Dynamically identify EWS-G1 horizontal offset

[nullpainter]

Possibly use the median position of medium to high intensity pixels?

[ZbychuButItWasTaken]

That's one way to do it, but the offsets change every 12h, and from what I've seen it is a constant change.. Maybe it would be more reliable to just have the offset in px and time of the change in the config?

[creinemann]

According to NOAA, EWS is centered in their imagery by the McDias software that finds the center of the current images disc, then crops the image . The reported center point of the image shifts.

IMGLIST EWS/FD.-99 TIME=7 9
Image file directory listing for:EWS/FD
Pos Satellite/ Date Time Center Band(s)
sensor Lat Lon

---

27 GOES-13 22 JAN 21022 08:45:00 0 -66 1-4,6
26 GOES-13 22 JAN 21022 08:15:00 0 -66 1-4,6
25 GOES-13 22 JAN 21022 07:45:00 0 -57 1-4,6
24 GOES-13 22 JAN 21022 07:15:00 0 -57 1-4,6
166 GOES-13 21 JAN 21021 08:45:00 0 -66 1-4,6
165 GOES-13 21 JAN 21021 08:15:00 0 -66 1-4,6
164 GOES-13 21 JAN 21021 07:45:00 0 -57 1-4,6
163 GOES-13 21 JAN 21021 07:15:00 0 -57 1-4,6
120 GOES-13 20 JAN 21020 08:45:00 0 -66 1-4,6
119 GOES-13 20 JAN 21020 08:15:00 0 -66 1-4,6
118 GOES-13 20 JAN 21020 07:45:00 0 -57 1-4,6
117 GOES-13 20 JAN 21020 07:15:00 0 -57 1-4,6
74 GOES-13 19 JAN 21019 08:45:00 0 -66 1-4,6
73 GOES-13 19 JAN 21019 08:15:00 0 -66 1-4,6
72 GOES-13 19 JAN 21019 07:45:00 0 -57 1-4,6
71 GOES-13 19 JAN 21019 07:15:00 0 -57 1-4,6
IMGLIST: done
IMGLIST EWS/FD.-99 TIME=19 21
Image file directory listing for:EWS/FD
Pos Satellite/ Date Time Center Band(s)
sensor Lat Lon

---

4 GOES-13 21 JAN 21021 20:45:00 0 -57 1-4,6
3 GOES-13 21 JAN 21021 20:15:00 0 -57 1-4,6
2 GOES-13 21 JAN 21021 19:45:00 0 -66 1-4,6
185 GOES-13 21 JAN 21021 19:15:00 0 -66 1-4,6
143 GOES-13 20 JAN 21020 20:45:00 0 -57 1-4,6
142 GOES-13 20 JAN 21020 20:15:00 0 -57 1-4,6
141 GOES-13 20 JAN 21020 19:45:00 0 -66 1-4,6
140 GOES-13 20 JAN 21020 19:15:00 0 -66 1-4,6
97 GOES-13 19 JAN 21019 20:45:00 0 -57 1-4,6
96 GOES-13 19 JAN 21019 20:15:00 0 -57 1-4,6
95 GOES-13 19 JAN 21019 19:45:00 0 -66 1-4,6
94 GOES-13 19 JAN 21019 19:15:00 0 -66 1-4,6
51 GOES-13 18 JAN 21018 20:45:00 0 -57 1-4,6
50 GOES-13 18 JAN 21018 20:15:00 0 -57 1-4,6
49 GOES-13 18 JAN 21018 19:45:00 0 -66 1-4,6
48 GOES-13 18 JAN 21018 19:15:00 0 -66 1-4,6

[nullpainter]

That's one way to do it, but the offsets change every 12h, and from what I've seen it is a constant change.. Maybe it would be more reliable to just have the offset in px and time of the change in the config?

Yes and no, and I think my quick Github issue may have suggested that I was intending to calculate the pixel offset based on these calculations. I'm actually just thinking of identifying whether to apply a left, or a right offset (based on a px constant).

To your point, if for a given timestamp we can be guaranteed that we know when the offset changes and in which direction and this never changes - sure, config all the way. But if there is any uncertainty, my approach should work.

[nullpainter]

Fixed in v1.0.18.