Straylight figure 1

Anomalies on Meteosat images

Explaining why some images appear degraded

Straylight figure 1
Straylight figure 1

Sometimes Meteosat images experience degradation, due to straylight (extra rays hitting SEVIRI's optical path) or light spots caused by intrusion of sunlight into the radiometer.

Last Updated

03 September 2020

Published on

12 May 2020

What is straylight?

The SEVIRI instrument on board Meteosat Second Generation (MSG) satellites consists of a scan mirror; a three-mirror telescope, and a set of filters and lenses that illuminate a series of infrared and visible light detectors. The observed Earth scene is projected on top of the detector, so that — in principal — any light ray meeting the detector originates from a specific location of the target scene (see right on the top image).

An additional type of ray reaches the detector, sometimes referred to as 'parasitic light'. Parasitic light does not arrive by the regular optical path, but by:

  • Diffuse scattering off the mirrors. Top image, centre)
  • Reflection off non-optical parts of the telescope such as baffles etc
  • Other effects, such as diffraction.

A good telescope design ensures that such straylight contributions remain negligible. However, a very bright light source, such as the Sun, can easily contribute straylight to an image.

Straylight patterns in MSG images

Straylight figure 2
Figure 1: Direct images of the Sun. WV6.2 channel (left), IR3.9 channel (right)

The prime source of straylight in SEVIRI images is the Sun. A secondary source is the Moon, which is more faint. Artefacts caused by straylight of bright Earth objects, such as clouds, have not yet been identified.

The abundance of straylight in an image depends on various factors. The most important one is the position and angle of the Sun relative to the telescope. Straylight is only observed during spring and autumn seasons because the Sun is too far in the North or the South during summer and winter seasons. Further, straylight is observed only during the night (with respect to the satellite longitude) because the Sun is always behind the satellite when the day-time Earth image is acquired.

The SEVIRI spectral channels are variably sensitive to sunlight. The VIS06, VIS08, NIR16 and HRV channels are designed to measure reflected solar light so are sensitive to sunlight. The IR3.9 channels are sensitive to both reflected solar light and the thermal radiation of the Earth. All other IR channels are designed to measure the thermal radiation emitted from the Earth's atmosphere and will not respond to reflected solar radiation. Only direct illumination by the Sun (i.e. where the Sun is in the image) will have an effect on those channels (see Figure 1). Hence, straylight can only be expected in the VIS06, VIS08, NIR16, HRV and IR3.9 channels. Straylight only affects the IR3.9 channel, given that there is not much useful signal in the VIS06, VIS08, NIR16 and HRV channels during night-time.

Straylight figure 3
Figure 2: Straylight pattern around an image of the Sun in NIR1.6 channel
Straylight figure 4
Figure 3: Close-up of a patch in Figure 2, showing the effect on individual detectors. The image is made up by repeated lines of the three individual detectors only one is seriously affected

As discussed above, straylight may arise from various light paths, resulting in a variety of complex effects. Figure 2 shows patterns that depend on the distance to the Sun and visual artefacts in horizontal and vertical directions. The vertical extent of the image is achieved by moving the scan mirror the same way as for Earth imaging. The horizontal extent is built up by the satellite rotation, again the same as Earth imaging.

Figure 3 shows a close-up of Figure 2, demonstrating that straylight not only depends on the channel, but also on the individual detectors (9 for HRV, 3 for non-HRV channels). The striped effect in Figure 3 can be explained by a patch of straylight on the focal plane meeting only one individual detector.

In summary, the factors describing the intensity of straylight include:

  • The angle between satellite (telescope), rotational plane and the Sun, given by date — and to a lesser degree the satellite attitude.

  • The angle between the Sun and the telescope's line of sight in the rotational plane of the satellite.

  • The scan mirror position.

  • The individual detector and, thereby, the spectral channel.

Mitigation of straylight effects in the Image Processing Facility (IMPF)

The IMPF cannot predict the complex straylight pattern, as seen in Figure 2, but allows for a correction based on a simple distance law towards the Sun. Moreover, the correction is activated only for the IR3.9 channel. Other IR channels do not show a significant straylight effects.

During image data processing straylight estimation is performed before the image lines are resampled from the so-called Level 1.0 co-ordinate frame to the Level 1.5 co-ordinate frame, i.e. the GEOS projection. Straylight contribution is calculated for various tie points on the L1.0 grid and are interpolated using a two-dimensional interpolation. This interpolation is then used to calculate straylight contribution for any one detector line, after which it is subtracted from the line. Finally, the corrected line data is then subject to the resampling process to create the final Level 1.5 image.

In order to record all the necessary radiometric processing of the data the two-dimensional interpolation is stored in the Level 1.5 image header, which describes the assumed straylight field in the data processing chain, seen as an image of the straylight without the Earth’s disk.

Other known image anomalies

Occasionally some images taken by Meteosat experience degradation due to the intrusion of sunlight into the radiometer. This occurs mainly when the Sun is not in complete eclipse, so that at least a part of the Sun is visible from the satellite behind the northern or southern hemisphere of the Earth.

The light spots detected in the visible image are caused by double reflections (secondary reflection straylight) due to the optical design of the Meteosat radiometer. Direct sunlight reaches the satellite and is reflected and diffused by the mechanical structure of the radiometer, so that reflected light enters the radiometer field of view and reaches the detectors to create the anomalous intense spots in some images. This effect is observed particularly around midnight and especially during the time of the eclipses in spring and autumn. Some water vapour and infrared images are also affected during these periods for the same reasons. The effect is demonstrated below in a sequence of six visible images and six water vapour images taken around midnight (click image to see animation).

Image Anomaly MFG WV Animation Large 01
Figure 4: Six visible images taken around midnight on 25–26 August 1997 (slots 46 to 3)
Image Anomaly MFG WV Animation Large 02
Figure 5: Six WV images taken around midnight on 14–15 April 1997 (slots 46 to 3)
Image Anomaly MFG WV Animation Large 03
Figure 6: Seven WV images taken just after midnight on 4 August 2002 (Met-6 scans between 00:30–01:30 UTC)
Image Anomaly MFG Large 04
Figure 7: A dark area can be seen over the Baltic Sea on 16 April 1997 at 24:00 (Slot 48)
Image Anomaly MFG Large 06
Figure 8: An anomaly caused by the midnight effect on WV images. Depending upon the Sun-Earth-satellite geometry 'dark spots' and 'bow-shaped structures' can be seen
Image Anomaly MFG Large 08
Figure 9: An midnight effect — a 'striping pattern'

All Meteosat images shown on our website are subject to EUMETSAT copyright. If you wish to re-use these Meteosat images, EUMETSAT's copyright credit should be shown by displaying the words "copyright {year} EUMETSAT" (where {year} is the current year) on each of the images shown.