janic, 1999). Tropospheric (ground level)

Please cite this article in press as: Pham, V.V., et al., Aviation emissiondoi:10.1016/j.envsoft.2010.04.004

ozone is a damaging oxidant. It is also considered as one of thegreenhouse gases in the troposphere. The concentration of NOx inthe upper troposphere has been large, and during winter monthsaircraft have been identified as dominant sources of the NOx atlocations in the Northern Hemisphere. At higher altitudes in thestratosphere (where supersonic flights have taken place), anincrease in concentration of NOx has depleted the ozone layer(Penner et al., 1999) which absorbs much of the negative ultravioletsunlight waves that would otherwise reach the surface of the earthand affect the biosphere. The crossover surface between productionand destruction of the ozone layer lies roughly between the lowerstratosphere and high troposphere d the tropopause, where long-range aircraft cruise (Wesoky et al., 1996). Aviation-inducednitrogen oxides (NOx) make up 2% of all human generated NOx

emissions (Penner et al., 1999). NOx has 10 times higher residualtime in the tropopause than at ground level.

2.3. Greenhouse effects

CO2, H2O emitted by aircraft absorb thermal infrared radiation,emitted by the Earth’s surface and by the atmosphere. This causesglobal warming of the earth (IPCC, 2007). The aviation sectorcontributes 1.6e2.2% to all human generated CO2 emissions(Schumann, 2002). H2O emissions from aviation are smallcompared to the water evaporating at the Earth’s surface. H2Oresulting from aviation emission is mainly emitted into the tropo-sphere, where it is removed by precipitation. Besides that, contrailsalso tend to warm the Earth’s surface, similar to thin high clouds(Penner et al., 1999). The condensation trails are triggered from thewater vapour emitted by aircraft and their optical propertiesdepend on the particles emitted or formed in the aircraft plume andon the ambient atmospheric conditions. Aircraft line-shapedcontrails are estimated to cover about 0.1% of the Earth’s surface onan annually averaged basis with larger regional values. The contrailcover is projected to grow to 0.5% by 2050.

3. Aviation emission inventories

There are two main approaches in the literature for developingaviation emission inventory. One is on a global basis for deter-mining the impact of aviation emission on climate change, and theother is on a local basis, which focuses on the level of air qualitysurrounding an airport. Emissions of carbon dioxide (CO2) andwater vapour (H2O) are of particular interest in global warming,whereas investigations of hydrocarbons (HC) and carbonmonoxide(CO) are done for their effects on air quality.

Emission inventories for global aviation are available fromdifferent sources, most prominently the U.S. National Aeronauticsand Space Administration (NASA), the European Community (EC)Working Group, and FAA.

The NASA study, commonly known as “Scheduled Civil AircraftEmission Inventories for 1999” (Baughcum et al., 1996; Sutkus et al.,2001), was done by the Boeing Group. Three-dimensional data ofaircraft fuel burn and emissions of NOx, CO and HC were deter-mined for the year 1999.

The AERO2K emission inventories (Michot et al., 2003) weredeveloped by the European Community in a common project byQinetiQ, EUROCONTROL, Manchester Metropolitan University, theNational Aerospace Laboratory (NLR) of the Netherlands and theGerman Aerospace Centre (DLR). Emissions of CO2, H2O, NOx, COand HC were calculated along with the fuel-use and distancetravelled.

The System for Assessing Aviation’s Global Emissions (SAGE)(FAA, 2005b) has been developed by FAA. Currently SAGE is


V.V. Pham et al. / Environmental Modelling & Software xxx (2010) 1e16 3

ARTICLE IN PRESS

a research tool and not available in the public domain. However,preliminary results of the model are available.

All the models are used to estimate the amount and the distri-bution of different emissions from aviation. All of them aredesigned to calculate aircraft emissions over theworld’s airspace bylatitude, longitude, and altitude. Historical inventories of aviationemissions and future trend forecasts have been produced by thesemodels.

There is a common approach in all of the aforementioned 3Demissions models. All of them operate in off-line mode. They allinvolve combining a database of global air traffic (fleet mix, city-pairs served, and flight frequencies) with a set of assumptionsabout flight operations (flight profiles and routing) and amethod tocalculate altitude-dependent emissions of aircraft/engine combi-nations. Idealized flight routings and profiles, with no winds orsystem delays are assumed. Thus there can be a vast discrepancybetween actual fuel burn and emissions happened, and estimationof fuel burn and aviation emission using idealized scenarios.

The trajectory of a flight in these models is also not very accu-rate. It is modeled via a number of waypoints. For example, SAGEmodels the trajectory of a flight by 30e40 chords, AERO2K uses20e30 waypoints to represent the trajectory. This affects theaccuracy of fuel and emission computation.

In the emission inventory developed by NASA, aircraft emissionsof CO2, H2O and SO2 were determined directly from fuelconsumption. Since great circle routes were assumed between city-pairs, the results are more suited for trend analyses. A majordrawback, however, is the incomplete coverage of global aviation,since only scheduled air traffic was accounted for. Furthermore, theoutput data does not contain any four-dimensional results (emis-sion data with time) limiting any seasonal and temporal trendanalysis.

In AERO2K, the generation of a movement database from AirTraffic Control (ATC) data is highly time-consuming and cumber-some, due to legacy systems and different data formats, and canonly be done with historic air traffic data.

Table 1 provides a comparative analysis of the above mentionedstate of the art aviation emission inventories, together with ourmethodology which is used to develop the emission inventory. Thistable is developed based on Schaefer (2006) and compares ourmethodology with others in the literature.

The main advantage in our study is that the system is online andprocesses air traffic data in real time, capturing not only effects ofwind & flight delays (holdings) but also complex vectoringmaneuvers during approach by Air Traffic Controllers (ATCs). The4D emission database is updated in real time, giving a live picture ofemission both for local air quality as well as greenhouse impact. Thesystem uses actual 4D flight trajectories, thus it provides a moreaccurate estimate of fuel burn and emissions.

4. Australian emission inventory development

The emission inventory is in the form of a relational databasewhich records emissions data in four dimensions. Latitude, longi-tude and altitude divide Australian Airspace into a 3D grid; thefourth dimension is time. The dimensions of the grid are theNational Airspace’s start latitude, start longitude, end latitude, andend longitude, with altitude ranging from ground level to 61,000 ft.The size of one cell in the grid is 1� (longitude)� 1� (latitude)�1000 ft (altitude). This can be changed to increase or decrease theresolution of the grid. Fuel and emissions are calculated for all ofthe grid cells as and when the flights transit through them.

Calculation of fuel and emissions is based on accurate flighttrajectories, constructed from real time data on latitude, longitudeand altitude at a number of intervals throughout each flight. This


allows appropriate fuel and emission data for the actual flight routeand altitude to be estimated. Computing fuel and emissions fromreal time radar data is a significant improvement over previousmodels. Previous models usually assume a great circle routebetween origin and destination airports and also a typical cruisealtitude. There can be a large difference between the actual routeflown and the great circle route, due to Air Traffic Control(congestions, flight corridor), restriction on airspace (SUAs),weather (turbulence, storm), or avoiding/taking advantage of pre-vailing winds.

There are two estimates available for the aviation fuel andemissions in Australia. U.S FAA’s System for assessing Aviation’sGlobal Emissions (SAGE) estimated the global aviation fuel andemissions from 2000 to 2004 and National Greenhouse GasInventory Committee (NGGIC), Department of Climate Change,Australia from 1990 to 2007. However, SAGE is a global aviationemission inventory where the focus is on aggregated numbers forfuel burn and emissions using flight plan data alone for AustralianAirspace. NGGIC’s estimation is undertaken using a Tier 2 meth-odology (IPCC, 2001) and was based on aviation fuel sold ratherthan flight trajectories or even flight plans (NGGIC, 2007). There-fore, these estimates provide only an aggregate number for fuelburn and aviation emission in Australia. The aviation emissioninventory developed in this paper is not only a step ahead (for ituses actual aircraft trajectory data) but also attempts to quantifydata from spatial & temporal perspectives.

4.1. Proposed methodology

The essence of our approach is that radar, flight plan andaircraft-specific aerodynamic models are used to predict andconstruct accurate flight trajectories. A flight trajectory consists ofa set of chords where the start and end of the chord representsa valid radar position for that flight. This can be as small as 30 s or aslong as 2 h depending upon radar coverage. For every chord the fuelflow and emission is computed by forward integrating the aircraftposition over time using the aerodynamic model for that aircraftand its associated engine. This allows for fine-grained computationof fuel flow and emission at every time step between two validradar signatures.

Fig. 1 illustrates the computation process used for developingthe aviation emission inventory. It consists of four main compo-nents, described in the following sections.

4.2. Real time Data Processing

Real time Data Processing performs the assimilation of flightplan, flight route and radar data; structuring of air traffic and flightplan data; elimination of military, helicopters and recreationalflights; assigning flight type (over flying, inbound, outbound,inland); aircraft mapping; aircraft-engine assignment; runwayallocation; estimating time of departure; and building the initialflight plan and route profile. This component continuously updatesthe initial flight plan information regarding the runway, arrival anddeparture route as data for the respective flight is received.

Air Traffic Control (ATC) data from the EUROCAT radar system isused for generating flight trajectories. Both flight plan and trajec-tory information from the EUROCAT radar data need to be alignedin space & time, get processed for data integrity, data validation,and mining data connection. As a consequence, checks andassessments are required before the data can be used for baselineinventory production. The processing of flight plan and radar dataincludes the substitution of missing data (departure route, arrivalroute, departure runway, arrival runway, cruising altitude, aircraft


Table 1State of the methodologies in aviation emission.

NASA AERO2K SAGE This study

Generalinformation

Years 1976, 1984, 1992, forecast for2015

2002, forecast for 2025 2000e2004, forecasts indevelopment

Real time emission database2008 onwards

Coverage Scheduled aviation, Charteraviation, the former SovietUnion/China unreportedflights, Military aviation

Scheduled and unscheduledaviation, Military aviation

Scheduled and unscheduledaviation

Scheduled and unscheduledaviation

Working mode Off-line Off-line Off-line OnlineProcessed data Historical data Historical data Historical data Real time data

Movementsdata

Sources OAG flight schedules ATC data from ETMS & AirTraffic Flow ManagementModelling Capabilities(AMOC), BACK flight schedules

ATC data from ETMS, OAGflight schedules

EUROCAT (Thales Radar DataProcessing Software) radardata, Flight plans

Data collectionperiod

Each month of 1992, 4 monthsof 1976 and 1989

6 representative weeks of2002 (ATC), Each month of2002 (schedules)

Each month of 2000 - 2004 Real time data

Contents ofdatabase

Scheduled information (nowaypoints/trajectories)

4D trajectories from ATC data,Artificial routing for scheduledflights

4D trajectories from ATC data,Dispersed great circles forscheduled flights

Flight plan, GroundMovementReport, 4D AutomaticDependent Surveillance e

Broadcast (ADS-B) fused Radartrack data

Performance RepresentativeAircraft/Engine(AC/Eng)combinations

27/36/76 for 1976/1984/1992(jets and generic turboprops)

40 (jets and turboprops) 91 (jets and turboprops) 91 (jets and turboprops)mapping to 200 aircraft

Performance data,Performance model

Boeing proprietary data,Boeing Mission AnalysisProgram (BMAP)

PIANO aircraft models, PIANOperformance software

BADA & Integrated NoiseModel (INM) aircraft models,BADA & INM performancemethodologies

BADA & aerodynamicperformance model

Selected missionassumptions

Great-circle routes, 70%passenger load factor,Continuous climb cruise

Real routing, 60.9% load factor(by mass), Cruise with stepclimbs

Real routing for ETMS flights,Take-off weight estimatedfrom INM, Cruise with stepclimbs, Delay modelling

Real routing from flighttrajectory data, actual flightmission plan, airport-runway-aircraft-specific Taxi-in andTaxi-out Time, nominalaircraft weight with fuel burnreduction.

Emissions Emission data,Emission model

ICAO emission indicesþ industry data, Boeing-2 fuelflow method (NOx, CO, HC)

ICAO emission indicesþ industry data, DLR fuel flowmethod (NOx), DLR Omegamethod (CO, HC), DLR Sootmethod (particles)

ICAO emission indicesþ Emissions and Dispersionmodelling System (EDMS)data, Boeing-2 fuel flowmethod (NOx, CO, HC)

ICAO emission indicesþ industry data, Boeing-2 fuelflow method (HC, CO, NOx,CO2, SOx)

Assumptions Atmosphere &wind ISA atmosphere and zero wind ISA Atmosphere and zerowind

ISA Atmosphere and zerowind

Actual atmosphere and actualwind

Results Allocation software Boeing-developed GlobalAtmospheric Emissions Code(GAEC)

AERO2 k data integration tool SAGE fuel burn and emissionmodule

TOP-LAT

Species covered Fuel burn, NOx, CO, HC Fuel burn, CO2, H2O, NOx, CO,HC, sootþ distance per gridcell

Fuel burn, CO2, H2O, NOx, CO,HC, SOxþ distance per grid cell

Fuel burn, HC, CO, NOx, CO2,SOx

Resolution 3D data in a 1� � 1� � 1 kmworld grid

3D data in a 1� � 1� � 500 ftworld grid, 4D data ina 1� � 1� � 500 ft� 6 h grid

3D data in a 1� � 1� � 1 kmworld grid, 4D raw data at anyresolution required

3D data in a 1� � 1� � 1 kftgrid, 4D raw data at anyresolution required


ARTICLE IN PRESS

performance data) from other available sources, and the identifi-cation and removal of inconsistent or redundant information.

The system focuses on civilian air traffic only. Some flights, suchas small recreational flights and helicopters, are not processed bythe system as aircraft/engine data is not available yet. Military flightdata is also not taken into account because they are classified. Theamount of emissions produced by these flights is relatively smallerwhen compared with commercial flights.

4.3. Trajectory prediction and reconstruction

This is mainly performed by a system called Trajectory Opti-mization and Prediction of Live Air Traffic (TOP-LAT), developed bythe authors. The system reconstructs the trajectory of a flight basedon every information about the flight including flight plans, flightroutes (the system can receive several flight plans and flight routesof the flight; representing different updates for the flight plan) andall radar data of the flight.


The trajectory prediction component of TOP-LAT predicts theaircraft trajectory from one radar point to the next radar point.When it reaches a radar point it computes the time of arrival at thenext waypoint and computes the acceleration/deceleration rate aswell as the climb/descent profile to meet the next radar pointrequirements. These computations are based on the aircraft’scurrent position, altitude, speed and aircraft performance param-eters. The trajectory reconstruction component of TOP-LAT updatesthe flight trajectory for flights that have missing/incomplete radardata. It then forms the completed flight plans which are compliantwith the flight plan structure. The radar data positions are con-verted into the route of the flight planwith necessary modifications(considering distances, heading, speed, and time between twoconsecutive radar signatures).

Trajectories based on EUROCAT radar data often need to besmoothed, i.e. kinks and altitude spikes need to be removed.Trajectory kinks may result from the limited accuracy of radarstations, if closely spaced waypoints originate from different radar


Fig. 1. Aviation emission inventory development process.


ARTICLE IN PRESS

centers. Similarly, altitude spikes are identified and smoothed bydefining a maximum rate-of-climb (ROC) value between successivewaypoints.

To determine the taxi-in time of a flight, firstly the departureroute of the flight is estimated, based on the flight’s currenttrajectory which includes waypoints and radar signatures, theaircraft type, the operator of the flight and the matrices of depar-ture route probability by waypoints, destination, operator, andaircraft. Then the departure runway is estimated based on depar-ture route, the destination, waypoints in the trajectory, anddeparture runway frequency matrices by destination, waypoints,and aircraft type. If arrival runway and taxi-in time of the arrivalrunway is available in the database, taxi-in & taxi-out time of theflight is determined, else a default value is assigned for taxi-in time.The database contains taxi-in times for all major airports inAustralia.

The trajectory of a flight in the system is constructed based onthe actual flight plan and actual radar data. Therefore it is moreaccurate than a trajectory that is based upon a single great circleroute in other inventory and modelling studies such as those byNASA/Boeing (Sutkus et al., 2001).

4.4. Fuel flow and emission computation

4.4.1. Aircraft performance and fuel flowFuel flow is computed between two consecutive radar positions

by forward integrating the aircraft position over time, using theaerodynamic parameters of the specific aircraft.

For aircraft performance parameters, we use the Base of AircraftData (BADA), which is an aircraft performance database developedby the EUROCONTROL Experimental Centre (Nuic, 2004). Version3.6 includes detailed information on 91 supported aircraft types,gathered from reference sources like flight and operating manuals,which are mapped to another 200 aircraft. BADA includes a total of51 parameters per aircraft, which specify the aircraft’s mass andflight envelope together with its aerodynamic and engine capa-bilities. The BADA aerodynamic model is used to calculate lift and


drag as well as thrust and fuel flow in all flight phases. It is used inair traffic simulation models of several organizations, researchinstitutes and government organizations in ATM research such asFuture ATM Concepts Evaluation Tool (FACET) of NASA (Bilimoriaet al., 2000), Early Demonstration & Evaluation Platform (eDEP)of EUROCONTROL (eDEP project team, 2007), and the Air TrafficOperations and Management Simulator (ATOMS) (Alam et al.,2008). Most aircraft types currently modeled in BADA demon-strate mean error in vertical speed lower than 100 feet per minute(fpm), over previously specified normal operation conditions. Thiscorresponds to mean error of vertical speed lower than 5% foroperational range of flight envelope under the assumption that2000 fpm is an average vertical speed value. Fuel consumption ismodeled with mean error lower than 5% for the same conditions(Nuic et al., 2005).

For each position of the aircraft in a discrete time interval of 1 s,the fuel flow is computed based on BADA thrust specific fuelconsumption model. For fuel flow computation, assuming nominalaircraft mass, the thrust specific fuel consumption h in kg/s/kN isspecified as a function of true airspeed, VTAS (knots) for the jetengines. The nominal fuel flow,

Rnom

(kg/s), can then be calculated

for jet engine aircraft using the thrust, Th as:

Znom

¼ h� Th (1)

where:

h ¼ Cf1

1þ VTAS

Cf2

!(2)

Cf1 (kg/s/kN) is first thrust specific fuel consumption coefficient,and Cf2 (knots) is a second thrust specific fuel consumptionobtained from aircraft performance tables. The absolute amount offuel burn in a flight segment is calculated by multiplying fuel flowwith time.



ARTICLE IN PRESS

The effect of wind on aircraft during flight is intrinsicallyincorporated as they impact directly on flight time. The real take-offweight of an aircraft is not provided by airlines, thus the nominalmass, as suggested by ICAO, is used. It is around 60.9% of maximumtake-off weight (ICAO average value for scheduled air traffic). Theweight of the aircraft is reduced as a result of fuel burn while it isflying. This is done by subtracting aircraft take-off weight with theinstantaneous fuel burn.

4.4.2. Emission computationEmission is calculated using Boeing Emission Method 2

(BEM2) methodology and ICAO Engine Exhaust Emissions DataBank. SOx, NOx, CO, HC and CO2 emissions are computed for eachflight.

The emission computation process is divided into two phases,below 3000 ft and above 3000 ft, as shown in Fig. 2.

Below 3000 ft, the emission calculation is based on the ICAOEngine Exhaust Emissions Data Bank (ICAO, 1995). The fuel burncalculation is based on the Landing and Take-Off Cycle (LTO)defined by the ICAO Engine Certification specifications. ICAO LTOcovers four engine operation modes, which are used to model thesix phases of aircraft operations. In order to calculate fuel andemissions for an aircraft/airframe type using the ICAO data bank,we firstly find the engine corresponding to the aircraft type byusing Airplane/Engine Substitution tables as used in the NASAstudy (Sutkus et al., 2001, Appendix B). Then we use the ICAOEngine Exhaust Emissions Data Bank for the given engine tocalculate fuel and emissions for a flight while it’s in idle, approach,climb-out and take-off states. It can be seen that in the ICAO EngineExhaust Emissions Data Bank, the HC and CO emission indices of jetengines in idle state are higher than in the other states (ICAO,1995).The emission and fuel data is given for single engine, it is thenmultiplied by the number of engines depending upon the type ofaircraft it is fitted on.

In this study the time-in mode is derived from actual flighttrajectory data, and runway timings are specific for each airport andaircraft type. This gives the emission computation a high level offidelity as compared to using generic ICAO LTO cycle timings.

Above 3000 ft, the emission calculation is also based on the ICAOEngine Exhaust Emissions Data Bank, but emission factors areadopted to the atmospheric conditions at altitude by Boeing Emis-sion Method 2 (BEM2) (Baughcum et al., 1996; FAA, 2005b), which

3000ft

Taxi-out Taxi-in

Cruise tnecseDfOpoTbmilCfOpoT

Origin Destination

ICAO Fuel FlowICAO Emissions

BADA Fuel FlowBEM2 Emissions

tt3000ft

tnecseDfOpoTbmilCfOpoT

Origin Destination

owICAO Fuel FloonsICAO Emissio

owBADA Fuel FloonsBEM2 Emissio

Fig. 2. Emission computation phases.


calculates emission indices based on fuelflowand ICAO certificationdata. ICAO data at the four certified power settings at sea-levelconditions are used to compute the resulting emissions, while cor-recting for atmospheric conditions. BEM2 allows for the estimationof emissions for pollutants such asNOx, HC, and CO. The emission forCO2 is directly proportional to fuel burnwith CO corrections. ICAO’sCommittee on Aviation Environmental Protection (CAEP) has rec-ommended the adoption of Boeing Method 2 as a standard methodfor calculating emissions (ICAO, 2004).

BEM2 method for emission computation is described here asdetailed in (Baughcum et al., 1996; DuBois and Paynter, 2006).BEM2 computes flight emissions using, as a base, the measured fuelflow and the engine ICAO data sheets. It accounts for ambientpressure, temperature and humidity as well as Mach number.

Since the ICAO fuel flow data does not account for the instal-lation effects of engine air bleed for aircraft use, a correction mustbe made. Therefore the ICAO fuel flow is modified by a correctionfor aircraft installation effects. This is known as the corrected fuelflow (Wff) of installed fuel flow (Wf) from airplane performancemanual or cockpit instrumentation (kilogram/second). Once theconversions and corrections are made, the emission indices (EI)are plotted against the corrected fuel flow on a log10elog10 scale.The data points are curve fitted to show trends of EI for differentfuel flows. The HC and CO are bi-linear least square fitted curves.The NOx curve is a simple point-to-point linear fit, on the log10paper, between the ICAO emission data points, as shown in Fig. 3.

BEM2 steps of emission computation are as follows:

1. Curve fit the ICAO engine emission dataThe Emission Indices (NOx, HC, CO) are to be plotted (logelog)against the corrected fuel flow (Wf).

2. Compute fuel flow factor(a) Calculate the values vamb (ambient pressure correction factor)

and qamb (ambient temperature correction factor) where:� vamb ¼ Pamb=14:696 (Pamb¼ ambient(inlet) pressure) and� qamb ¼ ðTamb þ 273:15Þ=288:15 (Tamb¼ ambient(inlet)temperature)

� qboiling ¼ ðTamb þ 273:15Þ=373:15(b) The fuel flow values are further modified by the ambient values

and are denoted by Wff:� Wff ¼

hWf=vamb q3:8ambe

0:2M2i

(kg/s) where M is the Machnumber.

(c) Calculate the humidity correction factor H:� H¼�19.0(u� 0.0063), u¼ specific humidity assumed to be60% for the entire flight duration.

� u ¼ ð0:62198� F� PvÞ=ðPamb � F� PvÞ (pounds mass ofwater/pounds mass of air), where F is relative humidity andPv¼ saturation vapour pressure in psia (pounds per squareinch absolute (including atmospheric pressure) wherePv¼ 0.014504�10b

� b ¼ 7:90298ð1� 1qboiling

Þ þ 3:00571þ 5:03808logð 1qboiling

Þþð1:3816� 10�7Þ�1� 1011:344ð1�qboilingÞ þð8:1328� 10�3Þ��½103:4919ð1�ð1=qboilingÞÞ � 1�

3. Compute emission indices (EI)Calculate the emission indices of HC, CO and NOx:� EIHC ¼ REIHCq3:8amb=v

1:02amb (pound mass of HC/1000 pound

mass of fuel)� EICO ¼ REICOðq3:8amb=v

1:02ambÞ (pound mass of CO/1000 pound

mass of fuel)� EINOx ¼ REINOxeHh

v1:02amb

q3:3amb

ið1=2Þ (pound mass of NOx/1000pound mass of fuel)

4. Total emission� Total(HC)¼Number_of_engines� Sum(EIHC)�Wff� ti-me_in_Mode (kg)


10010-4

10-3

10-2

10-1

100

101

102

a EIHC

Fuel Flow (kg/s)

E

IH

C (g

/kg

)

10010-1

100

101

102

103 b EICO

Fuel Flow (kg/s)

E

IC

O(g

/kg

)

100100

101

102c EINOx

Fuel Flow (kg/s)

E

IN

Ox (g

/kg

)

Fig. 3. Variation of emission index (g/kg) with fuel flow (kg/s) for RB211e524 D 4 engine: (a) Emission index for HC (EIHC); (b) Emission index for CO (EICO); (b) Emission index forNOx (EINOx).


ARTICLE IN PRESS

� Total(CO)¼Number_of_engines� Sum(EICO)�Wff� ti-me_in_Mode (kg)

� Total(NOx)¼Number_of_engines� Sum(EINOx)�Wff� ti-me_in_Mode (kg)

CO2 emission methodology (Eyers et al., 2004, p. 32): carbondioxide and water are the main products of the combustion ofhydrocarbon fuel and are therefore a function of fuel flow. One kgof Jet fuel produces 3156 g of CO2 and 1237 g of H2O (Eyers et al.,2004, p. 31). For accurate computationof CO2, partially or unburntspecies of CO and HC must be subtracted. Since the amount of


hydrocarbons emitted is less than 1% of CO2 and H2O emissionsand their hydrogen-carbon ratio is unknown, they are neglectedin our computations. The formula for CO2 computation is given by

� EICO2¼ EICO2,ideal� (44/28)� EICO, where EICO2,ideal is 3156(grams CO2/kg fuel burned) and 44/28 ratio represents theirdifferent molar masses.

SOx emission methodology (Eyers et al., 2004, p. 64): Themaximum sulphur content in kerosene according to internationalregulations is 0.3 mass percent with actual values often below this



ARTICLE IN PRESS

limit. For a sulphur content of between 0.001 and 3 g per kg fuelone can expect SOx production during combustion of 0.6e1 g/kg.The formula for SOx computation is given by:

� SOx¼ 0.8�Wff (kg)

This value is based on (Hadaller and Momenthy, 1989) and wasused in the NASA and SAGE inventories.

4.5. Emission database

The database acts as a dynamic repository of flight state infor-mation and emission data. The database is divided into two mainparts: flight database, and gridded emission database.

The flight database stores all the flight-related information (e.g.trajectory, flight plan).

The emission database stores aggregated fuel burn and emis-sions per grid cell. Lambert Canonical Projection system is utilizedbased on longitude and latitude; these are supplemented by analtitude coordinate which is oriented orthogonally to the earth’ssurface. Based on this coordinate system, the three-dimensionalairspace is divided into cells, which span the whole AustralianNational Airspace from ground level up to a maximum altitude of61,000 ft. Fuel burn and emissions of each flight are calculated andallocated to a grid cell. For the purpose of emission allocation, theintersections of the flight path with the grid cell boundaries aredetermined by their three-dimensional coordinates. The process isrepeated for all flights processed from the radar data, and therespective emissions per grid cell are summed.

4.6. Software and data availability

This paper presents the aviation emission inventory developed byusing Trajectory Optimization and Prediction of Live Air Traffic (TOP-LAT) software. However, the emission database is owned by Air-services Australia and TOP-LAT is currently being commercialized.TOP-LAT is developed in Microsoft Visual C# and it is based on

E110 E115 E120 E125 E130 E1S45

S40

S35

S30

S25

S20

S15

S10

Long

Latit

ude

Fig. 4. One-day domestic and international air traffic in Australian National Airspace;


a distributed client-server architecture. The main server processesreal time air traffic data, and other sub-systems compute fuel, emis-sion, safety, etc.

The system provides a comprehensive emission inventory in theform of emissions and fuel data as well as flight specific informationwhich is updated by TOP-LAT and stored in a 4D gridded databasein real time which represents national airspace in the form of a 4Dgrid, representing the latitude, longitude, altitude and time whichcan be used for spatial and temporal analysis of emission.

This emission inventory provides valuable information on thequantity and location of emissions in the Australian NationalAirspace. The previous international studies only used publishedflight plans and assumed a great circle route between origin anddestination airports. The use of radar data has considerablyenhanced the knowledge and accuracy of the actual position(latitude, longitude and altitude) at which these emissions actuallyoccur. This provides Air Navigation Service Providers with a higherfidelity and more accurate estimate of aviation emissions. Thisemission database provides a major new data source for aviationclimate impact assessment.

From the database, we can query the emission data in any areaat any time. Emission data can also be presented with any requiredresolution by the system.

5. Aviation emission results

We report the key summary results for the period October2008eMarch 2009. The results are only a subset of queries that canbe used to extract desired data from the emission database.

5.1. Summary of flight & emission data

In the study period there were 413,359 domestic flights and79,577 international flights. Domestic flights burned a total of1048.57 kt of fuel and international flights burned a total of1467.26 kt of fuel. Domestic flights flew a distance of 173,074 knmand international flights flew a distance of 122,229 knm.

35 E140 E145 E150 E155 E160itude

trajectory of latitude and longitude, south (S) and East (E), respectively for one.


Table 2Flight movements, distance travelled, fuel used, and emissions of domestic and international Flights in Australian National Airspace during the 6-month study period.

Flights Distance Fuel HC CO NOx CO2 SOx

Total 492,936 295,303 2515.83 114.59 200.95 45.92 7929.89 2.11Domestic 83.86% 58.61% 41.68% 96.57% 96.34% 32.37% 41.68% 41.68%International 16.14% 41.39% 58.32% 3.43% 3.66% 67.63% 58.32% 58.32%

Distance is measured in knm, and fuel and emissions in kt.


ARTICLE IN PRESS

Fig. 4 presents a one-day snapshot of the domestic and inter-national flights in Australian Airspace. Some of the trajectories inthe figure are discontinued over the sea, because they belong tointernational flights. The terminal points are the flights’ first or thelast radar signature (waypoints) in the Flight Information Region(FIR) of Australia, the system received.

Table 2 presents the total numbers of flights, distance flown, fuelused and emissions produced during the 6-month study periodstarting from October, 2008, and breaks them down into percent-ages for domestic and international flights.

Table 3 shows the averages of flight movements, fuel andemissions in one month for the study period. The standard devia-tions show that the number of flights, fuel burn and emissions arenot much different between the months.

Fig. 5 presents temporal distribution in the study period. Itshows a small decrease of CO2, NOx and SOx in February due toreduced demand for air travel after the Christmas holiday period.We also can see that the emissions of HC, CO, NOx and SOx are smallin comparison with CO2 emission.

5.2. Fuel and emissions of different aircraft categories

Table 4 presents the fuel and emissions data for different aircraftcategories. It can be seen that almost all HC and CO emissions areproduced by turbo prop and piston aircraft, while NOx, CO2 and SOx

are mainly emitted by jet aircraft. 81.66% of HC and 63.25% of CO areemitted by piston flights, 13.07% of HC and 31.47% of CO by turboprop flights, while 95.74% of NOx, 92.46% of CO2, and 92.46% of SOx

are produced by jet flights.Although the total distance travelled by international flights is

less than that by domestic flights, the fuel burn by internationalflights is more than that by domestic flights. This is because 92.51%of international aircraft types are jet, only 4.27% and 3.23% are turboprop and piston respectively. In contrast, in domestic aircraft, jetaccounts for 45.50%, followed by turbo prop 34.01%, and piston20.50%. As a result of this aircraft distribution, most HC and CO isemitted by domestic flights, while international flights emit moreNOx, CO2, and SOx than domestic ones.

In general, CO and HC emissions are produced when engines areat low thrust settings (e.g. Idle, Approach, etc.); thus in turbopropsand piston driven aircraft these emissions are high, as they aremostly short-distance flights. The majority of fuel is consumed atcruise flight level of altitude, which for long and medium distanceflights can be up to 70% of flight time. Thus for jet aircraft, CO2emissions is higher as it is directly proportional to fuel burn. NOx

and SOx also follow the trend in fuel consumption in jet aircraft.

Table 3Monthly averages of flight movements, fuel and emissions.

Flights Fuel HC CO NOx CO2 SOx

Average 82,156 419.30 19.10 33.49 7.65 1321.65 0.35Stdev 3,336 26.89 1.14 1.40 0.54 84.74 0.02

Fuel and emissions are measured in kt.


In Table 5, the fuel and emissions of each aircraft type arecalculated as the average of all flights of the aircraft type for first 14days of a month, 2008. From the table, we can see that there isa difference in fuel consumption and emission between differentaircraft types. Some aircraft/engine types consume fuel andproduce emissions twice of others.

5.3. Emissions distribution in different flight phases

Fig. 6 presents the aviation emissions distribution in differentflight phases: Fig. 6(a)e(c) present the emission distribution indifferent phases for a short, a medium and a long distance flightrespectively, and Fig. 6(d) shows the emission phase distributionfor all the flights in the study period. The reason for presenting thealtitude range of cruise phase as in the figures is that the cruisephase of a flight is determined from the top of climb to the top ofdescent, and the cruise altitude fluctuates slightly around a certainvalue. From the figures, we can see that NOx, CO2 and SOx in climbphase are much higher than in the other phases in the short-distance flight, while in themedium distance flight, CO2 and SOx arehighest in cruise phase following by climb phase; NOx in cruisephase is not much lower than in climb phase. In both short andmedium distance flights, HC and CO are highest in descent phase.For the long distance flight, all emissions are mostly produced incruise phase due to the long distance of this flight phase. Overall,Fig. 6(d) shows that the cruise phase of flight consumes the mostfuel and generates the most pollutants (more than 60%). Thedescent and climb phases come next, and the remaining phasesaccount for a small amount of the total emissions. Althoughemissions from the descent and climb phases are much less thanfrom the cruise phase, they are highly concentrated aroundairports, so the impact of these emissions is considerable. Duringcruise the rate of emission of HC is low. However the cruise phaseaccounts for much more of a flight than other phases that thepercent of HC emitted in the cruise phase is still high compared toother phases. The same applies to emissions of CO.

5.4. Result comparison

In Table 6, we compare the fuel burn and emission in one yearfor domestic and international flights in Australia from our studywith SAGE in 2004 and NGGIC in 2007. In “This Study 2008”, Thevalues of variables such as “Flights”, “Distance”,. in one year fordomestic and international flights in Australia are extrapolatedfrom those in the 6-month study period in Table 2. In SAGE2004, the number of flights (“Flights”) is estimated from the totalfuel burn (“Fuel”) and fuel per flight (“Fuel/F”); total distanceflights travelled (“Distance”) is estimated from fuel per distance(“Fuel/Dis”) and the total fuel burn (“Fuel”).

From Table 6, we can see that the number of flights in “Thisstudy 2008” is higher than in the two other studies. This is becauseair traffic grows by 4% per year. Further the definition of inter-national flights between our study and SAGE is different. In SAGE,international flights are flights with departure airport in Australia,and arrival airport in another country, while in our study


10−2008 11−2008 12−2008 1−2009 2−2009 3−2009

1200

1300

1400

Emis

sion

s (k

iloto

nne)

Month

30323436

Emis

sion

s (k

iloto

nne)CO

HC

CO2

NOx

SOx

Fig. 5. Temporal distribution of HC, CO, NOx, CO2, and SOx (kt) from October 2008 to March 2009.

Table 5Fuel burn & emissions in kg for seven representative jet aircraft.

Aircraft Fuel HC CO NOx CO2 SOx


ARTICLE IN PRESS

international flights are all flights whose departure airport orarrival airport is in another country. Therefore, the number ofinternational flights in our study is much higher than in SAGE.However, the distance travelled by international flights in ourstudy is not much higher than in SAGE in comparison with thedifference in the number of flights. The reason is that our studyincludes more international flights with short distance. The higher“Flights” in our study results in higher “Distance”, “Fuel”, “CO2”

and “NOx” than in the other study. Another reason is that NGGICconsiders the fuel consumption as the same in every segment inthe whole cruise phase (NGGIC, 2007). However, the fuelconsumption is not constant in different segments in the cruisephase. Therefore, total emissions summed in every segment arelikely to be different from the total emission estimated by usingthe fuel consumption in the whole cruise phase. This model alsodoes not take into account international flights and other phasesof flight, for instance taxi-out, climb, climb-out, approach, landingand taxi-in. As a result, the amount of aviation emissions calcu-lated by this methodology are likely to be an underestimate andcan not provide required insight into emissions, needed fora deeper analysis.

“Distance/F” in this study is less than in SAGE for domesticflights and especially for international flights. For internationalflights, the reason is the difference in the definition of internationalflights which results in more short-distance international flights inour study. For domestic flights, there may be more short-distanceflights in Australia in 2008 than in 2004. This leads to “Fuel/F”,”Fuel/Dis”, “CO2/F”, “CO2/Dis”, “NOx/F”, and “NOx/Dis” in our studybeing less than in SAGE.

Table 4Percentages of number of flights, fuel burn & emissions based on aircraft categories.

Aircraft Flights Fuel HC CO NOx CO2 SOx

Jet 53.08% 92.46% 5.27% 5.28% 95.74% 92.46% 92.46%Turbo Prop 29.21% 4.83% 13.07% 31.47% 2.11% 4.83% 4.83%Piston 17.71% 2.70% 81.66% 63.25% 2.15% 2.70% 2.70%


6. Geographical distribution of emission

In this section, we present the spatial analysis of emissions inAustralian Airspace with cumulative emission values on a planarand a vertical display. Here, we only show the distributions of COand CO2 emissions, because the distribution of HC is similar to CO,and the distributions of SOx and NOx are similar to CO2.

6.1. Spatial distribution of aviation emission

Figs. 7 and 8 show that the areas around Sydney, Melbourne,Brisbane, Perth, Darwin, Cairns, and Adelaide are most impacted byaviation emissions. The emission distribution is concentrated alongthe South coast line because flight paths are mainly concentratedthere.

Fig. 7 shows HC and CO emissions are mainly between S31� andS36�, and below 10,000 ft. They are concentrated around Brisbane,Cairns, Adelaide, Perth, Sydney, and Melbourne. HC and CO areemitted due to incomplete combustion process of jet engines at lowthrust level, which explains the high concentration of CO and HCduring descent and approach phases of flight.

From Fig. 8 we can see that NOx, CO2, and SOx emissions affectboth the local air quality and the upper airspace. Sydney,

AC1 2876.66 1.45 21.75 40.40 9067.24 2.42AC2 5240.87 2.36 13.03 144.58 16519.21 4.40AC3 2710.29 2.40 17.05 31.19 8542.83 2.28AC4 3016.98 2.14 12.07 34.75 9509.52 2.53AC5 5749.47 2.34 11.61 239.74 18122.32 4.83AC6 9208.64 31.51 65.39 219.92 29025.64 7.74AC7 5762.91 2.37 13.33 225.81 18164.70 4.84Average 4937.97 6.37 22.03 133.77 15564.50 4.15Stdev 2329.92 11.09 19.45 96.85 7343.91 1.96

Fuel and emissions are measured in kilograms.


Fig. 6. Emissions distribution in different flight phase. The numbers below phases in(a), (b), (c) are the altitude range of each phase, measured in kft.

Table 6Result comparison with other methodologies.

This study SAGE NGGIC

2008 2004 2007

Value Value %Diff Value %Diff

Flights D 826,718 454248 82.00% NA NAI 159,154 49600 220.88% NA NA

Distance D 346148 194945 77.56% NA NAI 244458 160179 52.62% NA NA

Distance/F D 0.4187 0.4292 �2.44% NA NAI 1.5360 3.2294 �52.44% NA NA

Fuel D 2097.13 1390 50.87% NA NAI 2934.52 2480 18.33% NA NA

Fuel/F D 0.0025 0.00306 �17.10% NA NAI 0.0184 0.05 �63.12% NA NA

Fuel/Dis D 0.0061 0.0071 �15.03% NA NAI 0.0120 0.0155 �22.47% NA NA

CO2 D 6610.16 NA NA 5291.94 19.94%I 9249.62 NA NA NA NA

CO2/F D 0.0080 NA NA NA NAI 0.0581 NA NA NA NA

CO2/Dis D 0.0191 NA NA NA NAI 0.0378 NA NA NA NA

NOx D 29.73 19.9 49.40% NA NAI 62.12 39.9 55.69% NA NA

NOx/F D 0.000036 0.000044 �18.45% NA NAI 0.000390 0.000797 �51.03% NA NA

NOx/Dis D 0.000086 0.000103 �16.44% NA NAI 0.000254 0.000256 �0.57% NA NA

In the table, “Flights” is the number of flights. “Distance” is the total distance flightstravelled, measured in knm; “Distance/F” is the knm per flight; “Fuel” (“CO2” or“NOx”) is the total fuel burn (“CO2” or “NOx”) measured in kt; “Fuel/F” (“CO2/F”or “NOx/F”) is the kt of fuel (“CO2” or “NOx”) per flight; “Fuel/Dis” (“CO2/Dis” or“NOx/Dis”) is the fuel (“CO2” or “NOx”) per 1 knm; These variables are calculated fordomestic (D) flights and international (I) flights separately; “Value” is the value ofvariables; “%Diff” is the percentage difference of variables between our study withother study.


ARTICLE IN PRESS


Melbourne and Brisbane airports shows high density of NOx, CO2,SOx emissions. Besides that, the NOx, CO2, and SOx distributions aremainly from S31� to S36�, and from 35,000 to 40,000 ft.

6.2. Distribution of aviation emissions by altitude

Fig. 9 shows that most emission of NOx, CO2 and SOx is in highaltitude, while emission of HC and CO is mainly in low altitude.About 50% of NOx, CO2 and SOx are between the altitude band35,000e40,000 ft and 20% between the altitude band30,000e35,000 ft. Around 30% of HC and CO emissions are between5000e10,000 ft and 0e5000 ft.

The reason for this is most jet flights, which emit a lot of NOx,CO2 and SOx, cruise between the altitude band 30,000e40,000 ft,and the number of jet flights is much more than that of the otherflights. On the contrary, piston aircraft, which produce a highamount of HC and CO, usually fly in low cruising altitude. Manypiston flights have cruising altitude equal or less than 10,000 ft.Therefore, the percentage of HC and CO is large in altitude bands0e5000 ft and 5000e10,000 ft. As a result the percentages of HCand CO in the other altitude bands are lower than the two altitudebands. In addition, flights with cruising altitude higher than10,000 ft produce a lot of HC and CO while they are at low altitude.

Because there are few flights with cruising altitude between40,000 and 45,000 ft (high altitude band), not only the percentagesof HC and CO in this band are low but the other emissions as well.

7. Discussion

From the aviation emissions data, we found that the amount ofemitted CO2 at several cells is much higher than the others. Such


Fig. 7. CO spatial emission distribution in the study period.


ARTICLE IN PRESS

cells are identified in Table 7, where the amount of CO2 of one cell issummed from ground level to an altitude of 14 km. This altitude isused, because the depth of the troposphere in Australia varieslatitudinally, seasonally, and daily. For example, it is about 16 kmabove Australia at year-end, and between 12 and 16 km at midyear,being lower at the higher latitudes (Sturman and Tapper, 2005, p.23). Therefore we adopt a mean depth of 14 km for the tropo-sphere to calculate CO2 from the ground to this altitude. If the airtraffic growth of 4% per year (Shepherd et al., 2007) and otheremissions sources are taken into account, the temperature aroundthese cells may increase faster than in other regions, potentiallywith significant impact on the local people’s health. Because the


main species for ozone change in aviation emissions is NOx, we tryto found some cells where flights emit NOx much more than atothers. Furthermore, most aviation emissions in Australia occurbelow 45,000 ft (about 13.7 km) (see Fig. 9), we only consider cellsin the upper troposphere in Australia between 10 km and 14 km(about 32,000e46,000 ft), where ozone is one important green-house gas. This means the NOx emission of aviation in Australiadoes not impact directly on the ozone layer in the stratosphere,where ozone absorbs the sun’s high frequency ultraviolet light.Understanding how ozone varies will reveal how this constituentbuffers the temperature increase. In Table 8, we found some cellsbetween 10 km and 14 km with much more NOx emitted than the


Fig. 8. CO spatial emission distribution in the study period.


ARTICLE IN PRESS

other cells at the same altitude level. In the future, we will considerother factors such as wind effect and chemical interactions whichaffect the concentration and distribution of emissions to estimatethe impact of aviation on the environments as a whole and at thesecells we found above. That may provide a better view of aviationemissions impact. The system’s output of gridded emissions withhigh temporal resolution may also be incorporated with generalemission processing systems or specific chemical-transport modelsto create a general picture about emissions in Australia. This willhelp environmental specialists to analyze the impact of emissionsas well as the air quality in Australia.


8. Limitations and assumptions

The first limitation in this study arises from lack of availability ofhigh fidelity emission & fuel flow data. An aviation engine’s fuelconsumption and emissions at higher altitudes is commerciallysensitive. The second limitation arises from some of the assump-tions underlying the study. In this section, the most significantassumptions are summarized and discussed.

Flight with a standard payload and no fuel tinkering: Payloadaffects the fuel flow and hence the emission computation. However,data about payloads or actual fuel amounts carried by the


0% 10% 20% 30% 40% 50% 60%

0-5

5-10

10-15

15-20

20-25

25-30

30-35

35-40

40-45

45-50

50-55

55-60

60-65

Altitu

de

B

an

d (1

00

0 fe

et)

Fraction of Emissions

SOxCO2NOxCOHC

Fig. 9. Distribution of aviation emissions by altitude in the study period.


ARTICLE IN PRESS

individual flights are not available from ATC data. Thus we haveassumed an average payload and do not account for fuel tankering.

No consideration of delays and holdings: Ground based (taxi)delays may be approximated by the use of aircraft-specific times-in-modes, whereas airborne delays, such as holding, are moredifficult to consider if not captured in trajectory data.

Simplified routing and trajectory modelling: Flight routing doesnot account for different mixing heights at different airports andare assumed fixed at 3000 ft. Further noise abatement proceduresare not considered, simplifying the landing and take-off cycle. Noaircraft and engine deterioration: Since such data is difficult toobtain and model, thus the effects due to deterioration and engineaging are excluded in our study.

Military, helicopters, and recreational flights are not: Apart fromthe fact that the military flights data is classified and data on engineand aircraft type is not available, military flights mostly operate inspecial use airspace and use upper airspace (above 41,000 ft).Emission modelling for helicopters is not yet developed and they

Table 7Five grid cells with largest amount of CO2 emitted in troposphere layer.

Long (deg) Lat (deg) CO2 (kt) Airports

E144 S38 196.17 Melbourne, Point Cook, Yarra Bank,Penfield, Essendon

E151 S34 177.46 SydneyE153 S28 162.87 Brisbane, Archerfield, DunwichE150 S34 97.6 Bankstown, Westmead, Richmond,

Glenbrook, KatoombaE115 S32 63.79 Perth, Gingin

“Long (deg)” is the longitude of a cell in degree; “Lat (deg)” is the latitude of a cell indegree; “CO2 (kt)” is the kilotonnes of CO2 emitted at a cell in troposphere layer inthe 6-month study period; “Airports” is the list of airports at a cell.


constitute a very small number of operations. Data on the engine ofrecreational aircraft is also not available.

A series of parametric studies were conducted to evaluate theeffects of wind, temperature, payload, tankering, and cargo on thecalculated fuel use (Baughcum et al., 1996). These studies show thatthe main uncertainty is in the assumption about fuel tankering, inwhich the actual fuel burn is 8% more than the fuel estimated withno fuel tankering. Next in uncertainty are assumptions aboutpassengers load factor and wind, where the uncertainty is about2.5%. The uncertainty caused by assumption about standardtemperature is less than 1%. Therefore, the overall uncertainty inour assumptions and limitations is small (likely less than 10%).

Another limitation of the inventory is that it does not take intoaccount that emissions can be moved around by wind after theyhave been emitted. Particle emissions are also not available in thisinventory currently. These are areas for possible future work.

Table 810 grid cells with largest amount of NOx emitted in upper troposphere layer.

Long (deg) Lat (deg) NOx (kt)

E148 S34 0.14E119 S14 0.13E118 S13 0.13E146 S37 0.13E149 S33 0.12E120 S15 0.12E148 S36 0.12E150 S32 0.12E152 S30 0.11E121 S16 0.10

“Long (deg)” is the longitude of a cell in degree; “Lat (deg)” is the latitude of a cell indegree; “NOx (kt)” is the kilotonnes of NOx emitted at a cell between 10 km and 14km altitude in the 6-month study period.



ARTICLE IN PRESS

9. Recommendations

In terms of air traffic management, several options exist to slowthe increase of CO2 concentration at polluted cells. Current majorairports might be replaced with several other airports in differentareas. Passengers will travel to the new airports by other trafficvehicles which consume less fuel. As a result, the emissions will bedistributed more evenly to different areas. While the developmentof multi-airport systems involves and is influenced by a wide rangeof factors (Neufville, 1995), empirically verifiable concerns about airtransport noise and atmospheric pollution are certainly one ofthem (Graham and Guyer, 1999).

Rerouting flights may help to reduce the emission concentrationat congested waypoints and flight paths. Because the depth of thetroposphere in Australia varies regionally, seasonally, and evendaily, using flight levels suitable to the regions and seasons mightreduce the impact of NOx emissions on the ozone layer.

Flights can use direct routes to save time and fuel burn. TheDirect-To Controller Tool identifies that aircraft can save at least oneminute of flying time by flying direct to a down-stream fix along itsroute of flight (TSC, 2002). Our system (TOP-LAT) also providesUser-Preferred Trajectories tool for optimizing flight trajectories interms of time and distance travelled, fuel burn, emissionsproduced, and comfortability for passengers.

Efficient departure and arrival planning can also help to reducethe time aircraft are in runways. As a result, aviation emissions willbecome less in runways. For example, the Departure EnhancedPlanning And Runway/Taxiway-Assignment System (DEPARTS)developed by the MITRE Corporation’s Center for Advanced Avia-tion System Development can generate optimized recommenda-tions on runway assignment, departure sequencing and departurefix loading for the Air Traffic Control tower to reduce taxi-out times(Cooper et al., 2001).

In terms of aviation technology, inventing new engine typeswhich consume less fuel, or another energy which does notproduce emissions, would also be a solution.

Such decisions could not be made without careful considerationof all their effects. The emission inventory presented herewill be anessential element in that process.

10. Conclusions

In this paper we present an aviation emission inventory forAustralian Airspace, and some discussion based on a subset of dataderived from the inventory. This is the first attempt to develop anemission inventory using real time trajectory and flight plan data.The emission database is prepared by TOP-LAT. This system isonline and processes air traffic data to compute aviation emissionsin real time. The emission inventory is an ongoing process, makinguse of real time data as it arrives. The inventory is in the form ofa 4D database which provides resolution of 1� �1�1000 ft, fortemporal and spatial emission analysis.

The aviation emissions are calculated based on the actual 4Dtrajectories which are mixed of actual flight routes and radar data.This can help provide more accurate emission data in comparisonwith other emissions inventories which only use flight plans andassume a great circle route between origin and destination airports.Some information of a flight is not given sometimes. They are alsoestimated to complete information about the flight. This furtherimproves the accuracy of emission computation.

The specific emission inventory reported here covers a period ofsix months fromOctober 2008. The study processed 492,936 flightsfor the given period, which burned 2,515.83 kt of fuel and produced114.59 kt of HC, 200.95 kt of CO, 45.92 kt of NOx, 7929.89 kt of CO2,and 2.11 kt of SOx.


For emission computation, the BEM2 methodology along withICAO engine emission data bank is used. Fuel flow is computedusing aircraft-specific aerodynamic and thrust model, with weightcorrection and accounting for atmospheric data.

The emission results show that domestic flights produce muchmore HC and CO than international flights, conversely internationalflights emit more NOx, CO2, and SOx than domestic ones. The resultsalso show that HC and CO emissions are mostly produced by turboprop and piston, while NOx, CO2 and SOx are emitted by jet aircraft.81.66% of HC and 63.25% of CO are emitted by piston flights, and13.07% of HC and 31.47% of CO by turbo prop flights, while 95.74% ofNOx, 92.46% of CO2, and 92.46% of SOx are produced by jet flights.

The study found that the major international airports such asSydney, Melbourne and Brisbane airports show high density ofaviation emissions. In addition, the coast line is much moreimpacted by aviation emission than the other regions.

All the emission distributions are mainly between S31� and S36�

latitude.Most emission of NOx, CO2 and SOx is in high altitude (the

cruising altitude of most of jet flights), while emission of HC and COis mainly in low altitude (the cruising altitude of most of pistonflights). About 50% of NOx, CO2 and SOx are between the altitudeband 35,000e40,000 ft and 20% between the altitude band30,000e35,000� ft. Around 30% of HC and CO emissions arebetween 5000e10,000 ft and 0e5000 ft.

The study found that several cells around major airports such asMelbourne,SydneyandBrisbanehaveCO2emittedbyair trafficmuchmore than theother regions, and somecells in theupper tropospherewith much more aviation emissions of NOx than the others. Theseareas may be environmentally impacted more than the other areas.

It is expected that with the availability of a real time aviationemission database for Australian Airspace, environmental analystsand aviation experts will have an indispensable source of infor-mation to support timely decision-making regarding expansion ofrunways, building new airports, applying route charges based onenvironmentally congested airways, and restructuring air trafficflow to achieve sustainable air traffic growth.

In the future, when we have more information about theassumptions that we use in this study, wewill update the system tocompute fuel burn and emissions more accurately. We will alsoanalyze the impact of aviation emissions on the environment inAustralia such as the effect on the ozone layer, and greenhouseeffects, inwhichwind effect and chemical interactions will be takeninto account.

Acknowledgements

The authors wish to acknowledge the contributions fromcolleagues at Airservices Australia with a particular mention for Dr.Rob Porteous and Mr. Udo Bucher. We also wish to thank ourcollaborators from EUROCONTROL Experimental Centre (EEC), Mr.Steve Kirby and Mr. Mohamed Ellejmi, for their hospitality duringour 2008 visit to EEC and the discussions they facilitated for us withcolleagues from EUROCONTROL on environmental modelling andsafety.

References

Alam, S., Abbass, H., Barlow, M., 2008. ATOMS: air traffic operations and managementsimulator. IEEETransactions on IntelligentTransportation Systems9 (2), 209e225.

Baughcum, S.L., Tritz, T.G., Henderson, S.C., Pickett, D.C., 1996. Scheduled CivilAircraft Emission Inventories for 1992: Database Development and Analysis.NASA Contractor Report 4700. Tech. rep.. NASA, Washington, D.C.

Bilimoria, K.D., Sridhar, B., Chatterji, G., Sheth, K.S., Grabbe, S., 2000. FACET: futureATM concepts evaluation tool. In: Proceedings of Third USA/Europe Air TrafficManagement Research and Development Seminar, Napoli, Italy.



ARTICLE IN PRESS

Borge, R., Lumbreras, J., Rodríguez, E., 2008. Development of a high-resolutionemission inventory for Spain using the SMOKE modelling system: a case studyfor the years 2000 and 2010. Environmental Modelling & Software 23 (8),1026e1044.

Brasseur, G.P., Cox, R.A., Hauglustaine, D., Isaksen, I., Lelieveld, J., Lister, D.H.,Sausen, R., Schumann, U., Wahner, A., Wiesen, P., 1998. European scientificassessment of the atmospheric effects of aircraft emissions. AtmosphericEnvironment 32 (13), 2329e2418.

Cooper, J.W.W., Cherniavsky, E.A., DeArmon, J.S., Glenn, J.M., Foster, M.J., Mohleji, S.C., Zhu, F.Z., 2001. Determination of minimum push-back time predictabilityneeded for near-term departure scheduling using departs. In: Proceedings ofthe Fourth USA/Europe Air Traffic Management Seminar ATM-2001, Santa FeNM. EUROCONTROL and the US Federal Aviation Administration.

DuBois, D., Paynter, G., 2006. Fuel flow method 2 for estimating aircraft emissions.In: Sae Technical Paper Series-2006-01-1987. The Boeing Company.

eDEP Project Team, 2007. Early Demonstration & Evaluation Platform. eDEP ProjectTeam. http://www.eurocontrol.fr/projects/edep/.

Eyers, C.J., Addleton, D., Atkinson, K., Broomhead, M.J., Christou, R., Elliff, T., Falk, R.,Gee, I., Lee, D.S., Marizy, C., Michot, S., Middel, J., Newton, P., Norman, P.,Raper, D., Stanciou, N., 2004. AERO2K Global Aviation Emissions Inventories for2002 and 2025. Tech. Rep. QinetiQ/04/01113. QinetiQ, UK.

FAA, 2005a. System for Assessing Aviation’s Global Emissions (SAGE), Version 1.5.Office of Environment and Energy (AEE-300), Federal Aviation Administration,800 Independence Ave., SW, Washington DC 20591, USA. Tech. Rep. FAA-AEE-2005-01.

FAA, 2005b. System for assessing Aviation’s Global Emissions (SAGE), version 1.5.Global Aviation Emissions Inventories for 2000 Through 2004. Office of Envi-ronment and Energy (AEE-300), Federal Aviation Administration, 800 Inde-pendence Ave., SW, Washington DC 20591, USA. Tech. Rep. FAA-EE-2005-02.

Graham, B., Guyer, C., 1999. Environmental sustainability, airport capacity, andeuropean air transport liberalization: irreconcilable goals? Journal of TransportGeography 7 (3), 165e180.

Hadaller, O.J., Momenthy, A.M., 1989. The Characteristics of Future Fuels. BoeingPublication D6-54940, Seattle.

ICAO, 1995. In: Engine Exhaust Emissions Databank, first ed. International CivilAviation Organisation, Document Sales Unit, 1000 Sherbrooke Street West,Suite 400, Montreal, Quebec, Canada ICAO Doc. 9646.

ICAO, 2004. Guidance on the Use of LTO Emissions Certification Data for theAssessment of Operational Impacts. Tech. Rep. CAEP/6-IP/5. International CivilAviation Organisation, Committee on Aviation Environmental Protection (CAEP)Working Group 3 (WG3), Alternative Emissions Task Group (AEMTG), Montreal,Quebec, Canada.

IPCC, 2001. IPCC report 2001, chapter 3. Tech. rep.. The Carbon Cycle and Atmo-spheric Carbon Dioxide CO2 Concentration Trends and Budgets. Intergovern-mental Panel on Climate Change, p. 3.

IPCC, 2007. Climate Change 2007 Synthesis Report e Appendix Glossary. Inter-governmental Panel on Climate Change. Tech. rep.

Janic, M., 1999. Aviation and externalities: the accomplishments and problems.Transportation Research 4D 4 (3), 159e180.

Keogh, D.U., Ferreira, L., Morawska, L., 2009. Development of a particle number andparticle mass vehicle emissions inventory for an urban fleet. EnvironmentalModelling & Software 24 (11), 1323e1331.

Michot, S., Stancioi, N., McMullan, C., Peeters, S., Carlier, S., 2003. AERO2k FlightMovement Inventory. Tech. Rep. EEC/SEE/2003/005. EUROCONTROL Experi-mental Centre, Bretigny-sur-Orge, France.

Monforti, F., Pederzoli, A., 2005. THOSCANE: a tool to detail CORINAIR emissioninventories. Environmental Modelling & Software 20 (5), 505e508.

Neufville, R., 1995. Management of multi-airport systems: a development strategy.Journal of Air Transport Management 2 (2), 99e110.

NGGIC, 2007. Australian methodology for the estimation of greenhouse gas emis-sions and sinks 2006. Workbook for Transport. National Greenhouse GasInventory Committee. Department of Climate Change, Canberra, Australia.

Nuic, A., 2004. User Manual for the Base of Aircraft Data (BADA) e Revision 3.6.EUROCONTROL Experimental Centre, Bretigny-sur-Orge, France.

Nuic, A., Chantal, P., Iagaru, M., Gallo, E., Navarro, F., Querejeta, C., 2005.Advanced aircraft performance modeling for ATM: enhancements to the


BADA model. In: Proceedings of the 24th Digital Avionics Systems Confer-ence, Washington, DC.

Penner, J.E., Lister, D., Griggs, D., Docken, D., MacFarland, M., 1999. Aviation and theGlobal Atmosphere. Cambridge University Press, New York.

Samaali, M., François, S., Vinuesa, J.-F., Ponche, J.-L., 2007. A new tool for pro-cessing atmospheric emission inventories: technical aspects and applicationto the escompte study area. Environmental Modelling & Software 22 (12),1765e1774.

Schaefer, M., 2006. Methodologies for aviation emission calculation e a comparisonof alternative approaches towards 4D global inventories. Ph.D. thesis, BerlinUniversity of Technology in cooperation with the German Aerospace Center(DLR).

Schumann, U., 2002. Aircraft emissions. In: Encyclopedia of Global EnvironmentalChange, vol. 3. Wiley, pp. 178e186.

Shepherd, G., Byron, B., Mrdak, M., Bills, K., 2007. WAYPOINT 2007 AirservicesAustralia Air Navigation Service Provision: Moving towards 2020. Availablefrom: http://www.airservicesaustralia.com/projectsservices/industryforums/waypoint/2007/docs/Air_Navigation_Service_Provision_Moving_towards_2020_Greg_Russell.pdf.

Sturman, A.P., Tapper, N.J., 2005. In: The Weather and Climate of Australia and NewZealand, second ed. Oxford University Press, Melbourne, Australia.

Sutkus, D.J.J., Baughcum, S.L., DuBois, D.P., 2001. Scheduled Civil Aircraft EmissionInventories for 1999: Database Development and Analysis. NASA, Washington,D.C. Tech. Rep. NASA/CR-2001-211216.

TSC, 2002. Direct-To (D2) Overview Description. Tech. Rep. NAS2-98005. TitanSystems Corporation, Billerica, Massachusetts, USA.

Wesoky, H., Schumann, U., Wurzel, D., 1996. Lingering uncertainty about aviation’simpact addressed by growing body of scientific data. ICAO Journal 51, 11e14.

Winiwarter, W., Schimak, G., 2005. Environmental software systems for emissioninventories. Environmental Modelling & Software 20 (12), 1469e1477.

Appendix e acronyms and nomenclature

AC/Eng: Aircraft/engineAMOC: Air traffic flow management modelling capabilitiesATC: Air Traffic ControlATCs: Air Traffic ControllersATM: Air traffic managementBADA: The base of aircraft dataBEM2: Boeing emission method 2BITRE: Bureau of infrastructure, transport and regional economicsDLR: The German Aerospace CentreEDMS: Emissions and dispersion modelling systemETMS: Enhanced traffic management systemEUROCAT: Thales radar Data Processing softwareFAA: The U.S. Federal Aviation AdministrationGAEC: Global Atmospheric Emissions CodeICAO: International Civil Aviation OrganisationINM: Integrated noise modelLTO: The landing and take-off cycleNASA: The U.S. National Aeronautics and Space AdministrationNLR: The National Aerospace Laboratory of the NetherlandsOAG: The Official Airline GuideROC: Rate-of-climbSUA: Special use airspacekm: kilometerft: feetkft: kilo feetknm: kilo nautical mileg: gramkg: kilogramkt: kilotonnes: secondfpm: feet per minutekN: kilo Newton


http://www.eurocontrol.fr/projects/edep/

http://www.airservicesaustralia.com/projectsservices/industryforums/waypoint/2007/docs/Air_Navigation_Service_Provision_Moving_towards_2020_Greg_Russell.pdf



Documents

2010 Aviation Emission Inventory Development and Analysis