GOES Image Scanline Shear
28 February 1999
GOES-8 Image Navigation and Registration (INR) requirements are tough.
Images must be constructed from line-by-line back-and-forth east-west scans to 0.00001 radian, or some line-to-line shearing will be observed.
The satellite and ground system do not always achieve this precision, and so the images "swim" a little if you run a fast, full-resolution time-series of high-contrast surface features.
However, much larger scanline displacements began to occur in GOES-8 early in 1997, apparently due to the brief loss of Image Motion Compensation (IMC) data from the Attitude and Orbit Control Electronics (AOCE).
GOES-8 image shear is observed in several classes:
- Major displacement of groups of 2 to 4 scan lines by 10's of kilometers. This has appeared daily in the GOES-8 images starting in 1997 and 1998, progressing to approximately once per hour in 1999. For example,
The underlying cause of this is known to be a temporary dropout of communication of the scan state from the Imager to the spacecraft, causing the image motion compensation to shut down for a few scans.
- Replication of previous lines. This has appeared occasionally in the GOES-8 images starting early in 1997. For example, Baja gets duplicated in two line-pairs on 8 April 1997.
- Shortening of a scanline by multiples of 8 samples. This can happen anywhere within a line. For example, the shorelines at the mouths of the Potomac River and of the Chesapeake Bay acquire "notches" in one frame of a three-frame MPEG animation. The ground software appears not to notice the that it missed a small block of data while constructing the scanlines for rebroadcast to users of the GVAR format.
- East-west misplacement of by a pixel. This affects the entire line within the frame, but it is hard to see without a high-contrast background. For example, one can see zigs and zags appearing and disapperaing in the canyons of the Sierra Madre Mountains in a series of images from the 11 micron channel. By mid-1997, this shear was endemic, as shown in this 20-frame MPEG movie of Baja for 3 hours on 6 June 1997, where the bad lines are marked in the lefthand margin with a white bar in each affected frame.
The underlying hardware and software problems are not yet known.
- Shear madness at the African limb, where GOES-8 makes a different shear pattern each day during the full-earth scans at 0845 UTC, for October 20-25, 1995. Some days, there is no problem at all. In early 1999, the san-to-scan interlace became noticably worse, such as on 14 January 1999. The problem does not seem to be in the spacecraft -- image motion compensation software for the east-/west-going scans is suspected.
- Failure by GOES-8 to interlace the east-going and west-going scans at the limb near the equator. For example, in the full-disk iamge at 1745 UTC on 17 February 1999, one sees significant "venetian blinding" at the east limb (slightly after sunset there), the west limb (slightly after sunrise there), the west-northwest limb (including one of the nowq-frequent "line slips" described above).
This limb shearing appears to be a more exaggerated form of the shearing seen at the African limb in the early years of GOES-8.