This is a comparison of the GOES-10, GOES-11, and GOES-12 imager 3.9 micrometer ("shortwave IR") and visible channels, showing large wildfires that were burning in the vicinity of the border between Alaska and the Yukon Territory during the 23:15 UTC / 19 June to 07:15 UTC / 20 June 2004 period (4:15 PM to midnight local time). All 3 satellite views are in their native projections. Unseasonably warm and dry conditions across that region (visible image with surface observations) helped to create an environment favorable for rapid fire growth (Aqua MODIS true color image).
This case is an excellent example to illustrate how sensitive the 3.9 micrometer IR channel can be to the reflected solar component from sun glint. Observe the rapid increase in the number and areal coverage of dark/warm "hot spots" on GOES-12 at 04:15 UTC (and then later on GOES-11 at 05:45 UTC) as the high satellite viewing angle co-inspires with the high latitude summer sun to light up the many lakes with sun glint (bright areas on the visible channel), thereby saturating the 3.9 micrometer channel on GOES-11 and GOES-12. Although the GOES-10 3.9 micrometer channel has a significantly lower saturation temperature, the sun glint effect was not observed over this region due to the more direct viewing angle of the GOES-10 satellite.
It's also interesting to note the change in appearance of the "anvil" of the pyrocumulus cloud feature that streams eastward from the largest/southernmost fire -- both GOES-12 and GOES-11 show this cloud feature as rather "warm" (darker gray enhancement) on the 3.9 micrometer channel early in the animation, due to solar reflection off the water-droplet-rich cloud; as the sun sets, this cloud element cools to a more appropriate brightness temperature (lighter gray enhancement). Again, this behavior is not noted on the GOES-10 imagery because of the more direct satellite viewing angle.