Unknow sky trail

François Guiot

Hi everyone,

I was playing around with data, trying to mix data from different setup (TelescopeLive CHI1 and SPA2 if you're curious). Everything was fine until the red channel which ended up with this:

I first thought it was due to a satellite, the culprit should be this frame:

But... whatever I tried (even excluded the frame), the result was almost still the same. And if we zoom in we can see this:

This is not a straight line at all. So one big question: what is that thing? Because whatever it is, it leaves something significant in the data itself.

(I suspect a meteor leaving hot stuff in the atmosphere as the trail could only be seen in the red channel).

Large scale rejection doesn't really help, I still don't know how could handle it (maybe playing with masks and linear fit stuff?).

Thanks in advance for making me smarter :)

Regards;

François

BlockHead

Francois,

There are many things to explain.

1. This is a single satellite trail. If you have more than 15-20 frames it will reject 100% completely. What are your rejection settings? Winsorized Sigma Clipping of 2.2 should do the job.

2. Satellites rotate (when not maintained). They can rotate many times per second. So you will see the light vary along a satellite's path.

3. As a satellite moves through your field your telescope is not tracking perfectly. This makes the trail look a little wiggly. It is your telescope... not the satellite, that is making the trail deviate from a straight line. 

4. This is not a meteor or anything strange.

5. There are some tools that you can select and removed trails directly (SKILL and SetiAstro's Blemish Remover). However, in this case I would expect proper rejection to do the job.

-the Blockhead

François Guiot

Hello Adam,

Thanks for replying. I forgot to specify what are my rejections settings and so on; sorry for this. There are 55 frames in total. 15 from chile scope (where the trail appears) and 40 from spain. I was using ESD with 0.15 outliers/0.05 significance. I played around with few values but didn't see any noticable difference. I also tried Winsorised with something like 2.3 or 2.4. But here's what it gives with 2.2:

Then I tried with only data from chile scope:

And finally, I removed the frame which containes the trails, une check "use cache", juste in case and...

Frames have been normalized, I use PSF Scale SNR for weighting. I made a test with PSF Signal but it gives me something similar.

Didn't thought about rotating satellites; interesting! (I actually have another frame where a trail is present but it didn't leave any mark, rejection has done a perfect job and the trail it self is wonderful straight line; that's why I'm really curious about what I see here :) )

BlockHead

15 frames is part of the problem. Not great for full rejection (of the faint part).

However, you can do a selective rejection.

Many ways to do it. If you provide me the 15 registered frames I can show you.

Here are the steps (for one method):

1. Use The GAME script on this particular frame. Create a rectangle (multipoint allows this) that completely surrounds the satellite trail.

2. Save this as a new file writing these values as Zeros.

3. Use this file with the black line as the one to integrate using ImageIntegration. 

4. The trail will be gone.

Review:

https://www.adamblockstudios.com/articles/Fundamentals_Selective_Rejection

(bottom video)

-the Blockhead

François Guiot

I've been able to figure out what the problem actually was.

Before sleeping last night, I suddenly thought: "Did I check my reference frame for normalization?" => No!

So when I checked it this mornign guess what? It contains the trail. So I carefully picked another frame, redo all normalization. Just enable large-scale rejection and voilà !

Lesson learned:

1. It is possible to see satellite rotation, awesome
2. Do not make "something quick just before going to bed"
3. Always check LN reference

BlockHead

Yes indeed.... that would be another place to look... but I am surprised! I would have thought you need a pretty small tiling scale for this to happen (and I assume you used the LN defaults). Thanks for showing this to me... I did not appreciate this could happen this strongly.

-the Blockhead