FWHMEccentricity maps

[Zocky]

I have a few more questions regarding those maps.

The maps show how your star sizes (FWHM) and shape (Eccentricity) varies over the FOV.  In a perfect world you'd have a consistent and small FWHM all over the field and lovely round stars everywhere, i.e. low Eccentricity - the SubframeSelector doc suggests that anything lower than 0.42 looks round.

It is clear that FWHM should be as small as possible, but what if you have a few very bright stars in your FOV. How does that affect on a overall FWHM?
If I understood correctly, ideal map would be the one that have just one grey tone with one numerical value. So the higher number of grey tones and the higher number of numerical values, represent poor image quality? What can be read from the position of the grey tone areas? Would be ideal map consist of concentric areas?
Here are maps from my Luminance image of M81/82. How does it look, good or bad?

[Geoff]

The ideal map for FWHM would be a uniform grey. Next best would be concentric circles centred on the chip. This shows you have  field curvature, but everything is perfectly aligned and collimated, so you are getting the best from your equipment. For eccentricity the best result would be a uniform value of zero right across the field. Your eccentricity map show that there is a slight tilt in your system--higher eccentricity in the bottom of the picture, but truth be told you would be hard put to tell the difference between an eccentricity of 0.45 and one of 0.38 (89% roundness vs 92% roundness)

To answer your question about bright stars in the field: as long as stars are not saturated they should all have the same FWHM irrespective of brightness. I presume the algorithm has some way of rejecting saturated stars based on their profile.
Geoff

[RickS]

Hi Zocky,

Geoff already answered most of your questions, so I'll just add a couple of additional comments...

Very bright (saturated) stars should be ignored for the purpose of FWHM calculation.  SubframeSelector has a parameter for this (Upper Limit.)

Your results look pretty good!  The variation in FWHM is quite small across the field and the Eccentricity values show round looking stars without a lot of variation.  Things are a little worse in the corners but that's quite typical.  Results will vary depending on the seeing and the star field so you probably should look at a few examples before forming a strong opinion.

Cheers,
Rick.

[Zocky]

Geoff, Rick, thnx for your answers. Things are much clearer now.
I do have one more question regarding SubframeSelector. When I click on Output maps, I only get Excel file. How can I get maps like the ones from FWHMEccentricity script?

[RickS]

I do have one more question regarding SubframeSelector. When I click on Output maps, I only get Excel file. How can I get maps like the ones from FWHMEccentricity script?

Output maps creates a _m.fit and a _m.csv file when I run it.  The FITS file contains three images, a map of the fitted stars and the FWHM and Eccentricity maps.  I can't see an obvious way to enable or disable this in the script GUI so I would have thought you'd get the same behaviour?

[sctall]

Zocky

I have almost the same issue.
I do get 2 files, but the _m.fit file only has 1 image in it.
I have never been able to get the 3 images.
I think it is the map of the fitted stars, but not sure.
Definitely not what is expected.

I put the 2 files on dropbox.
Could someone could look at it and tell me why this is?
I am thinking there is another component I am needing?
https://www.dropbox.com/sh/kawfzgw9c8uoz5e/AADJGxl3Lx1ozzY8BBinTMP9a?dl=0

-- Scott --

[Zocky]

I have reset the script and tried again. Now it`s OK. When I open newly created FITS file, 3 images are opened automatically. 

[james7]

It would help on the FWHM to know your pixel scale in arc seconds. The only thing you can really tell from the map you provided (in pixels, not arc seconds) is that you are getting close to being undersampled, meaning that the stars in your image are only being sampled over 2 pixels which is cutting it kind of close for really accurate FWHM measurements. Moreover, if you are imaging with a one-shot-color camera (with a Bayer pattern) then the sampling is effectively even worse (or lower).

In any case, consider the case where you are imaging at 2 arc seconds per pixel, that would give you a FWHM of about 4 arc seconds which is certainly fine but not great. However, if you were imaging at 4 arc seconds per pixel you'd have a FWHM of around 8 arc seconds which would be pretty bad (indicating, perhaps, a problem with focus). You also need to consider what happens if you have bad seeing or poor focus, in that case the FWHM (in arc seconds) will appear poor but the eccentricity may look surprisingly good (i.e. the stars will be "big" and round).

If I had to guess I'd say that you are okay (looking at the maps, since they seem to show fairly good uniformity across the field), but to be certain you really need to consider the FWHM in arc seconds, not just pixels.

[mschuster]

Scott,

Please check the console log. Sometimes you will see an error message stating not enough stars detected to generate maps. The script needs a sufficient number of stars distributed across the entire frame to generate the maps successfully. You can sometimes modify the star detection parameters to increase the number of stars detected. Alternatively, you can set cropping bounds to focus analysis on only field center to achieve the same thing, if most of the stars are centered.

Zocky,
The script uses robust statistics to reject outliers, such as saturated stars. Small blooms, likewise. As mentioned you can also use the threshold rejection parameter. I like to set this value small enough to so that brighter stars that are unsaturated but affected by detector anti-blooming are also rejected.

For each of my long 40 min Ha frames, I like to expose a short 1 min frame first for comparison. Ideally the results and maps for each pair of short and long frames are identical. If not, you can work on your setup to make them more similar. On my setup, I found I needed to improve tracking (reduce Dec axis tracking delay due to backlash) and reduce focus drift (due to temperature change).

Mike

[Zocky]

Thnx guys.

... but to be certain you really need to consider the FWHM in arc seconds, not just pixels.

Camera is SBIG ST8300m with 5,4u pixel size, exp. was 15 min., Luminance filter, focal length of the scope is 510 mm.
Which parameters should I change to get results in arc seconds?

[cfranks]

Arcseconds per pixel = (pixel size (microns) * 206.3) / Focal length (mm).

[sctall]

Thanks Mike --

Please check the console log. Sometimes you will see an error message stating not enough stars detected to generate maps. The script needs a sufficient number of stars distributed across the entire frame to generate the maps successfully.

Yep, It did not find enough stars.
I guess I am seeing this a lot.

Scott