jeffrey |
01/04/2009 02:49PM (Read 5927 times)
|
|
|
Status: offline
Registered: 01/04/2009
Posts: 5
|
Hi, I am an undergraduate just beginning to work with iraf and I was asked to find centroids for astronomical objects like supernova debris etc.I have tried [b:5af04e7531]daofind[/b:5af04e7531] to give me a value of xcenter, ycenter of the object, but is this means the centroids of it?
As I also found a [b:5af04e7531]center[/b:5af04e7531] in apphot, what is the difference between two?
My prof also asked me to consider the error in finding centroid, but seems all the tools in iraf just give the answer directly. Is there any methods?
Really sorry for asking stupid questions.
|
|
|
|
fitz |
01/04/2009 02:49PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
The CENTER task is appropriate when you have a list of initial positions and want to refine them for a select group of stars, DAOFIND is used when you want to generate a catalog for all objects in an image. There is a similar task called STARFIND in the IMAGES package which produces more shape information and is based on DAOFIND.In either case, errors aren't computed. I image these could be derived or estimated using the shape information (e.g. FWHM, position angle, ellipticity, etc), but this hasn't been asked before and off-handI couldn;t give you the equations. The centering itself is done by convolving an elliptical Gaussian to find a local maxima, so there is no direct least-squares fit on each star that would provide errors on the parameters.Cheers,
-Mike
|
|
|
|
jeffrey |
01/04/2009 02:49PM
|
|
|
Status: offline
Registered: 01/04/2009
Posts: 5
|
Thank you very much.Then if I dont consider error, i can use CENTER in apphot, STARFIND or DAOFIND and get the same centroid of the objects too?I just keep exploring iraf and I found [b:6def383c5a]imcntr[/b:6def383c5a] in proto, the help file said it can also find the center of an object by providing a x, y coodinate.
Will they get the same results?
|
|
|
|
fitz |
01/04/2009 02:49PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
For most uses the values produced will all be "accurate", however the various tasks use different algorithms to compute the center (e.g. convolution vs marginal centroids) so they won't necessarily all be exactly the same value out to N decimal places. Needing astrometric positions is different than needing an accurate center for photometry.-Mike
|
|
|
|
jeffrey |
01/04/2009 02:49PM
|
|
|
Status: offline
Registered: 01/04/2009
Posts: 5
|
ic, so I should use CENTER ,STARFIND and DAOFIND and do it serveral times.but i am quite confused about your saying that" Needing astrometric positions is different than needing an accurate center for photometry. ",Ain't they the same? I mean finding the centroid of photometry, then doing astrometry should find the astrometric positions. Sorry that I just know a bit... I am a beginner in these...
Moreover I try imcentroid this morning and it only gives up to 3 d.p. , can I get more?
I really cant thank you enough.
|
|
|
|
fitz |
01/04/2009 02:49PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
[quote:8d6a9b9cc9]Ain't they the same? I mean finding the centroid of photometry, then doing astrometry should find the astrometric positions. [/quote:8d6a9b9cc9]What I mean by 'astrometric position' is e.g. the RA/Dec of an object such as when you create a WCS for the image, try to find out if something has moved and by how much. Basically, the relative position of one source to everything else in the image. This is different than what you might need in photometry where what you want as the "center" is the middle of a Gaussian PSF you might be fitting and your concern is how much flux is enclosed within a radius of N pixels from that center (including fractional parts of pixels).You can't get any more precision out of IMCENTROID without modifying the task. Repeating runs of any of these tasks should give you the same result each time, however you can use the IMSHIFT task to fractionally shift an image -- doing this with repeated shifts should produce slightly different results (once you take into account the offset) and may help you get a believable error.
-Mike
|
|
|
|