Previous posts on this blog (and elsewhere) have detailed the co-registration misalignment that exists between the 3.9 µm and 10.7 µm channels on the GOES-13 Imager. Because of this diurnally-varying co-registration error, a 3.9 µm pixel may be offset to the right or left of a 10.7 µm pixel; if this occurs near a pronounced temperature gradient (such as along a lakeshore), a false brightness temperature difference signal can ensue.
Consider, for example, the toggle above from 26 September 2014. A strong brightness temperature difference exists at 1825 UTC along the shorelines of Lakes Michigan, Huron and Erie; it is gone five minutes later, at 1830 UTC. There is no discernible change in the visible image over the same 5-minute interval (Link).
NESDIS operations alters the GVAR signal just before 1830 UTC (when the 3.9 µm imagery is shifted one pixel to the West) and at 0630 UTC (when the 3.9 µm imagery is shifted one pixel to the East) to mitigate the effects of the diurnally-varying co-registrations differences between the 3.9 µm and 10.7 µm channels. The imagery above shows the visible and two infrared (10.7 µm and 3.9 µm) channels at 1825 and 1830 UTC (GOES-13 was in Rapid Scan Operations mode at this time). The 3.9 µm imagery shows a one-pixel westward shift that is especially evident if you look at the unchanging navigation along the eastern shore of Lake Michigan. (1825 UTC imagery: Visible, 3.9µm and 10.7µm; 1830 UTC imagery: Visible, 3.9µm and 10.7µm) A similar link between 1815 and 1830 UTC on 25 September shows the same shift in the shortwave IR. A toggle between 0615 and 0630 UTC on 29 September shows the eastward shift in the 3.9 µm imagery that occurs then.
NOAA/NESDIS continues to monitor this co-registration issue.