Posted: Mon Sep 26, 2005 10:14 am Post subject: PSF photometry
my name is Anestis Tziamtzis and i'm a PhD student at Stockholm's
observatory. At the moment i'm working with the daophot package and i'm
trying to do PSF photometry on a nebula. My task is to measure the
magnitude of a star within the nebula. I'm following your manual, but
when i check the final subtracted image it seems that the star is over
subtracted. Is it possible to give me some hints on what i'm doing wrong?
Here is a short description on what i'm doing. In the beginning i chose
a few isolated star in my image and performed photometry on them with
phot. These stars have a comparable magnitude with my star of interest.
Then i choose a few of these with pstselect and i run the psf task. I
choose the best candidates for the fit and then i run nstar. The final
script that i use is substar.
Here are some of the main parameters that i'm using. All of them are
scaled to my image.
Inner radius of sky annulus in scale units:5
Width of sky annulus in scale units:4
I also refit the centers and the sky value and my fitting algorithm is
set to gaussian.
I'm trying to fit a 0th order psf , none of my stars are saturated and i
have the nclean parameter equal to 0.
I would like to thank you for your time.
Post generated using Mail2Forum (http://www.mail2forum.com)
> Dear Sir/Madam,
> my name is Anestis Tziamtzis and i'm a PhD student at Stockholm's
> observatory. At the moment i'm working with the daophot package and
> i'm trying to do PSF photometry on a nebula. My task is to measure
> the magnitude of a star within the nebula. I'm following your manual,
> but when i check the final subtracted image it seems that the star is
> over subtracted. Is it possible to give me some hints on what i'm
> doing wrong?
> Here is a short description on what i'm doing. In the beginning i
> chose a few isolated star in my image and performed photometry on
> them with phot. These stars have a comparable magnitude with my star
> of interest. Then i choose a few of these with pstselect and i run
> the psf task. I choose the best candidates for the fit and then i run
> nstar. The final script that i use is substar.
> Here are some of the main parameters that i'm using. All of them are
> scaled to my image.
> Aperture size:2*FWHM
> Inner radius of sky annulus in scale units:5
> Width of sky annulus in scale units:4
> I also refit the centers and the sky value and my fitting algorithm
> is set to gaussian.
> I'm trying to fit a 0th order psf , none of my stars are saturated
> and i have the nclean parameter equal to 0.
The problem that you are seeing could be due to various reasons.
Unfortunately the maintainer of the daophot package has since left us,
but I found some email that may address your problems. I'll post
relevant snippets below.
I hope that helps. Please feel free to contact me if have further
questions or problems.
Do the PSF stars and their neighbors if any subtract cleanly from the
image, i.e. without severe holes or other features ? with or without
sky refitting ? The group file produced by the PSF task, NSTAR, and
SUBSTAR can be used to test this. AS the PSF model is constructed from
these stars this should be the case by definition. If it isn't then you
need to find out why it isn't before proceeding to fit the entire image.
PSF model problems can be caused by 1) poor choice of sky annulus
parameters in PHOT, 2) bad choice of PSF stars, 3) bad choice of PSF
fitting parameters (fitrad and / or psfrad too big or too small), as
well as more obvious things like data reduction problems (careless
cosmic ray removal can cause real trouble), linearity problems etc.
Undersampling will cause features in the subtractions. Averaging a large
enough sample of PSF stars beats this error down in the PSF model
but there will still be holes and peaks in the subtractions, i.e. a
sort of interpolation noise. I would not expect to see only holes.
Do all the subtractions show holes ? or only some ? Do only the brighter
or more crowded stars show holes ?
...one way that holes arise is that in very crowded regions very faint
stars in the wings can latch onto noise spikes or sometimes brighter
stars and cause holes and scatter in the photometry. Normally nstar has
more problem with this than allstar because in nstar the same stars are
fit together as a group, whereas in allstar the grouping is dynamic and
the brightest stars are usually fit first. A common source of this
problem is setting the detection threshold very low in daofind, which
then finds a lot of spurious faint stars. Most of these get filtered
out in the fitting process but some remain and cause problems. The fact
that setting recenter=no (which usually is a pretty non-standard thing
to do) may help improve things makes me wonder if there aren't a fair
number of spurious detections in your input list? I have not done much
with HST data but I have heard that there may be features in the PSF
that tend to get detected as stars, and that smoothing the image on
which the detection is done can help.
Usually if the field is very crowded or the backgrounds are very
complicated, i.e. galaxies and nebulosity refitting the sky, at least
in allstar tends to improve things rather than the reverse ... Do you
have groupsky = yes ?
Nstar refits the sky by making the sky an actual fitting parameter. To
get good values you may need to make fitrad larger so you actually
include pixels near the sky in the fit, which may then have the
unfortunate effect of increasing the scatter and the effects of
neigbors. In allstar the sky is refit every 3 iterations by computing
the median of the pixels inside the annulus defined by the daopars
sannulus and wannulus parameters, after the current best fit of all the
stars is subtracted from the data. This enables the user to use data
closer in than phot and actually underneath the star which can be good
in very crowded regions. You need to make sure the sky annulus is
reasonable, i.e. does not go all the way to the center because the
subtractions are noisy there, is big enough to get a reasonable sample
of pixels, but not too big to get too influenced by neighbors.
If the stellar groups get too strung out spatially however then the
group sky may not be representative of the stars and some stars on the
edge of the group may subtract poorly. In allstar making sure that
psfrad and fitrad are not larger than they need to be can help
alleviate this. In nstar it may be necessary to alter the grouping
Post generated using Mail2Forum (http://www.mail2forum.com)
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum