Variability of contrail formation conditions and the implications for policies to reduce the climate impacts of aviation

www.elsevier.com/locate/trd

Transportation Research Part D 10 (2005) 269–280

Variability of contrail formation conditionsand the implications for policies to reduce the

climate impacts of aviation

Victoria Williams, Robert B. Noland *

Centre for Transport Studies, Department of Civil and Environmental Engineering, Imperial College London,

London SW7 2AZ, United Kingdom

Abstract

This paper describes an approach to balance the climate benefits of contrail reduction against the pen-

alties incurred when cruise altitudes are restricted. Altitude restrictions are targeted by selecting, for each

6-h period, the altitude that provides the greatest reduction in contrail for the lowest increase in carbon

dioxide emission. Calculations are for western Europe. This paper discusses the variability in contrail for-mation conditions in the region and presents contrail reductions and carbon dioxide emission increases

obtained with this optimised approach, which compare favourably with fixed altitude restrictions. A new

method is also developed to estimate contrail fractions within three-dimensional grids. Conclusions discuss

potential operational issues associated with a varying altitude restriction policy.

� 2005 Elsevier Ltd. All rights reserved.

Keywords: Contrail; Climate change; Cruise altitude; Air transport

1. Introduction

Continued growth in air travel means aviation fuel accounts for a growing share of global fossilfuel use and of anthropogenic CO2 emissions, despite efficiency gains from improvements in

1361-9209/$ - see front matter � 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.trd.2005.04.003

* Corresponding author. Tel.: +44 20 7594 6036; fax: +44 20 7594 6102.

E-mail address: [email protected] (R.B. Noland).

mailto:[email protected]

270 V. Williams, R.B. Noland / Transportation Research Part D 10 (2005) 269–280

engine and airframe design. These emissions represent only one of the mechanisms through whichaviation can influence the global climate. At high altitudes, in appropriate atmospheric condi-tions, the expanding aircraft exhaust triggers the formation of a contrail, which can last for severalhours and can spread to form cirrus cloud.

Evidence for the climate impact associated with contrails and cirrus formation continues togrow. During the near-complete shutdown of air traffic over the United States following Septem-ber 11th 2001, the contrails from just six military aircraft spread to form cirrus cloud coverageover 20,000 km2 (Minnis et al., 2002). Contrails and the cirrus clouds they form are consideredto make a substantial contribution to the climate impact of aviation. The Intergovernmental Panelon Climate Change (IPCC) special report on aviation estimated the impact of linear contrail onthe radiative balance of the atmosphere to be similar in magnitude to that of CO2 emitted by air-craft, and identified considerable uncertainties in assessing the impact of cirrus clouds (Penneret al., 1999). Recent research has improved the understanding of the impacts of both contrailand cirrus cloud, identifying a smaller contribution from linear contrail than estimated by IPCC,but a larger impact from aged contrail spreading to form extensive cirrus cloud. One study sug-gests this to be at least 10 times that of the CO2 from aviation (Mannstein and Schumann, 2003).

The total climate impact of aviation is currently unregulated. The Kyoto Protocol includes car-bon dioxide emissions from domestic aviation in national targets, but other impacts specific toaviation are not included in the protocol and will remain unregulated in the absence of additionalagreements. These include high-altitude emissions of NOx and the formation of contrails and cir-rus clouds. In addition, policies to limit climate impacts of international aviation have not yetbeen agreed. Proposals have been explored at the European and global level, many focussingon limiting the growth in carbon dioxide emissions with current discussions focussed on carbontrading schemes.

Cruise altitude changes have been explored as a policy option to reduce the climate impact bypreventing or reducing contrails (Williams et al., 2002, 2003) or by reducing the impact on ozoneof aviation NOx emissions (Gauss et al., 2003). Reductions in cruise altitude have also been iden-tified as a way to reduce the impact of stratospheric water vapour emissions from a potentialhydrogen fuelled aircraft fleet (Svensson et al., 2004). However, reducing cruise altitude raises anumber of issues. The first is that it forces aircraft to fly through denser air, below their most effi-cient altitudes. This increases the fuel burn rate, increasing the amount of carbon dioxide emitted aswell as airline costs. In addition, this increased fuel requirement either increases the take-off weightof the aircraft or reduces the pay-load. Changes in the aircraft speed raise additional operationalissues associated with journey time. Air space congestion is a further constraint on altitude restric-tion policies. This is particularly relevant in regions where the density of air traffic is already high.

Previously, Williams et al. (2002) identified a policy design for altitude restrictions based onmonthly mean atmospheric conditions for the European 5-states region and calculated the asso-ciated penalties for CO2 emission, journey time and airspace congestion. Here, we present an anal-ysis of the short-term variability in the atmospheric conditions for contrail formation in the sameregion and develop an improved altitude restriction policy, with the aim of minimizing some ofthe adverse impacts. The carbon dioxide penalties for altitude restriction previously presented(Williams et al., 2002) were based on fixed altitude restrictions applied for each month with theseverity of restriction calculated from one year of monthly mean temperature and humidity data.This paper uses 5 years of six-hourly data and considers contrail formation for January, April,

V. Williams, R.B. Noland / Transportation Research Part D 10 (2005) 269–280 271

July and October. Using instantaneous, rather than monthly averaged, temperature and humiditydata improves the representation of contrail formation.

2. Estimating variability in contrail coverage

A measure of contrail sensitivity is developed using a parameterisation of the maximum poten-tial contrail coverage combined with air traffic density data. This uses a 1-day sample of air trafficin the European 5-states region and is based on previous methods for calculating potential con-trail fraction (Sausen et al., 1998) and actual recorded movements of aircraft within 3-dimensionalgrid cells.

The method is an improvement over previous techniques. First, detailed flight profile data al-lows distance, rather than fuel burned, to be used as the measure of air traffic density. In this way,aircraft burning more fuel per kilometre are not over-represented in the distribution of calculatedcontrail coverage. A second distinction is that no attempt is made to scale the calculated measureof contrail coverage to observed contrail. Previous studies have calculated this scaling factor usingsatellite observations of contrail coverage over Europe (Bakan et al., 1994) with calculated con-trail coverage to calculate global fractional contrail coverage (Sausen et al., 1998). The calculatedcontrail sensitivity used here is simply a measure of the distance of linear contrail formed per kilo-metre of flight in this traffic sample.

Potential contrail fraction is determined from a parameterisation to reflect the sub-grid scalevariability of relative humidity and temperature and describes the fraction of a grid box in whicha contrail could form.

The calculations use a single day of air traffic data, with the traffic divided into four 6-h periods.Cumulative distance travelled through each 3-D grid box in each period is calculated. Here, it isassumed that contrail cover is not saturated, so that the contrail amount for a given set of atmo-spheric conditions is linearly dependent on the amount of air traffic. For each day of atmosphericdata analysed, the cumulative distance travelled is multiplied by the potential contrail fraction foreach of the four periods. Summing this product over all grid boxes provides an indication of thetotal amount of contrail for each time step (CC), as follows:

CCt;d ¼Xi¼1;I

PCF i;t;d �Xn¼1;N

xn;i;t

!ð1Þ

Here, I is the total number of grid boxes, PCFi,t,d is the potential contrail fraction calculated forgrid box i, at time t for day d, N is the number of aircraft in the traffic sample and xn,i,t is the dis-tance travelled by aircraft n through grid box i during the time t. This provides a measure of thetotal contrail coverage associated with this traffic sample for each period of atmospheric data,which is then divided by total distance travelled to obtain the contrail sensitivity.

This measure of contrail sensitivity must be used with caution. It does not imply that increasingair traffic kilometre in the sample by 50% would produce a 50% increase in contrail; the change intotal contrail amount would depend on the distribution of the additional air traffic. Contrail sen-sitivity is used here to adjust contrail coverage calculated in response to the diurnal cycle in airtraffic. This contrail sensitivity, used to explore the variability in contrail production, is distinct

Fig. 1. Flight routes included in the air traffic sample for the European 5-states region.


from the ratio of contrail reduction to carbon dioxide emission increase, which is used in thedesign of the altitude restriction policy.

The traffic sample describes one day of air traffic movements for the European 5-states region(Dowdall, 2001; Williams et al., 2002). Fig. 1 shows the air traffic routes. All flights passingthrough this area are included in the calculations. For flights beginning and ending outside of thissimulation area, the distance and contrail calculations include only the section of the route shownon Fig. 1. The distance travelled in each atmospheric data grid box is used as a measure of airtraffic density and is obtained using the Reorganised Air traffic control Mathematical Simulator(RAMS)1. This is a fast time simulator which allows detailed calculation of aircraft trajectories,taking into account their performance characteristics, which are specified using the Eurocontrolbase of aircraft data (BADA) (Nuic, 2000).

The RAMS model is an event-based simulator, describing the position of each aircraft when-ever an air traffic control event takes place. The profile for each flight is retrieved using time andposition data from this event list. The flight is divided into flight segments, each described by twoevents. The position and time data for these two events is used to calculate the (great circle) dis-tance travelled and allocate that distance to the appropriate grid box. Where the two events do notfall within the same grid box or period, the distance is assigned according to the first event. Typ-ically, the segment distance and duration are much less than the grid box resolution, so the impactof this assumption on the calculated distribution of air traffic is small.

This data for the density of the air traffic is used with potential contrail fractions calculated usingtemperature and relative humidity data from the NCEP-II reanalysis dataset (Kanamitsu et al.,2002), with a resolution of 2.5� latitude by 2.5� longitude and vertical levels at 400, 300, 250,200 and 150 hPa at 6 h intervals. Five years of data for January, April, July and October is usedto identify inter-annual and seasonal differences. The sum of the calculated contrail coverage over

1 http://www2.isa-software.com.

http://www2.isa-software.com


all grid boxes and all layers is used as a measure of the total contrail coverage in each period, anddivided by the sum of the corresponding distance travelled to obtain the contrail sensitivity.

Fig. 2 shows a time series of calculated contrail sensitivity for each month (4 records per day),with data for the 5-years displaced along the x-axis. For each month, the horizontal line shows themean for that calendar month over the 5-year period. This mean value is highest in January and

Fig. 2. Time series (6-hourly) data of calculated contrail per kilometre of air travel for the month and year indicated.

Fig. 3. Contrail kilometre per kilometre travel. Mean (+), mean ± 1 standard deviation (solid line) and range (dashed

line), shown for four times of day.


April, with values approximately double that for July. This seasonal cycle in contrail sensitivity isconsistent with the greater probability of contrail formation in winter than in summer identifiedby previous authors through both atmospheric modelling and data analysis (Sausen et al., 1998)and observations using automated detection of contrails from the Advanced Very High Resolu-tion Radiometer (AVHRR) (Mannstein et al., 1999).

For each month, there are periods characterised by high or low contrail sensitivity values main-tained over several days, while other periods show higher frequency variability. This has implica-tions for policies to reduce contrail formation and for attempts to characterise the full climateimpact of different aircraft-route combinations. Because of the variability in the probability ofcontrail formation along a route, the net climate impact of an aircraft operating along an identicalflight trajectory may vary dramatically from day to day.

The mean, standard deviation and range of calculated contrail sensitivity values over the fiveyears of data for each calendar month are shown in Fig. 3. Values for each 6-h time period areindicated separately, so the range (dashed line) and standard deviation (solid line) correspondsonly to atmospheric variability. Contrail sensitivity shows no strong diurnal cycle. Total contrailproduction is highest for the air traffic samples for the 6 h from 6.00 am and from noon, due to thedaytime peak in air traffic over the 5 states region. The day-to-day variability in calculated con-trail sensitivity is considerable. For each time period, the maximum contrail sensitivity isapproaching double the mean value.

3. Altitude restriction policy

The altitude restriction policy described here applies a single maximum cruise altitude restric-tion across the region analysed, set every 6 h according to atmospheric conditions. The altituderestriction is selected from 5 options (31,000 ft, 29,000 ft, 26,000 ft, 24,000 ft or no restriction),and is chosen to maximise the ratio of reduction in contrail to the additional fuel required.

The method described above was used to calculate the total contrail amount for each period forthe control traffic sample, and then repeated for each of 4 altitude restriction scenarios. Profiles


for the traffic in the altitude restriction scenarios were obtained using the RAMS model, whichtakes into account the performance characteristics of aircraft operating at different altitudes. Ori-gin and destination combinations, the distribution of departures throughout the day, the air trafficcontrol sectors and the type of aircraft for each trip were kept the same. Each cruise altituderestriction is associated with an increase in fuel burn as some aircraft are forced to fly less effi-ciently. The increase for each of these restricted cruise altitude scenarios has been previously eval-uated using the RAMS model in combination with BADA data for aircraft performance and fuelconsumption rates (Williams et al., 2002).

For each calendar month considered, Fig. 4 shows the mean, standard deviation and range ofcontrail sensitivity values obtained from 5 years of data. The four times of day are plotted sepa-rately, with overlapping symbols indicating little or no diurnal cycle. For both January and April,mean contrail sensitivity is highest for cruise altitudes restricted to 29,000 ft and 26,000 ft; apply-ing these altitude restrictions would in fact increase the average contrail produced for this mix ofair traffic. Cruise altitudes must be reduced to 24,000 ft to obtain a lower mean contrail sensitivitythan in the control run. For July and October, any of the cruise altitude restrictions reduces meancontrail sensitivity. Even where very low values of contrail sensitivity are obtained, the range ofvalues is generally high, indicating that for some periods altitude restrictions may dramaticallyincrease the probability of contrail formation.

The fuel increases associated with the altitude restrictions were calculated using outputs fromthe RAMS model, with detailed information on aircraft performance data taken from the BADAperformance tables (Nuic, 2000). Table 1 shows the percent fuel increases for each 6-h period, andfor the full day. These fuel increases apply to the total fuel used by all aircraft in the simulationarea. Some aircraft, particularly on shorter routes, already fly at or below the altitude restrictionsand so are unaffected.

The variable altitude restriction policy is designed to minimise the penalties incurred while max-imising contrail reduction by selecting a varying sequence of cruise altitudes. For each period, theselected altitude restriction gives the largest ratio of contrail reduction to carbon dioxide emissionincrease. For times when each of the altitude restriction scenarios would increase contrail cover-age, no restriction is applied. Fig. 5 shows the frequency of selection of each cruise altitude. The

Fig. 4. Contrail sensitivity calculated for the control simulation and 4 altitude restrictions (5 years of data: 6 h from

6.00 am).

Table 1

Percent increase in total fuel burn for altitude restriction scenario

Time of day FL310 FL290 FL260 FL240

00:00–05:59 3.25 5.14 8.67 1.12

06:00–11:59 1.44 2.43 4.79 6.49

12:00–17:59 1.42 2.42 4.68 6.63

18:00–23:59 1.44 2.55 5.14 7.09

00:00–23:59 1.67 2.81 5.34 7.28

Increases are shown for the four 6-h periods, and for the full traffic sample.

Fig. 5. Frequency of selection of altitude restrictions for the combinations of altitude restrictions.


restriction to 24,000 ft (FL240) is applied more often in January and April. In July, this mostextreme restriction is only rarely applied, except in 2000. As expected, restrictions in the summerare generally less severe than in winter and spring, but importantly the imposition of one of thealtitude restrictions is always preferred. This contrasts with the other months, where for somedays applying any of the altitude restrictions would increase the amount of contrail. It is alsointeresting to note that the restriction to 26,000 ft is never identified as the preferred option.

Fig. 6 shows the change in fuel and contrail produced for this variable approach along with theresults for fixed cruise altitude restrictions. These values are shown as percentage changes from the(unrestricted) control scenario. For the fixed altitude restrictions, the contrail change shown foreach month is averaged over the 5 years. The calculations do not include any seasonal signal inair traffic amount or distribution, so the effect on fuel burn of applying a fixed altitude dependsonly on the severity of the restriction imposed. For January and April, fixed monthly altituderestrictions can increase contrail production by up to 30% compared to the unrestricted run, ascontrail formation conditions may peak below the normal cruise altitude of some of the aircraftin the sample. Only the most extreme restriction (maximum cruise altitude 24,000 ft) offers con-

Fig. 6. Contrail increase against fuel increase for each combination of altitude restrictions. For fixed restrictions,

contrail change is averaged over 5 years.


sistent reductions in the average contrail amount. For July and October, mean contrail reductionsare always obtained, with the reduction increasing with the severity of the restriction.

On the same figure, changes in contrail and fuel for the variable altitude restrictions are plottedfor each month and year. Broadly, the variable policy consistently reduces the amount of contrailpredicted by the model by between 65% and 95%. For January and April, the contrail reductionsobtained for the variable restrictions approach those associated with an altitude restriction fixedat 24,000 ft, but with a smaller increase in fuel. For July and October, the contrail reductions aresmaller than those obtained for the most extreme fixed altitude restriction (which virtually elim-inates contrail), but compared to the fixed scenarios at 26,000 and 29,000 ft the variable restric-tions offer improved contrail reduction for similar fuel increases.

In addition to the fuel burn and air traffic congestion penalties associated with imposing lowercruise altitudes, this variable policy would present additional difficulties associated with the tran-sition between restrictions. These issues have not yet been fully explored using the air traffic sim-ulator, but some insight into the distribution of these problems can be gained by considering the

Table 2

Number of transitions between flight altitude restrictions

99–00 00–01 01–02 02–03 03–04

JAN 10 27 19 30 35

APR 21 21 14 14 25

JUL 37 45 24 35 29

OCT 42 39 43 40 17


number of transitions between cruise altitude restrictions. Table 2 shows the number of transitionsfor each month and year considered. For January, the maximum cruise altitude imposed changeson average 24.2 times (out of 124 6-h time periods). April has the most stable conditions, with anaverage of only 19 transitions. There are 34 and 36.2 for July and October respectively, represent-ing more than one change in cruise altitude restriction per day. Although the restrictions requiredare generally less severe for July and October (Fig. 5) implying fewer airspace congestion prob-lems, the additional airspace complexity could present a significant challenge to implementation.The operational and fuel burn issues associated with the transition between cruise altitude regimesare discussed further in the following section.

4. Operational implications of variable altitude restrictions

Applying variable altitude restrictions presents new challenges for the management of air traf-fic. The calculations assume instantaneous changes between maximum cruise altitudes. To gainsome insight into the operational issues, additional RAMS simulations were conducted. To testa smoother transition between altitude restriction regimes, each aircraft was allocated a maximumcruise altitude according to the time in which it enters the traffic simulation. Two simulations wereundertaken. The first, PM240, applied no restriction to traffic for the first 12 h, with aircraft enter-ing the simulation after midday allocated a maximum cruise altitude of 24,000 ft. The second sim-ulation (AM240) reversed this, applying a 24,000 ft restriction before midday, but not restrictingaircraft entering the simulation in the afternoon. These two simulations were compared with twoof the simulations described above: the control traffic sample and the most restricted of the fixedcruise altitudes, with a maximum of 24,000 ft imposed for the full day.

The number of conflict events is a simplified measure of the complexity of the air traffic. Theseevents occur when aircraft separation conditions are violated; two aircraft are in conflict wheneach occupies the 3-dimensional protection zone defined around the other. The number of con-flicts in the whole simulation region increases with the severity of the altitude restriction applied.However, as the definitions of air traffic control sector and centre boundaries are fixed, high-levelcentres can experience a reduction in conflicts as traffic diverts to lower altitudes. As the two tran-sition scenarios both represent conditions in which the extreme altitude restriction is applied forhalf a day, assuming instantaneous transition between the two regimes would suggest that thetotal number of conflicts for each centre over the day would be between that obtained for the con-trol scenario and that for the full day 24,000 ft restriction. However, preliminary analysis of theconflict data for these simulations suggests that exceptions may occur. For example, for theKarlsruhe Centre, the AM240 simulation identifies more conflict events in the hour from noonto 1.00 pm than either the control scenario or the full day 24,000 ft restriction. This suggests that,for this centre at least, the change in altitude restriction can induce additional conflicts. For thesame hour in the PM240 simulation, there are fewer conflicts than in either the control scenario orthe full day 24,000 ft restriction.

The simulations AM240 and PM240 also allow the impact of the assumption of instantaneoustransition between maximum cruise altitudes to be tested. As there is a gradual transition betweenregimes, the impact of the altitude restriction selected will be diluted until air traffic entering thesimulation at an earlier time has completed its route, either by reaching its destination airport or


by leaving the simulation area. Compared to the control run simulation, AM240 increases fuel by4% over the whole day. For the PM240 simulation, the fuel increase is 3.2%. Using the controland the 24,000 ft restriction simulations and summing a.m. data from one and p.m. data fromthe other, it is possible to calculate the total fuel burn that would occur if the transition betweenaltitude restriction regimes took place instantaneously. Increases of 3.7% and 3.5% for AM240and PM240 respectively are obtained.

Further calculations are required to evaluate the extent of the assumption of an instantaneouschange between restrictions on calculated contrail. These calculations will be limited to the dayswith atmospheric conditions consistent with this combination of altitude restrictions. Of the 615days in the policy design presented here (5 years of data for 4 calendar months), 3 follow the alti-tude restrictions for PM240, while 1 follows AM240. Extending the analysis to other months andyears would be likely to yield further dates for comparison. The effects of less extreme and morefrequent transitions in altitude should also be considered.

5. Conclusion

These results highlight the considerable variability in the production of contrail. This has strongimplications for policies to address the impact of aviation on climate. First, while applying a blan-ket altitude restriction could reduce contrail, if the restriction is incorrectly set the effect could be acontrail increase. Even where a fixed policy would reduce contrail, significant further reductionscould be obtained using an adaptive policy, allowing restrictions to be altered when contrail sen-sitivity at low altitude is unusually high. This presents significant technical challenges for air trafficmanagement and raises operational issues related to the difficulty to predict precise journey timesin advance. A policy optimising the altitude restrictions required could reduce the penalties byensuring that unnecessary or counter-productive flight restrictions were not imposed. One exam-ple of such a policy is presented, and is shown to offer improved contrail reduction with a smallerincrease in carbon dioxide than fixed altitude restriction options. Further improvements to thispolicy, such as increasing the range and resolution of the altitude restriction options, or settinga threshold for the minimum contrail reduction to be obtained could further reduce the contrailand/or the fuel penalty. Similarly, the operational impacts could potentially be reduced by select-ing cruise altitude over a longer period, such as a day, allowing changes in air space to take placewhen there is less traffic.

The variability in contrail also presents a challenge for any scheme involving either tradablepermits or penalties/incentives based on net climate impact. This would need to include a measureof actual contrail production attributable to individual air traffic movements if the polluter paysprinciple is to be applied effectively.

Acknowledgement

This work was supported by the Engineering and Physical Sciences Research Council throughthe Advanced Research Fellowship scheme. The authors are grateful to Ralf Toumi from theSpace and Atmospheric Research Group at Imperial College London for helpful discussions,


Assistance from the RAMS Plus support group at ISA Software is also gratefully acknowledged.Ray Dowdall at Eurocontrol provided input data for the RAMS Plus simulation. NCEP Reanal-ysis data was provided by the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado,USA, from their Web site at http://www.cdc.noaa.gov/.

References

Bakan, S., Betancor, M., Gayler, V., Grassl, H., 1994. Contrail frequency over Europe from NOAA-satellite images.

Annales Geophysicae—European Geophysical Society 12, 962.

Dowdall, R., 2001. 5 States Fast-Time Simulation, Eurocontrol Experimental Centre, Bretigny, France, Report No.

361.

Gauss, M., Isaksen, I., Grewe, V., Kohler, M., Hauglustaine, D., Lee, D., 2003. Impact of aircraft NOx emissions:

effects of changing the flight altitude. In: Proceedings of the AAC-Conference, June 30–July 3, 2003,

Friedrichshafen, Germany.

Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S.K., Hnilo, J.J., Fiorino, M., Potter, G.L., 2002. NCEP-DOE

AMIP-II reanalysis (R-2). Bulletin of the American Meteorological Society 83, 1631–1643.

Mannstein, H., Meyer, R., Wendling, P., 1999. Operational detection of contrails from NOAA-AVHRR data.

International Journal of Remote Sensing 20, 1641–1660.

Mannstein, H., Schumann, U., 2003. Observations of contrails and cirrus over Europe. In: Proceedings of the AAC-

Conference, June 30–July 3, 2003, Friedrichshafen, Germany, pp. 244–248.

Minnis, P., Nguyen, L., Duda, D.P., Palikonda, R., 2002. In: 10th Conference on Aviation, Range, and Aerospace

Meteorology Portland.

Nuic, A., 2000. Aircraft Performance Summary Tables for the Base of Aircraft Data (BADA) Revision 3.3,

Eurocontrol Experimental Centre, Bretigny, France, Note 18/00.

Penner, J.E., Lister, D.H., Griggs, D.J., Dokken, D.J., McFarland, M. (Eds.), 1999. Aviation and the Global

Atmosphere: A Special Report of Intergovernmental Panel on Climate Change Working Groups I and III.

Cambridge University Press, Cambridge.

Sausen, R., Gierens, K., Ponater, M., Schumann, U., 1998. A diagnostic study of the global distribution of contrails

part I: Present day climate. Theoretical and Applied Climatology 61, 127–141.

Svensson, F., Hasselrot, A., Moldanova, J., 2004. Reduced environmental impact by lowered cruise altitude for liquid

hydrogen-fuelled aircraft. Aerospace Science and Technology 8, 307–320.

Williams, V., Noland, R.B., Toumi, R., 2002. Reducing the climate change impacts of aviation by restricting cruise

altitudes. Transportation Research Part D: Transport and Environment 7, 451–464.

Williams, V., Noland, R.B., Toumi, R., 2003. Air transport cruise altitude restrictions to minimize contrail formation.

Climate Policy 3, 207–219.

http://www.cdc.noaa.gov/

Documents

Variability of contrail formation conditions and the implications for policies to reduce the climate impacts of aviation