Channel alignment issues


David J Taylor GM8ARV 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇪🇺
 

This message was originally posted ny Alan Sewards in the SatSignal
group, on 2003 August 31:

-----------------------------------------------------------
David,
Today I was trying to use GSS (4.0.4.392) to look at the source
of the latest forest fire threatening us(!) when I ran into a
problem with the registration of the country boundary map and the
image, especially when I used "Fires". I observed that on the
Remapped image, the country boundary (in this case the coast) was
displaced to the south east of the true coast by about 0.2 degrees.
But the interesting thing was the that the red dot of the fire was
also displaced by about the same amount! The image ended up showing
the plume of smoke issuing from a different place from the red area
showing the hotspot. It seems as though there is a discrepancy
between the registration of the various images. I was originally
using the Channel 1 image for visible, but also tried using channel
12n, in case that registered differently, but it was the same.
I'm happy to say that the combined efforts of the firemen,
Canadair water bombers and various helicopters (including the big
Russian one) succeeded in mastering the fire about lunchtime.

Best regards - Alan
-----------------------------------------------------------

Alan,

I moved this to the MSG-1 list as I don't think it's a software
issue. I looked at the 1500 images from 2003 Aug 31, and tried a
number of ways of comparing them. As channel 4 is processed to a
negative image, and channels 1 and 2 are positive images, I found
that just using the image arithmetic function in Paint Shop Pro to
add the images together gave a easy way to judge the relative
position by eye. This also bypasses any effect of software
remapping errors (although these would be consistent from channel to
channel).

Adding either the channel 1 or the channel 2 image to the channel 4
image showed a quite distinctive edge effect with one image being
displaced relative to the other. I can only guess that this is one
of the side effects when Eumetsat say the "image quality may be
disturbed". There was nothing in the Admin report for the next day,
though. I tried to see if the edge effect differed over the image,
to see if there was a mapping size error, or whether there was
constant offset over the whole image, but I didn't find enough data
to check this. Perhaps using an image from earlier in the day might
show something?

I also tried the same test - adding channels 1 and 4 - on the 1500
cycle from today (2003 Sep 08) and the edge effect had disappeared -
so presumably the images are now in better alignment.

So the effect you saw was indeed real - channel 4 was correctly
mapped but not channel 1!

By the way: I cannot currently get channel 12 (HRV) to accurately
align with the published geolocation data, and I'm waiting to see if
this improves when the satellite moved from its present 10W position
to the new 3-4W position. There seems to be a non-linearity in the
alignment, not just simple gain and offset errors.

Cheers,
David

Join MSG-1@groups.io to automatically receive all group messages.