Aviation makes a significant contribution to anthropogenic
climate forcing. The impacts arise from emissions of greenhouse gases,
aerosols and nitrogen oxides, and from changes in cloudiness in the
upper troposphere. An important but poorly understood component of this
forcing is caused by ‘contrail cirrus’—a type of cloud that consist of
young line-shaped contrails and the older irregularly shaped contrails
that arise from them. Here we use a global climate model that captures
the whole life cycle of these man-made clouds to simulate their global
coverage, as well as the changes in natural cloudiness that they induce.
We show that the radiative forcing associated with contrail cirrus as a
whole is about nine times larger than that from line-shaped contrails
alone. We also find that contrail cirrus cause a significant decrease in
natural cloudiness, which partly offsets their warming effect.
Nevertheless, net radiative forcing due to contrail cirrus remains the
largest single radiative-forcing component associated with aviation. Our
findings regarding global radiative forcing by contrail cirrus will
allow their effects to be included in studies assessing the impacts of
aviation on climate and appropriate mitigation options.
Aviation-induced cloudiness consists of contrail cirrus (of
which a subset is line-shaped) and of changes in the occurrence or
properties of natural cirrus arising from both the presence of contrail
cirrus and increased ice-nuclei concentrations in the upper-troposphere
due to aircraft soot emissions. Observations indicate that these changes
may have a significant effect on cirrus cloudiness1.
Radiative forcing—a measure of the radiative imbalance of the
atmosphere caused by a particular forcing agent—due to aircraft-induced
cloudiness has been estimated from observed trends in cirrus cloudiness
to range approximately between 10 and 80mWm−2 for the year 2005 (refs 2, 3, 4). Contrail
cirrus initially form behind cruising aircraft as line-shaped contrails
and transform into cirrus-like clouds or cloud clusters in favourable
meteorological conditions, occasionally covering large horizontal areas5, 6, 7. They have been tracked for up to 17h in satellite observations6.
They remain line-shaped, and therefore easily distinguishable from
natural cirrus, for only a fraction of their lifetime. The impact of
aircraft soot emissions on cirrus in the absence of contrails depends on
the ice-nucleating properties and the ice-active number concentration
of soot-particle emissions. Both of these parameters are highly
uncertain8,
and whereas the impact of aircraft soot on cirrus has been shown to be
statistically significant in terms of cirrus ice-particle-number
concentrations9 in a climate model, at present this can not be shown for radiative forcing10. Contrail
cirrus are composed of ice crystals that—similarly to natural
cirrus—reflect solar short-wave radiation and trap outgoing long-wave
radiation11. For fixed ambient conditions, their radiative effect is mainly determined by their coverage and optical depth12. Contrail cirrus form and persist in air that is ice-saturated13, 14, whereas natural cirrus often require high ice supersaturation to form15.
This implies that in a substantial fraction of the upper troposphere,
contrail cirrus can persist in supersaturated air that is cloud-free16, 17, thus increasing high cloud coverage1, 11, 18.
Remote-sensing studies have estimated line-shaped-contrail coverages as
large as a few per cent in regions in which the levels of air traffic
are high19, 20, 21.
The coverage due to contrail cirrus is as yet unknown because they are
difficult to distinguish from natural cirrus in satellite observations11. The global radiative forcing due to line-shaped contrails has been estimated to amount to 10mWm−2 (6–15mWm−2) for 2005, with a low level of scientific understanding4. The global radiative-forcing estimates for line-shaped contrails22
rely on the scaling of simulated contrail-formation frequency to an
observed regional contrail coverage. Assuming the scaling coefficient to
be spatially and temporally constant, global contrail coverage can be
inferred16, 23. This methodology is not suited to studying the effect of contrail cirrus24. Present studies have been unable to provide a best estimate for the contrail-cirrus radiative forcing. We have developed a process-based contrail-cirrus module17, 25 (CCMod) in a global climate model, ECHAM4 (ref. 26; see Methods),
which enables the simulation of the life cycle of persistent contrails.
Contrail cirrus exist alongside and interact with natural clouds and,
depending on their overlap with natural clouds, can increase overall
cloud coverage. Here, we use our contrail-cirrus module to simulate
contrail-cirrus coverage, the associated radiative forcing and resulting
changes in the natural cirrus clouds.
Coverage due to contrail cirrus and that due to young contrails (defined here as up to 5h old) are shown in Fig. 1.
Coverage due to young contrails may be compared to the coverage
inferred from satellite observations, because young contrails are most
likely to be still line-shaped. Using a maximum random overlap scheme,
contrail-cirrus coverage (Fig. 1a) amounts to several per cent over large parts of the Northern Hemisphere. Coverage due to persistent young contrails reaches 2% over Europe and exceeds 1% over large parts of the US (Fig. 1b), which is in line with earlier estimates of line-shaped-contrail coverage obtained by an independent technique22. Over central Europe, contrail-cirrus coverage is largest, reaching up to 10%.
Although the level of air traffic over the east coast of northern
America is as large as over central Europe, contrail-cirrus coverage in
the former region is lower, reaching 6%. It is mainly the coverage due to contrails older than 5h
that is smaller over the USA than over Europe (see below). This is
mainly caused by the fact that many old contrails are advected into
central Europe from the North Atlantic flight corridor, an area
favourable to contrail formation and persistence, whereas there is
little contrail advection towards the eastern part of the US.
Furthermore, warmer temperatures over the US reduce the probability of
contrail formation so that coverage due to young contrails is also
slightly smaller than over central Europe. Over the east coast of
southeast Asia, the area in which air traffic density is third largest,
young-contrail coverage reaches 0.2% and contrail-cirrus coverage exceeds 1%. Globally averaged contrail-cirrus and young-contrail coverage amount to 0.61% and 0.07%,
respectively. Contrail-cirrus coverage is therefore approximately nine
times larger than coverage due to young contrails alone.
Figure 1: Contrail-cirrus and young-contrail coverage for the year 2002 as simulated by ECHAM4–CCMod.
a,b, Coverage due to contrail cirrus (a) and due to persistent young contrails with ages of up to 5h (b), considering young contrails and contrail cirrus of any optical depth. c, Coverage
due to visible contrail cirrus with a solar optical depth >0.02.
Coverages have been calculated by assuming maximum random overlap among
contrails or contrail cirrus alone. Only part of the
contrail/contrail-cirrus coverage leads to an increase in overall cloud
coverage.
Contrail-cirrus coverage (Fig. 1a) exceeds the coverage due to young contrails (Fig. 1b) significantly. The fraction of the total contrail-cirrus coverage that is due to young contrails (Fig. 2)
amounts to 0.11 globally and is spatially very variable. In the area of
the North Atlantic flight corridor only a small fraction (0.1–0.15) of
the contrail-cirrus coverage is due to young contrails. The fraction of
young contrails from the contrail-cirrus coverage lies between 0.15 and
0.25 over central Europe and between 0.25 and 0.4 over the eastern US.
Over southeast Asia, a large part of contrail-cirrus coverage is due to
young contrails.
Figure 2: Fraction of contrail-cirrus coverage that is identified as coverage due to young (age ≤5h) contrails.
A large fraction of contrail cirrus is optically very thin (solar
optical depth <0.02) and can therefore neither be detected by a
satellite nor seen with the human eye from the ground27. Owing to their abundance, the radiative effect of such optically thin contrail cirrus may not be negligible28, similar to the effect of optically thin natural cirrus29.
The fraction of optically thin contrail cirrus is larger in colder
areas (farther north), where ice-supersaturation frequency, at the main
flight level (230hPa),
is usually larger and the water content of the air is lower. This means
that the contrail-formation criterion is more frequently met, but
contrails forming and persisting in those areas are less likely to
become optically thick. Therefore, the fraction of optically thin
contrail cirrus is larger over Europe than over the US and especially
over southeast Asia. When considering only contrail cirrus that exceed a
threshold optical depth of 0.02, the coverage due to these visible
contrail cirrus is fairly similar over the eastern US and central Europe
and amounts to above 3% and up to 4%, respectively (Fig. 1c).
Over southeast Asia, most contrail cirrus are visible because of the
higher specific water content, and coverage due to visible contrail
cirrus amounts to 0.5%, exceeding locally 1%. Coverage due to visible young contrails reaches 1% over the southeast US and over central Europe and exceeds 0.2% over southeast Asia25. Globally, coverage due to visible contrail cirrus amounts to 0.23%, whereas coverage associated with visible young contrails amounts to 0.04%.
Stratosphere-adjusted radiative forcing due to both young
contrails and contrail cirrus was calculated online within the climate
model (see Methods).
The former estimate serves solely for comparison with previous results.
As CCMod does not allow the calculation of the optical depth of young
contrails alone, we assume that the optical depth of young contrails
(averaging over 5h) is
equal to that of contrail cirrus (averaging over all ages). There are no
data available to estimate the difference in optical depth between
young contrails and contrail cirrus, but we conjecture that this
assumption is likely to lead to an underestimation of young-contrail
optical depth and the associated net radiative forcing. Globally,
the long-wave radiative forcing due to contrail cirrus (after correcting
the scattering component of the long-wave forcing from the model30) amounts to 47.1mWm−2 and short-wave radiative forcing to −9.6mWm−2, resulting in a net radiative forcing of 37.5mWm−2.
This includes the effect of line-shaped contrails. Globally averaged
contrail-cirrus optical depth is 0.05. Net radiative forcing of contrail
cirrus (Fig. 3a) reaches values larger than 300mWm−2
over the eastern US and central Europe. Over most of the US, Europe,
over the North Atlantic flight corridor and also over parts of southeast
Asia, net radiative forcing exceeds 100mWm−2. Over much of the northern mid-latitudes contrail-cirrus radiative forcing exceeds 30mWm−2.
Maxima in radiative forcing are found in areas of maxima in
contrail-cirrus coverage, but radiative forcing is enhanced in areas
with large contrail-cirrus optical depth (Fig. 3b).
This means that for a fixed contrail-cirrus coverage, radiative forcing
is larger over southeast Asia than in the northern mid-latitudes and
slightly larger over the eastern US than over central Europe or the
North Atlantic flight corridor.
Figure 3: Contrail-cirrus radiative forcing and optical depth at 250hPa for the year 2002.
Net radiative forcing (without correcting for long-wave scattering) (a) and solar optical depth (b) of contrail cirrus.
Globally the long-wave radiative forcing due to young contrails amounts to 5.5mWm−2.
This estimate is in line with an earlier estimate of
line-shaped-contrail long-wave radiative forcing using the same model
but an independent technique22.
The latter estimate is slightly lower because the randomly overlapped
visible line-shaped-contrail coverage in the earlier study (0.06%) is slightly smaller than our visible young-contrail coverage using the same overlap assumption (0.07%). Short-wave radiative forcing due to young contrails amounts globally to −1.2mWm−2. This value is about 50%
larger than the value we obtain when neglecting the diurnal cycle of
air traffic, resulting from the fact that most flights are daytime
flights31, 32. Net radiative forcing due to young contrails amounts globally to 4.3mWm−2, which is at the lower bound of the range of radiative-forcing estimates for line-shaped contrails3, 31.
This may imply that our contrail-cirrus radiative forcing constitutes a
low estimate as well. Differences between our short-wave and long-wave
forcing estimates and those in the literature are likely to be due to
differences in: (1) the spatial and temporal distribution of contrail
coverage and its optical depth resulting from the differences in the
parameterization schemes; (2) the background cloud fields and their
overlap with the contrails; (3) the flight inventories and their
reference years used. The global net radiative forcing of
contrail cirrus is roughly nine times that of young contrails, making it
the single largest radiative-forcing component connected with aviation.
It is important to note that contrail cirrus have a much shorter
lifetime than long-lived greenhouse gases. This difference in lifetime
influences the relative importance of contrail cirrus and other forcing
agents for climate change when estimating their impact for remote time
horizons33. The uncertainty in the spreading rate effective in ECHAM4–CCMod (ref. 25) introduces an uncertainty in the estimate of contrail-cirrus net radiative forcing of ±5mWm−2.
The sensitivity of contrail-cirrus radiative forcing to optical depth
and ice-particle shape of contrail cirrus, to the flight inventory and
to the model’s radiation code is likely to be similar to that of
line-shaped contrails. Therefore, we estimate that the uncertainty
related to the first two variables amounts in ECHAM4 to an uncertainty
in contrail-cirrus radiative forcing of ~25% (ref. 34) and ~15% (ref. 35), respectively. The sensitivity of global radiative forcing to the flight inventory is probably small35. Furthermore, it has been shown that the radiative response due to contrails varies by ±22% around a multi-model mean36.
An estimate of the combined uncertainty of radiative forcing would need
to take into account the interdependence between the different
uncertainties.
Contrail cirrus change the water budget of the surrounding
atmosphere and therefore can have an impact on natural clouds. Water
vapour that is deposited on ice particles within contrail cirrus is not
available for formation and deposition in natural cirrus any longer.
Therefore, contrail cirrus have the potential to modulate the optical
properties of natural clouds, delaying their onset and replacing them,
which may partly offset the direct climate impact of contrail cirrus.
Virtually nothing is known about those cloud and humidity changes. We
have carried out long-term integrations using ECHAM4–CCMod with and
without prescribing air traffic. At the main flight level at 230hPa,
where we would expect a maximum impact of contrail cirrus on the
natural-cloud coverage due to the maximum in contrail-cirrus coverage,
the two simulations exhibit differences in natural-cirrus coverage in
the main traffic areas (Fig. 4a). Reductions in the natural-cirrus coverage of around 2%
can be found over northeastern Europe and over the east coast of the
US. Over the western part of the main transatlantic air-traffic route,
off the west coast of northwest Africa and west of the main
contrail-cirrus area of southeast Asia, reductions range between 1 and
1.5%. These changes are different from zero at the 95%
significance level (when taking into account serial correlations),
whereas the local decrease over southeast Asia is mostly significant
only at the 90% level. Furthermore, a
statistically significant increase in cirrus coverage can be found over
and north of eastern Siberia. In those areas, the 230hPa
level lies frequently within the stratosphere. Owing to a cold and
moist bias in the polar lower stratosphere, common to many climate
models37,
the model has a higher sensitivity to the moisture input than would be
expected in nature. Therefore, we will not discuss this signal any
further. Maxima of decrease in natural-cirrus coverage lie consistently
downstream of the maxima in contrail-cirrus coverage (compare with Fig. 1a). The change in cirrus coverage, calculated by vertically overlapping all ice clouds in a column (Fig. 4b),
confirms that the natural-cirrus coverage is decreased owing to the
presence of contrail cirrus. The decrease is slightly less significant
owing to the large natural variability in cloud coverage. We note that
over the North Pacific flight corridor natural-cirrus coverage is also
significantly changed and that the increase in cirrus coverage over
Siberia at 230hPa does not translate into an increase in overlapped cirrus coverage in the area.
Figure 4: Change in natural-cirrus coverage due to the presence of contrail cirrus.
Change in natural-cirrus coverage at the main flight level of 230hPa (a) and when overlapping all ice clouds within one vertical column (b). Dotted lines indicate the 90% confidence interval and solid lines the 95% confidence interval.
Locally, the decrease in natural-cirrus coverage (over Europe and the US) amounts to up to 10% of the natural-cirrus coverage or up to 20%
of the contrail-cirrus coverage. Furthermore, in the main
contrail-cirrus areas of North America and Europe, the optical depth of
natural clouds is significantly (at the 95% significance level) reduced by up to 10% owing to the presence of contrail cirrus. Both
changes in natural-cirrus coverage and optical depth exert a negative
net radiative forcing (cooling), partly offsetting the positive net
radiative forcing due to contrail cirrus. The large natural variability
in albedo, sea surface temperature, natural clouds and so on, all of
which influence the radiative fluxes in the atmosphere, impedes the
isolation of the relatively small change in natural-cirrus radiative
forcing. Assuming that the decrease in natural-cirrus coverage amounts
to approximately one fifth of the global contrail-cirrus coverage (as we
have found a maximum of 2% (1.5%) decrease of natural-cirrus coverage downstream of the areas of 10% (6%)
contrail-cirrus coverage over Europe (US)), the feedback due to this
change in natural-cirrus cloudiness would induce a cooling of
approximately a fifth of contrail-cirrus radiative forcing, that is −7mWm−2. This estimate is very uncertain and further work is needed to more reliably quantify the feedback.
We report a model-based estimate of the global climate
impact of contrail cirrus, comprising not only young or line-shaped but
also aged, irregularly shaped contrails, and including resulting changes
in cirrus cloudiness. Defining the radiative forcing due to
contrail-induced cloudiness (CIC) as the contrail-cirrus radiative
forcing offset by the natural-cloud feedback results in a radiative
forcing by CIC of about 31mWm−2.
Net radiative forcing due to CIC constitutes one of the largest single
aviation-related radiative-forcing components. This radiative forcing
due to CIC together with the timescale on which the climate impact is to
be evaluated or reduced needs to be considered33 in aviation climate-impact assessments or mitigation studies, respectively. Besides
the uncertainty due to the treatment of contrail cirrus, our
radiative-forcing estimates are also affected particularly by
uncertainties related to the model’s representation of
upper-tropospheric humidity and clouds. Clouds are influenced by
small-scale processes that cannot be resolved by a large-scale climate
model and which therefore need to be parametrized. The representation of
clouds is a major source of uncertainty in climate simulations4. The same problems also affect the representation of contrail cirrus. The
uncertainty of the radiative-forcing estimates should be evaluated from
independent studies based on different models and contrail-cirrus
parameterizations. Reducing the uncertainty in the evaluation of
contrail-cirrus radiative forcing requires more and better observational
data sets24, 38.
In the long term, progress in this research area requires advanced
representation of natural clouds and humidity in climate models and
appropriate data sets for their validation39, 40.
The contrail-cirrus module, CCMod, introduces a new cloud
class ‘contrail cirrus’ in the global climate model ECHAM4. It is based
on a prognostic treatment of fractional coverage, length and ice water
mixing ratio of contrail cirrus25. The processes controlling contrail-cirrus coverage and properties, which are contrail formation below a threshold temperature14,
advection, spreading and water deposition, sublimation and
precipitation, are parametrized physically consistent with the
parametrization of natural clouds25.
Of the flight distance, only a fraction (given by the supersaturated
area fraction) results in persistent contrails. CCMod simulates the life
cycle of those persistent contrails. Contrails are advected by the wind
field and remain in (and are limited by) the ice-supersaturated
fraction of a grid box, assuming that persistent contrail cirrus
predominantly form in large persistent ice-supersaturated areas, such as
prefrontal areas, in which they remain for a long time. Supersaturated
areas are inferred from the assumptions of subgrid-scale variability
given by the cloud scheme17.
Contrail cirrus spread proportional to the vertical wind shear and
their vertical extent. In nature, the vertical extent is dependent on
ice-particle sedimentation and is limited by the thickness of a
supersaturated layer. After 1h, the contrail’s vertical extent is set in CCMod to the model’s layer depth, approximately 700m, which is roughly in line with observations41.
Contrails dissipate as their ice water content is reduced by
sublimation and precipitation. Within the contrail-cirrus cloud class,
fractional coverage and length of young contrails (up to 5h
old) are tracked independently, allowing the analysis of the coverage
due to young contrails for purely validational purposes. The ice water
content due to young contrails has not been tracked independently,
prohibiting the analysis of the optical depth of young contrails. The ECHAM4 diagnostic cloud-coverage scheme is relative-humidity based and the cloud water content is prognostic26.
Cloud particle fall speeds are dependent on the cloud water content.
The model’s water budget was changed to accommodate for the new cloud
class25,
enabling the simulation of the competition for available water vapour
between natural and contrail cirrus. Water vapour deposition,
sublimation, precipitation and optical depth of natural cirrus and
contrail cirrus are dependent on their respective ice water content.
CCMod has been evaluated using satellite and in situ measurements of ice supersaturation (ref. 17 and N. Lamquin et al., manuscript in preparation) and regional observations of line-shaped-contrail coverage25.
As only observational data sets of line-shaped contrails and none of
contrail cirrus are available, coverage and optical properties of
contrail cirrus could not be validated. Stratosphere-adjusted
radiative forcing has been calculated as a difference between different
calls of the radiation scheme at each time step in a model run42,
allowing the online calculation of radiative forcing due to contrail
cirrus. For the radiation calculations, natural clouds and contrail
cirrus have been randomly overlapped in the vertical at each model time
step, except when clouds existed in neighbouring model levels, in which
case clouds were stacked above each other (maximum random overlap). This
allows natural clouds and contrail cirrus to overlap each other in the
vertical. The coverage due to contrail cirrus shown in Fig. 1
was calculated by assuming maximum random overlap among contrail cirrus
alone. Only part of this coverage leads to an increase in overall cloud
coverage. Simulations have been conducted using an hourly resolved version of the global air traffic inventory AERO2k (ref. 43) for the year 2002. Integrations of 10 and 35yr with the ECHAM4–CCMod climate model (using a time step of 30min,
a horizontal resolution of T30 and 39 vertical levels) have been
carried out to estimate contrail-cirrus coverage and radiative forcing
and the feedback of contrail cirrus on natural clouds.
Schumann, U. & Wendling, P. in Air Traffic and the Environment—Background, Tendencies and Potential Global Atmospheric Effects (ed. Schumann, U.) 138–153 (Lecture Notes in Engineering, Springer, 1990).
Haywood, J. M.et al. A case study of the radiative forcing of persistent contrails evolving into contrail-induced cirrus. J. Geophys. Res.114, D24201 (2009).
Kärcher, B., Möhler, O., DeMott, P. J., Pechtl, S. & Yu, F.Insights into the role of soot aerosols in cirrus cloud formation. Atmos. Chem. Phys.7, 4203–4227 (2007).
Hendricks, J., Kärcher, B., Lohmann, U. & Ponater, M.Do aircraft black carbon emissions affect cirrus clouds on the global scale?Geophys. Res. Lett.32, L12814 (2005).
Liu, X., Penner, J. E. & Wang, M.Influence
of anthropogenic sulphate and black carbon on upper tropospheric clouds
in the NCAR CAM3 coupled to the IMPACT global aerosol model. J. Geophys. Res.114, D03204 (2009).
Koop, T., Luo, B. P., Tsias, A. & Peter, T.Water activity as the determinant for homogeneous ice nucleation in aqueous solutions. Nature406, 611–614 (2000).
Meyer, R., Mannstein, H., Meerkötter, R., Schumann, U. & Wendling, P.Regional radiative forcing by line-shaped contrails derived from satellite data. J. Geophys. Res.107, 4104 (2002).
Palikonda, R., Minnis, P., Duda, D. P. & Mannstein, H.Contrail coverage derived from 2001 AVHRR data over the continental United States of America and surrounding areas. Meteorol. Z.14, 525–536 (2005).
Marquart, S., Ponater, M., Mager, F. & Sausen, R.Future development of contrail cover, optical depth, and radiative forcing: Impacts of increasing air traffic and climate change. J. Clim.16, 2890–2904 (2003).
Ponater, M., Marquart, S. & Sausen, R.Contrails in a comprehensive global climate model: Parameterization and radiative forcing results. J. Geophys. Res.107, 4164 (2002).
Sassen, K. & Cho, B. S.Subvisual–thin cirrus lidar dataset for satellite verification and climatological research. J. Appl. Meteorol.31, 1275–1285 (1992).
McFarquhar, G. M., Heymsfield, A. J., Spinhirne, J. & Hart, B.Thin and subvisual tropopause tropical cirrus: Observations and radiative impacts. J. Atmos. Sci.57, 1841–1853 (2000).
Stuber, N., Forster, P., Rädel, G. & Shine, K.The importance of the diurnal and annual cycle of air traffic for contrail radiative forcing. Nature441, 864–867 (2006).
Kärcher, B., Burkhardt, U., Ponater, M. & Frömming, C.Importance of representing optical depth variability for estimates of global line-shaped contrail radiative forcing. Proc. Natl Acad. Sci. USA107, 19181–19184 (2010).
Zhang, M. H.et al. Comparing clouds and their seasonal variations in 10 atmospheric general circulation models with satellite measurements. J. Geophys. Res.110, D15S02 (2005).
We thank M. Ponater for providing us with a code for
calculating stratosphere-adjusted radiative forcing and for comments and
U. Schumann for the diurnal cycle of air traffic. This work was carried
out within the DLR project ‘Climate compatible air transport system’.
![Fraction of contrail-cirrus coverage that is identified as coverage due to young (age [le]5[thinsp]h) contrails.](http://www.nature.com/nclimate/journal/v1/n1/carousel/nclimate1068-f2.jpg)
![Contrail-cirrus radiative forcing and optical depth at 250[thinsp]hPa for the year 2002.](http://www.nature.com/nclimate/journal/v1/n1/carousel/nclimate1068-f3.jpg)
![Fraction of contrail-cirrus coverage that is identified as coverage due to young (age [le]5[thinsp]h) contrails.](http://www.nature.com/nclimate/journal/v1/n1/images_article/nclimate1068-f2.jpg)
![Contrail-cirrus radiative forcing and optical depth at 250[thinsp]hPa for the year 2002.](http://www.nature.com/nclimate/journal/v1/n1/images_article/nclimate1068-f3.jpg)
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.