III. 3. A. Characterizing the frame

• Source code for character3.cl may be obtained here.
• Source code for pickone.f may be obtained here.
• Source code for pickround.f may be obtained here.

In order to make this practical, we've decided that autophot will be a "header driven" routine. So, the first step is to characterize the frame and put the important goodies in the header so that other scripts can then grab these numbers and use them.

What do we need, and how do we find them? The following are entered as parameters to character:

• gain
• readnoise
• saturation
• the filter key word
• the time key word

However, that's only a start. The character.cl script then figures out the following:

• "mode", the sky value. This comes from using imstat to find the mode of all the data values below 20,000, in order to avoid saturation as the modal value.
• "sigma", the mean error in the sky. This is computed using Poisson statistics from the modal sky value; i.e.: (sqrt(mode*gain+readnoise**2))/gain
• "datami", the minimum data value. Unfortunately the words "datamin" and "datamax" are reserved FITS words, so we call this "datami". We compute this as the mode - 10 * sigma
• "datama", the maximum data value, computed as saturation - 500.
• "fwhm", the typical full-width at half-maximum of a star. This has been the largest headache to find. We run daofind using approximate parameters (fwhm=4.0) to find most stars, and then reply on the apphot routine "fitpsf" to determine the typical fwhm's. Unfortunately, the output fitpsf depends upon having a good starting point for the fwhm to begin with. Following a suggestion from Wayne, I modified the routine so fitpsf is run with different "trial" fwhm's, from 2.0 to 6.0 pixels. Which one worked best is then determined from the scatter of the output sigma's from fitpsf. That seems to work great now, and stops the many problems we were having with this.

Note that the way sigma, and hence datami, are determined assumes that the image is a single exposure. If you are working from a stacked image, the effective gain will be different. You can fix sigma and datami in the headers using IRAF's hedit after the program has calculated them. This is important later when daofind uses n*sigma as the threshold for searching for stars.

PACKAGE = user
   TASK = character

PACKAGE = user
   TASK = character

img     =                       Image
gain    =                  3.1  ccdgain value
rdnoise =                   6.  readnoise
saturate=               30000.  saturation
fname   =               filter  filter key word
timestam=             time-obs  time key word
(inlist =                     )
(mode   =                   ql)



Here's the actual script:

# Written by Phil Massey, Lowell Obs.  March 21, 2002
        procedure characterize (img,gain,rdnoise,saturate,fname,timestamp)
        string img {prompt="Image"}
        real gain {3.1,prompt="ccdgain value"}
        real rdnoise {6., prompt="readnoise"}
        real saturate {30000., prompt="saturation"}
	string fname {"filter", prompt="filter key word"}
	string timestamp {"time-obs", prompt="time key word"}

	string *inlist

        begin

        string s1, s2, s3, s4, s5, s6, s7


	real fwhmv,gainv,rdnoisev,saturatev,f1,fd1,fd2,fd3,rms,bot,top

# Get parameters
	s1=img
	gainv=gain
	rdnoisev=rdnoise
	saturatev=saturate
	s6=fname
	s7=timestamp
	hedit(s1,"gain",gainv,add+,verify-,show+)
	hedit(s1,"rdnoise",rdnoisev,add+,verify-,show+)
	hedit(s1,"saturate",saturatev,add+,verify-,show+)
	hedit(s1,"fname",s6,add+,verify-,show+)
	hedit(s1,"timestamp",s7,add+,verify-,show+)
# Now find mode:
	delete("temp")
	imstat(s1,fields="mode",lower=INDEF,upper=20000,format-,>"temp")
	inlist="temp"
	i=fscan(inlist,f1)
	hedit(s1,"mode",f1,add+,verify-,show+)
	rms=sqrt(f1*gainv+rdnoisev**2)/gainv
	hedit(s1,"sigma",rms,add+,verify-,show+)
	bot=f1-max(10.*rms,30.)
	hedit(s1,"DATAMI",bot,add+,veri-,show+)
	top=saturatev-500.
	hedit(s1,"DATAMA",top,add+,veri-,show+)
	delete("tempxy")
	print("running daophot in character")
daofind(s1,output="tempxy",sigma=rms,fwhm=4.0,thresh=20.0,datamin=bot,datamax=top,interact-,verify-,verbose-,filter=s6,obstime=s7)
	delete("temprad")
	print("running fitpsf in character:",(rms),(bot),(top))
fitpsf(s1,10,sigma=rms,coords="tempxy",output="temprad",datamin=bot,datamax=top,interact-,verify-,filter=s6,obstime=s7,fwhmpsf=3.0)
	print("done running fitpsf")
	delete("allrad")
	txdump("temprad",fields="rsigma",expr="rsigma!=INDEF",>"allrad")
	delete("tempfwhm")
        radmode
	inlist="tempfwhm"
	i=fscan(inlist,fwhmv)
	hedit(s1,"fwhm",fwhmv,add+,ver-,show+)
end