Contrails seen a plane

Contrails - when do we see them from satellites?

May, June, December 2020 and January 2021

Contrails seen a plane
Contrails seen a plane

Looking at the contrails in satellite images and investigating supportive atmospheric conditions.

Last Updated

08 February 2021

Published on

02 February 2021

By Natasa Strelec Mahovic (EUMETSAT) and Ivan Smiljanic (CGI)

Contrails or condensation trails form when the water vapour, a product of exhaustion in the aircraft motor, mixes with the freezing cold air at cruising altitude. The water vapour condenses or instantly freezes into ice crystals, forming an elongated ice cloud, with cloud base usually at heights from 7000 to 12000 m above ground.

Looking from the Earth's surface, contrails are visible by far more frequently then from the satellites. One reason is the spatial resolution of the satellite sensors, especially the ones onboard geostationary satellites, which cannot distinguish features of very small dimensions. The other reason is the duration of these clouds, which can be shorter than the interval between two satellite scans, leaving the contrails undetected by satellite sensors. However, sometimes the contrails become broad and persistent enough to be seen in the satellite images, even for many hours.

Contrails in satellite images

Since contrails are ice clouds, at high altitudes, they appear in the satellite image the same as cirrus clouds. Therefore, to see them in the satellite images, the best channels and combinations are the same ones that are used to distinguish thin cirrus clouds, i.e. window channels 8.7, 10.8 and 12.0 micron. Since cirrus clouds are optically thin (semi-transparent), in the window channels also the information from the ground is contained, which makes the measurement of the cloud temperature difficult in singe channels. Therefore, difference or combination of channels is better for providing quantitative information. Contrails can most clearly be seen in brightness temperature differences 12.0-10.8 micron and, alternatively, 10.8–8.7 micron and 3.9–10.8 micron diff images (Figure 1).

Brightness temperature difference, 16 Dec 2020
Figure 1: Meteosat-11 16 December 2020, 17.30 UTC. Contrails north of Canary Islands as seen in 10.8-12.0 (left) and 8.7-10.8 (right) micron brightness temperature difference (BTD) images. Left image BTD ranges from -6 to +1 K, right image from -3 to +5 K.

The reason for this is difference is in the absorption on ice particles between these channels, where difference in absorption between 12.0 and 10.8 micron channel is larger for contrail clouds than between 10.8 and 8.7 micron channels (Figure 2).

Absorption graph
Figure 2: Absorption of water (blue curve) and ice (orange curve) in different parts of spectra. Absorption on ice in 8.7, 10.8 and 12.0 micron channels is marked by arrows.
(Source: MSG Interpretation Guide)

Comparing the same BTD ranges and colour scheme for GOES-16 satellite channels shows that the contrast between contrails and background is much better for 12.3–10.4 micron difference (Figure 3). The values for the contrails for 10.4–8.5 micron difference are slightly positive due to higher positive values of the surface.

Brightness temperature comparison

Brightness temperature difference (BTD) 10.4-8.5 micron compare1

Figure 3: Brightness temperature difference (BTD) images: 12.3-10.4 micron (left) and 10.4-8.5 micron (right). Groups of contrails are encircled in cyan. GOES-16 imagery, centered over Texas, 16 January 2021, 22:00 UTC. Associated BTD 12.3-10.4 animation from 12:10 to 23:50 UTC on the same day.

In RGB combinations composed with BTD 12.0-10.8 or BTD 10.8-8.7 (or both), in particular the Dust RGB or 24-hour Microphysics RGB (see the EUMeTrain RGB quick guides), contrails are well distinguished. In 24-hour Microphysics RGB contrails appear in shades from dark-blue, over dark violet to black colours, depending on the background emission and optical thickness (Figure 4).

24h microphysics RGB four panel
Figure 4: Examples of Meteosat 11 24-hour Microphysics RGB images showing contrails.

On some days, contrails can be observed for many hours and can expand over hundreds of kilometres (Figure 5).

Figure 5:  Meteosat 11 Dust RGB images from 22 June 2020, 04:30 UTC to 23 June 2020, 15:45 UTC, showing the persistence of numerous contrails first over Spain and Bay of Biscay, and later over France.

If contrails form a longer lasting and thicker high cloud, they can be seen also in the IR 10.8, WV 7.3 images, as well as Airmass RGB, and, sometimes, also in HRV images, but to lower extent (Figure 6). With solar channels contrails are much less apparent, due to their transparency, so it is easier to detect them over uniform water bodies, or through their shadows cast over uniform background (especially underlying stratiform clouds, sometimes also ground).

Meteosat image comparison

Meteosat-11 HRV Clouds RGB compare1

Figure 6: Contrails as seen in Meteosat-11 WV 7.3 micron image compared to HRV Clouds RGB image on 23 June 2020, 09:00 UTC.

This animation shows how contrails are seen not only in BTD 12.0-10.8 but also in BTD 6.2-7.3 and in single WV channels 6.2 and 7.3 due to contrast of high-level ice cloud with warmer background.

The Meteosat-11 Dust RGB animation (Figure 7) shows contrail formation, mostly over England, seen by geostationary satellite. When looking closer on these contrails, through advanced resolution of polar-orbiting instruments (Figure 8) one can observe that not all the contrails were resolved by geostationary imagery.

Figure 7: Meteosat-11 Dust RGB, 10 April 06:15-18:00 UTC

Also, polar-orbiting instrument has lower scanning angle at these latitudes, hence the observed position of contrails is closer to reality. With advanced resolution of VIIRS instrument (375m, vs 3000m for SEVIRI) it is possible to determine rough width of the observed contrails (some being >25 km wide). Depending also on optical thickness, minimum width of contrail detected by SEVIRI instrument is roughly 1 km. Figure 8 reveals, again, the fact that visible imagery (True Colour RGB) performs good detection of contrails over less reflective, uniform background (water), whilst IR-based imagery (Dust RGB) has better resolving power in respect to various contrail backgrounds.

Suomi-NPP and Meteosat-11 image comparison

Meteosat-11 Dust RGB compare1

Figure 8: Contrail comparison over the similar domain, similar time, high-resolution visible LEO imagery vs 'standard' infra-red GEO imagery. Suomi-NPP VIIRS True Coluor RGB at 375m (left) vs Meteosat-11 SEVIRI Dust RGB at 3km (right), 10 April 2020, 12:30 UTC.

Derived satellite products, such as NWC SAF products, reveal some characteristics of the contrails that are not directly seen in the RGB images. The Cloud Type product shows that contrails are classified as high semi-transparent clouds, thin or moderately thick (Figure 9, left) which are (in this case) mostly located at the heights between 8 and 12 km (Figure 9, right).

NWC SAF Cloud Type Product
Figure 9: NWC SAF CT Cloud Type (left) and CTTH Cloud Top Altitude (right) for 23 June 2020, 10:30 UTC.

Atmospheric conditions that support contrails

The loop of 24-hour Microphysics RGB images in Figure 10 shows the persistence of the contrails in the area with relative humidity at 250 hPa larger than 80%. In all investigated cases it was found that numerous and persistent contrails in the satellite images are closely related to humidity content, i.e. relative humidity, in the upper tropospheric layers, in particular around 250 hPa to 150 hPa.

Figure 10: Meteosat 11 24-hours Microphysics RGB, overlaid by ECMWF relative humidity field at 250 hPa (RH 250hPa), from 23 June 2020, 03:00 UTC to 23 June 2020, 18:00. Thick blue lines denote the areas with RH 250hPa > 80% and thin blue lines stand for RH 250hPa = 100%. Credit: EUMeTrain.

Looking at the vertical profile of the humidity, in the areas where larger number of contrails is found in the satellite image, it can be seen that there is a layer of high relative humidity values (moist layer) in upper troposphere (300-200 hPa) with a maximum of relative humidity around 250 hPa. At the same time, as seen in Figure 11, mid troposphere beneath that moist layer is rather dry.

24h microphysics with ECMWF equivalent potential temperature
Figure 11: Left top: Vertical cross section of ECMWF equivalent potential temperature (black lines) and relative humidity (brown lines, up to 70% and green lines, above 80%), Left bottom: Cross section of equivalent potential temperature (black lines) and temperature (red above 0 C and blue below 0 C): Right: Meteosat-11 24-hour Microphysics RGB, 23 May 2019, 06:00 UTC, showing the location of the cross-section across the area with contrails over France.  Source: EUMeTrain

Comparison of the cross-sections across the area where the contrails are seen to the one across the area without any contrails show that the humidity distribution in the vertical is very different, with high upper-level humidity accompanying the contrails and very dry air in the adjacent area without them (Figure 12).

Comparison of the vertical cross-sections of relative humidity
Figure 12: Comparison of the vertical cross-sections of relative humidity in the area where contrails are observed (above) to the cross-section in contrails-free area. Notice the lack of upper-tropospheric moist layer in the lower cross-section, Meteosat-11 24-hours Microphysics RGB, 23 June 2020, 09:00 UTC. Credit: EUMeTrain.

Wind regime/shear could also play the role when it comes to detection of the contrail clouds - stretching the initial contrail width from few tens of meters to possible few thousands of meters (as suggested by Figure 13 and Figure 16).

Meteosat-11 Dust RGB with ECMWF model
Figure 13: Vertical cross-section of wind and potential temperature (ECMWF model, left) over the contrail clouds (Meteosat-11 Dust RGB, right). Humidity maxima was also observed in the contrail region (not shown here). 5 December 2019, 09:00 UTC

Contrails in new MTG channels

Besides the information on thin cirrus clouds that can be retrieved from the SEVIRI channels, additional input will be provided by solar channel 1.38 micron, one of the new channels that will be available from FCI on MTG. This channel is very sensitive to high clouds, therefore, during daytime contrails will be detectable in 1.38 micron image and the RGBs containing it. Figure 14 shows GOES-16 (GOES-EAST) 1.37 micron image (used as proxy for 1.38 channel) in which contrails are seen.

GOES-17 Channel 13
Figure 14: GOES-16 1.37 micron image on 19 January 2021, 18:30 UTC. 

The albedo of the high-level cirrus clouds is very low (contrails is only about 2-3%), therefore, even when combining with other channels, the 1.37 channel is highly enhanced (up to 0-3% range, or even lower). The RGB combination utilising this channel is the Cloud Type RGB, in which contrails, as well as other cirrus clouds, are seen in bright red colours (Figure 15). It is also seen that in the area where contrails are present relative humidity at 200 hPa was above 80%.

GOES-16 image comparison

GOES-16 Cloud Type RGB compare1

Figure 15: GOES-16 24 hour Microphysics RGB overlaid with ECMWF Relative Humidity at 200 hPa, compared to Cloud Type RGB which is used as proxy for Cloud Type RGB that will be available with MTG. Images show contrails over Gulf of Mexico on 19 January 2021, 18:00 UTC.


Contrails seen from ground level
Figure 16: Photo of contrails taken in the central Austria (ca. north view) at roughly the same time as for Figure 13, 5 December 2019, 08:43 UTC. Credit: Ivan Smiljanic. 


Related content

Contrails (CIMSS/University of Wisconsin-Madison)
Contrails or condensation trails (Met Office)
SEVIRI 24-hour Microphysics RGB Quick Guide (EUMeTrain)