wcs residuals

[Jerry Zhang]

Hi Dustin and all,

Do you know that does astrometry.net have residual or uncertainty information when solving the wcs? I mean the X,Y rms coming from the distortion or whatever.

Best,

Jerry

[Dustin Lang]

Hi,

I don't think we report that directly.  You can look in the "corr" file for the field and index x,y coordinates.

cheers,

dusitn

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/6a452ddc-c26c-48df-b2e4-d884f6e345d6n%40googlegroups.com.

[Jerry Zhang]

Hi Dustin,

Thank you very much. I calculated average x error and average y error and added them together like this: rms = sqrt(mean(|index_x-field_x|)^2+mean(|index_x-field_x|)^2).

I think the method I found here is not correct, cuz their step of taking mean and square are reversed...: https://web.njit.edu/~gary/322/assets/Astrometry.net_uncertainties.pdf

I have two questions,  

1. what are the rows of the corr.fits file ? all the referenced sources in the index in the field of view? 
   
2. I have a quite large rms for one of my image, and I indeed found a row with |index-field| 10 times larger than the other. Does that mean a mismatching of that star? So should I discard this star when calculating the rms since it is not the 'astrometry uncertainty'?

Best,

Jerry

[Dustin Lang]

Hi,

The corr file lists stars that we think match.  There is a "match_weight" column that says how likely we think each match is.  Hopefully that one row you found with a much larger value has a small match_weight!

cheers,

dustin

To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/62a136c5-0ff9-4083-8377-025faee14407n%40googlegroups.com.

[Jerry Zhang]

Thanks Dustin, I will restrict the matched sample to match_weight > 0.99 to do the calculation!

Thanks,

Jerry

[Jerry Zhang]

Hi Dustin,

Now I think what I get is not what I want... the rms calculated is the 'rms of difference between index and the real detection', which includes the uncertainty coming from the source position measurement (centroiding) and motion of the star between the index and the observation ...     

I just want a rms about the fit of the coordinate transform, just like IRAF task GEOMAP does. Does astrometry.net provides any info about it?

Thank you again!

Jerry

在2023年4月19日星期三 UTC+1 02:10:54<dstn...@gmail.com> 写道：

[Jerry Zhang]

Well, now I figured out that https://web.njit.edu/~gary/322/assets/Astrometry.net_uncertainties.pdf is somehow correct, but just missing to divided by the square root of N (number of rows), to get the variation of the mean of the pixel error at sub-pixel level. I guess it is what I want, the rms of the fit of the coordinate transformation  (The mean should around 0 and more reference stars the uncertainty gets lower).  While what I have in the above reply is the the expect pixel error of single detections, which involves error of star extraction, and star motion, which is at several pixel level.

I hope I explained clearly and please correct me if I am wrong.

best,

Jerry