Initial commit

2 files changed, 8295 insertions, 0 deletions

diff --git a/noao/digiphot/daophot/doc/userdocs/daophot.usr.tex b/noao/digiphot/daophot/doc/userdocs/daophot.usr.tex
new file mode 100644
index 00000000..ce4ed623
--- /dev/null
+++ b/noao/digiphot/daophot/doc/userdocs/daophot.usr.tex
@@ -0,0 +1,2005 @@
+%de 21 (summer solstice, midnight, la serena) (this is IT, Phil!)
+%more done on Dec 27th or so
+%typos fixed 28 Dec 1pm
+%more done on Dec 28th (1st 4-m night morning)
+%copy xfered from Chile on Jan 1, 1990
+%modifications made after the AAS meeting, Jan 15 1990:
+% end itemize problems fixed
+% buku aperture photometry stuff added Jan 15/16
+% started to make Lindsey's changes Jan 28th
+% Next set of changes made Feb. 18th when we SHOULD have been
+% off with marcia having a good time.
+% more modifications Monday Feb 19
+% march 20/21 mods made in Boulder----Lindsey's comments
+% march 20/21 mods made in Boulder---beginning of JB's comments
+% march 27 mods back in Tucson
+% may 11th, fixed the sumed/average read-noise problem!
+\documentstyle[11pt,moretext]{article}
+\begin{document}
+\title{A User's Guide to Stellar CCD Photometry with IRAF}
+\author{Philip Massey \and Lindsey E. Davis}
+\date{March 29, 1990}
+\maketitle
+\begin{abstract}
+This document is intended to guide you through the steps for obtaining
+stellar photometry from CCD data using IRAF.  It deals both with the
+case that the frames are relatively uncrowded (in which case simple
+aperture photometry may suffice) and with the case that the frames
+are crowded and require more sophisticated point-spread-function
+fitting methods (i.e., {\bf daophot}).  In addition we show how one
+goes about obtaining photometric solutions for the standard stars, and
+applying these transformations to instrumental magnitudes.
+\end{abstract}
+\tableofcontents
+\eject
+\section{Introduction}
+This user's guide deals with both the ``relatively simple" case of
+isolated
+stars on a CCD frame (standard stars, say, or uncrowded program stars)
+and the horrendously more complicated case of crowded field
+photometry. We describe here all the steps needed to obtain instrumental
+magnitudes and to do the transformation to the standard system.  There
+are, of course, many possible paths to this goal, and IRAF provides no
+lack of options.  We have chosen to illuminate a straight road, but many
+side trails are yours for the taking, and we will occasionally point
+these out (``let many flowers bloom").  This Guide is {\it not} intended
+as a reference manual; for that, you have available (a) the various
+``help pages" for the routines described herein, (b) ``A User's Guide
+to the IRAF APPHOT Package"  by Lindsey Davis, and (c) ``A Reference
+Guide to the IRAF/DAOPHOT Package" by
+Lindsey Davis.  (For the ``philosophy" and algorithms of DAOPHOT, see
+Stetson 1987 {\it PASP} {\bf 99}, 111.) 
+What {\it this} manual is intended to be is a real
+``user's guide", in which we go through all of the steps necessary to go
+from CCD frames to publishable photometry. (N.B.: as of this writing
+the IRAF routines for determining the standard transformations and
+applying those transformations are still being written.)  We outline
+a temporary kludge that will work with Peter Stetson's CCDRED VMS
+Fortran package.  Hopefully the PHOTRED package currently under
+development at Cerro Tololo will be available by Spring 1990, and 
+this manual will then be revised. 
+  
+The general steps involved are as follows: (1) fixing the header
+information to reflect accurate exposure times and airmasses,
+(2) determining and cataloging the characteristics of your data (e.g.,
+noise, seeing, etc.), 
+(3) obtaining instrumental magnitudes for all the standard stars
+using aperture photometry, (4) obtaining instrumental magnitudes for
+your program stars using IRAF/daophot, (5) determining the aperture
+correction for your program stars, (6) computing the transformation
+equations for the standard star photometry, and (7) applying these
+transformations to your program photometry.  We choose to illustrate
+these reductions using {\it UBV} CCD data obtained with an RCA chip on the 0.9-m
+telescope at Cerro Tololo, but the techniques are applicable to data
+taken with any detector whose noise characteristics mimic those of a
+CCD.  
+ 
+If you are a brand-new IRAF user we strongly recommend first reading the
+document ``A User's Introduction to the IRAF
+Command Language" by Shames and Tody, which can be found in Volume 1A
+of the 4 blue binders that compose the IRAF documentation. (Actually
+if you are a brand-new IRAF user one of us recommends that you find
+some simpler task to work on before you tackle digital stellar photometry!)
+The procedures described here will work on any system supported by IRAF;
+for the purposes of discussion, however, we will assume that you are
+using the image display capabilities of a SUN.  If this is true you then
+may also want to familiarize yourself with the ins and outs of using
+the SUN Imtool window; the best description is to be found in ``IRAF
+on the SUN".
+ 
+We assume that your data has been read onto disk, and that the basic
+instrumental signature has been removed; i.e., that you are ready
+to do some photometry.  If you haven't processed your
+data this far yet, we
+refer you to ``A User's Guide to Reducing CCD Data with IRAF" by Phil
+Massey.
+ 
+\section{Getting Started}
+ 
+\subsection{Fixing your headers}
+You're going to have to this some time or another; why not now?  There
+are two specific things we may need to fix at this point: (a) Add any
+missing header words if you are reducing non-NOAO data, (b) correct the
+exposure time for any shutter opening/closing time, and (c) correct the
+airmass to the effective middle of the exposure.
+ 
+Two things that will be useful to have in your headers are the exposure
+time and the airmass.  If you are reducing NOAO data then you will
+already have the exposure time (although this may need to be corrected
+as described in the next paragraph) and enough information for the {\bf
+setairmass} task described below to compute the effective airmass of the
+observation.  You can skip to the ``Correcting the exposure time"
+section below.
+If
+you are reducing non-NOAO data you should examine your header with a
+ 
+\centerline{ {\bf imhead} imagename{\bf l+ $|$ page} }
+ 
+\noindent
+and see exactly what information {\it is} there.  If you are lacking the
+exposure time you can add this by doing an
+ 
+\centerline{ {\bf hedit} imagename{\bf ``ITIME"} value {\bf add+ up+
+ver- show+} }
+ 
+\noindent
+If you know the effective airmasses you can add an ``AIRMASS" keyword in
+the same manner, or if you want to compute the effective airmass
+(corrected to mid-exposure) using {\bf setairmass} as described below,
+you will need to have the celestial coordinates key words ``RA" and
+``DEC", as well as the siderial-time (``ST"), 
+and preferably the coordinate ``EPOCH" and the date-of-observation
+(``DATE-OBS"), all of which should have the form shown in Fig.~\ref{header}.
+
+You may want to take this opportunity to review the filter numbers in the
+headers, and fix any that are wrong.  If you are lacking filter numbers
+you may want to add them at this point.
+ 
+\subsubsection{Correcting the exposure time}
+The CTIO 0.9-m has an effective exposure time that is
+25-30ms longer than the requested exposure time (Massey et al. 1989 {\it
+A.J.} {\bf 97}, 107; Walker 1988 {\it NOAO Newsletter} {\bf No. 13},
+20).  First see what "keyword" in your header gives the exposure time:
+ 
+\centerline{
+{\bf imhead} imagename{\bf.imh l+ $|$ page} }
+ 
+\noindent
+will produce a listing such as
+given in Figure~\ref{header}.
+\begin{figure}
+\vspace{3.2in}
+\caption{\label{header}Header information for image n602alu.imh}.
+\end{figure}
+The exposure time keyword in this header is ``ITIME".  In this case
+we wish to add a new exposure time to each of the headers;  we will call
+this corrected exposure time
+EXPTIME, and make it 25 ms larger than whatever value is listed as
+ITIME.  To do this we use the {\bf hedit} command as follows:
+ 
+\centerline{
+{\bf hedit *.imh EXPTIME ``(ITIME+0.025)" ver- show+ add+}.}
+ 
+\noindent
+An inspection of the headers will now show a new keyword EXPTIME.
+(Walker lists a similar correction for the CTIO 1.5-m shutter, but the
+CTIO 4-m P/F shutters have a negligible correction.
+The direct CCD shutters on the Kitt Peak CCD cameras give
+an additional 3.5ms of integration on the edges but 13.0ms in the
+center [e.g., Massey 1985 {\it KPNO Newsletter} {\bf 36}, p. 6];
+if you have any 1 second exposures you had best correct these by
+10ms or so if you are interested in 1\% photometry.)
+  
+\subsubsection{Computing the effective airmass}
+The task {\bf setairmass} in the {\bf astutil} package will compute
+the effective airmass of your exposure, using the header values of RA,
+DEC, ST, EPOCH, and DATE-OBS, and whatever you specify for the observatory
+latitude.  An example is shown in Fig.~\ref{setairmass}.
+\begin{figure}
+\vspace{2.5in}
+\caption{\label{setairmass} The parameter file for {\bf setairmass}.}
+\end{figure}
+The default for the latitude is usually the IRAF
+variable {\bf observatory.latitude}.  To by-pass this ``feature", simply
+put the correct latitude in the parameter file
+(e.g., $-30.1652$ for CTIO,
+$+31.963$ for KPNO; $+19.827$ for Mauna Kea.).
+ 
+\subsection{{\bf imexamine:} A Useful Tool}
+ 
+The {\bf proto} package task {\bf imexamine} is a powerful and versatile task
+which can be used to interactively examine image data at all stages of
+the photometric reduction process. In this section we discuss and
+illustrate those aspects of {\bf imexamine} which are  most useful to
+photometrists with emphasis on three different applications of the task:
+1) examining the image, for example plotting lines and columns
+2) deriving  image characteristics, for example computing the
+FWHM of the point-spread function 3) comparing the same region
+in different images.
+ 
+The task 
+{\bf imexamine} lives within the {\bf proto} package, and you will also need
+to load {\bf images} and {\bf tv}.  Then 
+{\bf display}  the image, and type {\bf imexamine}.
+When the task is ready to accept input the image cursor will begin blinking
+in the display window, and the user can begin executing various keystroke
+and colon commands.  The most useful data examining commands are summarized
+below.  The column, contour, histogram, line and surface plotting commands
+each have their own parameter sets which set the region to be plotted and
+control the various plotting parameters. All can be examined and edited
+interactively from within the {\bf imexamine} task using the
+appropriate {\bf :epar} command.
+ 
+\begin{description}
+ \item[c] - Plot the column nearest the image cursor
+ \item[e] - Make a contour plot of a region around the image cursor
+ \item[h] - Plot the histogram of a region around the image cursor
+ \item[l] - Plot the line nearest the image cursor
+ \item[s] - Make a surface plot of a region around the image cursor
+ \item[:c N] - Plot column N
+ \item[:l N] - Plot line N
+ \item[x] - Print the x, y, z values of the pixel nearest the image cursor
+ \item[z] - Print a 10 by 10 grid of pixels around the image cursor
+ \item[o] - Overplot
+ \item[g] - Activate the graphics cursor
+ \item[i] - Activate the image cursor
+ \item[?] - Print help
+ \item[q] - Quit {\bf imexamine}
+ \item[:epar c]  - Edit the column plot parameters
+ \item[:epar e]  - Edit the contour plot parameters
+ \item[:epar h]  - Edit the histogram plot parameters
+ \item[:epar l]  - Edit the line plot parameters
+ \item[:epar s]  - Edit the surface plot parameters
+ 
+\end{description}
+ 
+ 
+Example 1 below shows how a user can interactively
+make and make hardcopies of image line plots using {\bf imexamine} and at the same time
+illustrates many of the general features of the task.
+ 
+The {\bf imexamine} task also has some elementary image analysis capability, including
+the capacity to do simple aperture photometry, compute image statistics
+and fit radial profiles.  The most useful image analysis commands are
+listed below.
+ 
+\begin{description}
+\item[h] - Plot the histogram of a region around the cursor
+\item[r] - Plot the radial profile of a region around the cursor
+\item[m] - Plot the statistics of a region around the cursor
+\item[:epar h]  - Edit the histogram parameters
+\item[:epar r]  - Edit the radial profile fitting parameters
+\end{description}
+ 
+Example 2 shows how a photometrist might use {\bf imexamine}
+and the above commands to estimate the following image characteristics:
+1) the full width at
+half maximum (FWHM) of the point-spread function, 2) the background sky level
+3) the standard deviation of the background level 4) and the radius at which
+the light from the brightest star of interest disappears into the noise
+(this will be used to specify the size of the point-spread-function,
+e.g.,PSFRAD).
+ 
+Finally {\bf imexamine} can be used to compare images. Example 3
+shows how to compare regions in the original image and in the
+same image with all the fitted stars subtracted out. The example
+assumes that the target image display device supports multiple frame buffers,
+i.e. the user can load at
+least two images into the display device at once.
+ 
+The {\bf imexamine} task offers even more features than are discussed here and the
+user should refer to the manual page for more details.
+ 
+\vspace{12pt}
+{\bf Example 1:} Plot and make hardcopies of image lines within {\bf imexamine}.
+ 
+\begin{itemize}
+\item {\bf display} the image and then type {\bf imexamine}.
+\item move the image cursor to a star and tap {\bf l} to plot the image
+line nearest the cursor
+\item tap the {\bf g} key to activate the graphics cursor
+\item type {\bf :.snap} to make a hardcopy of the plot on your default device
+\item expand a region of interest by first moving the graphics
+cursor to the lower left corner of the region and typing {\bf E},
+and then moving the graphics cursor to the upper right corner
+of the region and typing anything
+\item type {\bf :.snap} to make a hardcopy of the new plot
+\item tap the {\bf i} key to return to the image cursor menu
+\item type {\bf :epar l} to enter the line plot parameter set, change the
+value of the logy parameter to yes and type {\bf CNTL-Z} to exit and
+save the change
+\item  repeat the previous line plotting commands
+\item type {\bf q} to quit {\bf imexamine}
+\end{itemize}
+ 
+{\bf Example 2:} Compute some elementary image characteristics using
+{\bf imexamine}.
+ 
+\begin{itemize}
+\item {\bf display} the image and then type {\bf imexamine}.
+\item move to a bright star and tap the {\bf  r} key
+\item examine the resulting radial profile plot and note the final
+number on the status line which is the FWHM of the best fitting
+Gaussian
+\item repeat this procedure for several stars to estimate a good
+average value for the FWHM
+\item set the parameters of the statistics box ncstat and nlstat
+from 5 and 5 to 21 and 21 with  {\bf :ncstat 21} and {\bf :nlstat 21}
+commands so that the sizes of the statistics and histogram
+regions will be identical
+\item  move to a region of blank sky and tap the {\bf m} key to get an
+estimate of the mean, median and standard deviation of the
+sky pixels in a region 21 by 21 pixels in size around the
+image cursor
+\item leave the cursor at the same position and tap the {\bf h} key to
+get a plot of the histogram of the pixels in the same region
+\item tap the {\bf g} key to activate the graphics cursor, move the
+cursor to the peak of the histogram and type {\bf C} to print out
+the cursor's value.  The ``x" value then gives you a good estimate of
+the sky.  Similarly, you can move the cursor to the
+half-power point of
+the histogram and type {\bf C} to estimate the standard deviation
+of the sky pixels.  Tap the {\bf i} key to return to the
+image cursor menu
+\item compare the results of the h and m keys
+\item repeat the measurements for several blank sky regions and note
+the results
+\item move to a bright unsaturated star and turn up the zoom and
+ contrast of the display device as much as possible
+\item using the {\bf x} key mark the point on either side of the center
+where the light from the star disappears into the noise
+and estimate PSFRAD 
+\item type {\bf :epar r} to edit the radial profile fitting parameters
+and set rplot to something a few pixels larger than PSFRAD
+and tap the {\bf r} key
+\item note the radius where the light levels off and compare with
+the eyeball estimate
+\item repeat for a few stars to check for consistency
+\item type {\bf q} to quit imexamine
+\end{itemize}
+ 
+\noindent
+{\bf Example 3:} Overplot lines from two different images.
+ 
+\begin{itemize}
+\item {\bf imexamine image1,image2}
+\item move the image cursor to a star and type {\bf z} to print the
+pixel values near the cursor
+\item tap the {\bf n} key to display the second image followed by {\bf z}
+to look at the values of the same pixels in the second
+image
+\item tap the {\bf p} key to return to the first image
+\item tap {\bf l} to plot a line near the center of the star and tap
+the {\bf o} key to overlay the next plot
+\item tap the {\bf p} key to return to the second image and without
+moving the image cursor tap the l key again to overplot
+the line
+\item type {\bf q} to quit imexamine
+\end{itemize}
+ 
+\subsection{Dealing with Parameter Files (Wheels within Wheels)}
+ 
+The {\bf daophot} (and {\bf apphot}) packages are unique in IRAF in that
+they obtain
+pertinent information out of separate ``parameter files" that can be
+shared between tasks.  As anyone that
+has used IRAF knows, each IRAF command has its own parameter file that
+can
+be viewed by doing an {\bf lpar} {\it command} or edited by doing an
+{\bf epar} {\it command}.
+However, in {\bf daophot} and {\bf apphot} there are ``wheels within
+wheels"---some of the parameters are in fact parameter files themselves.
+For instance, the aperture photometry routine {\bf phot} does not
+explicitly
+show you the methods and details of
+the sky fitting in its parameter file.  
+However, if you do an {\bf lpar phot}
+you will see a parameter
+called ``fitskypars" which
+contains, among many other things, the radii of the annulus to be used
+in determining the sky value.
+You will also find listed ``datapars" (which specifies the properties
+of your data, such as photons per ADU and read-noise), ``centerpars"
+(which
+specifies the centering algorithm to be used), and ``photpars" (which gives
+the
+size of the digital apertures and the zero-point magnitude).
+The contents of any of these parameter files can be altered either by
+{\bf epar}ing them on their own or by typing a ``:e" while on that
+line of the main parameter file.  If you do the latter, a control-z
+or a ``:q" will bring you back. 
+For example, to examine or edit {\bf fitskypars}, you can 
+do an explicit {\bf lpar fitskypars}
+or {\bf epar fitskypars}, or you can do an {\bf epar phot}, move the
+cursor down to the ``fitskypars" line, and then type a {\bf :e} to edit
+(see Fig.~\ref{wheels}).
+\begin{figure}
+\vspace{4.2in}
+\caption{\label{wheels}Changing the Sky Annulus in {\bf fitskypars}.}
+\end{figure}
+Confusing?  You bet!
+But once you are used to it, it is a convenient and powerful way to
+specify a whole bunch of things that are used by several different
+commands---i.e., you are guaranteed of using the same parameters in
+several different tasks.  If there is only one thing that you want to
+change in
+a parameter file you {\it can} enter it on the command line when
+you run the command, just as if it were a ``normal" (hidden) parameter,
+i.e., {\bf phot imagename dannulus=8.} does the same as
+running {\bf epar fitskypars} and changing the ``width of sky annulus"
+{\bf dannulus} to 8.0.
+ 
+Mostly these things are kept out of the way (``very hidden" parameters)
+because you {\it don't} want to be changing them, once you have set them
+up for your data.  There are exceptions, such as changing the PSF radius
+in making a point-spread function in a crowded field (Sec. 4.6).
+However,
+you are well protected here if you leave the {\bf verify} switch on.
+A task will then give you an opportunity to take one last look at
+anything
+that you really care about when you run the task.  For instance, if we
+had simply run {\bf phot} on an image (we'll see how to do this shortly)
+it would have said ``Width of sky annulus (10.)", at which point we
+could
+either have hit [CR] to have accepted the 10., or we could have
+entered a new value.
+ 
+ 
+\section{Aperture Photometry on your Standards}
+ 
+Standard stars provide a good example of relatively uncrowded
+photometry,
+and in this section we will describe how to obtain instrumental
+magnitudes for your standards using {\bf phot}. 
+The basic steps are 
+\begin{itemize}
+	\item Decide what aperture size you wish to use for measuring your
+	standards {\bf (this should be the same for all the frames).} At the
+	same time we will pick a sky annulus. 
+	\item Set up the various parameter files ({\bf datapars,
+		centerpars, fitskypars, photpars}) to have the correct values.
+	\item For each frame:
+		\begin{enumerate}
+		\item Identify the standard star(s) either 
+		       interactively using a cursor
+			or by using the automatic star finding algorithm
+			{\bf daofind}.
+		\item Run the aperture photometry program {\bf phot}
+		      on each of your standard star frames.
+\end{enumerate}
+\end{itemize} 
+Although the routines you will need to use are available both in the
+{\bf daophot} and {\bf apphot} packages, we strongly advise you to run
+them from the {\bf daophot} package: the default setup is somewhat different,
+and the two packages each have their own data parameter files.
+
+\subsection{Picking an Aperture Size}
+Unfortunately, there are not good tools available with IRAF to do this
+yet, and we will restrict our discussion here to some of the
+considerations before telling you to just go ahead and use a radius that
+is something like 4 or 5 times the FWHM of a stellar image; e.g.,
+12 or 15
+pixels as a radius, assuming you have the usual sort of ``nearly
+undersampled" FWHM$\approx3$ data.
+You might naively expect (as I did) that you wish to pick an aperture
+size
+that will ``contain all the light" from your standard stars, but in fact
+this is impossible: the wings of a star's profile extend much further
+than you imagine at a ``significant" level.  King (1971 {\it Publ.
+A.S.P.} {\bf 83}, 199) and Kormendy (1973 {\it A.J.} {\bf 78}, 255)
+discuss the fact that on photographic plates the profile of a star
+extends out to {\it arcminutes} at an intensity level far exceeding the
+diffraction profile; Kormendy attributes this to scattering off of dust
+and surface irregularities on the optical surfaces.
+Massey {\it et al}.\ (1989 {\bf 97}, 107) discusses
+this in regards to CCD's and standard star solutions using the very data
+we are using here as an example (which is not exactly a coincidence).
+Although the intensity profile falls off rapidly, the increase in area
+with radius increases rapidly, and in practical terms Massey {\it et
+al.}
+found that in cases where the FWHM was typically small (2.5-3 pixels),
+increasing the digital aperture size from a diameter of 18 pixels to
+one of 20 pixels resulted in an additional 1-2\% increase in light
+for a well-exposed star, and that this increase continues
+for larger apertures until masked by the photometric errors.
+ 
+Given that you presumably want 1\% photometry or better, what should you
+do?  
+Well, the fact that photoelectric photometery through fixed apertures
+in fact does
+work suggests that there is some radius beyond which the same fraction
+of
+light is excluded, despite variations in the seeing and guiding.  You do
+not want to choose a gigantic aperture ($>$ 20 pixels, say) because the
+probability of your having a bad pixel or two goes up with the area.
+But you do not want to choose too small an aperture ($<$10 pixels, say)
+or you will find yourself at the mercy of the seeing and guiding.  Most
+photoelectric photometrists will use an aperture of at least 10
+arcseconds in diameter, but remember you have one advantage over them:
+you are not sensitive to centering errors, since any digital aperture can
+be exactly centered. 
+If you
+have enough standard star observations (I used about 300 obtained over a
+10 night run) you can
+compute magnitude differences between a large aperture (20 pixels),
+and a series of smaller apertures (8, 10, 12, 15, 18) for each filter,
+and then see for which radius the difference (in magnitudes) becomes
+constant.  Unfortunately, there are no tools currently available within
+IRAF for taking the differences between two apertures, or for conveniently
+plotting these differences, so you are on your own.  My recommendation
+would be that if you have typical data with a
+FWHM of $\leq 4$ pixels, that you use something like an aperture of 12 to 15
+pixels in radius for your standard stars.  {\bf You can save yourself a lot
+of trouble if you simply adopt a single radius for all the standards
+from all the nights for all filters.}
+  
+\subsection{Setting Things Up}
+ 
+As discussed in ``Dealing with Parameter Files" (Section 2.1) we must
+setup the parameter files from which {\bf phot} will get the details of
+what it is going to do.  The easiest way to do this is to simply
+{\bf epar phot}, and on each of the four parameter lists to do a
+{\bf :e}.  Mostly we will leave the defaults alone, but in fact you will
+have to change at least one thing in each of the four files.
+ 
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{photdatapars} Parameters for {\bf datapars}.}
+\end{figure}
+In {\bf datapars} (Fig.~\ref{photdatapars}) we need to specify both
+the FWHM
+of a star image ({\it fwhmpsf}) and the 
+threshold value above sky ({\it threshold}) if we are going to use the
+automatic star-finding routine {\bf daofind}; the choices for these
+are discussed further below. In order to have
+realistic
+error estimates for our aperture photometry we need to specify 
+the CCD readnoise {\it readnoise} in electrons and the
+gain (photons per ADU) for the CCD {\it epadu}.  
+In order to
+correct the results for the exposure time we need the exposure time
+keyword {\it
+exposure}. Do an 
+ 
+\centerline{{\bf imhead} {\it imagename} {\bf l+ $|$ page}}
+ 
+\noindent
+to see a
+listing of all the header information (Fig.~\ref{phothead}).
+\begin{figure}
+\vspace{4.0in}
+\caption{\label{phothead} Header information for std159.imh}
+\end{figure}
+By specifying the (effective) airmass and filter keywords,
+these can be carried along in the photometry file for use when we do
+the standards solution ({\it airmass} and {\it filter}).  Finally we use
+{\it datamin} and {\it datamax} so we will know if we exceeded the
+linearity of the CCD in the exposure, or whether there is some anomalously
+low valued pixel on which our star is sitting.
+Since the value of the sky on our standard exposures is
+probably nearly zero, {\it datamin} should be set to a negative value
+about three times the size of the readnoise in {\it ADU's}; e.g., $-3 \times
+65. \div 2.25 \approx -90$ in this example.  Note that although we will
+later argue that the shape of the PSF changes a little about 20000 
+ADU's (presumably due to some sort of charge-transfer problem),
+for the purposes of simple aperture photometry we are happy
+using 32000 ADU's as the maximum good data value. (We do not really
+want to use 32767 as afterall the overscan bias was probably at a
+level of several hundred.)
+ 
+\begin{figure}
+\vspace{3.0in}
+\caption{\label{photcenterpars} Parameters for {\bf centerpars}.}
+\end{figure}
+In {\bf centerpars} (Fig.~\ref{photcenterpars}) we need to 
+change the centering algorithm {\it calgorithm}
+from the default value of ``none" to
+``centroid".  If the FWHM of your frames are unusually large ($>4$, say,
+you would also do well to up the size of {\bf cbox} to assure that the
+centering works well; make it something like twice the FWHM. In this
+case the FWHM is 3 pixels or a bit smaller, and we are content to leave
+it a the default setting of 5 pixels.
+ 
+\begin{figure}
+\vspace{2.7in}
+\caption{\label{photfitskypars} Parameters for {\bf fitskypars}.}
+\end{figure}
+In {\bf fitskypars} (Fig.~\ref{photfitskypars})
+the only things we must specify are the size and
+location of the annulus in which the modal value of the sky will be
+determined.  If you are going to use a value of 15 for your photometry
+aperture, you probably want to start the sky around pixel 20.  Keeping
+the width of the 
+annulus large (5 pixels is plenty) assures you of good sampling, but
+making it too large increases the chances of getting some bad pixels in
+the sky.
+ 
+\begin{figure}
+\vspace{2.7in}
+\caption{\label{photphotpars} Parameters for {\bf photpars}.}
+\end{figure}
+In {\bf photpars} (Fig.~\ref{photphotpars}) 
+we merely need to specify the size (radius) of the
+aperture we wish to use in measuring our standards.
+ 
+\subsection{Doing It}
+There are basically two ways of proceeding in running photometry on the
+standard stars, depending upon how you are going to identify the
+relevant star(s) on each frame.  If you have only one (or two)
+standard stars on each frame, and it is always one of the brightest
+stars present, then you can avoid a lot of the work and use the
+automatic star-finding program {\bf daofind} to find all your standards
+and the whole thing can be done fairly non-interactively.  However,
+if you are one of the believers in cluster field standards, then you
+may actually want to identify the standards in each field using the
+cursor on the image display so that the numbering scheme makes sense.
+We describe below each of the two methods.  
+ 
+\subsubsection{Automatic star finding}
+First let's put the name of each frame containing standard stars into
+a file; if you've put the standard star exposures into a separate
+directory this can be done simply by a {\bf files *.imh $>$ stands}.
+This will leave us with funny default output file 
+names for a while (we advise against
+including the ``.imh" extension when we discuss crowded field photometry
+in the next section), but this will only be true for a short
+intermediate
+stage. 
+ 
+We want to run {\bf daofind} in such a way that it finds only the
+brightest
+star or two (presumably your standard was one of the brightest stars
+in the field;
+if not, you are going to have to do this stuff as outlined below in
+the ``Photometry by eye" section).  We will delve more fully into the
+nitty-gritty of {\bf daofind} in the crowded-field photometry section,
+but here we are content if we can simply find the brightest few stars.
+Thus the choice of the detection
+threshold is a critical one.  If you make it too low you will find all
+sorts of junk; if you make it too high then you may not find any stars.
+You may need to run {\bf imexamine} on a few of your images: first
+{\bf display} the image, and then {\bf imexamine}, using the ``r" cursor
+key to produce a radial profile plot.  Things to note are the
+typical full-width-half-maximum and the peak value.  If your sky is
+really around zero for your standard exposures, then using a value
+that is, say, twenty times the readnoise (in ADU's) is nearly guaranteed to
+find only the brightest few stars; do your radial plots in {\bf
+imexamine} show this to be a reasonable value?  In the example here we
+have decided to use 500 ADUs as the threshold ($20 \times 65 \div 2.25
+\approx 500$).
+ 
+Now {\bf epar daofind} so it resembles that of Fig.~\ref{photdaofind}.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{photdaofind} Parameter file for {\bf daofind}.}
+\end{figure}
+Go ahead and execute it (Fig. ~\ref{daoout}). 
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{daoout} Screen output from a {\bf daofind} run.}
+\end{figure}
+Note that since {\it verify} is on that you
+will be given a chance to revise the FWHM and detection threshold.  By
+turning verbose on you will see how many stars are detected on each
+frame.  
+%Probably the best way of doing this is to write the output from
+%{\bf daofind} into a file; do a
+% 
+%\centerline{ {\bf daofind @stands $|$ tee starsfound} }
+% 
+%\noindent
+%to put the output into the file ``starsfound" as well as on the screen.
+Make a note of any cases where no stars were found; these you will have
+to
+go back and do with a lower threshold.
+ 
+The run of {\bf daofind} produced one output file named {\it
+imagename.imh.coo.1} for each input file.  If you {\bf page} one of
+these you will find that it resembles that of Fig.~\ref{photcooout}.
+\begin{figure}
+\vspace{3.7in}
+\caption{\label{photcooout} Output file from {\bf daofind}.}
+\end{figure}
+The file contains many lines of header, followed by the {\it x} and {\it
+y} center values, the magnitudes above the threshold value, the ``sharpness"
+and ``roundness" values, and finally an ID number.
+In the example shown
+here in Fig.~\ref{photcooout} two stars were found: one 2.9 mags
+brighter than our detection threshold, and one about 0.4 mag brighter
+than our detection threshold.
+ 
+In a few cases we doubtlessly found more than one star; this is a good
+time to get rid of the uninteresting non-standards in each field.
+If things went by too fast on the screen for you to take careful notes
+while running {\bf daofind} we can find these cases now: do a 
+
+\centerline{ {\bf txdump *coo* image,id,x,y yes }}
+
+ 
+\noindent
+to get a listing of the location and number of stars found on each image.
+If you have cases where there were lots of
+detections (a dozen, say) you may find it easier to first {\bf sort
+*.coo* mag} in order to resort the stars in each file by how bright they
+are.  Of course, your standard may not be the brightest star in each
+field; you may want to keep an eye on the {\it x} and {\it y} values to
+see if it is the star you thought you were putting in the middle! 
+To get rid of the spurious stars you will need to {\bf edit} each of the
+output files (e.g., {\bf edit std148.imh.coo.1} ) and simply delete the
+extras.
+ 
+Finally we can run aperture photometry on these frames, using the 
+``.coo" files to locate the standard star in each frame.  {\bf epar
+phot} until it resembles that of Fig.~\ref{photphot}.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{photphot} The parameter file for a run of {\bf phot}.}
+\end{figure}
+Note that we are specifying a {\it single} output file name
+(``standstuff" in this example); {\it all} the photometry output will be
+dumped into this single file, including things like the airmass and filter
+number.  Go ahead and execute {\bf phot}.
+You should see something much like that of Fig.~\ref{photrun} on the
+screen.
+\begin{figure}
+\vspace{5.5in}
+\caption{\label{photrun} Running {\bf phot} non-interactively
+on the standard stars.}
+\end{figure}
+We will discuss the output below under ``Examining the results".
+ 
+\subsubsection{Photometry by Eye}
+In this section we will discuss the case of selecting stars {\it
+without}
+running the automatic star-finding program, using the image display
+window and the cursor.  The first step is to {\bf epar phot} so it
+resembles that of Fig.~\ref{photeye}.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{photeye} Parameter file for {\bf phot} when stars will
+be selected interactively.}
+\end{figure}
+Note that we have replaced the {\bf coords} coordinate list with the
+null string (two adjacent double-quotes) and turned ``interactive" on.
+ 
+We need to display the frame we are going to work on in the imtool
+window:
+ 
+\centerline { {\bf display std145 1} }
+ 
+\noindent
+will display image {\bf std145.imh} in the first frame buffer.
+ 
+Now let's run {\bf phot}.  We are not likely to be {\it too} accurate
+with where we place the cursor, so to be generous we will increase the
+allowable center shift to 3 pixels; otherwise we will get error messages
+saying that the ``shift was too large":
+ 
+\centerline{ {\bf phot std145 maxshift=3.}  }
+ 
+\noindent
+(Note that even though {\bf maxshift} is a parameter of {\bf centerpars}
+we can change it on the command line for {\bf phot}.) Also note that we
+left off the ``{\bf .imh}" extension for a reason: we are going to take
+the default names for the output files, and they will be given names
+such as {\bf std145.mag.1} and so on. If we had included the {\bf .imh}
+extension would would now be getting {\bf std145.imh.mag.1} names.
+ 
+At this point I get a flashing circle in my {\bf imtool} window; I don't
+know what you get (it depends upon how your defaults are set up) but
+there should be some sort of obvious marker on top of your image.
+Put it on the first star you wish to measure and hit the space bar.  The
+coordinates and magnitude should appear in the {\bf gterm} window, and
+you are ready to measure the next star on this frame.  Proceed until all
+the stars on this frame are measured, and then type a ``q" followed by
+another ``q".  Display the next frame, and run {\bf phot} on it.
+ 
+When you get done you will have  kerjillions of files.  
+ 
+\subsection{Examining the Results: the power of {\bf txdump }}
+ 
+Depending upon which of the two methods you selected you will either
+have a single file {\bf standstuff} containing the results of all your
+aperture photometry, or you will have a file for each frame ({\bf
+stand145.mag.1}, {\bf stand146.mag.1} \ldots)containing the stars
+on each frame.  In either event the file will pretty much resemble that
+shown in Fig.~\ref{photphotout}.
+\begin{figure}
+\vspace{7.5in}
+\caption{\label{photphotout} Output file from {\bf phot}.}
+\end{figure}
+The file begins with a large header describing the parameters in
+force at the time that {\bf phot} was run.  There is, however, a real
+subtlety to this statement.  If you had changed a parameter in {\bf
+datapars}, say, (or any of the other parameters) between running {\bf
+daofind} and {\bf phot}, the header in {\bf phot} will reflect only the
+setting that was in force at the time that {\bf phot} was run---in other
+words, it does not take the values of what was used for the {\bf
+threshold} from the coordinate file and retain these, but instead simply
+copies what value of {\bf thresh} happens to be in {\bf datapars} at the
+time that {\bf phot} is run.  To those used to the
+``self-documenting" feature of VMS DAOPHOT this is a major change!
+ 
+Once we get past the header information we find that there are 5 lines
+per star measured.  The ``key" to these five lines of information are
+found directly above the measurement of the first star. On the first
+line we have ``general information" such as the
+image name, the beginning x and y values, the id,
+and the coordinate file.  On the next line we have all the centering
+information: the computed x and y centers,
+the x and y shift, and any centering errors.  On the third line of the
+file we have information about the sky.  On the fourth line we have some
+information out of the image header: what was the integration time, what
+was the airmass, and what was the filter. Note
+that {\bf phot} has used that integration time in producing the
+magnitude---the exposures are now normalized to a 1.0 sec exposure.
+The fifth line gives the actual photometry, including the size of the
+measuring aperture, the total number of counts within the aperture, the
+area of the aperture, and the output magnitude, photometric error, and
+any problems encountered (such as a bad pixel within the aperture).
+ 
+We can extract particular fields from this file (or files) by using the
+{\bf txdump} command.  For instance, are there any cases where there
+there were problems in the photometry?  We can see those by saying
+ 
+\centerline{\bf txdump standstuff image,id,perror}
+ 
+\noindent
+(If you did ``Photometry by eye" you can substitute {\bf *mag*} for {\bf
+standstuff}.)
+When it queries you for the ``boolean expression" type
+ 
+\centerline{ {\bf perror!$=$"No\_error"} }
+ 
+\noindent
+The ``!$=$" construction is IRAF-ese for "not equal to"; therefore, this
+will select out anything for which there was some problem in the
+photometry. 
+ 
+We can create a single file at this point containing just the
+interesting results from the photometry file(s): do a
+ 
+\centerline{ {\bf txdump standstuff
+image,id,ifilt,xair,mag,merr yes $>$ standsout} }
+ 
+\noindent
+to dump the image name, id-number, filter, airmass, magnitude,
+and magnitude error into a file {\bf standsout}.  (Again, if you did
+``Photometry by Eye" substitute {\bf *mag*} for {\bf standstuff}).
+Unfortunately, what you do with this file is up to you right now until
+the standard reductions routines become available.  In the example shown
+here we have selected the fields in the same order as used in Peter
+Stetson's VMS CCDCAL software, and at the end of this manual we will
+describe a (painful) kludge that nevertheless {\it will} let you use
+these numbers with that software. 
+ 
+\section{Crowded Field Photometry: IRAF/daophot} 
+\subsection{Historical Summary}
+ 
+In the beginning (roughly 1979) astronomers 
+interested in obtaining photometry from stars in ``relatively" crowded fields
+would make the journey to Tucson in order to use Doug Tody's RICHFLD
+program which ran on the IPPS display system. 
+RICHFLD allowed the user to define a
+point-spread-function (PSF), and then fit this PSF to the brightest star
+in a group, subtract off this star, and then proceed to the next
+brightest star, etc.  This represented a giant qualitative improvement
+over the possibilities of aperture photometry, and allowed stars
+separated by a few FWHM's to be accurately measured.  
+ 
+Beginning in 1983, a group of RICHFLD users at the DAO (including
+Ed Olszewski and Linda Stryker) began modifications to the ``poorman"
+program of Jeremy Mould. This was largely motivated by the
+implementation of the ``Kitt Peak CCD" at the prime-focus of the Tololo
+4-m, and the idea was to design a crowded-field
+photometry
+program that (a) allowed simultaneous PSF-fitting,  (b) made
+use of the {\it known noise characteristics of a CCD} to do the fitting
+in a
+statistically correct manner (i.e., to make ``optimal" use of the data),
+and (c) to be largely batch oriented.
+In mid-1983 Peter Stetson arrived at the DAO, and took over
+the effort.  The result was
+DAOPHOT, which did all these things and more. 
+By 1986 DAOPHOT was well distributed within the astronomical community.
+The basic algorithms and philosophy can be found in Stetson 1987 (PASP
+{\bf 99}, 111).
+ 
+DAOPHOT (and its companion program ALLSTAR) were not part of a
+photometry
+package; they were instead stand-alone Fortran
+programs which did not deal in any way with the issue of image display
+or what to do with the instrumental magnitudes once you had them.  They
+were also only supported on VMS, although several ``frozen" versions
+were translated into UNIX by interested parties around the country.  
+There was therefore
+much to be gained from integrating the algorithms of daophot 
+with IRAF in order to make use of
+the image display capabilities and general tools for manipulating
+images.  Also, since many astronomers were now reducing their CCD data
+with IRAF, it avoided the necessity of translating the IRAF files into
+the special format needed by VMS DAOPHOT.  Dennis Crabtree began this
+translation program while at the DAO; it was taken over by Lindsey Davis
+of the IRAF group in early 1989, and taken to completion in early 1990.
+Pedro Gigoux of CTIO is currently hard at work on the photometry
+reduction package, scheduled for completion sometime during the spring.
+ 
+\subsection{{\bf daophot}
+Overview}
+The steps involved in running daophot are certainly more involved than
+in simple aperture photometry, but they are relatively straightforward.
+The following sections will lead you through the necessary procedures.
+Alternative routes will be noted at some points, and more may be gleaned
+from reading the various "help" pages.  A general outline is given here
+so that you have some overview in mind; a detailed step-by-step summary
+is provided at the end of this section.  
+ 
+\begin{itemize}
+\item Before you reduce the first frame, {\bf imexamine}  your data to
+determine FWHM's and the radius at which the brightest star you wish to
+reduce blends into the sky.  Run {\bf imhead} to find the ``key-words"
+in your data headers for exposure times, filter number, and airmass.
+Enter these, along with the characteristics of your chip (read-noise,
+photons per ADU, maximum good data value) 
+into the parameter sets {\bf datapars} and {\bf
+daopars}.
+\item Use {\bf daofind} and {\bf tvmark} 
+to produce a list of x and y positions of most
+stars on the frame.
+\item Use {\bf phot} to perform aperture photometry on the identified
+stars.  This photometry will be the basis of the zero-point of
+your frame via the PSF stars. This is also the only point where sky
+values are determined for your stars.
+\item Use {\bf psf} to define the PSF for your frame.  If your PSF stars are crowded this
+will require some iteration using the routines {\bf nstar} and {\bf
+substar}.
+\item Use {\bf allstar} to do simultaneous PSF-fitting for all the stars
+found on your frame, and to produce a subtracted frame.
+\item Use {\bf
+daofind} on the subtracted frame to identify stars that had been
+previously hidden.  
+\item Run {\bf phot} {\it on the original frame} to obtain aperture photometry
+and sky values for the stars on the new list.
+\item Use {\bf append} to merge the two aperture photometry lists.
+\item Run {\bf allstar} again on the merged list.
+\end{itemize}
+When you have done this for your {\it U, B,} and {\it V} frames it is
+then time to
+\begin{itemize}
+\item Use {\bf txdump}, {\bf tvmark}, and the image display
+capabilities to come up with a consistent matching between the frames.
+If there are additions or deletions then you will need to re-run
+{\bf phot} and {\bf allstar} one more time.
+\end{itemize} 
+Finally you will need to
+\begin{itemize}
+\item Determine the aperture correction for each frame by subtracting
+all but the brightest few isolated stars on your frames and then running
+{\bf phot} to determine the light lost between your zero-point aperture
+and the large aperture you used on your standard stars.
+\end{itemize}
+ 
+\subsection{How Big Is A Star: A Few Useful Definitions}
+
+The parameter files {\bf datapars} and {\bf daopars} contain three
+``size-like" variables, and although this document is not intended as
+a reference guide, there is bound to be confusion over these three
+parameters, particularly among those new to DAOPHOT.  In the hopes
+of un-muddying the waters, we present the following.
+
+\begin{description}
+\item[fwhmpsf] This is the full-width at half-maximum of a stellar object
+(point-spread function, or psf).  The value for {\bf fwhmpsf} gets used
+only by the automatic star-finding algorithm {\bf daophot}, unless you
+do something very bad like setting {\bf scale} to non-unity.
+
+\item[psfrad] This is the ``radius" of the PSF.  When you construct a PSF,
+the PSF will consist of an array that is
+$$(2 \times psfrad +1) \times
+(2 \times psfrad + 1)$$
+on a side.  The idea here is that ``nearly all" of the light of the brightest
+star you care about will be contained within this box.  If you were to construct
+a PSF with some large value of {\bf psfrad} and then run {\bf nstar} or
+{\bf allstar} 
+specifying  
+a smaller value of {\bf psfrad}, the smaller value would be used.  Making
+the {\bf psfrad} big enough is necessary to insure that the wings of some
+nearby bright star are properly accounted for when fitting a faint star.
+
+\item[fitrad] This is how much of the psf is used in making the fit
+to a star. The ``best" photometry will be obtained (under most circumstances)
+if this radius is set to something like the value for the fwhm.
+
+\end{description}
+
+\subsection{Setting up the parameter files ``daopars" and ``datapars" }
+ 
+The first step in using IRAF/daophot is to determine and store the
+characteristics of your data in two parameter files called ``datapars"
+and ``daopars"; these will be used by the various daophot commands.
+In Section 1 we discussed how to deal with parameter files, and 
+in Section 2 we went through setting up ``datapars" for the standard
+star solutions; at the risk of repeating ourselves, we will go through
+this again as the emphasis is now a little different.
+
+ 
+First inspect your headers by doing an {\bf imhead} imagename {\bf long+
+$|$ page}.  
+This will produce a listing similar to that shown in Fig.~\ref{newhead}.
+\begin{figure}
+\vspace{3.0in}
+\caption{\label{newhead}Header for image n602alu.imh.}
+\end{figure}
+The things to note here are (a) what the filter keyword is (we can
+see from Fig.~\ref{newhead} that the answer is F1POS; while there is
+an F2POS also listed, the second filter bolt was not used and was always
+in position ``zero"),
+(b) what the effective exposure
+time keyword is (EXPTIME in this example), and (c) what the effective
+airmass keyword is (AIRMASS in this example).
+ 
+Next you need to examine some ``typical" frames in order to determine
+the FWHM ({\bf fwhmpsf}) and the radius of the brightest star for which
+you plan to do photometry ({\bf psfrad}).
+First {\bf display} an image, and use the
+middle button of the mouse (or whatever you need to do on your image
+display) to zoom on a few bright stars.  On the SUN the "F6" key will
+let you see x and y values.  The ``default" PSF radius is 11 pixels:
+are your stars bigger than 23 pixels($23=2 \times 11 + 1$)
+pixels from one side to the other?  The FWHM is undoubtably variable
+from frame to frame, but unless these change by drastic amounts (factors
+of two, say) using a ``typical" value will doubtless suffice.  You can
+use the {\bf imexamine} routine to get some idea of the FWHM; do
+{\bf imexamine} filename and then strike the ``r" key (for radial
+profile) after centering the cursor on a bright (but unsaturated) star.
+The last number on the plot is the FWHM of the best-fit Gaussian.
+ 
+We are now ready to do an {\bf epar datapars}.  This parameter file
+contains information which is data-specific.  We set {\bf fwhmpsf} to the FWHM
+determined above, and we enter the names of the keywords determined from
+the header inspection above.  The ``gain" and ``read-noise" are values
+you have either determined at the telescope (using the Tololo routines)
+or which are carved in stone for your chip.  Choosing the value
+for datamax, the ``Maximum good data value",
+(in ADU's, NOT electrons) is a little bit trickier.  In the case of
+aperture photometry we were satisfied to take the nominal value for
+the chip, but point-spread-function fitting is a bit more demanding
+in what's ``linear".  The data obtained
+here was taken with an RCA chip, and we all know that RCA chips are
+linear well past 100,000 e-.  Thus, naively, we would expect that
+with a gain of 2.25 that the chip was still linear when we hit the
+digitization limit of 32,767 ADU's.  Subtract off 500 for the likely
+bias, and we {\it might} think that we were safe up to 32,200.  However,
+we would be wrong.  Experience with PSF fitting on these data shows that
+something (presumably in those little silver VEB's) has resulted in
+these data being non-linear above 20,000 ADU's.  My suggestion here is
+to start with the nominal value but be prepared to lower it if the
+residuals from PSF fitting appear to be magnitude dependent (more on this
+later).  The value for
+{\bf datamin}, the
+``Minimum good
+data value", will be different for each frame (depending what the sky
+level is) and there is not much point in entering a value for that yet.
+Similarly the value we will use for threshold will change
+from frame to frame depending upon what the sky level is.
+When you are done your {\bf datapars} should resemble that of
+Fig.~\ref{datapars}.
+\begin{figure}
+\vspace{2.7in}
+\caption{\label{datapars} A sample {\bf datapars} is shown.}
+\end{figure}
+ 
+Next we will {\bf epar daopars}.  This parameter file contains
+information specific to what you want {\bf daophot} to do.  The only things here
+we might want to change at this point are the ``Radius of the psf" {\bf psfrad}
+(if your experiment above showed it should be increased somewhat), and
+you might want to change the fitting radius {\bf fitrad}.  Leaving the fitting
+radius to ``something like" the FWHM results in the best SNR (you can
+work this out for yourself for a few different regimes if you like to
+do integrals).  The ``standard values" are shown in Fig.~\ref{daopars}.
+\begin{figure}
+\vspace{2.7in}
+\caption{\label{daopars} A sample {\bf daopars} is shown.}
+\end{figure}.
+ 
+\subsection{Finding stars: {\bf daofind} and {\bf tvmark} }
+The automatic star finder {\bf daofind} convolves a Gaussian of
+width FWHM with the image, and looks for peaks greater than some
+threshold in the smoothed image.  It then keeps only the ones that are
+within certain roundness and sharpness criteria in order to reject
+non-stellar objects (cosmic rays, background galaxies, bad columns,
+fingerprints).  We have already entered a reasonable value for the FWHM
+into {\bf datapars}, but what should we use as a threshold?  We expect
+some random fluctuations due to the photon statistics of the sky
+and to the read-noise of the chip. You can calculate this easily by
+first
+measuring the sky value on your frame by
+using {\bf imexamine} and the ``h" key to produce a histogram of
+the data ({\bf implot} and the ``s" key is another way).  In the example
+shown in Fig~\ref{hist} we see that the sky value is roughly 150.
+\begin{figure}
+\vspace{3.6in}
+\caption{\label{hist} The {\bf imexamine} histogram (``h" key) indicates
+that the sky value is roughly 150.}
+\end{figure}
+In general, if $s$ is the sky value in ADU, $p$ is the number of
+photons per ADU, and $r$ is the read-noise in units of electrons,
+then the expected $1\sigma$ variance in the sky
+will be 
+$$\left(\sqrt{s\times p + r^2}\right)/p$$ 
+in units of ADU's.  For the example here we expect
+$1\sigma=\left(\sqrt{150.\times 2.25 + 65^2}\right)/2.25=30$ ADU's.
+Of course, if you have averaged N frames in producing your image,
+then you should be using
+$N\times p$ as the gain both here and in the value entered in
+{\bf datapars}; similarly the readnoise is really just $r \times \sqrt{N}$.
+If instead you summed N frames then the gain is just {\it p} and the
+readnoise is still $r\times \sqrt{N}$.
+ 
+In the example shown here the expected $1\sigma$ variation of the sky is
+30 ADU's; we might therefore want to set our star detection threshold to
+3.5 times that amount.  That won't guarantee that every last star we
+find is real, nor will it find every last real star, but it should do
+pretty close to that!
+ 
+We should use this opportunity to set datamin in {\bf
+datapars} to some value like $s-3\sigma$.  In this case we will set it
+to 60.  This is not currently used by {\bf daofind} but will be used
+by all the photometry routines.  Fig.~\ref{ndatapars} shows the data
+parameters with the appropriate values of threshold and datamin now
+entered.
+\begin{figure}
+\vspace{3.0in}
+\caption{\label{ndatapars}  Datapars with {\bf threshold} and {\bf datamin}
+entered.} 
+\end{figure}
+ 
+We now can {\bf epar daofind} so it resembles that of
+Fig.~\ref{daofind}.
+\begin{figure}
+\vspace{3.0in}
+\caption{\label{daofind}  Parameters for {\bf daofind}.}
+\end{figure}
+Note that although nothing appears to be listed under {\bf datapars} the
+default name is ``datapars"; you could instead have created a separate
+data parameter file for each ``type" of data you have and have called
+them separate names (you could do this by doing an {\bf epar datapars}
+and then exiting with a ``:w newnamepar").  This might be handy if
+all your {\it U} frames were averages, say, but your {\it B} and {\it V} 
+frames were
+single exposures; that way you could keep track of the separate
+effective gain and readnoise values.  In that case you would enter the
+appropriate data parameter name under {\bf datapars}.  As explained earlier,
+you could also do a
+``:e" on the {\bf datapars} line and essentially do the {\bf epar datapars} from
+within the {\bf epar daofind}. 
+For normal star images, the
+various numerical values listed are best kept exactly the way they are;
+if you have only football shaped images, then read the help page for
+{\bf daofind} for hints how best to find footballs.
+
+We can now run {\bf daofind} by simply typing {\bf daofind}. 
+As shown in Fig.~\ref{daofind} that we were asked for the FWHM and threshold
+values; this is a due to having turned ``verify" on in the parameter
+set.  This safeguards to a large extent over having forgotten to set
+something correctly.  A [CR] simply takes the default value listed.
+ 
+Running {\bf daofind} produced an output file with the (default)
+filename of {\bf n602csb.coo.1}.
+(Do {\it not} give the {\bf .imh} extension
+when specifying the image name, or the default naming
+process will get very confused!) We can page
+through that and see the x and y centers, the number of magnitudes
+brighter than the cutoff, the sharpness and roundness values, and the
+star number.  However, of more immediate use is to use this file
+to mark the found stars on the image display and see how we did.
+If we have already displayed the frame in frame 1, then we can {\bf epar
+tvmark} to make it resemble Fig.~\ref{tvmark}.
+\begin{figure}
+\vspace{2.7in}
+\caption{\label{tvmark} Parameter file for {\bf tvmark}.}
+\end{figure}
+This will put red dots on top of each star found.
+ 
+We can see from Fig.~\ref{dots} that {\bf daofind} did a pretty nice
+\begin{figure}
+\vspace{7.0in}
+\caption{\label{dots} Stars found with {\bf daofind} and marked with
+{\bf tvmark}.}
+\end{figure}
+job.  If we didn't like what we saw at this point we could rerun
+{\bf daofind} with a slightly higher or slightly lower threshold---try
+varying the threshold by half a sigma or so if you are almost right.
+As you may have guessed, subsequent runs will produce output files with
+the names n602csb.coo.2, n602csb.coo.3,...
+If you are using a very slow computer, or are exceedingly impatient,
+ you could have saved some
+time by putting a ``c" (say) under ``convolv" in your first run of
+{\bf daofind}---this would have saved the
+smoothed image as cn602csb.imh, and would drastically reduce
+the number of cpu cycles needed to rerun {\bf daofind} with
+a different threshold value.
+If you really very happy with what {\bf daofind} did but you
+just want to add one or two stars at this point, you
+can in fact do that quite readily using {\bf tvmark}.  Set the
+parameters as in Fig.~\ref{tvmark}, but turn interactive on.  
+Position the cursor on top of the star you wish to add and strike
+the ``a" key.  Note that this will ``disturb" the format of the file,
+but we really don't care; it will still work just fine as the input to
+{\bf phot}.
+ 
+Note that it is fairly important that you do a good job at this stage.
+If you have used too low a threshold, and have a lot of junk marked as
+stars, these fictitious objects are likely to wander around during the
+PSF-fittings until they find something to latch onto---{\it not} a good
+idea.  However, you also do not want the threshold to be so high that
+you are missing faint stars.  Even if you are not planning to publish
+photometry of these faint guys, you need to have included them in the
+list of objects if they are near enough to affect the photometry of
+stars for which you do have some interest.  If you find that varying the
+threshold level does not result in a good list, then something is
+wrong---probably you have badly over- or under-estimated the FWHM.
+When you are close to the ``perfect" value of the threshold, 
+changing its value by as little as half a sigma will make a substantial
+difference between getting junk and real stars.
+ 
+\subsection{Aperture Photometry with {\bf phot} }
+The next step is to do simple aperture photometry for each of the stars
+that have been found.  These values will be used as starting points in
+doing the PSF fitting, and this is the only time that sky values will be
+determined.  
+ 
+{\bf One of the few ways of ``crash landing" in the current
+implementation of the software is to forget to reset ``datamin" in the
+datapars file before running phot on a new frame.  It is the only
+critical parameter which is not queried when verify is turned on.  Therefore,
+this is a good time to check to see that ``datamin" is really set to
+several sigma lower than the sky value of this particular frame.}
+ 
+The aperture photometry routine {\bf phot} has more parameters than all
+the others put together: there are the parameter files 
+{\bf centerpars}, {\bf fitskypars}, and {\bf photpars}.  
+Fortunately the ``verify"
+option frees you from having to look at these, and helps prevent you
+from making a mistake. If this is your first pass through DAOPHOT it is
+worth your while to do the following:
+ 
+\centerline{ {\bf unlearn centerpars} }
+ 
+\centerline{ {\bf unlearn fitskypars} }
+ 
+\centerline{ {\bf unlearn photpars} }
+ 
+\noindent
+If you have used {\bf phot} for measuring standard stars, then this will
+reset the defaults to reasonable values for crowded-field photometry; 
+in particular, we want to make sure that the centering
+algorithm in {\bf centerpars} is set to ``none".  
+Do an {\bf epar phot} and make it look like that of Fig.~\ref{phot}.
+Since we have the ``verify" switch turned on, we can be happy, not
+worry, and simply type {\bf phot}.
+{\bf phot} will then prompt you as shown in
+Fig.~\ref{phot}.
+\begin{figure}
+\vspace{7.0in}
+\caption{\label{phot} Questions and answers with {\bf phot}.}
+\end{figure}
+Note that the answers were particularly simple: we told it the name of
+the frame we wished to work with, we accepted the default for the coordinate
+list (it will take the highest ``version" of image.coo.NUMBER) and the
+default for the output photometry list (n602csb.mag.1 will be produced
+in this case.)  We accepted the centers from {\bf daofind} as being
+``good enough" to not have to recenter (they are good to about one-third
+of a pixel, plenty good enough for aperture sizes of 2.5 pixels and
+bigger; when we run this routine later on the second pass we would make
+a Big Mistake by turning centering on here, so leave it off). 
+The sky
+values will be taken from an annulus extending from a radius of 10
+pixels to a radius of 20 pixels, and it will determine the standard
+deviation of the sky from the actual data.  Note that this is probably a
+lot closer in than you used on your standard stars; in crowded regions
+of variable background keeping this annulus relatively close in will
+help.
+Finally, we used a measuring
+aperture of 3 pixels.  The number of counts within this aperture will be
+what defines the zero-point of your frame, as we will see in Section 4.9,
+and keeping this value {\it fixed} to some value like your typical FWHM
+will keep you safe.
+ 
+\subsection{Making the PSF with {\bf psf} }
+ 
+If you are used to the VMS version of DAOPHOT, you are in for a pleasant
+surprise when it comes to making a PSF within the IRAF version.
+Nevertheless, just because it's easy doesn't mean that you shouldn't be
+careful.
+ 
+What constitutes a good PSF star?  Stetson recommends that a good PSF
+star meets the following criteria:
+\begin{enumerate}
+\item No other star at all contributes any light within one fitting
+radius of the center of the candidate star. (The fitting radius will be
+something like the FWHM.)
+\item Such stars as lie near the candidate star are significantly
+fainter. (``Near" being defined as, say, 1.5 times the radius of the
+brightest star you are going to measure.)
+\item There are no bad columns or rows near the candidate star; there
+should also be no bad pixels near the candidate star.
+\end{enumerate}
+ 
+ 
+In making a PSF, you wish to
+construct a PSF which is free from bumps and wiggles (unless those
+bumps and wiggles are really what a single isolated star would look like.)
+First off, does it matter if we get the PSF ``right"?  If we had
+only isolated stars, then the answer would be no---any
+old approximation to the PSF would give you
+good relative magnitudes, and there are programs in the literature
+which do exactly this.  However, if your stars are relatively isolated
+you are not going to gain anything by PSF-fitting over aperture photometry
+anyway, so why bother?  If you are dealing with crowded images, then the
+PSF has to be right {\it even in the wings}, and for that reason we
+construct a PSF empirically using the brightest and least crowded stars
+in our frame.
+If you are very, very
+lucky you will find that your brightest, unsaturated star is well
+isolated, and has no neighbors about it---if that's the case, use that
+one and forget about the rest.  Usually, however, you will find that
+it isn't quite that easy, and it will be necessary to construct the PSF
+interatively.  The steps involved will be
+\begin{enumerate}
+	\item Select the brightest, least-crowded stars for the zeroth-order
+	      PSF.
+	\item Decrease the size of the PSF radius and fit these stars
+	      with their neighbors using {\bf nstar}.  
+	\item Subtract off the PSF stars and their neighbors using
+	      {\bf substar} to see
+	      if any of the PSF stars are ``funny"; if so, go back to
+	      the step 1 and start over.
+	\item Edit the {\bf nstar} results file ({\bf imagename.nst.N})
+	      and delete the entries for the PSF stars.  You are left
+	      with a file containing the magnitudes and positions of just
+	      the neighbors.
+	\item Subtract off just the neighbors using this file as input
+	      to {\bf substar}.  Display
+	      the results, and examine the region around each PSF star.
+	      Are the neighbors cleanly removed?
+	\item Increase the PSF radius back to the original value.
+	      Construct an improved PSF using the new frame (the one with the
+	      neighbors gone.)
+        \item Run {\bf nstar} on the PSF stars and their neighbors again, and
+	      again subtract these using {\bf substar}.  Examine the results.
+	      If you are happy, proceed; otherwise, if the neighbors need
+	      to be removed a bit more cleanly go back to step 4.  
+\end{enumerate}
+ 
+First {\bf display} the frame, and put dots on all the stars you've found
+using {\bf tvmark} as discussed above.  Next {\bf epar psf} and make sure
+it looks like that of Fig.~\ref{psfparams}.
+\begin{figure}
+\vspace{2.5in}
+\caption{\label{psfparams} Parameter file for {\bf psf}}
+\end{figure}
+We have set this up so we can choose the stars interactively from the
+display window.
+ 
+Next run {\bf psf}.  The defaults that you will be asked to {\bf verify}
+are probably fine, but pay particular attention to {\bf psf radius}
+and {\bf fitting radius}.  The {\bf psf radius} should be as large
+as you determined above (11 usually works well on ``typical" CCD
+frames whose star images have FWHM's $\approx 3$).  The ``fitting radius"
+should be relatively generous here---maybe even larger than what you
+want to use on your program stars.  A reasonable choice is approximately
+that of the FWHM. 
+ 
+You will find that the cursor has turned into a circle and is sitting
+on your image in the display window.  Position it on a likely looking
+PSF star, and strike the ``a" key. You will be confronted with a mesh
+plot that shows the star and it surroundings.  To find out more
+about the star (such as what the peak data value is you can type
+an ``s" while looking at the mesh plot.  To reject the star type an
+``x", to accept the star type an ``o".  In the latter case, you will
+next see a mesh plot that
+shows you the star with a two-dimensional Gaussian fit removed from the
+star.
+Again, exit this with a ``o".  If you don't find these mesh
+plots particularly useful, you can avoid them by setting {\bf showplot=no}
+in the {\bf psf} parameters (see Fig.~\ref{psfparams}).  
+At this point you will be told what the star number was, what the
+magnitude was, and what the minimum and maximum data values within
+the PSF were.   (If you picked a star whose peak intensity was greater
+than ``datamax" it will tell you this and not let you use this star.)
+When you are done selecting stars, type a ``w" (to write the PSF to
+disk) followed by a ``q". 
+ 
+If in making the PSF you noticed that there were stars you could have
+used but didn't because they had faint neighbors not found in the earlier
+step of star finding, you can add these by hand by simply 
+running {\bf tvmark} interactively and marking the extra stars.  First
+{\bf epar tvmark} so it resembles that of Fig.~\ref{tvmark}. Then:
+ 
+\centerline{ {\bf display n602csb 1} }
+ 
+\centerline{ {\bf tvmark 1 n602csb.coo.1 interactive+} }
+ 
+\noindent
+ 
+Striking the ``l" key will mark the stars it already knows about onto
+the display (as red dots this time around); positioning the cursor on the
+first star you wish to add and type an ``a".  When you are done adding
+stars exit with a ``q" and re-run {\bf phot}.
+ 
+Now that you have made your preliminary PSF, do a {\bf directory}.  You'll
+notice that in addition to the image {\bf n602csb.psf.1.imh} that the
+{\bf psf} routine has also added a text file {\bf n602csb.psg.1}.  If
+you {\bf page} this file you will see something like that of Fig.~\ref{psg}.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{psg} The ``point spread function group" file 
+{\bf n602csb.psg.1}}
+\end{figure}
+This contains the aperture photometry of each PSF star plus its neighbors,
+with each set constituting a ``group".  Running the psf-fitting photometry
+routine {\bf nstar} will fit PSF's to each of the stars within a group
+simultaneously.
+ 
+Before we run {\bf nstar}, however, we must decide what psf radius to use.
+Why not simply keep it set to the value found above (e.g., something like 11
+pixels)?  The answer to this is a bit subtle, but understanding it will
+help you diagnose what is going wrong when you find a PSF going awry (and
+don't worry, you will).  Let's consider the case that you construct a PSF
+from a single star with one neighbor whose center is 12 pixels away from
+the center of the PSF star, and let's have the PSF radius be 11 and the PSF
+fitting radius be 3. The PSF looks something like that of Fig.~\ref{bump}.
+\begin{figure}
+\vspace{5.0in}
+\caption{\label{bump} The zeroth order PSF of a star with a neighbor 12 pixels
+away.}
+\end{figure}
+The light from the neighbor star ``spills
+over" into the PSF. 
+ 
+What happens when you try to fit two PSF's simultaneously?  The bump from the
+PSF of the brighter star sits within the fitting radius of the fainter star,
+and it is the sum of the PSF's which are being fit to each star (that's
+what ``simultaneous" means).  Thus there is an ``implicit subtraction" of
+the fainter star simply from fitting the bumpy PSF to the brighter star,
+and the brightness of the fainter star will be underestimated.  The way
+to avoid this is to see that the PSF of the brighter star does not come
+within the fitting radius of the fainter star, and {\it that} we can 
+accomplish easily by truncating the PSF size to something like the separation
+of the two stars minus the fitting radius.  Thus in the example here
+we would want to fit the two stars using PSF's that were only ($12-3=9$)
+pixels in radius. It's true that there may still be light of the PSF
+star beyond this radius, but that will matter only if the PSF star is still
+going strong when you get within the {\it fitting radius} of the fainter
+star.  
+ 
+Now that we understand all that, run {\bf nstar}.  Specify the appropriate
+image name for ``image corresponding to photometry" and  give it
+the ``.psg" file {\bf n602csb.psg.1} for the ``input group file".
+Remember to decrease
+the {\bf psf radius} when it tries to verify that number.  {\bf nstar}
+will produce a photometry output file {\bf n60csb.nst.1}.
+You can
+subtract the fitted PSF's from these stars now by running {\bf substar}.
+Again, {\bf verify} the PSF radius to the smaller value.  When the routine
+finishes, {\bf display} the resultant frame {\bf n60csb.sub.1.imh} and
+take a look at the PSF stars...or rather, where the PSF stars (and their
+neighbors) were.  Are they subtracted cleanly?  Does one of the PSF
+stars have residuals that look the reverse of the residuals of the others?
+If so, it would be best to reconstruct the PSF at this point throwing out
+that star---possibly it has a neighbor hidden underneath it, or has something
+else wrong with it.  Are the variations in the cores of the subtracted image
+consistent with photon statistics?  To answer this you may want to play
+around with {\bf imexamine} on both the original and subtracted images,
+but if the stars have cleanly disappeared and you can't even tell where
+they were, you are doing fine. 
+ 
+The worst thing to find at this point
+is that there is a systematic pattern with position on the chip.  This
+would indicate that the PSF is variable.  There is the option for making
+a variable PSF, but the assumption is that the PSF varies smoothly in x
+and
+y; usually this is not the case.  (In the case of the non-flat TI chips
+the variations are due to the potato-chip like shape.)  If you {\it do}
+decide the PSF is variable, be sure to use plenty of stars in making the
+PSF.  As it says in the ``help page",
+twenty-five to thirty is then not an unreasonable number.  If that
+doesn't scare you off, nothing will.
+ 
+If the brightest stars have residuals that are systematically different than 
+those of the fainter stars, maybe that chip wasn't quite as linear as you
+thought, or perhaps there are charge transfer problems.  This proved to
+be the case for the RCA CCD data being reduced here.  In Fig.~\ref{yuko}
+we show the residuals that result when we based our PSF on a star whose
+peak counts were 30000 ADUs.  
+Empirically we found that stars with peaks of 18K ADUs (a mere 40K electrons)
+were safe to use, with the result that the dynamic range of our data
+was simply not quite as advertised.  Although the PSF function broke down
+above 18K, the chip remained ``linear" in the sense that aperture photometry
+continued to give good results---the total number of counts continued to
+scale right up to the A/D limit of 32,767 ADUs (72K electrons after bias
+is allowed for).   This appears to be a subtle charge transfer
+\begin{figure}
+\vspace{7.0in}
+\caption{\label{yuko} A ``before" and ``after" pair of images, where the
+PSF was constructed with a star that was too bright.  Note the systematic
+residuals for the two bright stars.  A ``bad" PSF star would result in a
+similar effect; however, in these data we found that there was always a
+systematic effect if the PSF stars were about 18000 ADU.}
+\end{figure}
+problem.
+ 
+We will assume that you have gotten the PSF to the point where
+the cores of the stars disappear cleanly, although there may be residuals
+present due to the neighbors.  Our next step is to get rid of these neighbors
+so that you can make a cleaner PSF.  Edit the {\bf nstar} output file
+{\bf n602csb.nst.1} and delete the lines associated with the PSF stars,
+leaving only the neighbors behind.  You can recognize the PSF stars, as
+they are the first entry in each group.  When you are done with this
+editing job, re-run {\bf substar}, using the edited ``.nst" file as the
+photometry file.  Again in running {\bf substar} make sure you {\bf verify}
+the PSF radius to the smaller value you decided above.  Examine the results
+on the image display.  Now the PSF stars should be there but the neighbors
+should be cleanly subtracted.  Are they?  If so, you are ready to proceed.
+If not, re-read the above and keep at it until you get those neighbors
+reasonably well out of the frame.
+ 
+We can now run {\bf psf} on the subtracted frame---the one with only the
+neighbors gone.  We have added some noise by doing the subtraction, and
+so we should reset {\bf datamin} to several sigma below the previously
+used
+value. We are going to have to do more typing this time when
+we run it, as the defaults for things will get very confused when we
+tell it that the ``Image for which to build PSF" is actually
+{\bf n60csb.sub.1}.  For the ``Aperture photometry file" we can tell
+it the original photometry file {\bf n602csb.mag.1} if we want, or
+even the old ``.psg" file {\bf n602csb.psg.1} since every star that
+we are concerned about (PSF star plus neighbor) is there.  Go ahead
+and give it the next `version" number for the ``Output psf image"
+{\bf n602csb.psf.2} and for the ``Output psf group file" 
+{\bf n602csb.psg.2}.  
+We can of course do this all on the command line:
+ 
+\centerline{ {\bf psf n602csb.sub.1 n602csb.mag.1 n602csb.psf.2
+n602csb.psg.2 datamin=-150.} }
+ 
+\noindent
+An example is shown in Fig.~\ref{psf1}.  
+{\it This time make sure you take the
+large psf radius.}
+\begin{figure}
+\vspace{7.0in}
+\caption{\label{psf1} Making the first revision PSF using the frames with the
+neighbors subtracted. Compare this to Fig. 23, which shows the
+same region before the neighbors have been removed.}
+\end{figure}
+Make a new PSF using the cursor as before.
+ 
+How good is this revised PSF?  There's only one way to find out: run
+{\bf nstar} on the original frame, this time keeping the psf radius large.
+Then do {\bf substar} and examine the frame with both the PSF stars and
+neighbors subtracted.  Does this show a substantial improvement over the
+first version?  Now that you have a cleaner PSF it may be necessary to repeat
+this procedure (edit the {\bf n602csb.nst.2} file, remove the PSF stars,
+run {\bf substar} using this edited file to produce a frame with the
+just the neighbors subtracted this time using a better PSF, run {\bf psf}
+on this improved subtracted frame) but probably not.
+ 
+\subsection{Doing the psf-fitting: {\bf allstar}.}
+The next step is to go ahead and run simultaneous PSF-fitting on all
+your stars, and produce a subtracted frame with these stars removed.
+To do both these things you need only run {\bf allstar}.  The defaults
+are likely to be right: see Fig.~\ref{allstar}.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{allstar} Running {\bf allstar}.}
+\end{figure}
+As you may imagine, {\bf allstar} produces a photometry file
+{\bf n602csb.als.1}, and another subtracted image: {\bf imagename.sub.N}.
+ 
+Display the subtracted frame, and blink it against the original.  Has
+IRAF/daophot done a nice job?  If the stars are clearly gone with a few
+hidden ones now revealed, you can be proud of yourself---if the results
+are disappointing, there is only one place to look, and that is in the
+making of the PSF.  Assuming that all is well, it is now time to 
+add those previously hidden stars into the photometry.  
+The easiest way to do this is to run {\bf daofind} on the subtracted
+image. 
+Set the value of {\bf datamin} to a value several sigma lower
+than what you had used earlier in case the subtraction process generated
+some spuriously small values, and you will want to {\it increase} the
+value of threshold by 1 or 2 sigma above what you used previously.
+Why?  Because the subtraction process has certainly added noise to the
+frame, and if you don't do this you will be mainly adding spurious
+detections.  Use {\bf tvmark} as before to examine the results of {\bf
+daofind}; remember that the coordinate file name will be 
+{\bf imagename.sub.N.coo.1} this time around.  If you are really close,
+but want to add a couple of stars, re-run {\bf tvmark} on this file
+using
+{\bf interactive+}; this will allow you to add (and delete) coordinates
+from the file.
+ 
+Now run {\bf phot} using this new coordinate file as the input list.
+However, you do want to use the {\it original} frame for this photometry;
+otherwise the sky values for the newly found stars will be very messed
+up owing to the many subtracted images.  A new aperture photometry file
+{\bf n602csb.mag.2} will have been produced.  Use {\bf append} to 
+concatenate these two files: {\bf append n602csb.mag.1,n602csb.mag.2
+n602csb.mag.3}.  You can now re-run {\bf allstar} using this combined
+photometry file as the input.
+ 
+\subsection{Matching the frames}
+In the example here we have been reducing the {\it B} frame of 
+a set of {\it UBV}.  Once all three frames have been reduced it is often
+necessary to do a little fiddling. Have the same stars been identified
+in each group?  In many cases you don't want the same stars to have been
+identified in each clump---afterall, some stars are red, some are blue
+(that's presumably why you are doing this afterall, right?), but in some
+cases you may find that a clump was identified as three objects on the
+{\it U} and the {\it V} frames and clearly should have been three on the
+{\it B} frame but instead is four or two. What to do?
+ 
+Using {\bf tvmark} it is relatively easy to set this right.  First we
+need to use {\bf txdump} to produce a file for each frame that can be
+displayed.  Do something like an 
+ 
+\centerline{ {\bf txdump n602csu.als.2 $>$ tvu}}
+ 
+\noindent
+followed by an 
+ 
+\centerline{ {\bf txdump n602csb.als.2 $>$
+tvb}} 
+ 
+\noindent
+and a 
+ 
+\centerline{ {\bf
+txdump n602csv.als.2 $>$ tvv}}
+ 
+\noindent
+In each case select {\bf xc,yc} and use
+{\bf MAG!=INDEF} as a selection criteria.  Thus you will then have three text
+files that contain only the x's and y's of the stars with photometry.
+ 
+Next display the three frames ({\bf display n602csu 1}, {\bf display
+n602csb 2}, {\bf display n602csv 3}) and put colored dots up to denote
+the different allstar stars: 
+ 
+\centerline{ {\bf tvmark 1 tvu color=204 inter-},}
+ 
+\centerline{
+{\bf tvmark 2 tvb color=205 inter-},}
+ 
+\noindent
+and 
+ 
+\centerline{ {\bf tvmark 3 tvv color=206
+inter-}}
+ 
+\noindent
+will give pleasing results.  Zoom, pan, register, and blink
+around the frames until you are convinced that you really do want to
+add or delete a star here or there.  If you want to add or delete a star to the
+{\it U} frame list, do a 
+ 
+\centerline{ {\bf tvmark 1 tvu color=203 inter+}}
+ 
+\noindent  
+You are
+now in interactive mode, and centering the cursor on the star you want
+to add and striking the ``a" key will append the x and y value of the
+cursor the tvu list.  Similarly, striking the ``u" key
+will delete a star from the list if you are using IRAF v2.9 or later.  
+(For earlier versions you are just going to have to do a little
+editing by hand, good luck!) The star you add or delete will have
+a white dot appear on top of it. 
+If you need to switch to a different coordinate file, simply exit the
+interactive {\bf tvmark} with a ``q" and re-execute it specifying, for
+example, {\bf tvmark 3 tvv color=203 inter+}. 
+ 
+When you are done with adding and deleting stars, then it is time to
+redo the photometry.  Do a {\bf phot n602csu coords=tvv datamin=100}
+in order to generate new aperture photometry and sky values.  These
+can then be run through {\bf allstar}, and the procedure repeated for
+each
+of the frames.
+ 
+\subsection{Determining the Aperture Correction}
+ 
+The zero-point of your magnitudes have been set as follows.  When you
+ran {\bf phot} using a small aperture (3 pixels in the example above)
+magnitudes were defined as -2.5 * log{(Counts above sky)/(Exposure
+time)} + Const.
+(The constant Const was hidden away in {\bf photpars} and is the
+magnitude assigned to a star that had a total of  one ADU per second 
+within the measuring aperture you used.)  When you defined your PSF the
+magnitudes of the PSF stars determined from the aperture photometry were
+then used to set the zero-point of the PSF.  However, your standard
+stars were presumably measured (if you did things right) through a much
+larger aperture, and what we must do now is measure how much brighter
+the PSF would have been had its zero-point been tied to the same size
+aperture used for the standard stars.  
+ 
+We need to determine the aperture correction from the brightest, 
+unsaturated stars (so there will still be reasonable signal above sky
+at the size of the large aperture); if you can pick out stars that are
+reasonably well isolated, so much the better.  If this sounds vaguely
+familiar to you, you're right---this is basically what you did for
+selecting PSF stars, and these would be a good starting point for
+selecting stars for determining the aperture correction.  Ideally you
+would like to use at least five such stars, but since when is data
+reduction ideal?  Nevertheless, it is in the determination of the
+aperture correction the largest uncertainty enters in doing CCD
+photometry on crowded fields.
+ 
+We will first need to pick out the brightest, isolated stars and then
+to subtract off any stars that might affect their being measured through
+the large ``standard star" aperture (e.g., something like 15 pixels).
+To do this we need good photometry of any of these neighbor stars, and
+we describe two ways to do this (1) the very long complicated way, and
+(2) the very short easy way:
+ 
+\begin{enumerate}
+ 
+\item {\bf Method 1: Using the image display}
+We can also use {\bf tvmark} to mark the stars that we wish to use for
+aperture photometry.  First we should remind ourselves what are multiple
+stars and what aren't: {\bf display} the image, and then use {\bf
+tvmark} to mark the stars with {\bf allstar} photometry:
+ 
+\centerline{ {\bf display n602csb 1} }
+ 
+\centerline{ {\bf txdump n602csb.als.2 xc,yc yes $>$ tvb} }
+ 
+\centerline{ {\bf tvmark 1 tvb color=204 interact-} }
+ 
+\noindent
+Now go through and mark the stars you want to use as the aperture
+correction stars {\it plus any neighbors that might contribute light
+to a large aperture centered on the bright stars:}
+ 
+\centerline{ {\bf tvmark 1 bapstars color=203 interact+ }}
+ 
+\noindent
+Use the ``a" key to generate a list ({\bf bapstars}) of the approximate
+{\it x} and {\it y} positions of these stars.  Next run this list
+through {\bf phot} to generate improved centers and good sky values:
+ 
+\centerline{ {\bf phot n602csb bapstars bapphot calgor=``centroid" } }
+ 
+\noindent
+Next run the photometry output file {\bf bapphot} through {\bf group}:
+ 
+\centerline{ {\bf group n602csb bapphot default default crit=0.2} }
+ 
+\noindent
+This will have generated a ``group" file {\bf n602csb.grp.1}.
+ 
+\noindent 
+Finally (!) run this group file through {\bf nstar}:
+ 
+\centerline{ {\bf nstar n602csb default default default} }
+ 
+\item {\bf Method 2: Using the ``.psg" files}
+If you used a goodly number ($>3-5$, say) stars in
+making the PSF, then we will simply use these stars as the aperture
+correction stars.  Your last {\bf nstar} run should have produced an
+``{\bf .nst}" file that contains good photometry for the PSF stars {\it
+and} their neighbors. (If you don't remember if you did this, run {\bf
+nstar} using the ``{\bf .psg}" as the input group file.)  Note that this
+method relies upon the assumption that the sum of the psf radius and psf
+fitting radius is about as large as the size of the large aperture you
+will use, so that all the important neighbors have been included in the
+point-spread-function group, but this is probably a reasonable
+assumption.
+ 
+\end{enumerate}
+ 
+Now that we are done with the preliminaries (!!),
+we now want to produce two files: one of them containing only the
+neighbors that we wish to subtract off, and another containing only the
+bright isolated stars which we want to use in computing the aperture
+correction.  To do this we will use {\bf group} to divide up the ``{\bf
+.nst}" file (we could simply use the editor but that would be a lot of
+work).  First we will use {\bf txdump} on the {\bf nstar} file to see the magnitude
+range covered by the PSF stars and their neighbors: hopefully there
+won't be any overlap.  To do this try
+ 
+\centerline{ {\bf txdump n602csb.nst.3 id,group,mag yes} }
+ 
+\noindent
+In the example shown in Fig.~\ref{grouping} we see that the PSF stars
+\begin{figure}
+\vspace{2.0in}
+\caption{\label{grouping} The three PSF stars and their groups.}
+\end{figure}
+have magnitudes of 13.9, 15.0, and 16.5 in the three groups; all the
+neighbor stars are fainter than 17.0.  Thus we can use {\bf select} 
+to create a file containing the
+photometry of the faint stars:
+ 
+\centerline{ {\bf select n602csb.nst.3 n602csbsub} }
+ 
+\noindent 
+and answer {\bf MAG$>$17.0} when you are queried for the ``Boolean
+expression".  This will put the photometry of the stars you wish to get
+rid of into the file {\bf n602csbsub}.  Next do an 
+ 
+\centerline{ {\bf txdump n602csb.nst.3 xc,yc $>$ n602csbap} }
+ 
+\noindent
+and answer {\bf MAG$<$17.0} in response to ``Boolean expression".  This
+will put the {\it x} and {\it y} values of the stars we wish to use for
+the aperture correction into the file
+{\bf n602csbap}.  Next subtract the stars in the first file:
+ 
+\centerline{ {\bf substar n602csb n602csbsub} }
+ 
+\noindent and accept the defaults.  This will result in the subtracted
+image {\bf n602csb.sub.N}.  It is this file on which we wish to run
+the aperture photometry to determine the aperture correction:
+ 
+\centerline{ 
+{\bf phot n602csb.sub.N n602csbap n602csbapresults apertures=3.,15. annulus=20. dannu=5.} }
+ 
+\noindent
+You will see something like Fig.~\ref{apcor1} on your terminal.
+In this example we've made the assumption that the aperture size that
+set your zero-point in making the PSF was 3 pixels (i.e., what you used
+with {\bf phot} Way Back When), and that the aperture size used on your
+standard stars was 15 pixels.
+\begin{figure}
+\vspace{3.0in}
+\caption{\label{apcor1} The aperture correction run of {\bf phot}.}
+\end{figure}
+It is time to drag out your hand calculator.  Using all three stars we
+find an average aperture correction of $-0.371$ with a standard
+deviation of the mean of 0.012 mag; given the large range in magnitude,
+I might have been tempted to ignore the two fainter stars and keep the
+aperture correction based only upon the brightest star (the frame is
+sparsely populated, and there isn't a whole heck of a lot else we can
+do).  By an amazing coincidence, the aperture correction based just on
+the brightest star is also $-0.371$.
+ 
+ 
+\subsection{{\bf daophot} summary}
+\begin{itemize}
+\item Set up {\bf datapars} and {\bf daopars}.
+	\begin{enumerate}
+	\item Do an {\bf imhead} on some image and note the keywords for the
+	filter position, the effective exposure time, and the effective
+	airmass.
+	\item Use {\bf display} and {\bf imexamine} on a few frames to 
+	determine the typical full-width-half-max
+	of stars and what would be a good
+	value to use for the radius of the psf (i.e., what radius will
+	contain the brightest star for which you wish to do photometry.)
+        \item Enter these into {\bf daopars} (psfrad) and {\bf datapars}
+	(header key words, fwhm).  Also check that the correct values
+	are entered in {\bf datapars} for the gain (photons per ADU)
+	and read-noise (in electrons), as well as the ``maximum good data
+	value".
+	\end{enumerate}
+\item Find stars.
+	\begin {enumerate}
+	\item Do an {\bf implot} or {\bf imexamine} to determine the sky
+	level on your frame.  Calculate the expected $1\sigma$ error.
+	\item Enter the sky value minus 3$\sigma$ as your value for
+	{\bf datamin} in {\bf datapars}.
+	\item Run {\bf daofind} using as a threshold value 3 to 5 $\sigma$.
+	\item Use {\bf tvmark} to mark the stars found ({\bf imagename.coo.1}).
+	If you need to, rerun {\bf daofind} with a larger or small
+	threshold.
+	\end {enumerate}
+\item Run aperture photometry using {\bf phot}.  
+\item Generate a PSF. Run {\bf psf} and add stars using the ``a" key.  Try
+      to select bright, uncrowded stars. Then:
+	\begin {enumerate}
+	\item Run {\bf nstar} using the file {\bf imagename.psg.1} as the
+	``input photometry group" file.  If there are neighbors, be sure
+	to decrease the psf radius as explained above.
+	Run {\bf substar}  (also using the smaller sized psf radius)
+	and display the
+	resultant subtracted frame {\bf imagename.sub.1}.  Do the residuals
+	of the PSF stars look consistent, or is one of them funny?  If need
+	be, start over. 
+	\item Remove any neighbor stars by editing the PSF stars out of the
+	``.nst"  file, and rerunning {\bf substar}.  Run
+	{\bf psf} on the subtracted file, using the normal psf radius again.
+	You will have to over-ride the defaults for the input and output file
+	names now that you are using the subtracted image.  Rerun {\bf nstar}
+	on the original frame using the normal psf radius and the revised
+	PSF. Run {\bf substar} and display the results.  Are the PSF stars
+	nicely removed, and do the areas around the PSF stars look clean?
+	It may be necessary to remove neighbors again using this revised
+	PSF.
+	\end {enumerate}
+\item Run {\bf allstar}.  Display the subtracted frame and see if your stars
+	have been nicely subtracted off.
+\item Run {\bf daofind} on the subtracted frame, using a value for 
+       {\bf threshold} which is another $\sigma$ or two larger than before,
+       and a value for {\bf datamin} which is several $\sigma$ lower than
+       before.  Use {\bf tvmark} to examine the results, and if need be
+       run {\bf tvmark} interactively so that you may add any extra stars.
+\item Run aperture photometry using {\bf phot} {\it on the original frame},
+      using the new coordinate list produced above.
+\item {\bf append} the two aperture photometry files.
+\item Run {\bf allstar} using the combine photometry file.
+\item Repeat all of the above for each frame in your ``set" (e.g., all short
+      and long exposures in each filter of a single field, say.
+\item Use {\bf txdump} to select the stars from the allstar files which
+      have magnitudes not equal to ``INDEF".  Mark these stars using
+      {\bf tvmark}, and then use the capabilities of the image display
+      and {\bf tvmark} to match stars consistently from frame to frame.
+      Rerun {\bf phot} and {\bf allstar} on the final coordinate lists.
+\item Determine the aperture corrections.
+\item Transform
+	to the standard system (see the next section) and then
+      publish the results.
+\end{itemize}
+\section{Transforming to the Standard System}
+ 
+This section will eventually tell you how to easily and painless obtain
+the transformation equations for going from your instrumental magnitudes
+to the standard system, and how to apply these transformation equations
+to your program fields.  Unfortunately, the IRAF routines for doing this
+are still under construction.
+In the meanwhile, we are providing here a kludge solution that can be
+used by initiates of Stetson's VMS CCDCAL routines.  If you haven't been
+made a member of the club yet, and don't feel like waiting until the
+IRAF routines are become available before you get results, then I would
+recommend getting a hold of the good Dr. Stetson and bribing him until he
+offers to send you a copy of CCDCAL.  There is an excellent manual that
+comes along with it, and we will not attempt to repeat any of that
+material here.
+ 
+\subsection{Standard Star Solution}
+First we will describe how to get output good enough to fool
+the CCDCAL software into believing the photometry was produced by CCDOBS
+(for the standard magnitudes), and what modifications need to be made
+to CCDSTD.FOR
+ 
+On the standard file do a {\bf txdump standstuff lid,ifilt,xair,mag,merr
+$>$ foolit} to dump the star number, filter number, airmass, and
+instrumental magnitudes and errors into the file {\bf foolit}.
+Unfortunately, you are now going to have to edit this file and stick in
+the star name (in what ever form you have it in creating the library of
+standard stars with CCDLIB) in place of the image name and star ID.
+(These were simply placed in the file to help guide you).  While you are
+at it, line up the filter numbers, airmasses, and magnitudes into nice,
+neat columns.  When you get done, stick in a line at the top that gives
+the number of instrumental magnitudes and their names, using a
+i1,13x,n(6x,a6) format.  For instance, in the case shown here there
+are 3 instrumental magnitudes, U, B, and V.  Finally, the filter numbers
+have to be edited so they agree with these (e.g., they must denote 
+instrumental magnitude 1, 2, and 3...now aren't you sorry you didn't
+decide to wait until the IRAF routines were finished?).  In
+Fig~\ref{groan} we show an example of the ``before" and ``after" file.
+\begin{figure}
+\vspace{3.5in}
+\caption{\label{groan}The output of {\bf txdump} and the final file
+ready for {\bf ccdstd}.  Note the switching of the filter number ``5"
+with  ``1".}
+\end{figure}
+ 
+CCDOBS.FOR itself now needs to be modified.  Search for line statement
+``1120" (which will say JSTAR=JSTAR+1).  Add a line that sets the
+integration time to 1 (tint=1.).  Modify the READ statement as shown
+in Fig.~\ref{ccdobs}, and finally modify the 213 FORMAT statement
+so it actually matches your data file.
+\begin{figure}
+\vspace{2.5in}
+\caption{\label{ccdobs} Modifications to CCDOBS.FOR}
+\end{figure}
+You should now be able to compile, link, and run this modified
+version of CCDOBS and have it work on your standard star data.
+ 
+\subsection{Program Stars}
+The work required for faking ``CCDCAL" is actually a lot less. The data
+files are easily produced.  Do a
+ 
+\centerline{{\bf txdump n602csu.als.2
+id,xc,yc,mag,merr,nit,chi $>$ csu} }
+ 
+\centerline{{\bf txdump n602csb.als.2 id,xc,yc,mag,merr,nit,chi $>$
+csb}}
+ 
+\centerline{{\bf txdump n602csv.als.2 id,xc,yc,mag,merr,nit,chi $>$
+csv}}
+ 
+\noindent
+answering {\bf MAG!=INDEF} to ``boolean expression" each time.  
+These three files ({\bf csu}, {\bf csb}, {\bf csv} can be used
+with CCDCAL once a single modification is made to CCDCAL.FOR: on
+statement number 2020 change the format to ``free format", e.g.,
+2020 IF(NL(IOBS).NE.2) READ(2,*,END=2040).  When CCDCAL queries
+you for an integration time, be sure to tell it 1.0, as your data have
+already been corrected for exposure times. 
+ 
+\section{Acknowledgements}
+We are grateful to Jeannette Barnes and Carol Neese for critical
+readings of this document, although final blame for style and content
+of course rests with the authors.
+\end{document}
diff --git a/noao/digiphot/daophot/doc/userdocs/daoref.ms b/noao/digiphot/daophot/doc/userdocs/daoref.ms
new file mode 100644
index 00000000..6ef3d1f6
--- /dev/null
+++ b/noao/digiphot/daophot/doc/userdocs/daoref.ms
@@ -0,0 +1,6290 @@
+.LP
+\0
+.de XS
+.DS
+.ps -1
+.vs -1p
+.ft CB
+..
+.de XE
+.DE
+.ft R
+.ps
+.vs
+..
+.de YS
+.nf
+.ps -1
+.vs -1p
+.ft CB
+..
+.de YE
+.fi
+.ft R
+.ps
+.vs
+..
+.RP
+.TL
+A Reference Guide to the IRAF/DAOPHOT Package
+.AU
+Lindsey E. Davis
+.AI
+IRAF Programming Group
+.K2
+.ce
+.TU
+.br
+.ce
+January 1994
+.AB
+.PP
+DAOPHOT is a software package for doing stellar photometry in crowded stellar
+fields
+developed by Peter Stetson (1987) of the Dominion Astrophysical
+Observatory. IRAF/DAOPHOT uses the task structure and
+algorithms of DAOPHOT to do crowded-field stellar photometry within the
+IRAF data reduction and analysis environment.
+.PP
+This document briefly describes the principal similarities and differences
+between DAOPHOT and IRAF/DAOPHOT, the data preparation required to
+successfully use IRAF/DAOPHOT, how to examine and edit the IRAF/DAOPHOT
+algorithm parameters, how to run the IRAF/DAOPHOT package tasks interactively,
+non-interactively, or in the background, and how to examine
+and perform simple database operations on the output photometry files.
+.PP
+This document is
+intended as a reference guide to the details of using and
+interpreting the results of IRAF/DAOPHOT not a user's cookbook or a general
+guide to doing photometry in IRAF. Its goal is to take the user
+from a fully reduced image of a crowded stellar field to aperture
+corrected instrumental magnitudes using a small artificial image as a
+sample data set.
+First time IRAF/DAOPHOT users
+should consult \fIA User's Guide to Stellar Photometry With IRAF\fR, by
+Phil Massey and Lindsey Davis. Detailed descriptions of the DAOPHOT photometry
+algorithms can be found in Stetson (1987, 1990, 1992).
+.AE
+.ds CH
+.bp
+\0
+.bp
+.PP
+.na
+.LP
+\fBContents\fP
+.sp 1
+1.\h'|0.4i'\fBIntroduction\fP\l'|5.6i.'\0\01
+.sp
+2.\h'|0.4i'\fBDAOPHOT and IRAF/DAOPHOT\fP\l'|5.6i.'\0\01
+.sp
+3.\h'|0.4i'\fBPreparing Data for DAOPHOT\fP\l'|5.6i.'\0\03
+.sp
+4.\h'|0.4i'\fBSome IRAF Basics for New IRAF and DAOPHOT Users\fP\l'|5.6i.'\0\04
+.br
+.sp
+\h'|0.4i'4.1.\h'|0.9i'\fBPre-loaded Packages\fP\l'|5.6i.'\0\04
+.br
+\h'|0.9i'4.1.1.\h'|1.5i'The DATAIO Package\l'|5.6i.'\0\05
+.br
+\h'|0.9i'4.1.2.\h'|1.5i'The PLOT Package\l'|5.6i.'\0\05
+.br
+\h'|0.9i'4.1.3.\h'|1.5i'The IMAGES Package\l'|5.6i.'\0\05
+.br
+\h'|0.9i'4.1.4.\h'|1.5i'The TV Package\l'|5.6i.'\0\05
+.br
+\h'|0.4i'4.2.\h'|0.9i'\fBOther Useful Packages and Tasks\fP\l'|5.6i.'\0\05
+.br
+\h'|0.4i'4.3.\h'|0.9i'\fBImage Types, Image Directories, and Image Headers\fP\l'|5.6i.'\0\05
+.br
+\h'|0.4i'4.4.\h'|0.9i'\fBThe Image Display and Image Cursor\fP\l'|5.6i.'\0\06
+.br
+\h'|0.4i'4.5.\h'|0.9i'\fBThe Graphics Device and Graphics Cursor\fP\l'|5.6i.'\0\07
+.sp
+5.\h'|0.4i'\fBSome DAOPHOT Basics for New DAOPHOT Users\fP\l'|5.6i.'\0\08
+.br
+.sp
+\h'|0.4i'5.1.\h'|0.9i'\fBLoading the DAOPHOT Package\fP\l'|5.6i.'\0\08
+.br
+\h'|0.4i'5.2.\h'|0.9i'\fBLoading the TABLES Package\fP\l'|5.6i.'\0\08
+.br
+\h'|0.4i'5.3.\h'|0.9i'\fBRunning the Test Script\fP\l'|5.6i.'\0\08
+.br
+\h'|0.4i'5.4.\h'|0.9i'\fBOn-line Help\fP\l'|5.6i.'\0\09
+.br
+\h'|0.4i'5.5.\h'|0.9i'\fBEditing the Package Parameters\fP\l'|5.6i.'\010
+.br
+\h'|0.4i'5.6.\h'|0.9i'\fBEditing the Task Parameters\fP\l'|5.6i.'\011
+.br
+\h'|0.4i'5.7.\h'|0.9i'\fBInput and Output Image Names\fP\l'|5.6i.'\011
+.br
+\h'|0.4i'5.8.\h'|0.9i'\fBInput and Output File Names\fP\l'|5.6i.'\012
+.br
+\h'|0.4i'5.9.\h'|0.9i'\fBAlgorithm Parameter Sets\fP \l'|5.6i.'\012
+.br
+\h'|0.4i'5.10.\h'|0.9i'\fBInteractive Mode and Non-Interactive Mode\fP \l'|5.6i.'\014
+.br
+\h'|0.4i'5.11.\h'|0.9i'\fBImage and Graphics Cursor Input\fP\l'|5.6i.'\014
+.br
+\h'|0.4i'5.12.\h'|0.9i'\fBGraphics Output\fP\l'|5.6i.'\015
+.br
+\h'|0.4i'5.13.\h'|0.9i'\fBVerify, Update, and Verbose\fP\l'|5.6i.'\015
+.br
+\h'|0.4i'5.14.\h'|0.9i'\fBBackground Jobs\fP\l'|5.6i.'\015
+.br
+\h'|0.4i'5.15.\h'|0.9i'\fBTiming Tests\fP\l'|5.6i.'\016
+.sp
+6.\h'|0.4i'\fBDoing Photometry with DAOPHOT\fP\l'|5.6i.'\016
+.br
+.sp
+\h'|0.4i'6.1.\h'|0.9i'\fBThe Test Image\fP\l'|5.6i.'\016
+.br
+\h'|0.4i'6.2.\h'|0.9i'\fBTypical Analysis Sequence\fP\l'|5.6i.'\017
+.br
+\h'|0.4i'6.3.\h'|0.9i'\fBCreating and Organizing an Analysis Directory\fP\l'|5.6i.'\019
+.br
+\h'|0.4i'6.4.\h'|0.9i'\fBReading the Data\fP \l'|5.6i.'\019
+.br
+\h'|0.4i'6.5.\h'|0.9i'\fBEditing the Image Headers\fP\l'|5.6i.'\019
+.br
+\h'|0.9i'6.5.1.\h'|1.5i'The Minimum Image Header Requirements\l'|5.6i.'\019
+.br
+\h'|0.9i'6.5.2.\h'|1.5i'The Effective Gain and Readout Noise\l'|5.6i.'\019
+.br
+\h'|0.9i'6.5.3.\h'|1.5i'The Maximum Good Data Value\l'|5.6i.'\021
+.br
+\h'|0.9i'6.5.4.\h'|1.5i'The Effective Exposure Time\l'|5.6i.'\022
+.br
+\h'|0.9i'6.5.5.\h'|1.5i'The Airmass, Filter Id, and Time of Observation\l'|5.6i.'\022
+.br
+\h'|0.9i'6.5.6.\h'|1.5i'Batch Header Editing\l'|5.6i.'\024
+.br
+\h'|0.4i'6.6.\h'|0.9i'\fBEditing, Checking, and Storing the Algorithm Parameters\fP\l'|5.6i.'\024
+.br
+\h'|0.9i'6.6.1.\h'|1.5i'The Critical Algorithm Parameters\l'|5.6i.'\024
+.br
+\h'|0.9i'6.6.2.\h'|1.5i'Editing the Algorithm Parameters Interactively with Daoedit \l'|5.6i.'\024
+.br
+\h'|1.5i'6.6.2.1.\h'|2.2i'The Data Dependent Algorithm Parameters \l'|5.6i.'\025
+.br
+\h'|1.5i'6.6.2.2.\h'|2.2i'The Centering Algorithm Parameters\l'|5.6i.'\028
+.br
+\h'|1.5i'6.6.2.3.\h'|2.2i'The Sky Fitting Algorithm Parameters\l'|5.6i.'\029
+.br
+\h'|1.5i'6.6.2.4.\h'|2.2i'The Aperture Photometry Parameters\l'|5.6i.'\029
+.br
+\h'|1.5i'6.6.2.5.\h'|2.2i'The Psf Modeling and Fitting Parameters\l'|5.6i.'\030
+.br
+\h'|1.5i'6.6.2.6.\h'|2.2i'Setting the Algorithm Parameters Graphically\l'|5.6i.'\031
+.br
+\h'|0.9i'6.6.3.\h'|1.5i'Checking the Algorithm Parameters with Daoedit\l'|5.6i.'\031
+.br
+\h'|0.9i'6.6.4.\h'|1.5i'Storing the Algorithm Parameter Values with Setimpars\l'|5.6i.'\032
+.br
+\h'|0.9i'6.6.5.\h'|1.5i'Restoring the Algorithm Parameter Values with Setimpars\l'|5.6i.'\032
+.br
+\h'|0.4i'6.7.\h'|0.9i'\fBCreating a Star List\fP\l'|5.6i.'\032
+.br
+\h'|0.9i'6.7.1.\h'|1.5i'The Daofind Task\l'|5.6i.'\033
+.br
+\h'|1.5i'6.7.1.1.\h'|2.2i'The Daofind Algorithm\l'|5.6i.'\033
+.br
+\h'|1.5i'6.7.1.2.\h'|2.2i'The Daofind Algorithm Parameters\l'|5.6i.'\033
+.br
+\h'|1.5i'6.7.1.3.\h'|2.2i'Running Daofind Non-Interactively\l'|5.6i.'\034
+.br
+\h'|1.5i'6.7.1.4.\h'|2.2i'Running Daofind Interactively\l'|5.6i.'\034
+.br
+\h'|1.5i'6.7.1.5.\h'|2.2i'The Daofind Output\l'|5.6i.'\036
+.br
+\h'|1.5i'6.7.1.6.\h'|2.2i'Examining the Daofind Output\l'|5.6i.'\037
+.br
+\h'|0.9i'6.7.2.\h'|1.5i'Rgcursor and Rimcursor\l'|5.6i.'\038
+.br
+\h'|0.9i'6.7.3.\h'|1.5i'User Program\l'|5.6i.'\039
+.br
+\h'|0.9i'6.7.4.\h'|1.5i'Modifying an Existing Coordinate List\l'|5.6i.'\039
+.br
+\h'|0.4i'6.8.\h'|0.9i'\fBInitializing the Photometry with Phot\fP\l'|5.6i.'\039
+.br
+\h'|0.9i'6.8.1.\h'|1.5i'The Phot Algorithm\l'|5.6i.'\039
+.br
+\h'|0.9i'6.8.2.\h'|1.5i'The Phot Algorithm Parameters\l'|5.6i.'\040
+.br
+\h'|0.9i'6.8.3.\h'|1.5i'Running Phot Non-interactively\l'|5.6i.'\040
+.br
+\h'|0.9i'6.8.4.\h'|1.5i'Running Phot Interactively\l'|5.6i.'\042
+.br
+\h'|0.9i'6.8.5.\h'|1.5i'The Phot Output\l'|5.6i.'\043
+.br
+\h'|0.9i'6.8.6.\h'|1.5i'Examining the Results of Phot\l'|5.6i.'\044
+.br
+\h'|0.4i'6.9.\h'|0.9i'\fBCreating a Psf Star List with Pstselect\fP\l'|5.6i.'\044
+.br
+\h'|0.9i'6.9.1.\h'|1.5i'The Pstselect Algorithm\l'|5.6i.'\045
+.br
+\h'|0.9i'6.9.2.\h'|1.5i'The Pstselect Algorithm Parameters\l'|5.6i.'\045
+.br
+\h'|0.9i'6.9.3.\h'|1.5i'How Many Psf Stars Should Be Selected ?\l'|5.6i.'\046
+.br
+\h'|0.9i'6.9.4.\h'|1.5i'Running Pstselect Non-interactively\l'|5.6i.'\047
+.br
+\h'|0.9i'6.9.5.\h'|1.5i'Running Pstselect Interactively\l'|5.6i.'\047
+.br
+\h'|0.9i'6.9.6.\h'|1.5i'The Pstselect Output\l'|5.6i.'\048
+.br
+\h'|0.9i'6.9.7.\h'|1.5i'Examining and/or Editing the Results of Pstselect\l'|5.6i.'\048
+.br
+\h'|0.4i'6.10.\h'|0.9i'\fBComputing the Psf Model with Psf\fP\l'|5.6i.'\049
+.br
+\h'|0.9i'6.10.1.\h'|1.5i'The Psf Algorithm\l'|5.6i.'\049
+.br
+\h'|0.9i'6.10.2.\h'|1.5i'Choosing the Appropriate Analytic Function\l'|5.6i.'\050
+.br
+\h'|0.9i'6.10.3.\h'|1.5i'The Analytic Psf Model\l'|5.6i.'\050
+.br
+\h'|0.9i'6.10.4.\h'|1.5i'The Empirical Constant Psf Model\l'|5.6i.'\051
+.br
+\h'|0.9i'6.10.5.\h'|1.5i'The Empirical Variable Psf Model\l'|5.6i.'\051
+.br
+\h'|0.9i'6.10.6.\h'|1.5i'Rejecting Bad Data from the Psf Model\l'|5.6i.'\051
+.br
+\h'|0.9i'6.10.7.\h'|1.5i'The Model Psf Psfrad and Fitrad\l'|5.6i.'\052
+.br
+\h'|0.9i'6.10.8.\h'|1.5i'Modeling the Psf Interactively Without a Psf Star List\l'|5.6i.'\052
+.br
+\h'|0.9i'6.10.9.\h'|1.5i'Fitting the Psf Model Interactively Using an Initial Psf Star List\l'|5.6i.'\054
+.br
+\h'|0.9i'6.10.10.\h'|1.5i'Fitting the Psf Model Interactively Without an Image Display\l'|5.6i.'\055
+.br
+\h'|0.9i'6.10.11.\h'|1.5i'Fitting the Psf Model Non-interactively\l'|5.6i.'\056
+.br
+\h'|0.9i'6.10.12.\h'|1.5i'The Output of Psf\l'|5.6i.'\057
+.br
+\h'|0.9i'6.10.13.\h'|1.5i'Checking the Psf Model\l'|5.6i.'\059
+.br
+\h'|0.9i'6.10.14.\h'|1.5i'Removing Bad Stars from the Psf Model\l'|5.6i.'\062
+.br
+\h'|0.9i'6.10.15.\h'|1.5i'Adding New Stars to a Psf Star Group\l'|5.6i.'\062
+.br
+\h'|0.9i'6.10.16.\h'|1.5i'Refitting the Psf Model With the New Psf Star Groups\l'|5.6i.'\062
+.br
+\h'|0.9i'6.10.17.\h'|1.5i'Computing the Final Psf Model\l'|5.6i.'\063
+.br
+\h'|0.9i'6.10.18.\h'|1.5i'Visualizing the Psf Model with the Seepsf Task\l'|5.6i.'\063
+.br
+\h'|0.9i'6.10.19.\h'|1.5i'Problems Computing the Psf Model\l'|5.6i.'\064
+.br
+\h'|0.4i'6.11.\h'|0.9i'\fBDoing Psf Fitting Photometry with Peak, Nstar, or Allstar\fP \l'|5.6i.'\065
+.br
+\h'|0.9i'6.11.1.\h'|1.5i'Fitting Single Stars with Peak\l'|5.6i.'\065
+.br
+\h'|1.5i'6.11.1.1.\h'|2.2i'The Peak Algorithm\l'|5.6i.'\065
+.br
+\h'|1.5i'6.11.1.2.\h'|2.2i'Running Peak \l'|5.6i.'\065
+.br
+\h'|1.5i'6.11.1.3.\h'|2.2i'The Peak Output\l'|5.6i.'\066
+.br
+\h'|0.9i'6.11.2.\h'|1.5i'Fitting Stars with Group, Grpselect, Nstar and Substar\l'|5.6i.'\067
+.br
+\h'|1.5i'6.11.2.1.\h'|2.2i'The Group and Nstar Algorithms\l'|5.6i.'\067
+.br
+\h'|1.5i'6.11.2.2.\h'|2.2i'Running Group, Grpselect, and Nstar\l'|5.6i.'\068
+.br
+\h'|1.5i'6.11.2.3.\h'|2.2i'The Nstar Output\l'|5.6i.'\070
+.br
+\h'|0.9i'6.11.3.\h'|1.5i'Fitting Stars With Allstar\l'|5.6i.'\071
+.br
+\h'|1.5i'6.11.3.1.\h'|2.2i'The Allstar Algorithm\l'|5.6i.'\071
+.br
+\h'|1.5i'6.11.3.2.\h'|2.2i'Running Allstar\l'|5.6i.'\072
+.br
+\h'|1.5i'6.11.3.3.\h'|2.2i'The Allstar Output\l'|5.6i.'\073
+.br
+\h'|0.4i'6.12.\h'|0.9i'\fBExamining the Output Photometry Files\fP\l'|5.6i.'\073
+.br
+\h'|0.4i'6.13.\h'|0.9i'\fBProblems with the Photometry\fP\l'|5.6i.'\074
+.br
+\h'|0.4i'6.14.\h'|0.9i'\fBDetecting Stars Missed By Daofind\fP\l'|5.6i.'\075
+.br
+\h'|0.4i'6.15.\h'|0.9i'\fBInitializing the Missing Star Photometry with Phot\fP\l'|5.6i.'\075
+.br
+\h'|0.4i'6.16.\h'|0.9i'\fBMerging Photometry Files with Pfmerge\fP\l'|5.6i.'\076
+.br
+\h'|0.4i'6.17.\h'|0.9i'\fBRefitting the Stars with Allstar\fP\l'|5.6i.'\076
+.br
+\h'|0.4i'6.18.\h'|0.9i'\fBExamining the Subtracted Image\fP\l'|5.6i.'\076
+.br
+\h'|0.4i'6.19.\h'|0.9i'\fBComputing an Aperture Correction\fP\l'|5.6i.'\076
+.sp
+7.\h'|0.4i'\fBReferences\fP\l'|5.6i.'\077
+.sp
+8.\h'|0.4i'\fBAppendices\fP\l'|5.6i.'\077
+.br
+.sp
+\h'|0.4i'8.1.\h'|0.9i'\fBThe Instrumental Magnitude Scale\fP\l'|5.6i.'\077
+.br
+\h'|0.4i'8.2.\h'|0.9i'\fBThe Analytic Psf Models\fP\l'|5.6i.'\077
+.br
+\h'|0.4i'8.3.\h'|0.9i'\fBThe Error Model\fP\l'|5.6i.'\078
+.br
+\h'|0.4i'8.4.\h'|0.9i'\fBThe Radial Weighting Function\fP\l'|5.6i.'\078
+.br
+\h'|0.4i'8.5.\h'|0.9i'\fBTotal Weights\fP\l'|5.6i.'\078
+.br
+\h'|0.4i'8.6.\h'|0.9i'\fBBad Data Detection\fP\l'|5.6i.'\078
+.br
+\h'|0.4i'8.7.\h'|0.9i'\fBStellar Mergers\fP\l'|5.6i.'\079
+.br
+\h'|0.4i'8.8.\h'|0.9i'\fBFaint Stars\fP\l'|5.6i.'\079
+.br
+.bp
+\0
+.ds CH - % -
+.bp 1
+\0
+
+.TL
+A Reference Guide to the IRAF/DAOPHOT Package
+.AU
+Lindsey E. Davis
+.AI
+IRAF Programming Group
+.K2
+.ce
+.TU
+.br
+.ce
+January 1994
+
+.NH
+Introduction
+
+.PP
+DAOPHOT is a software package for doing stellar photometry
+in crowded fields  developed by Peter Stetson of the DAO (1987, 1990, 1992).
+The IRAF/DAOPHOT package uses the task structure and algorithms of DAOPHOT
+to do crowded field photometry within the IRAF data reduction and
+analysis environment.
+.PP
+Input to IRAF/DAOPHOT consists of an IRAF image file, numerous parameters
+controlling the analysis algorithms and, optionally, graphics cursor and/or
+image display cursor input. IRAF/DAOPHOT produces output photometry files
+in either text format or STSDAS binary table format. Some IRAF/DAOPHOT tasks
+also produce image output and graphics output in the form of plot metacode
+files.
+.PP
+Separate tasks are provided for examining, editing, storing, and recalling
+the analysis parameters, creating and editing star
+lists, computing accurate centers, sky values and initial magnitudes
+for the stars in the list, computing the point-spread function,
+grouping the stars into physical associations, fitting the stars either
+singly or in groups, subtracting the fitted stars from the original image,
+and adding artificial test stars to the original image. A set of tools are
+also provided for examining and editing the output photometry files.
+
+.NH
+DAOPHOT and IRAF/DAOPHOT
+
+.PP
+The principal similarities and differences between DAOPHOT and IRAF/DAOPHOT
+are summarized below.
+.IP [1]
+The structure of IRAF/DAOPHOT is very similar to the
+structure of DAOPHOT. All the DAOPHOT photometry tasks and many of
+the utilities tasks are present in
+IRAF/DAOPHOT and in many cases the DAOPHOT task names have been preserved.
+A listing of the DAOPHOT photometry tasks and their closest IRAF/DAOPHOT
+equivalents is shown below. 
+
+.TS
+l l
+l l
+l l.
+DAOPHOT\tIRAF/DAOPHOT
+TASK\tEQUIVALENT
+
+add*\taddstar
+allstar\tallstar
+attach\tN/A
+append\tpfmerge,pconcat
+find\tdaofind
+group\tgroup
+monitor\tdaophot.verbose=yes
+nomonitor\tdaophot.verbose=no
+nstar\tnstar
+offset\tpcalc
+options\tdaoedit
+peak\tpeak
+photometry\tphot
+pick\tpstselect
+psf\tpsf
+select\tgrpselect
+sort\tpsort,prenumber
+sub*\tsubstar
+.TE
+
+.IP [2]
+Some DAOPHOT utilities tasks are missing from IRAF/DAOPHOT.
+The DAOPHOT tasks \fBdump\fR, \fBexit\fR, \fBfudge\fR,
+\fBhelp\fR, \fBlist\fR, and \fBsky\fR
+have been replaced with general IRAF tasks, or with IRAF system facilities
+that perform the equivalent function. The missing DAOPHOT utilities tasks
+and their IRAF equivalents are shown below.
+
+.TS
+l l
+l l
+l l.
+DAOPHOT\tIRAF/DAOPHOT
+TASK\tEQUIVALENT
+
+dump\tlistpixels,imexamine
+exit\tbye
+fudge\timreplace,fixpix,imedit
+help\thelp daophot
+list\timheader
+sky\timstatistics,phistogram,imexamine
+.TE
+
+.IP [3]
+The IRAF/DAOPHOT default algorithms are the DAOPHOT II algorithms
+(Stetson 1992).
+.IP [4]
+Users have more choice of and control over the algorithms
+in IRAF/DAOPHOT than they do in DAOPHOT. For example the
+IRAF/DAOPHOT aperture photometry task \fBphot\fR offers several
+sky fitting algorithms besides the default "mode" algorithm,
+and full control over the sky fitting algorithm parameters.
+.IP [5]
+The algorithm parameters in IRAF/DAOPHOT are grouped by function into
+six parameter sets or psets rather than three as in DAOPHOT.
+The six IRAF/DAOPHOT parameter sets with their DAOPHOT equivalents
+in brackets are:
+1) \fBdatapars\fR, the data definition parameters (daophot.opt),
+2) \fBfindpars\fR, the detection algorithm parameters (daophot.opt),
+3) \fBcenterpars\fR, the aperture photometry centering algorithm parameters
+(no equivalent),
+4) \fBfitskypars\fR, the aperture photometry sky fitting parameters (photo.opt),
+5) \fBphotpars\fR, the aperture photometry parameters (photo.opt),
+6) \fBdaopars\fR, the IRAF/DAOPHOT psf fitting parameters (daophot.opt,
+allstar.opt).
+.IP [6]
+The IRAF/DAOPHOT algorithm parameter sets unlike the DAOPHOT parameter sets
+can be interactively examined,
+edited and saved with the \fBdaoedit\fR task using the image display
+and radial profile plots.
+.IP [7]
+The IRAF/DAOPHOT algorithm parameter sets unlike the DAOPHOT parameter sets
+can be saved and restored as a function of image using the \fBsetimpars\fR task.
+.IP [8]
+Memory allocation in IRAF/DAOPHOT is dynamic not static as in
+DAOPHOT.  IRAF/DAOPHOT allocates and frees memory as required
+at run-time subject to the physical memory and swap space limitations of
+the host computer.
+.IP [9]
+The IRAF/DAOPHOT point-spread function look-up table is stored in an
+IRAF image not an ASCII table as in DAOPHOT.
+.IP [10] 
+Unlike DAOPHOT, the IRAF/DAOPHOT tasks \fBdaofind\fR, \fBphot\fR,
+\fBpstselect\fR
+and \fBpsf\fR can be run interactively using the image display and graphics
+window or non-interactively. Display and graphics capabilities were
+deliberately omitted from DAOPHOT to minimize portability problems.
+.IP [11]
+The IRAF/DAOPHOT output photometry files can be written in either text
+format as in DAOPHOT or STSDAS binary table format.
+.IP [12]
+Unlike DAOPHOT, fields or columns in both IRAF/DAOPHOT text and
+STSDAS binary table photometry files are identified
+by name and have an associated units and format specifier.
+The IRAF/DAOPHOT photometry file input routines search for column
+names, for example "GROUP,ID,XCENTER,YCENTER,MAG,MSKY" as
+appropriate but are independent
+of their placement in the input file.
+.IP [13]
+Several general purpose IRAF/DAOPHOT tasks are available for performing
+operations on the final output photometry catalogs. In addition to 
+\fBpcalc\fR, \fBpconcat\fR, \fBpfmerge\fR, \fBprenumber\fR,
+and \fBpsort\fR which are
+also available in DAOPHOT, there are three photometry file editing tasks which
+have no analog in DAOPHOT \fBpdump\fR, \fBpexamine\fR, and \fBpselect\fR.
+All these tasks work on IRAF/DAOPHOT output text files or STSDAS binary
+tables. An IRAF/DAOPHOT task is supplied  for converting output text files to
+STSDAS binary tables so as to make use of the even more general STSDAS
+tables manipulation tools in the TABLES package.
+.IP [14]
+The IRAF/DAOPHOT output files are self-documenting.
+All the information required to comprehend the history of or decode the
+output photometry file is in the file itself, including the IRAF version
+number, host computer, date, time, and names of all the
+input and output files and the values of all the parameters.
+.PP
+For the remainder of this document IRAF/DAOPHOT will be referred to
+as DAOPHOT.
+
+.NH
+Preparing Data for DAOPHOT
+
+.IP [1]
+DAOPHOT assumes that the images to be analyzed exist on disk in IRAF
+image format. DAOPHOT can read and write old IRAF format ".imh" images
+and ST IRAF format ".hhh" images.
+When the IRAF FITS kernel becomes available DAOPHOT will be able
+to read FITS images on disk as well.
+QPOE IRAF format ".qp" images must be rasterized before they can
+be input to DAOPHOT.
+.IP [2]
+All internal DAOPHOT calculations are done in real precision.
+The pixel type of the image data on disk may be any of the following
+data types: short integer, unsigned short integer, integer, long integer,
+real or double. Users should realize that the extra precision in
+images of type double will not be used by DAOPHOT.
+.IP [3]
+The instrumental signature must be removed from the input images
+prior to running DAOPHOT.  All CCD images should be overscan
+corrected, bias corrected, dark current corrected and flat-fielded.
+Users should be aware of the IRAF CCDRED package for reducing CCD data.
+.IP [4]
+DAOPHOT assumes that the input pixel data is linear.
+If the data is non-linear over a large fraction of its total dynamic range,
+the data must be linearized before running DAOPHOT.
+.IP [5]
+Saturated pixels or pixels distinguishable from good data by intensity,
+do not need to be removed from the image prior to running DAOPHOT.
+For example if the data
+is non-linear only above 25000 counts, DAOPHOT can be instructed to 
+ignore pixels above 25000 counts.
+.IP [6]
+Extreme-valued pixels should be removed from the images prior to running
+DAOPHOT. Extreme-valued pixels include those with values at or near
+the floating point limits of the host machine and host machine special
+numbers produced by operations like divide by zero, floating point
+underflows and overflows, etc. The latter category of extreme-valued
+pixels should not be produced by IRAF software, but may be produced by
+user programs including imfort programs.
+Floating point operations involving such numbers will frequently cause
+arithmetic exception errors, since for efficiency and portability reasons
+the DAOPHOT package and most IRAF tasks do not test for
+their presence.
+The \fBimreplace\fR task in the PROTO  package can be used to remove extreme-
+valued pixels.
+.IP [7]
+The background sky value should NOT be subtracted from the image prior
+to entering the DAOPHOT package. The DAOPHOT fitting routines use an optimal
+weighting scheme which depends on the readout noise, the gain, and the
+true counts in the pixels. If the mean sky has been subtracted
+then the counts in the image are not the true counts and the computed weights
+will be incorrect. For similar reasons users should not attempt to
+correct their magnitudes for exposure time by dividing their images
+by the exposure time.
+.IP [8]
+Cosmic ray and bad pixel removal programs should be used with caution. If the
+data and parameter values are set such that the cosmic ray and bad pixel
+detection and
+removal algorithms have difficulty distinguishing between stars and bad
+pixels or cosmic rays,
+the peaks of the stars may be clipped, altering the point-spread function
+and introducing errors into the photometry.
+.IP [9]
+DAOPHOT assumes that the local sky background is approximately flat in the
+vicinity of the object being measured. This assumption is equivalent to
+requiring that the local sky region have a unique mode. Variations
+in the sky background which occur on the same scale as the size of the
+local sky region will introduce errors into the photometry.
+.IP [10]
+The point spread function must be constant or smoothly
+varying with position over the entire image. This is the fundamental
+assumption
+underlying all of DAOPHOT. All stars in the image must be indistinguishable
+except for position and magnitude. The variable point spread function
+option is capable of handling second order variability as a function of
+position in the image.
+.IP [11]
+The input images should not have undergone any operations which fundamentally
+alter the image point spread function or the image statistics in a non-linear
+way. For example, non-linear image restoration tasks must not be run on
+the image to prior to running DAOPHOT.
+.IP [12]
+The gain, readout noise, exposure time,
+airmass, filter, and observing time should be present and correct in the
+image headers before DAOPHOT reductions are begun.
+DAOPHOT tasks can extract this information from the image headers, use it
+in the computations, and/or store
+it in the output photometry files, greatly simplifying the analysis
+and subsequent calibration procedures.
+.fi
+
+.NH
+Some IRAF Basics for New IRAF and DAOPHOT Users
+
+.NH 2
+Pre-loaded Packages
+
+.PP
+Under IRAF versions 2.10 and later the DATAIO, PLOT, IMAGES, TV and NOAO
+packages are pre-loaded so that all the tasks directly under them are
+available when
+IRAF is started. Each of these packages contains tasks which are useful
+to DAOPHOT users for various reasons, and each is discussed briefly below.
+
+.NH 3
+The DATAIO Package
+
+.PP
+DAOPHOT users should be aware of the DATAIO \fBrfits\fR and \fBwfits\fR tasks
+which are used to transport data into and out of IRAF. Any input 
+and output images, including point-spread function look-up table images,
+should normally be archived with \fBwfits\fR.
+The cardimage reader and writer tasks for archiving text files,
+\fBrcardimage\fR and \fBwcardimage\fR, are also located here.
+
+.NH 3
+The PLOT Package
+
+.PP
+Various general purpose image and file plotting utilities can be found
+in the PLOT packages. DAOPHOT users should be aware of the interactive image
+row and column plotting task \fBimplot\fR, the image contour plotting task
+\fBcontour\fR, the image surface plotting task \fBsurface\fR, image
+histogram plotting task \fBphistogram\fR, the image radial profile
+plotting task \fBpradprof\fR, and the general purpose graphing tool
+\fBgraph\fR. The tasks \fBgkidir\fR and \fBgkiextract\fR are also useful
+for extracting individual plots from the plot metacode files which may
+be produced by some DAOPHOT tasks.
+
+.NH 3
+The IMAGES Package
+
+.PP
+The IMAGES package contains a set of general purpose image operators. DAOPHOT
+users
+should be aware of the image header examining tasks \fBimheader\fR and
+\fBhselect\fR, the header editing task \fBhedit\fR, the coordinate and
+pixel value dumping task \fBlistpixels\fR, and the image statistics
+task \fBimstatistics\fR.
+
+.NH 3
+The TV Package
+
+.PP
+The TV package contains tasks which interact with the image display including
+the all important \fBdisplay\fR task for displaying images, the
+interactive image examining task \fBimexamine\fR, and the \fBtvmark\fR task
+for marking objects on the image display. DAOPHOT users should become 
+familiar with all three of these tasks.
+
+.NH 2
+Other Useful Packages and Tasks
+
+.PP
+The NPROTO package contains two useful tasks, \fBfindgain\fR,
+for computing the gain and readout noise of a CCD
+from a pair of biases and flats, and \fBfindthresh\fR for computing
+the standard deviation of the background in a CCD frame given the
+readout noise and gain. The ASTUTIL package contains the \fBsetairmass\fR
+task for computing and/or correcting the airmass given the appropriate
+input data. 
+Users might also wish to experiment with the tasks in the artificial
+data package ARTDATA, and run the resulting images through DAOPHOT.
+
+.NH 2
+Image Types, Image Directories, and Image Headers
+
+.PP
+The IRAF image environment is controlled by several
+environment variables. The most important of these for DAOPHOT users
+are: \fBimtype\fR the disk image format, \fBimdir\fR the default pixel
+directory, and \fBmin_lenuserarea\fR the maximum length of the image header.
+The values of these environment variables can be listed
+as shown below.
+
+.YS
+cl> show imtype
+imh
+cl> show imdir
+/data/davis/pixels/
+cl> show min_lenuserarea
+24000
+.YE
+
+.PP
+\fB"imh"\fR is the default image format for most IRAF users, \fB"hhh"\fR the
+default image format for ST users, and \fB"qp"\fR the photon counting format
+used for photon counting data. DAOPHOT will work transparently on
+"imh" and "hhh" images. "qp" event lists must be rasterized prior to using
+DAOPHOT. When IRAF supports FITS images on disk, image format "fits", DAOPHOT
+will be able to work directly on FITS images as well. IRAF uses the
+image name extension, e.g. "imh" to automatically sense the image
+disk format on input. The output disk format is set by: 1) the
+extension of the output image name if present e.g. "imh", 2) the cl
+environment variable \fBimtype\fR if the output image is opened as a new
+image, e.g. the output of the \fBrfits\fR task, 3) the type of the input
+image if the output image is opened as a new copy of an existing image,
+e.g. the output of the \fBimcopy\fR task.
+.PP
+\fBimdir\fR specifies the default image pixel directory for "imh" format
+files. The image header files are written to the current directory
+and the pixel files are written to imdir. imdir can be set
+to an existing directory on a scratch disk, the current
+directory "HDR$", or the subdirectory pixels under the current
+directory "HDR$pixels/". DAOPHOT users should keep both the intrinsic
+speed of a disk and its network configuration in mind when setting
+imdir.
+.PP
+\fBmin_lenuserarea\fR is the size of the image header area reserved
+in memory when a new or existing image is opened.
+The current default value of 24000 corresponds to space for approximately
+300 keywords.
+If an image on disk has a header larger than this the image header will
+be truncated when it is read.
+For most DAOPHOT users the default value is sufficient. However users whose
+images have large headers or who are
+creating a point-spread function using more than ~70 stars should set
+min_lenuserarea to a larger value, e.g. 40000.
+.PP
+The following example shows how to change the default pixel directory to
+HDR$pixels/ and set min_lenuserarea to 40000.  To avoid redefining these
+quantities for every session, users should enter the redefinitions into
+their login.cl or loginuser.cl files.
+
+
+.YS
+cl> reset imdir = "HDR$pixels/"
+cl> reset min_lenuserarea = 40000
+.YE
+
+.NH 2
+The Image Display and Image Cursor
+
+.PP
+Several DAOPHOT tasks are interactive tasks or have an interactive as well
+as a non-interactive mode. In interactive mode these tasks must be able to
+read the image cursor on a displayed image and perform various
+actions depending on the position of the image cursor and the keystroke
+command typed.
+.PP
+DAOPHOT will work with the display servers Imtool, Saoimage, and Ximtool.
+DAOPHOT users should be aware that both Imtool and Ximtool support multiple
+frame buffers while SAOimage does not. Multiple frame buffers  are an
+important feature for users who wish to compare their original
+images with the DAOPHOT output images from which all the fitted
+stars have been subtracted. Users running DAOPHOT on a remote machine, e.g.
+one with lots of memory and/or disk space, but displaying on their local
+machine also need to set the \fBnode\fR environment variable to
+the name of the local machine.
+
+.YS
+cl> show node
+ERROR: No such environment variable
+    show (node)
+cl> set node = mymachine
+.YE
+
+.PP
+The maximum size of the display server frame buffer is defined by the
+environment variable \fBstdimage\fR whose value can be printed as
+shown below.
+
+.YS
+cl> show stdimage
+imt512
+.YE
+
+In the previous example the default frames buffers are 512 pixels square.
+A user whose images are 2K square will want to reset the default frame
+buffer size as shown below.
+
+.YS
+cl> reset stdimage = imt2048
+cl> show stdimage
+imt2048
+.YE
+
+.PP
+In order for image cursor read-back to function correctly the environment
+variable \fBstdimcur\fR must be set to "stdimage" as shown below.
+
+.YS
+cl> show stdimcur
+stdimage
+.YE
+
+To check that image cursor read-back is functioning correctly the user
+should display an image and try to bring up the image display cursor
+as shown below.
+
+.YS
+cl> display image 1
+cl> =imcur
+.YE
+
+The image cursor should appear on the image display reading the correct
+image pixel coordinates and ready to accept a
+keystroke command. Any keystroke will terminate the cursor read.
+
+.NH 2
+The Graphics Device and Graphics Cursor
+
+.PP
+Some interactive DAOPHOT tasks have graphics submenus which require
+them to be able to read the graphics cursor on for example a radial
+profile plot and perform various
+actions based on the position of the graphics cursor in the
+plot and the keystroke
+command issued. The default graphics device is determined by
+the \fBstdgraph\fR environment variable as shown below.
+
+.YS
+cl> show stdgraph
+xgterm
+.YE
+
+To check that graphics cursor read-back is functioning correctly the user
+should draw a plot and try to bring up the graphics cursor as
+shown below.
+
+.YS
+cl> contour image
+cl> =gcur
+.YE
+
+The graphics cursor should appear in the graphics window ready to accept a
+keystroke command. Any keystroke will terminate the cursor read.
+
+.NH
+Some DAOPHOT Basics for New DAOPHOT Users
+
+.NH 2
+Loading the DAOPHOT Package
+
+.PP
+The DAOPHOT package is located in the digital stellar photometry package
+DIGIPHOT. To load DIGIPHOT and DAOPHOT the user types the package names
+in sequence as shown below,
+
+.YS
+cl> digiphot
+di> daophot
+.YE
+
+after which the following menu of tasks appears.
+
+.YS
+addstar       daotest       nstar         pexamine      psf
+allstar       datapars@     pcalc         pfmerge       psort
+centerpars@   findpars@     pconcat       phot          pstselect
+daoedit       fitskypars@   pconvert      photpars@     seepsf
+daofind       group         pdump         prenumber     setimpars
+daopars@      grpselect     peak          pselect       substar
+.YE
+
+Task names with a trailing "@" are parameter set tasks.
+The remaining tasks are script and/or compiled tasks.
+After the DAOPHOT package is loaded the user can redisplay 
+the package menu at any time with the command.
+
+.YS
+da> ? daophot
+.YE
+
+.NH 2
+Loading the TABLES Package
+
+.PP
+The DAOPHOT photometry tasks write their output photometry files in 
+either text format (the default) or ST binary tables format. Users wishing
+to use the ST binary tables format should acquire and install
+the ST TABLES external package. Without the TABLES package the DAOPHOT
+photometry tasks will read and write ST binary tables, but DAOPHOT
+utilities like \fBpsort\fR which call TABLES package
+tasks will not run on ST binary tables.
+.PP
+When DAOPHOT is loaded, it checks to see if the TABLES package is defined,
+and if so loads it. A warning message is issued if the TABLES package is
+undefined. The TABLES package tasks can be listed at any time after DAOPHOT
+is loaded with the following command.
+
+.YS
+da> ? tables
+.YE
+
+.NH 2
+Running the Test Script
+
+.PP
+The DAOPHOT package includes a script task \fBdaotest\fR which
+executes each of the core DAOPHOT photometry tasks in turn using a test
+image stored
+in  FITS format in the DAOPHOT test directory. \fBDaotest\fR is run as
+shown below.
+
+.YS
+da> daotest
+
+DAOTEST INITIALIZES THE DAOPHOT TASK PARAMETERS
+TYPE 'q' or 'Q' TO QUIT, ANY OTHER KEY TO PROCEED
+
+Name of the output test image: test
+
+INITIALIZE THE DAOPHOT PACKAGE
+
+TESTING THE DAOFIND TASK
+TESTING THE PHOT TASK
+TESTING THE PSTSELECT TASK
+TESTING THE PSF TASK
+TESTING THE PEAK TASK
+TESTING THE GROUP TASK
+TESTING THE GRPSELECT TASK
+TESTING THE NSTAR TASK
+TESTING THE ALLSTAR TASK (CACHE=YES)
+TESTING THE ALLSTAR TASK (CACHE=NO)
+TESTING THE SUBSTAR TASK
+TESTING THE ADDSTAR TASK
+
+DAOPHOT PACKAGE TESTS COMPLETED
+.YE
+
+On task completion the user will find the input image in
+test.imh, the psf image in test.psf.1.imh, the subtracted image produced
+by \fBallstar\fR in test.sub.1.imh, the input image with artificial stars
+added in test.add.1.imh, copies of all the output photometry files in
+test.log, and copies of the plots produced by the \fBpsf\fR task
+in test.plot on disk.
+.PP
+Users should be aware that the \fBdaotest\fR task will reset the DAOPHOT
+task and algorithm parameters to their default values before and after it
+is executed.
+
+.NH 2
+On-line Help
+
+.PP
+A one-line description of each DAOPHOT task can be obtained by typing
+the following command,
+
+.YS
+da> help daophot\fR
+.YE
+
+upon which the following package menu appears.
+
+.YS
+digiphot.daophot:
+   addstar - Add stars to an image using the computed psf
+   allstar - Group and fit psf to multiple stars simultaneously
+centerpars - Edit the centering algorithm parameters
+   daoedit - Review/edit algorithm parameters interactively
+   daofind - Find stars in an image using the DAO algorithm
+   daopars - Edit the daophot algorithms parameter set
+   daotest - Run basic tests on the daophot package tasks
+  datapars - Edit the image data dependent parameters
+  findpars - Edit the star detection parameters
+fitskypars - Edit the sky fitting algorithm parameters
+     group - Group stars based on position and signal/noise
+     nstar - Fit the psf to predefined groups of stars
+      peak - Fit the psf to single stars
+      phot - Compute skies and initial magnitudes for a star list
+  photpars - Edit the aperture photometry parameters
+       psf - Compute the point spread function
+    seepsf - Compute an image from the point spread function
+ setimpars - Save/restore parameter sets for a particular image
+   substar - Subtract the fitted stars from the original image
+
+     pcalc - Do arithmetic operations on list of daophot databases
+   pconcat - Concatenate a list of daophot databases
+  pconvert - Convert a text database to a tables database
+     pdump - Print selected fields from daophot databases
+   pfmerge - Merge a list of photometry databases
+ pstselect - Select candidate psf stars based on proximity
+ grpselect - Select groups from a daophot database
+  pexamine - Interactively examine and edit a daophot database
+ prenumber - Renumber stars in a daophot database
+   pselect - Select records from a daophot database
+     psort - Sort a daophot database\fR
+.YE
+
+.PP
+All the DAOPHOT tasks have on-line manual pages which can be
+listed on the terminal. The following command lists the help for the
+\fBphot\fR task on the terminal.
+
+.YS
+da> phelp phot\fR
+.YE
+
+Any section of the manual pages can be listed individually.
+For example the examples section of the \fBphot\fR manual page can be
+listed as follows.
+
+.YS
+da> phelp phot sections=examples\fR
+.YE
+
+The help page for \fBphot\fR can be piped to the local default printer as
+follows.
+
+.YS
+da> phelp phot | lprint\fR
+.YE
+
+Finally the manual pages for the whole DAOPHOT package can be printed
+by typing.
+
+.YS
+da> phelp daophot.* | lprint\fR
+.YE
+
+
+.NH 2
+Editing the Package Parameters
+
+.PP
+DAOPHOT has a package parameter set which defines the DAOPHOT
+package environment. The DAOPHOT package parameters can edited
+with epar as shown below.
+
+.YS
+da> epar daophot
+.YE
+
+.YS
+Image Reduction and Analysis Facility
+  PACKAGE = digiphot
+     TASK = daophot
+ (version = "Dec92")
+    (text = yes)            Text file on output ?
+  (verify = yes)            Verify critical parameters ?
+  (update = no)             Update critical parameters ?
+ (verbose = yes)            Print verbose output ?
+(graphics = "stdgraph")     Default graphics device
+ (display = "stdimage")     Default display device
+    (mode = "ql")
+.YE
+
+To edit a parameter simply move the cursor to the parameter in question,
+enter the new value, type return, and finally type \fB:wq\fR to quit and
+update the parameter set. Package parameters can also be edited on the
+command line as shown below.
+
+.YS
+da> daophot.text = yes
+.YE
+
+.PP
+The DAOPHOT package parameters control the operation of the DAOPHOT package
+as a whole.  For example the \fBtext\fR parameter specifies whether the
+output photometry files will be written in text or STSDAS binary tables format,
+the parameters \fBverify\fR, \fBupdate\fR, and \fBverbose\fR determine
+the default mode of operation of the DAOPHOT package tasks, and the parameters
+\fBgraphics\fR and \fBdisplay\fR determine the default graphics and display
+devices for the entire package.
+
+.NH 2
+Editing the Task Parameters
+
+.PP
+The DAOPHOT task level parameters specify the input and output images and
+files, the algorithm parameter sets, the graphics and image display input and
+output devices, and the mode of operation of each DAOPHOT task.
+.PP
+To enter and edit the parameter set for the DAOPHOT \fBphot\fR task
+the user types the following command,
+
+.YS
+cl> epar phot
+.YE
+
+after which the parameter set for the \fBphot\fR task appears on the
+terminal ready for editing as shown below.
+
+.YS
+Image Reduction and Analysis Facility
+PACKAGE = daophot
+   TASK = phot
+
+image   =                       Input image(s)
+coords  =              default  Input coordinate list(s)
+output  =              default  Output photometry file(s)
+skyfile =                       Input sky value file(s)
+(plotfil=                     ) Output plot metacode file
+(datapar=                     ) Data dependent parameters
+(centerp=                     ) Centering parameters
+(fitskyp=                     ) Sky fitting parameters
+(photpar=                     ) Photometry parameters
+(interac=                   no) Interactive mode ?
+(radplot=                   no) Plot the radial profiles?
+(verify =            )_.verify) Verify critical phot parameters ?
+(update =            )_.update) Update critical phot parameters ?
+(verbose=           )_.verbose) Print phot messages ?
+(graphic=          )_.graphics) Graphics device
+(display=           )_.display) Display device
+(icomman=                     ) Image cursor: [x y wcs] key [cmd]
+(gcomman=                     ) Graphics cursor: [x y wcs] key [cmd]
+(mode   =                   ql)
+.YE
+
+The \fBphot\fR parameters can be edited by moving
+the cursor to the line opposite the parameter name, entering the new value
+followed by a carriage return, and typing \fB:wq\fR to exit the
+\fBepar\fR task and update the parameters.
+.PP
+In the following sections the \fBphot\fR task is used to illustrate
+some general features of the DAOPHOT package.
+
+.NH 2
+Input and Output Image Names
+
+.PP
+The \fBphot\fR parameter \fIimage\fR
+defines the image to be analyzed. The
+root image name, the value of \fIimage\fR 
+stripped of directory and section information,
+sets up the default input and output image naming convention for the task.
+Users should avoid appending the ".imh" or ".hhh" extension
+to their image name specification as these extensions are not required by IRAF
+image i/o and become part of the default output image names.
+.PP
+The \fBphot\fR task does not create an output image but DAOPHOT tasks
+which do, will by default create an output image name of the form
+"image.extension.?" where image is the input image name
+stripped of directory
+and section information, extension is an id appropriate
+to the task, and ? is the next available version number.
+For example the first run of the \fBsubstar\fR task on the image "image"
+will create an image called "image.sub.1", the second an image
+called "image.sub.2", and so on.  The default output image naming convention
+can always be overridden by the user in any task.
+
+.NH 2
+Input and Output File Names
+
+.PP
+DAOPHOT uses a default input and output file naming convention based on the
+root image name or the input image name  with the directory and
+section specification removed.  Users should avoid appending the ".imh" or
+".hhh" extension to their input image name specification as these extensions
+are not required by IRAF image i/o and become part of the default input
+and output file names.
+.PP
+If a DAOPHOT task expects its input to have been written
+by another DAOPHOT task, and the input file parameter value is "default",
+the task will search for an existing
+file called "image.extension.?" where image is the root image
+name, extension identifies the task expected to have written the file,
+and version is the highest version number for that file. For example,
+if the user sets the \fBphot\fR parameters \fIimage\fR and
+\fIcoords\fR to "m92b" and "default", \fBphot\fR will search
+for a coordinate file called "m92b.coo.#" written by the
+\fBdaofind\fR task. The default input file naming convention
+can be over-ridden by the user at any point.
+.PP
+The output file naming convention works
+in an identical manner to the input file naming convention,
+although in this situation ? is the next available
+version number. For example if the user sets the \fBphot\fR task
+parameter \fIoutput\fR to "default", the output photometry file name
+will be "image.mag.?"
+where ? is 1 for the first run of \fBphot\fR, 2 for the second run, and so
+on.  The default output file naming convention can be over-ridden
+by the user at any point.
+
+.NH 2
+Algorithm Parameter Sets 
+
+.PP
+The DAOPHOT parameters have been grouped together into parameter sets
+or psets.
+The use of psets encourages the logical grouping of parameters, permits
+the various DAOPHOT tasks to share common parameters, and
+permits the user to optionally store the DAOPHOT algorithm parameters
+with the data rather than in the default uparm directory. 
+.PP
+Six DAOPHOT psets, \fBdatapars\fR, \fBfindpars\fR, \fBcenterpars\fR,
+\fBfitskypars\fR, \fBphotpars\fR and \fBdaopars\fR
+control the DAOPHOT algorithm parameters. The \fBphot\fR task 
+uses four of them, \fBdatapars\fR which specifies data dependent
+parameters like \fIfwhmpsf\fR (the full-width half-maximum of the psf),
+\fIsigma\fR (the standard deviation of
+the sky background), \fIepadu\fR and \fIreadout noise\fR
+(the gain and readout noise of the detector),
+and the \fBcenterpars\fR, \fBfitskypars\fR and \fBphotpars\fR parameter
+sets which define the centering algorithm, sky fitting algorithm
+and aperture photometry algorithm parameters respectively,
+used by phot to compute initial centers, sky values,
+and initial magnitudes for the stars to be analyzed. The \fBfindpars\fR pset
+controls the star detection algorithm parameters used by the \fBdaofind\fR
+task. The \fBdaopars\fR pset defines the psf model fitting
+and evaluation parameters including the radius of the psf, the fitting radius,
+and the grouping parameters used by all the psf fitting tasks.
+.PP
+By default the pset parameters can be examined, edited and stored 
+in the user's uparm directory, in the same manner as the task level
+parameters. For example to list the current \fBdatapars\fR
+pset the user types.
+
+.YS
+da> lpar datapars
+.YE
+
+To edit the \fBdatapars\fR parameter set, the user types either
+
+.YS
+da> epar datapars
+
+or
+
+da> datapars
+.YE
+
+and edits the parameter set in the usual manner with \fBepar\fR.
+All the DAOPHOT tasks which reference this
+pset will pick up the changes from the uparm directory, assuming
+that the \fIdatapars\fR parameter is specified as  "" in the calling task.
+The user can also edit the \fBdatapars\fR
+pset from within the \fBphot\fR
+task or any other task which calls it as shown below.
+
+.YS
+da> epar phot
+.YE
+
+Move the cursor to the \fBdatapars\fR parameter line and type \fB:e\fR.
+The menu for the 
+\fBdatapars\fR pset will appear ready for editing. Edit the desired
+parameters and type \fB:wq\fR. \fBEpar\fR will return to the main
+\fBphot\fR parameter set after which other psets or the main task parameters
+can be edited.
+.PP
+Psets may also be stored in user files providing
+a mechanism for saving a particular pset
+with the data.
+The example below shows how to store a  pset in a file in the same directory
+as the data and recall it for use by the \fBphot\fR task. The user types
+
+.YS
+da> epar phot
+.YE
+
+as before, enters the \fBdatapars\fR menu with \fB:e\fR and edits the
+parameters.  The command \fB:w data1.par\fR
+writes the parameter set to a file called "data1.par" and a \fB:q\fR
+returns to the main task menu.
+A file called "data1.par" containing the new \fBdatapars\fR parameters
+is written in the current directory. At this point the user is still in the
+\fBphot\fR parameter set at the line opposite \fBdatapars\fR. He/she
+enters "data1.par" on the line opposite this parameter.
+The next time \fBphot\fR is run the parameters will
+be read from "data1.par" not from the pset in the uparm directory.
+The new parameter set can be edited in the usual way by typing
+
+.YS
+da> epar data1.par
+
+or
+
+da> epar phot
+.YE
+
+Users should be sure to append a .par extension to any pset files they
+create as IRAF needs this extension to identify the file as a pset.
+.PP
+It is possible to develop quite efficient and creative schemes for using psets.
+For example a user might choose to copy each crowded stellar field
+image to its own directory, copy the default psets \fBdatapars\fR,
+\fBfindpars\fR, \fBcenterpars\fR, \fBfitskypars\fR, \fBphotpars\fR
+and \fBdaopars\fR to the files "datapars.par", "findpars.par",
+"centerpars.par", "fitskypars.par", "photpars.par" and "daopars.par" in
+each image directory, and then edit
+the parameter sets of the top level tasks to look for psets with those names.
+Once this is done the psets in each directory can be edited at will
+without ever needing to edit the names of the psets in the top
+level tasks.
+.PP
+The individual pset parameters themselves have the same attributes as
+task level parameters. Hidden pset parameters may be altered on the
+command line in the same way as task parameters.
+The only distinction between task level parameters and pset parameters
+is that the latter may be stored in or read from a user defined file.
+
+.NH 2
+Interactive Mode and Non-Interactive Mode 
+
+.PP
+The \fBphot\fR task's \fIinteractive\fR parameter
+switches the task between interactive and non-interactive mode.
+.PP
+In interactive mode user instructions in the form of single keystroke
+commands or colon commands are read from the image cursor.
+For example the \fBphot\fR task \fB'i'\fR keystroke command enters the
+interactive setup menu and the \fB'v'\fR keystroke command verifies the
+current parameters. The colon commands are used to show or set any parameter.
+For example, if the user does not like the fact that the full-width
+half-maximum of a star
+as measured with the cursor is 2.5368945 he/she can set it to 2.54 by
+typing \fB:fw 2.54\fR.
+.PP
+In non-interactive mode the input files and images are read,
+the parameters are read from the psets,
+and the output files are written,
+all, with the exception of an optional verification step, without the
+intervention of the user.
+.PP
+The DAOPHOT parameter editing task \fBdaoedit\fR and the photometry catalog
+examining task \fBpexamine\fR are interactive tasks.
+Four other DAOPHOT tasks, \fBdaofind\fR, \fBphot\fR, \fBpstselect\fR,
+and \fBpsf\fR
+have an interactive and a non-interactive mode. The default mode for
+\fBdaofind\fR, \fBphot\fR, and \fBpstselect\fR is non-interactive while
+for \fBpsf\fR
+it is interactive.
+The remaining DAOPHOT tasks are currently non-interactive tasks.
+
+.NH 2
+Image and Graphics Cursor Input
+
+.PP
+All tasks which can be run interactively accept commands from the logical image
+cursor parameter \fIicommands\fR. Logical image cursor commands can
+read from the logical image cursor, \fIicommands\fR = "" or a file,
+\fIicommands\fR = "filename". The logical image cursor is normally
+the physical image cursor and the value of the IRAF environment
+variable \fBstdimcur\fR is normally "stdimage". In cases where the image
+display device is non-existent or cursor read-back is not implemented for
+a particular device the logical image cursor may be reassigned globally to the
+the graphics cursor or the standard input
+by setting the IRAF environment variable \fBstdimcur\fR as follows.
+
+.YS
+da> set stdimcur = "stdimage"         (image cursor default)
+
+da> set stdimcur = stdgraph           (graphics cursor)
+
+da> set stdimcur = "text"             (standard input)
+.YE
+
+If logical image cursor commands are read from the standard input or a
+file, the commands must have the following format
+
+.YS
+[x y wcs] key [cmd]\fR
+.YE
+
+where x and y stand for the x and y position of the image cursor, wcs defines 
+the world coordinate system, key is
+a keystroke command, and cmd is an optional user command.
+Quantities in square brackets are optional. The necessity for their
+presence is dictated by the nature of the keystroke command. In the
+case of the \fBphot "i"\fR keystroke described above they are required, whereas
+in the case of the \fBphot "v"\fR keystroke they are not.
+.PP
+Some interactive commands require input from the logical graphics cursor
+parameter \fIgcommands\fR which may be the logical graphics cursor,
+\fIgcommands\fR = "", or a file of graphics cursor commands,
+\fIgcommands\fR = "filename".
+In DAOPHOT the logical graphics cursor must be set to the physical
+graphics cursor and the value of the IRAF environment variable
+\fBstdgcur\fR should be "stdgraph".
+
+.NH 2
+Graphics Output
+
+.PP
+The \fBphot\fR parameters \fIgraphics\fR and \fIdisplay\fR specify the
+default vector graphics and image display graphics devices.
+Vector graphics output is written to the user's
+graphics window, and image
+graphics is overlaid on the user's image display. window
+All interactive vector graphics output is written to
+the device specified by \fIgraphics\fR. An example of this type of graphics
+output is the
+radial profile plot of a star plotted by the \fBphot\fR interactive
+setup menu.
+Image graphics is written to the image display device
+specified by \fIdisplay\fR. 
+Examples of this type of output are the optional crosses 
+which mark the centers of the stars being measured by \fBphot\fR.
+\fBIRAF does not currently support writing interactive graphics
+to the image display device
+so the display marking features of DAOPHOT are not supported\fR.
+The single exception occurs in the situation
+where the user is running interactively
+off a contour plot as described in the \fBphot\fR help documentation.
+In this case marking will work if
+the parameter \fIdisplay\fR is set to "stdgraph".
+DAOPHOT tasks which reference \fIgraphics\fR or \fIdisplay\fR  will, in
+interactive mode, issue
+a warning if they cannot open either or both of these devices,
+and continue execution.
+.PP
+Some DAOPHOT tasks permit the user to save plots of the results
+for each measured star in a plot metacode file.
+For example. if the \fBphot\fR task parameter \fIplotfile\fR is defined,
+then for each star written to \fIoutput\fR
+a radial profile plot is written to the plot metacode file \fIplotfile\fR.
+\fIPlotfile\fR is opened in append mode and succeeding executions
+of \fBphot\fR will write to the end of the same file.
+Users should be aware plotfile can become very large and
+that writing radial profile plots
+to \fIplotfile\fR will greatly slow the execution of \fBphot\fR or any
+other task.
+
+.NH 2
+Verify, Update, and Verbose
+
+.PP
+In non-interactive mode the algorithm parameter values are read from the psets,
+critical parameters are verified if the \fIverify\fR switch is on, and 
+updated if both the \fIverify\fR and \fIupdate\fR switches are on.
+The \fIverify\fR and \fIupdate\fR options are also available as
+separate keystroke commands in interactive mode.
+Users must remember to turn off the
+\fIverify\fR switch if they submit a task to the background or the task
+will pause and wait indefinitely for input from the terminal.
+.PP
+In interactive or non-interactive mode a results summary and/or
+error messages are written
+to the standard output if the \fIverbose\fR switch is on.
+Users must remember to redirect
+any verbose output to a file if they submit the task to the background or
+it will be lost.
+
+.NH 2
+Background Jobs
+
+.PP
+Any DAOPHOT task can be run in background by appending an ampersand
+to the end of the command. For example the \fBphot\fR task can be run
+as a background job as shown below.
+
+.YS
+da> phot image image.coo.1 image.mag.1 verbose- verify- &
+.YE
+
+The user must be sure to turn off verbose mode
+and set the verify switch to no. VMS users may have to append a queue
+name after the trailing ampersand.
+If verbose output is desired it can be captured in a file as shown 
+in the example below below. The & after the > will ensure that any error
+output is also captured.
+
+.YS
+da> phot image image.coo.1 image.mag.1 verbose+ inter- verify- \\ 
+    >& listing &
+.YE
+
+.NH 2
+Timing Tests
+
+.PP
+Any DAOPHOT or IRAF task can be timed by prepending a $ sign to the
+command as shown below.
+
+.YS
+da> $phot image image.coo.1 image.mag.1 inter- verify- verbose- &
+.YE
+
+At task termination the computer will print the cpu and elapsed
+time on the terminal.
+.PP
+Care must be taken in using this feature
+to make timing comparisons between hosts or even between runs on the same host,
+as factors like which queue a task is submitted to (VMS), which version of
+the OS the host is running, which  version of the compiler
+two programs were compiled under, 
+whether the disks are local or networked, and the number of users on the
+machine will effect the elapsed time and/or the cpu time.
+
+.NH
+Doing Photometry with DAOPHOT
+
+.NH 2
+The Test Image
+
+.PP
+Each of the DAOPHOT analysis steps summarized in the following section
+and discussed in detail in succeeding
+sections uses the artificial image stored in fits format in
+the file "daophot$test/fits3.fits" as test data. This image is small,
+51 by 51 pixels, contains 10 stars whose coordinates and magnitudes
+are listed below, has, a mean background level of ~100, poisson noise
+statistics, a gain of 1.0,  and a readout noise of 0.0.
+
+.YS
+# Artificial Stars for Image Test
+
+    41.0   4.0  17.268
+    23.0   7.0  17.600
+    18.0   8.0  17.596
+    26.0  22.0  16.777
+    36.0  22.0  16.317
+     8.0  23.0  16.631
+    31.0  25.0  16.990
+    21.0  26.0  19.462
+    29.0  34.0  17.606
+    36.0  42.0  16.544
+.YE
+
+.PP
+Results for this test image are used to illustrate the text. It is hoped
+that users so inclined will be able to mimic the reductions on 
+their host machine. The fact that the image is small, means that
+the tasks execute quickly, it is possible to display all the
+important results in the manual, and it is possible for
+the user to track and examine  all the important numbers, something not
+easy with larger images. Users are encouraged to construct more
+challenging artificial images with the ARTDATA package, and to run
+them through DAOPHOT.
+.PP
+All the examples in the following text were run
+under IRAF 2.10.3 on a SPARCstation IPX. Users with different hardware
+may see minor deviations from the output shown here due to machine
+precision differences. 
+
+.NH 2
+Typical Analysis Sequence
+
+.PP
+The following sequence of operations summarizes the steps required to analyze
+a crowded stellar field with DAOPHOT.
+.IP [1]
+Create a directory in which to analyze the image and make it the current
+working directory. By default all output photometry and image files
+will be written there.
+.IP [2]
+Read the reduced image into the working directory with the DATAIO package
+task \fBrfits\fR.
+.IP [3]
+Check that the correct exposure time, airmass, filter id, time of
+observation, gain, and readout noise are present and correct
+in the image header with the \fBhselect\fR task. Enter / edit them
+with the \fBhedit\fR task if they are not. Correct the exposure time for
+shutter error, the airmass to mid-exposure, and the gain
+and readout noise to the effective gain and readout noise, using the
+\fBhedit\fR and/or \fBsetairmass\fR tasks.
+.IP [4]
+Edit the DAOPHOT algorithm psets with the interactive \fBdaoedit\fR
+task. The parameters that require editing at this point are:
+1) the numerical parameters
+\fIfwhmpsf\fR (full-width at half-maximum of the point-spread function),
+\fIsigma\fR (standard deviation of the background in counts), \fIdatamin\fR
+(the minimum good data value in counts), \fIdatamax\fR (the maximum good
+data value in counts), and the image header keyword parameters
+\fIccdread\fR, \fIgain\fR, \fIexposure\fR, \fIairmass\fR, \fIfilter\fR,
+and \fIobstimes\fR in the \fBdatapars\fR parameter set,
+2) \fIcbox\fR (the centering box width) in the \fBcenterpars\fR parameter set,
+3) \fIannulus\fR (inner radius of the sky annulus) and
+\fIdannulus\fR (width of the sky annulus) in the \fBfitskypars\fR parameter set,
+4) \fIapertures\fR (radii of the photometry apertures) in the \fBphotpars\fR
+parameter set, and 5) \fIpsfrad\fR (maximum radius of the psf model)
+and \fIfitrad\fR (psf model fitting radius) in the \fBdaopars\fR parameter set.
+.IP [5]
+Create an initial star list using the \fBdaofind\fR task.
+Mark the detected stars on the image display with the \fBtvmark\fR task
+and adjust the \fBfindpars\fR parameter \fIthreshold\fR until
+a satisfactory star list is created.
+.IP [6]
+Compute sky background values and initial magnitudes for
+the detected stars using the \fBphot\fR task and the star
+list written by the \fBdaofind\fR task in step [5].
+.IP [7]
+Create a psf star list using the \fBpstselect\fR task
+and the photometry file written by \fBphot\fR in step [6]. Mark
+the coordinates of the psf stars on the image display with the
+\fBtvmark\fR task
+and edit out any non-stellar objects, stars with 
+neighbors within \fIfitrad\fR pixels, or stars with obvious
+cosmetic blemishes, using the \fBpexamine\fR  task.
+.IP [8]
+Compute the current psf model using the
+\fBpsf\fR task, the input photometry file written by the \fBphot\fR task
+in step [6], and the psf star list written by the \fBpstselect\fR task
+in step [7].
+.IP [9]
+Fit the current psf model to the psf stars and their neighbors 
+using the \fBnstar\fR task, the psf star group photometry file
+written by the \fBpsf\fR task in step [8] or created by the user in step [11],
+and the current psf model written by the \fBpsf\fR task in steps [8] or [13].
+Subtract the fitted psf stars
+and their neighbors from the original image using the \fBsubstar\fR task,
+the photometry file written by the \fBnstar\fR task, and the current
+psf model.
+Display the subtracted image, mark the psf stars and their neighbors
+on the display with the \fBtvmark\fR task,
+and examine the \fBnstar\fR photometry
+file and the subtracted image with the \fBpexamine\fR task.
+If all the psf stars subtract out cleanly and none of them have any
+significant neighbors, skip directly to step [14]. If all the  psf stars
+and their neighbors subtract out cleanly, and one or more of the psf
+stars do have significant neighbors, skip directly to step [13].
+.IP [10]
+Reexamine the subtracted image written in step [9]. Remove any psf stars
+revealed by the subtraction to be  non-stellar, multiple, or to contain
+cosmetic blemishes,
+from the psf star list written by the \fBpsf\fR task in step
+[8] using the \fBpexamine\fR task.
+If any bad psf stars are detected recompute the psf model by returning to
+step [8] using the newly edited psf star list in place
+of the one written by the previous execution of the \fBpsf\fR task in step [8].
+.IP [11]
+Add any psf star neighbors too faint to be detected by the \fBdaofind\fR
+task in step [5] but bright enough to effect the computation of the
+psf model, to the original psf star group photometry file written
+by the \fBpsf\fR task in step [8],
+by estimating their positions, sky values, and magnitudes interactively
+with the \fBphot\fR task, merging the results with the original psf star group
+photometry file
+using the \fBpfmerge\fR task, and regrouping the stars with the \fBgroup\fR
+task. Refit the newly grouped psf stars and their neighbors using
+the current psf model by returning to step [9],
+replacing the original input group photometry file with the one
+including the new psf star neighbors.
+.IP [12]
+Using the subtracted image written by the \fBsubstar\fR task in step [9],
+note any systematic patterns in the psf star residuals with distance from
+the star (\fIthese indicate a poorly chosen value for the annulus,
+dannulus, function,  or psfrad parameters),
+position in the image (\fIthese suggest that the psf is variable
+and that the value of the varorder parameter should be increased\fR),
+or intensity (\fIthis suggests problems with the image data itself, e.g.
+non-linearity\fR). If the problem is in the sky fitting parameters
+edit the appropriate algorithm parameters and return to step [6]. If
+the problem is in the psf modeling and fitting parameters, edit the
+appropriate algorithm parameters and return to step [7]. I the problem
+appears to be in the data or the data reduction procedures, review the
+data taking and reduction history of the image before proceeding.
+.IP [13]
+Subtract the psf star neighbors but not the psf stars from the original
+image using the \fBsubstar\fR task,
+the photometry file written by the \fBnstar\fR task
+in step [9], and the psf star list and current psf model written by
+the \fBpsf\fR task in step [8].
+Recompute the current psf model using
+the psf neighbor star subtracted image, the psf star group photometry file
+written by the \fBpsf\fR task in step [8] or created by the user in step [11],
+and the psf star list written in step [8].
+If the \fIvarorder\fR parameter was changed
+return to step [9].
+Otherwise save the psf star neighbor subtracted image as it may be
+required for computing the image
+aperture correction in step [20], and proceed to step [14].
+.IP [14]
+Fit the final psf model computed in steps [8] or [13]
+to the stars in the photometry file written in
+step [6] using the \fBallstar\fR task.
+.IP [15]
+Run \fBdaofind\fR on the subtracted image produced by \fBallstar\fR in step
+[14] in order to pick up stars missed by the first pass of \fBdaofind\fR in
+step [5].
+.IP [16]
+Run \fBphot\fR on the original image using the new star list produced by
+\fBdaofind\fR in step [15] and the \fBphot\fR algorithm parameters used
+in step [6].
+.IP [17]
+Merge the photometry file produced by \fBallstar\fR in step [14] with
+the one produced by \fBphot\fR in step [16] using the \fBpfmerge\fR
+task.
+.IP [18]
+Rerun \fBallstar\fR on the original image using the merged photometry file
+created in step [17] and the psf model created in steps [8] or [13]. 
+.IP [19]
+Repeat steps [15]-[18] as required, remembering to run \fBdaofind\fR
+on the subtracted image produced by \fBallstar\fR and \fBphot\fR on the
+original image.
+.IP [20]
+If the psf model is constant, compute the aperture correction for the
+image using the original image and  a sample of bright well-isolated stars
+if possible, or the image with
+the psf neighbor stars subtracted if necessary, the
+\fBphot\fR task, and the PHOTCAL package \fBmkapfile\fR task.
+If the psf model is variable, compute the aperture correction by calculating
+the mean magnitude difference, for the psf stars with any
+the neighbors subtracted, between the psf model fitted magnitudes computed
+by the \fBnstar\fR task, and large aperture photometry magnitudes computed
+with the \fBphot\fR task.
+.IP [21]
+Archive the algorithm parameters for the image with the \fBsetimpars\fR task
+and proceed to the next image.
+
+
+.NH 2
+Creating and Organizing an Analysis Directory
+
+.PP
+By default DAOPHOT reads and writes data from and to the current working
+directory. To create and set a new working directory the user must
+execute the commands \fBmkdir\fR and \fBchdir\fR as shown below.
+
+.YS
+da> mkdir testim
+da> chdir testim
+.YE
+
+.PP
+DAOPHOT can in the course of reducing a single image,
+generate a large number of photometry catalogs and output images.
+Users should take a moment to consider how they wish to organize their data
+directories before beginning any DAOPHOT analysis.  Some possibilities for data
+directory organization are: 1) by night of observation for standard star fields,
+2) by star field for multi-filter observations of a crowded field, or
+3) by individual image for single filter observations of several fields,
+or any combination of the above.
+
+.NH 2
+Reading the Data 
+
+.PP
+DAOPHOT input images are normally read into IRAF from FITS files with
+the DATAIO package task \fBrfits\fR. The following example shows how to
+read the DAOPHOT test image stored in the FITS file "daophot$test/fits3.fits"
+into the IRAF image test.imh.
+
+.YS
+da> rfits daophot$test/fits3.fits 1 test
+File: test  Artificial Starfield  Size = 51 x 51
+.YE
+
+When IRAF supports FITS format images on disk this step will no longer be
+necessary, although for some images it may still be desirable for
+image i/o efficiency reasons.
+
+.NH 2
+Editing the Image Headers
+
+.NH 3
+The Minimum Image Header Requirements
+
+.PP
+Before beginning DAOPHOT reductions the user must gather
+all the data required to determine the following quantities:
+1) the effective readout noise of the detector in electrons, 2) the effective
+gain of the detector in electrons per count, 3) the maximum good data value
+of the detector in counts, 4) the effective exposure time in any units
+as long as these units are identical for all the images to be analyzed
+together,
+5) the filter id, 6) the effective airmass of the observation at mid-exposure,
+and 7) the time of the observation. 
+
+.NH 3
+The Effective Gain and Readout Noise
+
+.PP
+The DAOPHOT package tasks require correct effective
+gain and readout noise values for:
+1) the computation of the magnitude errors in the \fBphot\fR (gain only
+required), \fBpeak\fR, \fBnstar\fR and \fBallstar\fR tasks,
+2) the computation of the optimal weights used by the non-linear
+least-squares fitting code in the \fBpeak\fR, \fBnstar\fR, and 
+\fBallstar\fR tasks,
+3) the computation of the predicted signal-to-noise
+ratios in the \fBgroup\fR task,
+4) the computation of the sharpness and chi statistics in the \fBpeak\fR,
+\fBnstar\fR, and \fBallstar\fR tasks, and 5) the correct computation of
+the poisson noise (gain only required) in the \fBaddstar\fR task.
+.PP
+Nominal gain and readout noise values for a single image
+should be obtained from the instrument
+scientist. These values should also be determined/checked empirically with the
+PROTO package task \fBfindgain\fR using bias and flat-field frames that
+are unprocessed and uncoadded so that the noise characteristics of the
+original data are preserved.
+.PP
+If the input image is the sum or average of several frames
+the gain and readout noise values in the image headers must be edited
+from single frame to effective  gain and readout noise values
+as shown below. In the following examples
+gain and effective gain are in electrons / ADU,
+readout noise and effective readout noise are  in electrons, and N is the
+number of individual frames which
+have been summed, averaged, or medianed to create the input image.
+
+.nf
+	[1]. The image is the sum of N frames
+
+             effective gain = gain
+	     effective readout noise = sqrt (N) * readout noise
+
+	[2]. The image is the average of N frames
+
+             effective gain = N * gain
+	     effective readout noise = sqrt (N) * readout noise
+
+	[3]. The image is the median of N frames
+
+             effective gain = 2.0 * N * gain / 3
+	     effective readout noise = sqrt (2 * N / 3) * readout noise
+.fi
+
+.PP
+The following example shows how to add the correct values of gain and
+readout noise, which in this very artificial example are 1.0 and 0.0
+respectively, to the header of the test image with the \fBhedit\fR task.
+
+.YS
+da> imheader test l+
+test[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=71.00896, max=535.1335
+    Line storage mode, physdim [51,51], length of user area 163 s.u.
+    Created Mon 09:59:00 17-May-93, Last modified Mon 09:59:00 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+da> hedit test gain 1.0 add+ verify-
+add test,gain = 1.
+test updated
+da> hedit test rdnoise 0.0 add+ verify-
+add test,rdnoise = 0.
+test updated
+da> imheader test l+
+test[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=71.00896, max=535.1335
+    Line storage mode, physdim [51,51], length of user area 244 s.u.
+    Created Mon 09:59:00 17-May-93, Last modified Mon 09:59:00 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+    GAIN    =                   1.
+    RDNOISE =                   0.\fR
+.YE
+
+.PP
+The following example shows how to correct the single frame
+values of gain and readout noise, already present in the input image
+header, to account for the fact that the input image is actually the
+average of three frames (note that the frames are NOT actually independent
+in this example!). 
+
+.YS
+da> imsum test,test,test testav3 option=average
+da> hedit testav3 gain "(3.0*gain)" verify- 
+testav3,GAIN: 1. -> 3.
+testav3 updated
+da> hedit testav3 rdnoise "(rdnoise*sqrt(3.0))" verify-
+testav3,RDNOISE: 0. -> 0.
+testav3 updated
+da> imheader testav3 l+
+testav3.imh[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=unknown, max=unknown
+    Line storage mode, physdim [51,51], length of user area 244 s.u.
+    Created Mon 11:02:22 17-May-93, Last modified Mon 11:02:22 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/testav3.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    New copy of test
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+    GAIN    =                   3.
+    RDNOISE =                   0.
+.YE
+
+.NH 3
+The Maximum Good Data Value
+
+.PP
+Datamax is the maximum good data value in counts. Datamax
+is the count level at which the detector saturates or the count
+level at which it becomes non-linear, whichever is lower. DAOPHOT requires
+a correct value of datamax to: 1) identify bad data in the \fBdaofind\fR,
+\fBphot\fR, \fBpsf\fR, \fBpeak\fR, \fBgroup\fR, \fBnstar\fR,
+and \fBallstar\fR tasks, and 2) identify saturated stars in the \fBphot\fR,
+\fBpsf\fR, and \fBsubstar\fR tasks.
+.PP
+Users should be sure to allow adequate leeway for the detector bias level
+in their determination of datamax. Test is an artificial image
+linear over its entire data range. However as an example assume that it was
+actually observed with a detector which is linear from 0 to 25000 counts
+at a gain setting of 1.0, and that the mean bias level that was subtracted
+from the raw data was ~400 counts.
+In that case the user should set datamax to something like 24500 not 25000
+counts.
+.PP
+Datamax may be stored in the image header with \fBhedit\fR
+as shown below. The use of the header keyword gdatamax instead of
+datamax avoids any confusion with the reserved FITS keywords
+datamin and datamax should they already be present in the image header,
+or the IRAF keywords iraf-max and iraf-min which have the same meaning.
+
+.YS
+da> hedit test gdatamax 24500 add+ verify-
+add test,gdatamax = 24500
+test updated
+.YE
+
+.NH 3
+The Effective Exposure Time
+
+.PP
+The exposure time is used by the \fBphot\fR task to normalize the computed
+initial magnitudes to an effective exposure time of one time unit. The
+magnitude scale established in \fBphot\fR is preserved
+in all the subsequent DAOPHOT analysis. Setting the correct exposure
+time in the image headers before beginning DAOPHOT reductions will
+simplify the book-keeping required in the later calibration step
+significantly.
+.PP
+Exposure times should also be corrected
+for any timing errors in the instrument shutter, although this is normally
+important only for short exposure observations of standard stars.
+.PP
+The following example shows how to add the exposure time in seconds
+to the image header, and how to correct it for a known shutter error
+of 13 milli-seconds. Note that rather than overwrite the nominal exposure time
+exptime, the user has chosen to store the corrected exposure time in
+a new keyword cexptime.
+
+.YS
+da> hedit test exptime 1.0 add+ verify-
+add test,exptime = 1.
+test updated
+da> hedit test cexptime "(exptime+.013)" add+ verify- 
+add test,cexptime = 1.013
+test updated
+da> imheader test l+
+test[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=71.00896, max=535.1335
+    Line storage mode, physdim [51,51], length of user area 365 s.u.
+    Created Mon 09:59:00 17-May-93, Last modified Mon 09:59:00 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+    GAIN    =                   1.
+    RDNOISE =                   0.
+    GDATAMAX=                24500
+    EXPTIME =                   1.
+    CEXPTIME=                1.013
+.YE
+
+.NH 3
+The Airmass, Filter Id, and Time of Observation
+
+.PP
+The airmass, filter id, and time of observation are not used directly by
+any of the DAOPHOT tasks.  They are read from the image header and recorded
+in the output photometry files. Correctly setting the airmass,
+filter id,  and the time of observation in the image headers before running
+any DAOPHOT tasks will however significantly reduce the book-keeping
+required in the subsequent calibration step.
+.PP
+The airmass can be computed and/or corrected to mid-exposure with the
+ASTUTIL package task \fBsetairmass\fR. By default \fBsetairmass\fR requires
+that the name of the observatory, date of observation, ra and dec, epoch of
+the ra and dec, sidereal time, and exposure time be recorded
+in the image header in the appropriate units in the keywords
+observat, date-obs, ra, dec, epoch, st, and exptime. Hopefully most or
+all of this information is already in the image header but in case
+it is not, the following example shows how to edit it in and run
+\fBsetairmass\fR.
+
+.YS
+da> hedit test observat "CTIO" add+ verify- show-
+da> hedit test "date-obs" "12/10/88" add+ verify- show-
+da> hedit test ra "(str('21:51:59.0'))" add+ verify- show-
+da> hedit test dec "(str('02:33:31.0'))" add+ verify- show-
+da> hedit test epoch 1985.0 add+ verify- show-
+da> hedit test st "(str('20:47:55.0'))" add+ verify- show-
+da> setairmass test show-
+da> imheader test l+
+test[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=71.00896, max=535.1335
+    Line storage mode, physdim [51,51], length of user area 649 s.u.
+    Created Mon 09:59:00 17-May-93, Last modified Mon 09:59:00 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+    GAIN    =                   1.
+    RDNOISE =                   0.
+    GDATAMAX=                24500
+    EXPTIME =                   1.
+    CEXPTIME=                1.013
+    OBSERVAT= 'CTIO    '
+    DATE-OBS= '12/10/88'
+    RA      = '21:51:59.0'
+    DEC     = '02:33:31.0'
+    EPOCH   =                1985.
+    ST      = '20:47:55.0'
+    AIRMASS =             1.238106
+.YE
+
+The tortuous syntax required to enter the ra, dec, and st keywords is
+necessary in order to avoid \fBhedit\fR turning strings like
+"21:51:59.0" into numbers,
+e.g. 21.86639. \fBSetairmass\fR permits the user to change the
+default names for the date-obs and exptime image header keywords but
+not those of observat, ra, dec, epoch or st.
+To list the observatories in the IRAF observatory database and/or to find out
+how to deal with the case of data taken at an observatory not in the
+observatory database, the user should consult the help page for the
+\fBobservatory\fR task.
+.PP
+The filter id is a string defining the filter used to take the observations.
+It can be easily edited into the image header as shown below.
+
+.YS
+da> hedit test filters V add+ verify- show-
+.YE
+
+Users should be aware that any embedded blanks will be removed from the
+filter id after it is read from the image header, but before it is
+recorded in the photometry files. For example a filter id of "V band"
+in the image header will become "Vband" in the photometry file.
+.PP
+The time of observation is a string defining the time at which the
+observation was taken. The time of observation may be ut or local
+standard time. If the time of observation is not already recorded in
+the image header it can be entered in the usual fashion as shown below.
+
+.YS
+da> hedit test ut "(str('00:07:59.0'))" add+ verify- show-
+.YE
+
+.PP
+After editing the "final" image header should look something like the
+following.
+
+.YS
+da> imheader test l+
+test[51,51][real]: Artificial Starfield with Noise
+    No bad pixels, no histogram, min=71.00896, max=535.1335
+    Line storage mode, physdim [51,51], length of user area 730 s.u.
+    Created Mon 09:59:00 17-May-93, Last modified Mon 09:59:00 17-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.pix' [ok]
+    'KPNO-IRAF'           /
+    '10-05-93'            /
+    IRAF-MAX=           5.351335E2  /  DATA MAX
+    IRAF-MIN=           7.100896E1  /  DATA MIN
+    IRAF-BPX=                   32  /  DATA BITS/PIXEL
+    IRAFTYPE= 'REAL    '            /  PIXEL TYPE
+    GAIN    =                   1.
+    RDNOISE =                   0.
+    GDATAMAX=                24500
+    EXPTIME =                   1.
+    CEXPTIME=                1.013
+    OBSERVAT= 'CTIO    '
+    DATE-OBS= '12/10/88'
+    RA      = '21:51:59.0'
+    DEC     = '02:33:31.0'
+    EPOCH   =                1985.
+    ST      = '20:47:55.0'
+    AIRMASS =             1.238106
+    FILTER  = 'V       '
+    UT      = '00:07:59.0'\fR
+.YE
+
+
+.NH 3
+Batch Header Editing
+
+.PP
+The previous examples described in detail how to enter each of the required
+keyword and value pairs into the image header using the \fBhedit\fR task.
+Users with large number of header keywords to enter should consider using the
+more batch oriented alternative task \fBasthedit\fR.
+
+
+.NH 2
+Editing, Checking, and Storing the Algorithm Parameters
+
+.NH 3
+The Critical Algorithm Parameters
+
+.PP
+The critical DAOPHOT algorithm parameters that should be set
+before beginning any DAOPHOT analysis are:
+1) the 
+full-width at half-maximum of the psf \fIfwhmpsf\fR, the standard
+deviation of the sky background in counts \fIsigma\fR, the minimum and
+maximum good data values \fIdatamin\fR and \fIdatamax\fR, and the image
+header keyword parameters 
+\fIccdread\fR, \fIgain\fR, \fIexposure\fR, \fIairmass\fR, \fIfilter\fR,
+and \fIobstimes\fR in the \fBdatapars\fR parameter set,
+2) the default centering algorithm \fIcalgorithm\fR and centering box
+\fIcbox\fR parameters in the \fBcenterpars\fR parameter set, 3) the sky fitting
+algorithm \fIsalgorithm\fR, and the sky annulus \fIannulus\fR and
+\fIdannulus\fR parameters in the \fBfitskypars\fR parameter set,
+4) the \fIapertures\fR parameter in the \fBphotpars\fR parameter set,
+and 5) the psf radius \fIpsfrad\fR
+and fitting radius \fIfitrad\fR parameters in the \fBdaopars\fR parameter set.
+The reamining parameters should be left at their default values, at least
+initially.
+
+.NH 3
+Editing the Algorithm Parameters Interactively with Daoedit 
+
+.PP
+The DAOPHOT algorithm parameter editing task is \fBdaoedit\fR. \fBDaoedit\fR
+permits
+the user to edit all the algorithm parameter sets at once. It offers all the
+capabilities of the IRAF parameter editing task \fBepar\fR, plus the
+ability to set parameters using the displayed image and radial
+profile plots of isolated stars.
+.PP
+To run \fBdaoedit\fR the user displays the image, types \fBdaoedit\fR, and waits
+for the image cursor to appear ready to accept user commands. The following
+example summarizes a typical \fBdaoedit\fR parameter editing session.
+
+.YS
+da> display test 1 fi+ 
+da> daoedit test
+.YE
+
+.IP ...
+Execute the command \fB":epar datapars"\fR and enter the correct
+values for the \fIdatamax\fR parameter, and the image header
+keyword parameters \fIccdread\fR, \fIgain\fR, \fIexposure\fR, \fIairmass\fR,
+\fIfilter\fR, and \fIobstime\fR.
+.IP ...
+Choose a bright isolated star and execute the \fBr\fR cursor
+keystroke command to plot its radial profile.
+.IP ...
+From the information in the radial plot header and the plot
+itself estimate reasonable values for the full-width at
+half-maximum of the psf, the sky level, and the standard
+deviation of the sky level in the image.
+.IP ...
+Repeat the previous step for several stars in order to
+confirm that the original estimated values are reasonable.
+.IP ...
+Execute the \fB":epar datapars"\fR command once more and enter
+the estimated values of the full-width at half-maximum of the psf and
+the standard deviation of the sky background in the \fIfwhmpsf\fR
+and \fIsigma\fR parameters respectively.
+.IP ...
+Set the \fIdatamin\fR parameter to the estimated sky background level
+minus k times the standard deviation of the sky background, where
+k is a number between 5.0 and 7.0.
+
+.IP 
+then
+
+.IP ...
+Execute the command \fB":epar centerpars"\fR and set the \fIcbox\fR
+parameter to 5 pixels or ~ 2 * \fIfwhmpsf\fR whichever is
+greater.
+.IP ...
+Execute the command \fB":epar fitskypars"\fR and set the \fIannulus\fR
+parameter to ~ 4 * \fIfwhmpsf\fR and the \fIdannulus\fR parameter to a
+number between 2.5 * \fIfwhmpsf\fR and 4.0 * \fIfwhmpsf\fR.
+.IP ...
+Execute the command \fB":epar photpars"\fR and set the apertures
+parameter to ~ 1.0 * fwhmpsf or 3 pixels whichever is greater.
+.IP ...
+Execute the command \fB":epar daopars"\fR and set the \fIpsfrad\fR
+parameter to ~ 4 * \fIfwhmpsf\fR + 1 and the \fIfitrad\fR parameter to
+~ 1.0 * \fIfwhmpsf\fR or 3 pixels whichever is greater.
+
+.IP
+or alternatively
+
+.IP ...
+Move to a bright star and execute the \fBi\fR cursor keystroke
+command to enter the interactive setup menu.
+.IP ...
+Mark the \fIfwhmpsf\fR, \fIcbox\fR, \fIannulus\fR, \fIdannulus\fR,
+\fIapertures\fR, \fIpsfrad\fR, and \fIfitrad\fR parameters with the
+graphics cursor on the displayed radial profile plot, and verify and/or
+roundoff the marked values.
+.PP
+The following sections discuss in detail how to edit each of the
+parameter sets using the test image as a specific example.
+
+.NH 4
+The Data Dependent Algorithm Parameters 
+
+.PP
+A subset of the datapars parameters are used to specify the
+characteristics of the detector, including the saturation or linearity
+limit (\fIdatamax\fR) and noise model (\fIccdread\fR
+and \fIgain\fR), and the parameters of the observation, including
+exposure time (\fIexposure\fR),
+airmass (\fIairmass\fR), filter (\fIfilter\fR), and time of observation
+(\fIobstime\fR).
+.PP
+To edit the \fBdatapars\fR algorithm parameter set
+from within the \fBdaoedit\fR task the user enters the command
+\fB":epar datapars"\fR to invoke the \fBepar\fR task and edits the
+parameters in the usual manner.
+Editing is terminated with the usual \fB":wq"\fR command which returns the
+user to the main \fBdaoedit\fR command loop.
+.PP
+After the appropriate  \fIdatamax\fR, \fIccdread\fR, \fIgain\fR,
+\fIexposure\fR, \fIairmass\fR,
+\fIfilter\fR, and \fIobstime\fR parameter values for the
+test image are entered, the \fBdatapars\fR
+parameter should look as follows.
+
+.YS
+Image Reduction and Analysis Facility
+PACKAGE = daophot
+   TASK = datapars
+
+(scale  =            1.) Image scale in units per pixel
+(fwhmpsf=           2.5) FWHM of the PSF in scale units
+(emissio=           yes) Features are positive ?
+(sigma  =             0.) Standard deviation of background in counts
+(datamin=         INDEF) Minimum good data value
+(datamax=         24500) Maximum good data value
+(noise  =       poisson) Noise model
+(ccdread=       rdnoise) CCD readout noise image header keyword
+(gain   =          gain) CCD gain image header keyword
+(readnoi=            0.) CCD readout noise in electrons
+(epadu  =            1.) Gain in electrons per count
+(exposur=      cexptime) Exposure time image header keyword
+(airmass=       airmass) Airmass image header keyword
+(filter =        filter) Filter image header keyword
+(obstime=            ut) Time of observation image header keyword
+(itime  =            1.) Exposure time
+(xairmas=         INDEF) Airmass
+(ifilter=         INDEF) Filter
+(otime  =         INDEF) Time of observation
+(mode   =            ql)
+.YE
+
+.PP
+Users should realize that the values of the parameters \fIreadnoise\fR
+and \fIepadu\fR will be used for the gain and readout noise if the image
+header keywords specified by \fIccdread\fR and \fIgain\fR are not found
+in the image header or cannot be correctly decoded. Similarly the values of
+the \fIitime\fR,
+\fIxairmass\fR, \fIifilter\fR, and \fIotime\fR parameters will be used
+for the exposure time, airmass, filter id, and time of observation if
+the image header keywords specified by \fIexposure\fR, \fIairmass\fR,
+\fIfilter\fR, and \fIobstime\fR are not found in the image header
+or cannot be correctly decoded.
+.PP
+The \fBdatapars\fR parameters \fIfwhmpsf\fR, \fIsigma\fR, and \fIdatamin\fR 
+are used to: 1) determine the size of star for which the \fBdaofind\fR star
+detection algorithm is optimized (fwhmpsf), 2) define the \fBdaofind\fR
+algorithm detection threshold for faint objects (sigma),
+3) define the fwhm of the psf for the \fBphot\fR task centering algorithms 
+"gauss" and "ofilter" (fwhmpsf),
+4) supply a first guess for the true fwhm of the psf to the psf
+function fitting task \fBpsf\fR (fwhmpsf), 5) determine the
+minimum good data value
+in the \fBdaofind\fR, \fBphot\fR, \fBpsf\fR, \fBpeak\fR, \fBgroup\fR, 
+\fBnstar\fR, and \fBallstar\fR tasks (datamin).
+.PP
+Reasonable values for these parameters can be obtained by examining the
+radial profile plots of several isolated stars from within the
+\fBdaoedit\fR task as outlined below:
+
+.IP ...
+Move the image cursor on the displayed image to a
+reasonably bright isolated star (a good candidate is
+the star at pixel 8,23 in the test image) and execute
+the \fBr\fR keystroke command.
+A radial and integrated profile plot of the selected
+star will appear on the screen with the largest photometry aperture radius,
+inner and outer radii of the sky annulus, and median sky level in the sky
+annulus marked on the plot.
+.IP ...
+Assuming that the plot is normal, note the computed
+fwhmpsf (2.6 rounded to the nearest tenth of a pixel for
+the star at 8,23), median sky value (100 counts rounded
+to the nearest count for the star at 8,23), and standard
+deviation of the sky values (10 counts rounded to the
+nearest count for the star at 8,23) written in the plot header.
+These numbers suggest a value of ~50 for datamin (50 is ~5
+standard deviations of the background counts below the
+background count estimate)
+.IP ...
+Edit the estimated values into the datapars pset by
+typing the command \fB":epar datapars"\fR, entering the values,
+and typing \fB":wq"\fR to update the parameter set.
+
+.IP
+or
+
+.IP ...
+Enter them individually using the daoedit colon commands,
+e.g. \fB":fwhmpsf 2.5"\fR, \fB":sigma 10.0"\fR, and \fB":datamin 50.0"\fR.
+.IP ...
+Check the new values of \fIfwhmpsf\fR, \fIsigma\fR, and \fIdatamin\fR 
+by doing radial profile plots of several
+other isolated stars (the stars at 36,42 and 41,4 in
+the test image are good test stars).
+.IP ...
+On the basis of the estimated \fIfwhmpsf\fR of these stars change the
+fwhmpsf parameter back to 2.5 with the command
+\fB":fwhmpsf 2.5"\fR.
+.IP ...
+Check that the observed standard deviation of the sky
+background, sigma, agrees reasonably well with the
+predicted value, psigma, based on the median sky level,
+and the effective gain and readout noise of the image.
+For the test image these numbers are related as shown below.
+
+.nf
+	    psigma = sqrt (median sky / effective gain +
+	             (effective rdnoise / effective gain) ** 2)
+	           ~ sqrt (100.0 / 1.0 + (0. / 1.0) ** 2)
+	           ~ 10.0
+	           ~ sigma
+.fi
+
+.IP ...
+If psigma and sigma are significantly different check
+that the sky region  is uncrowded, that the effective
+gain and readout noise values are correct, and that
+earlier reduction procedures have not altered the image
+statistics in some fundamental manner
+
+.PP
+The \fIemission\fR parameter must be left at "yes",
+since DAOPHOT assumes that stars are local maxima not local minima.
+.PP
+The \fInoise\fR parameter must be left at "poisson" since poisson noise
+statistics are assumed throughout the DAOPHOT package.
+.PP
+The \fIscale\fR parameter defines the units in which radial distances
+in the image will be measured. For example if the image scale
+is 0.25 "/pixel, users can set \fIscale\fR to 0.25 if they wish
+to define the \fIfwhmpsf\fR, \fIcbox\fR, \fIannulus\fR, \fIdannulus\fR,
+\fIapertures\fR, \fIpsfrad\fR, \fIfitrad\fR and all the other algorithm
+parameters which are defined in terms of a radial distance in arc-seconds.
+For simplicity most users choose to leave scale set to 1.0 and
+work in pixels.
+.PP
+The final version of the \fBdatapars\fR parameter set should look something
+like the following.
+
+.YS
+Image Reduction and Analysis Facility
+PACKAGE = daophot
+   TASK = datapars
+
+(scale  =            1.) Image scale in units per pixel
+(fwhmpsf=           2.5) FWHM of the PSF in scale units
+(emissio=           yes) Features are positive ?
+(sigma  =           10.) Standard deviation of background in counts
+(datamin=           50.) Minimum good data value
+(datamax=         24500) Maximum good data value
+(noise  =       poisson) Noise model
+(ccdread=       rdnoise) CCD readout noise image header keyword
+(gain   =          gain) CCD gain image header keyword
+(readnoi=            0.) CCD readout noise in electrons
+(epadu  =            1.) Gain in electrons per count
+(exposur=      cexptime) Exposure time image header keyword
+(airmass=       airmass) Airmass image header keyword
+(filter =        filter) Filter image header keyword
+(obstime=       obstime) Time of observation image header keyword
+(itime  =           1.0) Exposure time
+(xairmas=         INDEF) Airmass
+(ifilter=         INDEF) Filter
+(otime  =         INDEF) Time of observation
+(mode   =            ql)
+.YE
+
+
+.NH 4
+The Centering Algorithm Parameters
+
+.PP
+The \fBcenterpars\fR parameter set controls the centering algorithms used by
+the \fBphot\fR aperture photometry task. DAOPHOT users should concern
+themselves with only two of these parameters, \fIcalgorithm\fR and \fIcbox\fR,
+and leave the remaining \fBcenterpars\fR parameters at their default values.
+.PP
+\fICalgorithm\fR specifies the default \fBphot\fR centering algorithm. Its value
+should be "none" if the input coordinate list is the output of the
+\fBdaofind\fR task,  or "centroid", "gauss", or "ofilter" if the input
+coordinate list was
+created with the image or graphics cursor list tasks \fBrimcursor\fR
+or \fBrgcursor\fR or the coordinates are
+read from the image cursor in interactive mode. The choice of centering
+algorithm is not critical since the centers are recomputed using accurate
+non-linear least-squares fitting techniques during the psf fitting
+process. The most efficient and simplest choice is "centroid", although
+more accurate results may be obtained with "gauss" which is
+very similar to the centering algorithm used in \fBdaofind\fR.
+.PP
+The \fIcbox\fR
+parameter determines the width in scale units of the data used to compute
+the center if \fIcalgorithm \fR is not "none".
+For reasonable results \fIcbox\fR should be set to the equivalent of 5 or 
+~ 2 * \fIfwhmpsf\fR in pixels whichever is larger.
+.PP
+\fBCenterpars\fR can be edited from within the \fBdaoedit\fR task
+with the command \fB":epar centerpars"\fR. After editing, the \fBcenterpars\fR
+parameter set should look like the example below. Note that for the test
+image \fIfwhmpsf\fR is ~2.5 pixels so \fIcbox\fR is left at 5.0. 
+
+.YS
+PACKAGE = daophot
+TASK = centerpars
+
+(calgori=          none) Centering algorithm
+(cbox   =            5.) Centering box width in scale units
+(cthresh=            0.) Centering threshold in sigma above background
+(minsnra=            1.) Minimum signal-to-noise ratio
+(cmaxite=            10) Maximum iterations
+(maxshif=            1.) Maximum center shift in scale units
+(clean  =            no) Symmetry clean before centering
+(rclean =            1.) Cleaning radius in scale units
+(rclip  =            2.) Clipping radius in scale units
+(kclean =            3.) K-sigma rejection criterion in skysigma
+(mkcente=            no) Mark the computed center
+(mode   =            ql)
+.YE
+
+.NH 4
+The Sky Fitting Algorithm Parameters
+
+.PP
+The \fBfitskypars\fR parameter set controls the sky fitting algorithm
+parameters used by the \fBphot\fR task. At this point DAOPHOT users should
+concern themselves with only three of these parameters: \fIsalgorithm\fR,
+\fIannulus\fR, and \fIdannulus\fR.
+.PP
+Users should realize that the \fBphot\fR task computes sky values
+for the individual stars, and that these values are
+used in the \fBpsf\fR task to compute the psf, averaged to form a group sky
+value in the \fBpeak\fR, \fBnstar\fR and \fBallstar\fR tasks if sky refitting
+is disabled (the default) or an initial sky value if sky refitting
+is enabled,  and used
+to compute the predicted signal-to-noise ratios in the \fBgroup\fR task.
+Although the option to refit the skies at a later stage of analysis exists,
+there are difficulties associated with this choice. It is
+in the user's best interest to determine the skies as accurately as
+possible as early as possible, since sky determination will probably be
+the single most important factor in doing good photometry.
+.PP
+In cases where contamination of the sky region is mostly due
+to crowding by neighboring stars users should use the default sky fitting
+algorithm "mode"; if the variations in the background are due instead
+to nebulosity or large contaminating objects so that the sky statistics are
+confused
+"median", "centroid", or "crosscor"  might be a better choice; in cases
+where the sky statistics
+are so poor that the histogram is aliased, undersampled, or sparse such
+as might be the case with
+very low sky backgrounds "mean" might be the best choice.
+When in doubt about the correct choice the user should leave \fIsalgorithm\fR
+at "mode" but examine the results carefully for accuracy at each step.
+.PP
+A good starting value for the inner radius of the sky annulus is ~ 4 *
+\fIfwhmpsf\fR
+or ~ 10 pixels for the test image. The width of the sky annulus should be
+sufficient to give a reasonable sample of sky pixels, >= 5 pixels. We have
+chosen a dannulus of ~4 * \fIfwhmpsf\fR or 10 pixels for the test image.
+.PP
+\fBFitskypars\fR can be edited from within the \fBdaoedit\fR task
+with the command \fB":epar fitskypars"\fR. After editing the \fBfitskypars\fR
+parameter set should look like the example below.
+
+.YS
+PACKAGE = daophot
+TASK = fitskypars
+
+(salgori=            mode) Sky fitting algorithm
+(annulus=             10.) Inner radius of sky annulus in scale units
+(dannulu=             10.) Width of sky annulus in scale units
+(skyvalu=              0.) User sky value
+(smaxite=              10) Maximum number of sky fitting iterations
+(sloclip=              0.) Lower clipping factor in percent
+(shiclip=              0.) Upper clipping factor in percent
+(snrejec=              50) Maximum number of sky fitting rejection iteratio
+(sloreje=              3.) Lower K-sigma rejection limit in sky sigma
+(shireje=              3.) Upper K-sigma rejection limit in sky sigma
+(khist  =              3.) Half width of histogram in sky sigma
+(binsize=             0.1) Binsize of histogram in sky sigma
+(smooth =              no) Boxcar smooth the histogram
+(rgrow  =              0.) Region growing radius in scale units
+(mksky  =              no) Mark sky annuli on the display
+(mode   =              ql)
+.YE
+
+.NH 4
+The Aperture Photometry Parameters
+
+.PP
+The \fBphotpars\fR parameter set controls the aperture photometry algorithm
+parameters used by the \fBphot\fR task. At this point DAOPHOT users should
+concern themselves with only one of these, \fIapertures\fR, the radius
+of the aperture through which the initial magnitudes will be computed.
+A good rule of thumb is to set the aperture radius to the maximum
+of 3 pixels or 1.0 * \fIfwhmpsf\fR pixels. Although magnitudes can be measured
+through more than one aperture at a time, it is the magnitude of the
+smallest aperture radius along with \fIzmag\fR and the exposure time
+which set the DAOPHOT instrumental magnitude scale, and
+the magnitudes through the other apertures contribute nothing to the
+DAOPHOT analysis until it comes time to compute accurate aperture
+corrections. Therefore it is in the user's best interest to set \fIapertures\fR
+to a single value at this point and carefully record it.
+.PP
+\fBPhotpars\fR can be edited from within the \fBdaoedit\fR task
+with the command \fB":epar photpars"\fR. After editing the \fBphotpars\fR
+parameter set should look like the example below. Note that in this example
+\fIfwhmpsf\fR is ~2.5 pixels so \fIapertures\fR is left at 3.0. 
+
+.YS
+PACKAGE = daophot
+TASK = photpars
+
+(weighti=         constant) Photometric weighting scheme
+(apertur=             3.0) List of aperture radii in scale units
+(zmag   =              25.) Zero point of magnitude scale
+(mkapert=               no) Draw apertures on the display
+(mode   =               ql)
+.YE
+
+.NH 4
+The Psf Modeling and Fitting Parameters
+
+.PP
+The \fBdaopars\fR parameter set controls the psf computation, star grouping,
+and psf fitting
+parameters used by the \fBpstselect\fR,  \fBpsf\fR, \fBpeak\fR,
+\fBgroup\fR, \fBnstar\fR,
+\fBallstar\fR, \fBsubstar\fR, and \fBaddstar\fR tasks. At this point
+DAOPHOT users should
+concern themselves with only two of these parameters \fIpsfrad\fR, the radius
+over which the psf will be defined, and \fIfitrad\fR, the radius 
+over which the psf will be fit to the individual stars. A good rule of thumb is
+to set \fIpsfrad\fR to the radius at which the radial profile of the brightest
+star of interest disappears into the noise plus 1, something like
+~ 4 * \fIfwhmpsf\fR + 1, and
+to set \fIfitrad\fR to the maximum of 3 pixels or ~ 1 * \fIfwhmpsf\fR in pixels.
+
+.PP
+\fBDaopars\fR can be edited from within the daoedit task
+with the command \fB":epar daopars"\fR. After editing the \fBdaopars\fR
+parameter set should look something like the example below for the
+test image.
+
+.YS
+PACKAGE = daophot
+TASK = daopars
+
+(functio=           gauss) Analytic component of psf
+(varorde=               0) Order of psf variation 
+(nclean =               0) Number of cleaning passes
+(saturat=              no) Use wings of saturated stars 
+(matchra=              3.) Matching radius in scale units
+(psfrad =             11.) Radius of psf in scale units
+(fitrad =              3.) Fitting radius in scale units
+(recente=             yes) Recenter stars during fit
+(fitsky =              no) Recompute group sky value during fit
+(sannulu=              0.) Inner radius of sky annulus in scale units
+(wsannul=             11.) Width of sky annulus in scale units
+(flaterr=            0.75) Flat field error in percent
+(proferr=              5.) Profile error in percent
+(maxiter=              50) Maximum number of iterations
+(clipexp=               6) Data clipping exponent
+(clipran=             2.5) Data clipping range in sigma
+(critove=              1.) Critical overlap group for membership
+(maxnsta=           10000) Maximum number of stars to fit
+(maxgrou=              60) Maximum number of stars to fit per group
+(mode   =              ql)
+.YE
+
+.NH 4
+Setting the Algorithm Parameters Graphically
+
+.PP
+Each of the radial distance dependent parameters \fIfwhmpsf\fR, 
+\fIcbox\fR, \fIannulus\fR, \fIdannulus\fR, \fIapertures,\fR,
+\fIpsfrad\fR, \fIfitrad\fR can be edited
+individually and interactively by marking the current radial profile
+plot with the
+graphics cursor after executing the appropriate keystroke command.
+For example the \fBf\fR keystroke command will prompt the user to
+mark the fwhm of
+the psf on the current radial profile plot,
+verify the marked value, and update the \fIfwhmpsf\fR parameter. 
+.PP
+All the radial distance dependent parameters listed above
+can be edited at once my moving the
+image cursor to a bright star, typing the \fB daoedit i\fR keystroke command
+to invoke the interactive graphics setup menu.
+The size of the radial profile plot and the sky regions
+are set by the \fIscale\fR, \fIannulus\fR, and \fIdannulus\fR parameters.
+The centering algorithm
+used is always "centroid" regardless of the value of the \fIcalgorithm\fR
+parameter, \fIcbox\fR and \fIscale\fR determine the centering box size,
+and the photometry is computed inside the largest aperture specified by
+the \fIapertures\fR parameter. After the user finishes marking all the
+parameters on the plot
+he/she is given an opportunity to verify or edit the results, e.g., change the
+value for fwhmpsf from 2.536 as read from the graphics cursor to 2.5. 
+
+.NH 3
+Checking the Algorithm Parameters with Daoedit
+
+.PP
+The purpose of setting all the critical algorithm parameters to reasonable
+values before beginning any DAOPHOT analysis, is to ensure that the user
+gets off to a good start. Although setting the parameters to unreasonable
+values often results in bizarre results which are immediately obvious,
+e.g., the detection of thousands of
+noise spikes, the problems can sometimes be more subtle.
+For example, a sky annulus that is too close to the star will result in
+measured sky values which are too high and poor subtractions of 
+the fitted stars which may not be discovered until the user has
+become thoroughly exasperated trying to produce good fits to the psf
+stars.
+.PP
+The current DAOPHOT algorithm parameters can be checked at any time with
+the \fBdaoedit\fR task and the \fB":lpar"\fR command. For example the
+\fBdatapars\fR parameters set can be listed with the \fBdaoedit
+":lpar datapars"\fR command. The remaining parameters sets \fBfindpars\fR,
+\fBcenterpars\fR, \fBfitskypars\fR, \fBphotpars\fR, and \fBdaopars\fR
+may be listed in the same way.
+.PP
+When listing the algorithm parameters users should check that:
+.IP [1]
+the \fBdatapars\fR image header keyword parameters \fIccdread\fR, \fIgain\fR,
+\fIexposure\fR, \fIairmass\fR, \fIfilter\fR, and \fIobstime\fR are 
+properly set.
+.IP [2]
+the \fBdatapars\fR \fIfwhmpsf\fR, \fIsigma\fR, \fIdatamin\fR, and \fIdatamax\fR parameters
+are appropriate for the image. Be especially careful of datamin as the
+correct value for this parameter varies with the mean sky.
+.IP [3]
+the \fBdatapars\fR parameter \fIscale\fR is 1.0 unless the user is
+thoroughly aware of the meaning of this parameter and the consequences
+of setting it to something other than 1.0, and \fIemission\fR is "yes".
+.IP [4]
+the \fBcenterpars\fR \fIcbox\fR parameter is
+appropriate for the image and the remaining \fBcenterpars\fR parameters
+are at their default values unless the user
+really understands the consequences of altering these parameters.
+.IP [5]
+the \fBfitskypars\fR  \fIannulus\fR, and \fIdannulus\fR
+parameters are appropriate for the image and the remaining \fBfitkskypars\fR
+parameters are at their default values unless the user really
+understands the consequences of altering these parameters.
+.IP [6]
+the \fBphotpars\fR \fIapertures\fR parameter is appropriate for the image
+and the remaining parameters are at their default values unless the user
+really understands the consequences of altering these parameters.
+.IP [6]
+the \fBdaopars\fR \fIpsfrad\fR and \fIfitrad\fR parameters are appropriate
+for the image and all the remaining \fBdaopars\fR parameters are at their
+default values unless the user really understands the consequences of
+altering these parameters.
+
+.NH 3
+Storing the Algorithm Parameter Values with Setimpars
+
+.PP
+The current values of all the algorithm parameters for a particular
+image may be saved in a file on disk at any
+point in the reduction sequence by executing the \fBsetimpars\fR task.
+The following command saves the current values of the parameters for
+the test image in a file called "test.pars".
+
+.YS
+da> setimpars test no yes
+.YE
+
+Repeating the previous command at any point in the reduction sequence will
+replace the stored parameter values with the current parameter values.
+
+
+.NH 3
+Restoring the Algorithm Parameter Values with Setimpars
+
+.PP
+At some point the user may wish to interrupt work on a particular image and
+begin work on a different image. This should be no problem as long as the
+user remembers to save the algorithm parameter sets  with \fBsetimpars\fR
+as described in the previous section.
+.PP
+The command to restore the algorithm parameter sets for the test image is:
+
+.YS
+da> setimpars test yes no
+.YE
+
+or
+
+.YS
+da> setimpars test yes no parfile=test.pars 
+.YE
+
+
+.NH 2
+Creating a Star List
+
+.PP
+The initial input to the DAOPHOT package is a star list.
+Star lists may be created with the DAOPHOT package task \fBdaofind\fR,
+interactively with the image or graphics cursor (the
+\fBrimcursor\fR and \fBrgcursor\fR tasks), by another IRAF task, or by
+any user program which writes a text file in the correct format.
+.PP
+Legal star lists are text files containing a list of stars, one star
+per line with the x and y coordinates in columns one and two.
+Blank lines, lines beginning with "#", and lines containing anything other
+than numbers in columns one and two are ignored.
+A sample DAOPHOT star list is shown below.
+
+.YS
+# Artificial Stars for Image Test
+
+	41.0   4.0  17.268
+	23.0   7.0  17.600
+	18.0   8.0  17.596
+	26.0  22.0  16.777
+	36.0  22.0  16.317
+ 	 8.0  23.0  16.631
+	31.0  25.0  16.990
+	21.0  26.0  19.462
+	29.0  34.0  17.606
+	36.0  42.0  16.544
+.YE
+
+.NH 3
+The Daofind Task
+
+.PP
+The \fBdaofind\fR task, searches for point sources 
+in an image whose peak intensities are above some user-defined threshold,
+computes approximate centers, magnitudes, and 
+shape characteristics for all the detected objects, and writes the results
+to the output star list.
+
+.NH 4
+The Daofind Algorithm
+
+.PP
+By default the \fBdaofind\fR algorithm performs the following steps:
+.IP[1]
+reads the \fBdaofind\fR task parameters, including the input image 
+and output star list names and the \fBdatapars\fR and \fBfindpars\fR
+algorithm parameters, and asks the user to verify the
+\fIfwhmpsf\fR, \fIsigma\fR,
+\fIthreshold\fR, \fIdatamin\fR, and \fIdatamax\fR parameters
+.IP[2]
+calculates the convolution kernel whose mathematical function when
+convolved with the input image is to
+compute the amplitude of the best-fitting Gaussian of
+full-width half-maximum \fIfwhmpsf\fR at each point in the input image
+.IP[3]
+convolves the input image with the convolution kernel after
+eliminating bad data with the \fIdatamin\fR and \fIdatamax\fR parameters,
+and writes the results to a temporary convolved image
+.IP[4]
+searches for local maxima in the convolved image whose amplitudes are greater
+than the detection threshold,  and greater than the amplitudes of any neighbors
+within a region the size of the convolution kernel
+.IP[5]
+computes approximate centers, magnitudes, and shape
+statistics for these local maxima
+.IP[6]
+eliminates local maxima whose centers are outside the image, and whose
+sharpness and roundness statistics are outside the limits set by the user
+.IP[7]
+writes the centers, approximate magnitudes, sharpness and roundness
+statistics, and id number for the remaining local maxima, to the
+output star list
+.IP[8]
+deletes the convolved image
+
+.NH 4
+The Daofind Algorithm Parameters
+
+.PP
+The critical \fBdaofind\fR algorithm parameters are \fIfwhmpsf\fR,
+\fIdatamin\fR, \fIdatamax\fR, \fIsigma\fR, and \fIthreshold\fR.
+These parameters are verified at startup time by \fBdaofind\fR.
+.PP
+The \fIfwhmpsf\fR parameter should be close
+to the true full-width at half-maximum of the psf in order
+to optimize the detection algorithm for stellar objects. If \fIfwhmpsf\fR
+is too far from the true value, stars may be omitted from the star list
+and/or non-stellar objects added to it.
+.PP
+The \fIdatamin\fR and \fIdatamax\fR parameters are used to flag and remove
+bad data from the convolved image.
+If \fIdatamin\fR and \fIdatamax\fR are
+too far from the true value stars may be omitted from the star list
+and/or non-stellar objects added to it.
+.PP
+The \fIsigma\fR parameter should be close to the true standard deviation of
+the sky background in an uncrowded region of the frame. This parameter
+in combination with \fIthreshold\fR
+determines the detection threshold in counts for faint objects. If it is
+incorrect either too few or too many objects will be detected.
+.PP
+The \fIthreshold\fR parameter should normally be set to some small
+number between 3.0 and 5.0.  If threshold is too big only
+the brightest stars will be detected. If threshold is too small too
+many noise spikes will be detected.
+
+.NH 4
+Running Daofind Non-Interactively
+
+.PP
+The following example shows how to run \fBdaofind\fR in non-interactive mode.
+
+.YS
+da> daofind test default 
+
+FWHM of features in scale units (2.5) (CR or value):
+        New FWHM of features: 2.5 scale units  2.5 pixels
+Standard deviation of background in counts (10.) (CR or value):
+        New standard deviation of background: 10. counts
+Detection threshold in sigma (4.) (CR or value):
+        New detection threshold: 4. sigma 40. counts
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Image: test.imh  fwhmpsf: 2.5  ratio: 1.  theta: 0.  nsigma: 1.5
+
+       40.97     4.02   -1.663   0.612   0.017     1
+       23.06     7.03   -1.214   0.636  -0.019     2
+       18.02     7.96   -1.318   0.622   0.010     3
+       25.99    22.01   -2.167   0.658   0.001     4
+       35.98    22.00   -2.499   0.572  -0.039     5
+        8.02    22.97   -2.239   0.550   0.068     6
+       30.97    25.01   -1.934   0.711  -0.044     7
+       28.96    33.92   -1.087   0.418   0.132     8
+       35.98    42.03   -2.332   0.639   0.108     9
+
+threshold: 40. relerr: 1.140  0.2 <= sharp <= 1.  -1. <= round <= 1.\fR
+.YE
+
+If this is the first time \fBdaofind\fR has been run the results will appear
+in the file "test.coo.1".
+.PP
+The detected objects can be marked on the image display using the
+\fBtvmark\fR task as shown below.
+
+.YS
+da> display test 1 fi+
+da> tvmark 1 test.coo.1 col=204
+.YE
+
+In this example the detected stars will be marked on the displayed image
+as red dots. If too many faints stars have been missed the user
+can rerun \fBdaofind\fR with a lower value of the \fIthreshold\fR
+parameter.
+
+.NH 4
+Running Daofind Interactively
+
+.PP
+\fBDaofind\fR may also be run in interactive mode.
+Most users will only exercise this option for small images which do not
+require long cpu/elapsed times to perform the convolution.
+.PP
+The following example shows how to run \fBdaofind\fR interactively.
+
+.YS
+da> display test 1 fi+
+
+da> daofind test default inter+
+.YE
+
+.IP ...
+Type the \fBv\fR keystroke command to verify the critical algorithm
+parameters.
+.LP
+
+.YS
+FWHM of features in scale units (2.5) (CR or value):
+        New FWHM of features: 2.5 scale units  2.5 pixels
+Standard deviation of background in counts (10.) (CR or value):
+        New standard deviation of background: 10. counts
+Detection threshold in sigma (4.) (CR or value):
+        New detection threshold: 4. sigma 40. counts
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+.YE
+
+.IP ...
+Type the \fBspacebar\fR keystroke command to detect the objects and write them
+out to the star list file.
+.LP
+
+.YS
+Image: test.imh  fwhmpsf: 2.5  ratio:  1.  theta:  0.  nsigma: 1.5
+
+       40.97     4.02   -1.663   0.612   0.017     1
+       23.06     7.03   -1.214   0.636  -0.019     2
+       18.02     7.96   -1.318   0.622   0.010     3
+       25.99    22.01   -2.167   0.658   0.001     4
+       35.98    22.00   -2.499   0.572  -0.039     5
+        8.02    22.97   -2.239   0.550   0.068     6
+       30.97    25.01   -1.934   0.711  -0.044     7
+       28.96    33.92   -1.087   0.418   0.132     8
+       35.98    42.03   -2.332   0.639   0.108     9
+
+threshold: 40. relerr: 1.140  0.2 <= sharp <= 1.  -1. <= round <= 1.
+
+Output file: test.coo.1
+.YE
+
+.IP ...
+Change \fIthreshold\fR to 3.0 with the colon command \fB:threshold 3.0\fR.
+.IP ...
+Type the \fBspacebar\fR keystroke command to detect the objects and write
+them out to a new star list file.
+.LP
+
+.YS
+Image: test.imh  fwhmpsf: 2.5  ratio:  1.  theta:  0.  nsigma: 1.5
+
+      40.97     4.02   -1.975   0.577   0.017     1
+      23.06     7.03   -1.526   0.604  -0.019     2
+      18.02     7.96   -1.631   0.587   0.010     3
+      25.99    22.01   -2.480   0.626   0.001     4
+      35.98    22.00   -2.811   0.537  -0.039     5
+       8.02    22.97   -2.551   0.515   0.068     6
+      30.97    25.01   -2.246   0.681  -0.044     7
+      21.27    25.94   -0.146   0.804  -0.558     8
+      28.96    33.92   -1.400   0.379   0.132     9
+      35.98    42.03   -2.645   0.606   0.108    10
+
+threshold: 30. relerr: 1.140  0.2 <= sharp <= 1.  -1. <= round <= 1.
+
+Output file: test.coo.2
+.YE
+
+.IP ...
+Change \fIthreshold\fR to 5.0 with the colon command \fB:threshold 5.0.
+.IP ...
+Type the \fBspacebar\fR keystroke command to detect the objects and write them
+out to a new coordinate file.
+.LP
+
+.YS
+Image: test.imh  fwhmpsf: 2.5  ratio:  1.  theta:  0.  nsigma: 1.5
+
+      40.97     4.02   -1.420   0.577   0.017     1
+      23.06     7.03   -0.972   0.604  -0.019     2
+      18.02     7.96   -1.076   0.587   0.010     3
+      25.99    22.01   -1.925   0.626   0.001     4
+      35.98    22.00   -2.257   0.537  -0.039     5
+       8.02    22.97   -1.997   0.515   0.068     6
+      30.97    25.01   -1.692   0.681  -0.044     7
+      28.96    33.92   -0.845   0.379   0.132     8
+      35.98    42.03   -2.090   0.606   0.108     9
+
+threshold: 50. relerr: 1.140  0.2 <= sharp <= 1.  -1. <= round <= 1.
+
+Output file: test.coo.3
+.YE
+
+.IP ...
+Type the \fBq\fR keystroke, first in the image display window then the
+text window to quit the task.
+.LP
+
+If this is the first run of \fBdaofind\fR,
+the three star list files for the \fIthreshold\fR values of
+4.0, 3.0, and 5.0 will be written to "test.coo.1", "test.coo.2", and 
+"test.coo.3" respectively.
+.PP
+The \fBdaofind\fR results for different thresholds can be evaluated by
+marking the detected objects on the image display using the \fBtvmark\fR
+task and different
+colors for each threshold. In the following example objects detected
+at threshold=3.0 are marked in red, at threshold=4.0 in green, at threshold=
+5.0 in blue.
+
+.YS
+da> display test 1 fi+
+da> tvmark 1 test.coo.2 col=204 point=3
+da> tvmark 1 test.coo.1 col=205 point=3
+da> tvmark 1 test.coo.3 col=206 point=3
+.YE
+
+Note that the identical stars were detected at thresholds 4.0 and 5.0 but
+the faint star at 21,26 was only detected at threshold=3.0. 
+.PP
+In this example the user decides that threshold = 4.0 is the
+"best" threshold, sets the \fIthreshold\fR parameter appropriately as shown
+below, and deletes the the star lists for threshold = 3.0 and threshold = 5.0.
+
+.YS
+da> findpars.threshold = 4.0
+da> delete test.coo.2,test.coo.3
+.YE
+
+.NH 4
+The Daofind Output
+
+.PP
+The quantities xcenter, ycenter, mag, sharpness, roundness, and id
+are recorded for each detected object.  Each is described briefly below.
+.IP [1]
+\fIXcenter\fR and \fIycenter\fR are the coordinates of
+the detected object in fractional pixel units. They are computed by
+fitting one-dimensional Gaussian functions  of full-width at half-maximum
+\fIfwhmpsf\fR to the x and y marginal pixel distributions centered on
+the star. The computed coordinates can be overlaid on the
+displayed image with the \fBtvmark\fR command.
+.IP [2]
+The estimated magnitude is measured relative to the detection threshold
+and is defined as
+
+.YS
+mag = -2.5 * log10 (density / (relerr * threshold * sigma))
+.YE
+
+where density is the peak density of the object in the convolved image,
+relerr an internally
+computed factor measuring the amount by which the standard error in one pixel
+in the input image must be multiplied to obtain the standard error
+in one pixel in the convolved image, and threshold and sigma
+are the values of the corresponding \fIthreshold\fR and \fIsigma\fR
+parameters. For stellar
+objects the computed magnitude is directly proportional to the true
+magnitude of the star. Stars with a peak density exactly equal to
+the detection threshold will have a magnitude of 0.0. The remaining
+stars will have negative magnitudes.
+.IP [3]
+The sharpness statistic is the ratio of the amplitude of the best fitting
+delta function at the position of a detected object to the amplitude of
+the best fitting gaussian at the same position as shown below.
+
+.YS
+sharpness = (data - <data>) / density
+.YE
+
+The amplitude of the best fitting gaussian is simply the density
+of the detected object in the convolved image.
+The amplitude of the best fitting delta function is defined
+as corresponding original image data value minus the average of all
+the neighboring
+pixels in the image <data>. Typical values of sharpness are of ~0.6 for
+approximately gaussian stars and \fInsigma\fR = 1.5.
+Hot pixels will have sharpness values >> 1 and cold pixels will have
+sharpness values
+of ~0, hence reasonable limits for the \fIsharphi\fR and \fIsharplo\fR
+parameters are 1.0 and 0.2 respectively.
+Increasing the size of convolution box defined by the
+\fInsigma\fR parameter from its default value of 1.5 to a larger value
+(smaller values should be avoided !) while keeping the \fIfwhmpsf\fR the
+same, will increase the average value of the sharpness statistic because
+more pixels further from the center of the star are included in the
+computation of <data>. If \fInsigma\fR is changed
+the \fBfindpars\fR parameters \fIsharphi\fR and \fIsharplo\fR
+will also need to be changed.
+.IP [4]
+The roundness statistic is computed by fitting a one-dimensional
+gaussian function of full-width at half-maximum \fIfwhmpsf\fR to the
+x and y marginal pixel distributions.
+
+.YS
+roundness =  2.0 * (hx - hy) / (hx + hy)
+.YE
+
+hx and hy are the heights of the best fitting one-dimensional
+gaussians in x and y. A totally
+round object will have a roundness of ~ 0.0. If the object is very
+elongated in x roundness will be a large negative number; a large positive
+number if it is elongated in y.  The roundness statistic is 
+effective at filtering out bad columns and rows of data. It is not
+effective at filtering out objects elongated at intermediate angles.
+.IP [5]
+Id is a sequence number which identifies the star.
+
+.NH 4
+Examining the Daofind Output
+
+.PP
+The easiest way to check that \fBdaofind\fR is performing correctly is to mark
+the detected stars on the image display with \fBtvmark\fR.
+.PP
+If the marked image suggests that \fBdaofind\fR is detecting too few or
+too many stars the first items to check are the the values of the
+\fIsigma\fR and \fIthreshold\fR parameters since these parameters determine the
+detection threshold. Sigma should be
+the standard deviation of the sky pixels in an uncrowded region of the image.
+Threshold should normally be some number between 3.0 and 5.0. If sigma
+and threshold are reasonable the user should compare the observed value of
+sigma with the predicted value derived from the median background level and the
+effective gain and readout noise values. If there is a significant
+mismatch in these numbers the user should check the reduction history of the
+image.  The number of spurious detections goes up dramatically for
+thresholds less than ~3.0 * sigma. A  plot of number
+of detections versus threshold will show a change in slope at some point below
+this threshold.  Users who wish to detect faint objects while keeping
+spurious detections at a manageable minimum should
+set the detection threshold to a value just above the threshold at
+which this change in slope occurs.
+.PP
+Users should also check the values of the
+parameters \fIsharplo\fR, \fIsharphi\fR,
+\fIroundlo\fR, and \fIroundhi\fR parameters to ensure that
+detected objects are not being unfairly filtered out. In particular the values
+of \fIsharplo\fR and \fIsharphi\fR should be changed if the \fInsigma\fR
+parameter is changed.
+.PP
+Finally the user should check
+the \fIfwhmpsf\fR, \fInsigma\fR, \fIdatamin\fR and \fIdatamax\fR parameters
+for correctness since these parameters  control the computation of the
+convolution kernel and the density enhancement image.
+.PP
+Histograms of the various columns in the \fBdaofind\fR output can be
+plotted using the \fBpdump\fR and \fBphistogram\fR tasks. The following
+example shows how to plot a histogram of the magnitudes.
+
+.YS
+da> pdump test.coo.1 mag yes | phistogram STDIN binwidth=.1
+.YE
+
+The various columns can also be plotted against each other.
+The following example shows how to plot magnitude error versus magnitude.
+
+.YS
+da> pdump test.coo.1 mag,merr yes | graph point+
+.YE
+
+.PP
+By setting the \fBdaofind\fR \fIstarmap\fR and \fIskymap\fR parameters the
+user can save and examine the density enhancement image and the corresponding
+background density image.  The sum of these two images should yield a
+close representation of the original image except for regions of
+bad data and edge pixels. Due to the nature of the convolution kernel
+the starmap image will have a mean value of
+~0.0 in the sky regions,  an rms \(~= relerr * sigma in the sky regions,
+and positive peaks of intensity surrounded by negative valleys at the positions
+of bright stars. The skymap image will have a mean value \(~= sky in 
+the sky regions, an rms \(~= sqrt (sigma ** 2 / N + K * (relerr * sigma) ** 2),
+(N is the number of pixels in the gaussian kernel and K is the average power
+in the gaussian kernel), and dips in intensity surrounded by
+bright rings at the position of the stars. 
+
+.NH 3
+Rgcursor and Rimcursor
+
+.PP
+The LISTS package tasks \fBrimcursor\fR and \fBrgcursor\fR can be used to
+generate coordinate lists interactively. For example a coordinate
+list can be created using the image display and the image display cursor
+as shown below.
+
+.YS
+da> display test 1 fi+
+
+da> rimcursor > test.coo
+.YE
+
+.IP ...
+Move cursor to stars of interest and tap the space bar.
+.IP ...
+Type <EOF> to terminate the list.
+
+.PP
+A coordinate list can also be created using a contour plot and the graphics
+cursor as shown below.
+
+.YS
+da> contour test
+
+da> rgcursor > test.coo
+.YE
+
+.IP ...
+Move the cursor to the stars of interest and tap the space bar.
+.IP ...
+Type <EOF> to terminate the list.
+
+.PP
+In both cases the text file "test.coo" contains the x and y coordinates of
+the marked stars in image pixel units. The output of \fBrimcursor\fR or
+\fBrgcursor\fR can
+be read directly by the DAOPHOT \fBphot\fR task. 
+
+.NH 3
+User Program
+
+.PP
+Any user program which produces a text file with the stellar coordinates
+listed one per line with x and y in columns 1 and 2, can be used to produce
+DAOPHOT coordinate files which can be read by the \fBphot\fR task.
+
+.NH 3
+Modifying an Existing Coordinate List
+
+.PP
+The LISTS package routine \fBlintran\fR 
+can be used to perform simple coordinate transformations on
+coordinate lists including shifts, magnifications, and rotations.
+
+.NH 2
+Initializing the Photometry with Phot
+
+.PP
+The \fBphot\fR task computes initial centers,
+sky values, and initial magnitudes for all the objects in the input
+star list. The centers and magnitudes are used as starting
+values for the non-linear
+least-squares psf computation and fitting routines in the \fBpsf\fR,
+\fBpeak\fR, \fBnstar\fR, and \fBallstar\fR tasks, and to estimate
+signal-to-noise
+values in the \fBgroup\fR task. The individual sky values
+computed by \fBphot\fR are used directly by the \fBpsf\fR task to compute
+the psf model, by the
+\fBpeak\fR, \fBnstar\fR,
+and \fBallstar\fR tasks to compute the group sky values, and by the
+\fBgroup\fR task to estimate signal-to-noise ratios.
+
+.NH 3
+The Phot Algorithm
+
+.PP
+By default the \fBphot\fR task performs the following functions:
+.IP [1]
+reads in the \fBphot\fR task parameters including the input image name,
+the input coordinate file name, the output photometry file name,
+the \fBdatapars\fR, \fBcenterpars\fR, \fBfitskypars\fR, and \fBphotpars\fR
+algorithm parameters, and determines whether the task mode of operation
+is interactive or non-interactive
+.IP [2]
+reads in the initial coordinates of a star from the coordinate list and/or
+the image cursor, and computes new coordinates for the star using the centering
+algorithm defined by the \fIcalgorithm\fR parameter (if \fIcalgorithm\fR
+is not "none") using data in a box whose size is defined by the \fIcbox\fR
+parameter
+.IP [3]
+computes the sky value for the star using the default algorithm
+specified by the \fIsalgorithm\fR parameter and the data in an annulus of
+pixels defined by the \fIannulus\fR and \fIdannulus\fR parameters
+.IP [4]
+computes the instrumental magnitude  and magnitude error for each star
+inside the aperture radii
+specified by the \fIapertures\fR parameter using fractional pixel techniques,
+the computed sky value, the standard deviation of the sky pixels, and the
+gain of the CCD
+.IP [6]
+sets the instrumental magnitude scale for the image using the
+\fBphotpars\fR \fIzmag\fR parameter and the exposure time specified by
+the \fBdatapars\fR \fIexposure\fR or \fIitime\fR parameters
+.IP [7]
+sets the magnitude(s) to INDEF for stars which are saturated or contain
+bad data,
+for which the aperture is partially off the image, for which a sky
+value could not be computed, or for which the
+signal is fainter than the background
+.IP [8]
+writes the results to the output photometry file
+
+.NH 3
+The Phot Algorithm Parameters
+
+.PP
+The critical \fBphot\fR algorithm parameters are \fIcalgorithm\fR,
+\fIsalgorithm\fR,
+\fIannulus\fR, \fIdannulus\fR, \fIapertures\fR, \fIdatamin\fR and
+\fIdatamax\fR. These parameters are verified by \fBphot\fR
+at startup time.
+.PP
+The \fIcalgorithm\fR parameter tells \fBphot\fR how to compute
+centers for the objects in the coordinate list. Calgorithm should be "none"
+if the coordinate
+list was computed by \fBdaofind\fR or the coordinates are known to
+be precise; otherwise calgorithm should one of "centroid", "gauss", or
+"ofilter". "centroid" is quick and sufficiently accurate in most cases;
+"gauss" and "ofilter" take longer but are more accurate. If calgorithm is
+not "none",
+\fBphot\fR will ask the user to verify the centering box size
+\fIcbox\fR. \fIcbox\fR should be set to 5 or ~2 * \fIfwhmpsf\fR pixels wide
+whichever is greater.
+.PP
+The \fIsalgorithm\fR parameter tells \fBphot\fR how to compute the sky
+values. If the
+fluctuations in the sky background are due primarily to crowding the
+default choice
+"mode" should be used. If the fluctuations in the sky background
+are due to nebulosity or large galaxies
+and the sky statistics are confused, "median", "centroid" or
+"crosscor" might be the best choice. In cases where the
+background is very low and the sky histogram is sparse or
+undersampled "mean" might be the best choice.
+.PP
+The \fIannulus\fR and \fIdannulus\fR parameters tell \fBphot\fR the
+position of the sky annulus with respect to the star. The sky region
+must be far enough away from the star to avoid contamination from
+the star itself, but close enough
+to be representative of the intensity distribution under the star. Values of
+~ 4.0 * \fIfwhmpsf\fR for both parameters are good starting values.
+.PP
+The \fIapertures\fR parameter tells \fBphot\fR the radius of the
+photometry aperture. The photometry through this aperture sets the
+instrumental magnitude scale for all the subsequent DAOPHOT
+reductions.  \fIApertures\fR should be ~ 1.0 * \fIfwhmpsf\fR.
+.PP
+The \fIdatamin\fR and \fIdatamax\fR parameters are used to detect
+bad data in the photometry and sky apertures. Bad data is removed from
+the sky pixel list before sky fitting takes place so it is important
+that datamax and datamin, but particularly datamin, be correct.
+Stars which have bad data in the photometry apertures will have their
+magnitudes set to INDEF and be flagged with an error.
+
+.NH 3
+Running Phot Non-interactively
+
+.PP
+The following example shows how to run \fBphot\fR in non-interactive
+mode using the results of \fBdaofind\fR as input.
+
+.YS
+da> phot test default default
+
+Centering algorithm (none) (CR or value):
+        New centering algorithm: none
+Sky fitting algorithm (mode) (CR or value):
+        Sky fitting algorithm: mode
+Inner radius of sky annulus in scale units (10.) (CR or value):
+        New inner radius of sky annulus: 10. scale units 10. pixels
+Width of the sky annulus in scale units (10.) (CR or value):
+        New width of the sky annulus: 10. scale units 10. pixels
+File/list of aperture radii in scale units (3.0) (CR or value): 3.0
+        Aperture radius 1: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+test     40.97     4.02  100.7955   17.218  ok
+test     23.06     7.03  100.3257   17.650  ok
+test     18.02     7.96  99.40262   17.484  ok
+test     25.99    22.01  101.3196   16.800  ok
+test     35.98    22.00  101.1601   16.373  ok
+test      8.02    22.97  98.89139   16.603  ok
+test     30.97    25.01  101.2904   17.051  ok
+test     28.96    33.92  100.6189   17.782  ok
+test     35.98    42.03   101.043   16.594  ok
+
+.YE
+
+\fBPhot\fR looks for an input star list called "test.coo.?",
+creates a file called "test.mag.?", and verifies
+the critical parameters. By default the verbose switch is set to "yes",
+so a short summary of the results for each star is printed on the
+terminal as it is computed.
+.PP
+\fBPhot\fR may also be run non-interactively from a coordinate list created
+with the image cursor list task \fBrimcursor\fR as shown below. Note
+that centering has been turned on, and the verify switch has been turned off.
+
+.YS
+da> display test 1 fi+
+
+da> rimcursor > cursor.coo
+
+da> page cursor.coo
+
+41.02   4.033 101 \040
+22.918  6.969 101 \040
+18.123  7.849 101 \040
+25.951 21.939 101 \040
+35.736 21.744 101 \040
+ 7.947 23.016 101 \040
+30.843 24.777 101 \040
+28.984 33.779 101 \040
+36.127 41.705 101 \040
+
+da> phot test cursor.coo default calg=centroid verify- 
+
+test     40.92     4.04  100.8871   17.222  ok
+test     23.17     6.97  100.6163   17.666  ok
+test     18.04     7.92  99.55305   17.487  ok
+test     25.96    21.97  101.4161   16.801  ok
+test     35.94    21.98  101.2101   16.373  ok
+test      8.05    23.00  98.74371   16.601  ok
+test     30.94    25.02  101.3224   17.052  ok
+test     28.91    33.85  100.6207   17.786  ok
+test     35.96    42.08  100.9039   16.591  ok\fR
+.YE
+
+The "centroid" algorithm
+computes a new center by doing an intensity-weighted sum of the
+x and y marginals, whereas the \fBdaofind\fR algorithm
+fits a 1D gaussian to the marginal pixel distributions in x and y.
+The following example shows the results for the almost equivalent
+\fBphot\fR centering algorithm "gauss".
+
+.YS
+da> phot test cursor.coo default calg=gauss verify-
+
+test     41.00     4.03  100.8698   17.219  ok
+test     23.11     7.03  100.3567   17.653  ok
+test     18.02     7.96  99.40262   17.484  ok
+test     25.98    21.97  101.4021   16.801  ok
+test     35.96    21.99  101.2101   16.373  ok
+test      8.02    22.98  98.77907   16.601  ok
+test     30.97    25.02  101.2904   17.051  ok
+test     28.93    33.94  100.6726   17.783  ok
+test     35.97    42.02   100.976   16.593  ok\fR
+.YE
+
+The positions produced by the "gauss" algorithm are closer to
+the positions computed by the \fBdaofind\fR task, 
+than those computed by the "centroid" algorithm.
+However as the positions computed by \fBphot\fR are used as initial
+positions by the DAOPHOT tasks,
+it is usually not necessary to go to the more expensive "gauss" algorithm.
+
+.NH 3
+Running Phot Interactively
+
+.PP
+\fBPhot\fR can also be configured to run interactively using the
+image display and image
+cursor for coordinate input. In this mode the user loads the image
+into the display and runs
+\fBphot\fR interactively by turning the interactive switch on as
+shown below.
+When the program is ready to accept input the cursor will begin
+blinking in the display window. The following series of steps will
+do photometry on stars selected with the image cursor.
+
+.YS
+da> display test 1 fi+
+
+da> phot test "" default interactive+ calgorithm=centroid
+.YE
+
+.IP ...
+Execute the \fBv\fR keystroke command to verify the critical parameters.
+.LP
+
+.YS
+Centering algorithm (centroid) (CR or value):
+        New centering algorithm: centroid
+Centering box width in scale units (5.) (CR or value):
+        New centering box width: 5. scale units  5. pixels
+Sky fitting algorithm (mode) (CR or value):
+        Sky fitting algorithm: mode
+Inner radius of sky annulus in scale units (10.) (CR or value):
+        New inner radius of sky annulus: 10. scale units 10. pixels
+Width of the sky annulus in scale units (10.) (CR or value):
+        New width of the sky annulus: 10. scale units 10. pixels
+File/list of aperture radii in scale units (3.) (CR or value):
+        Aperture radius 1: 3. scale units 3. pixels
+Standard deviation of background in counts (10.) (CR or value):
+     	New standard deviation of background: 10. counts
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+.YE
+
+.IP ...
+Move the cursor to the stars of interest and tap the \fBspacebar\fR to do
+the photometry.
+.LP
+
+.YS
+test     40.92     4.04  100.8871   17.222  ok
+test     23.17     6.97  100.6163   17.666  ok
+test     18.04     7.92  99.55305   17.487  ok
+test     25.96    21.97  101.4161   16.801  ok
+test     35.94    21.98  101.2101   16.373  ok
+test      8.05    23.00  98.74371   16.601  ok
+test     30.94    25.02  101.3224   17.052  ok
+test     28.91    33.85  100.6207   17.786  ok
+test     35.96    42.08  100.9039   16.591  ok
+.YE
+
+.IP ...
+Type \fBq\fR to quit image, and \fBq\fR again to exit task.
+.LP
+
+The coordinate file name has been set to "" so that initial
+positions for the stars to be measured will be read from the image
+cursor, and the centering algorithm has been temporarily changed
+on the command line
+from "none" to "centroid" so that new centers will be computed.
+The user simply points the cursor to the stars to be measured and taps the
+space bar to measure the star. This option is often useful for
+picking up stars missed by \fBdaofind\fR in a previous iteration,
+or in cases where the user only wishes to measure a small group of stars.
+
+.NH 3
+The Phot Output
+
+.PP
+\fBPhot\fR produces a large output file containing many parameters, 
+intermediate and final results. The principle
+quantities of interest are: 1) the position
+of the star xcenter and ycenter, 2) the sky value,
+its standard deviation, and the number of pixels used to compute
+it, msky, stdev, and nsky 3) the total
+counts inside the aperture and the effective area of the aperture,
+sum and area 4) the magnitude and magnitude error in the 
+aperture, mag and merr, 
+and 5) the exposure time, airmass, filter, and time of observation, itime,
+xairmass, ifilter, and otime.
+.IP [1]
+\fIXcenter\fR and \fIycenter\fR are the computed coordinates for
+the detected objects in fractional pixels. They can be overlaid on the
+displayed image with the \fBtvmark\fR command. These numbers should be compared
+with the initial coordinates xinit and yinit, to which they will be equal
+if the centering algorithm was "none", or to which they should be close
+if the centering algorithm is "centroid", "gauss", or "ofilter" assuming
+that the original x and y positions were reasonable.
+.IP [2]
+\fIMsky\fR, \fIstdev\fR and \fInsky\fR are the estimated sky value in counts,
+its standard deviation in counts, and the number of pixels used to compute it.
+Users should, check that the position of the sky annulus is reasonable
+and, check that the msky, stdev, and nsky values are reasonable
+for a few isolated stars before proceeding. 
+.IP [3]
+\fISum\fR and \fIarea\fR are the total counts (star + sky) in the 
+photometry aperture and area is area of the aperture in pixels squared
+and should be roughly
+equal to PI * r ** 2 where r is the radius of the photometry aperture
+in pixels.
+.IP [4]
+\fIMag\fR and \fImerr\fR are the magnitude and magnitude error respectively
+computed as follows.
+
+.YS
+ mag = zmag - 2.5 * log10(sum - area * msky) + 2.5 * log10(itime)
+
+merr = 1.0857 * sqrt((sum - area * msky) / gain + area * stdev ** 2
+       + area ** 2 * stdev ** 2 / nsky) / (sum - area * msky)
+.YE
+
+Users should check that the exposure time \fIitime\fR is correct since it
+is used to determine the instrumental magnitude scale. The correct value
+of gain is also required in order to get a correct estimate of the
+magnitude error. Stdev is the observed standard deviation
+of the sky pixels not the predicted value.
+.IP [5]
+The remaining quantities itime, \fIxairmass\fR, \fIifilter\fR, and \fIotime\fR
+should be checked for correctness, e.g., were they read correctly from the image
+header.
+
+.NH 3
+Examining the Results of Phot
+
+.PP
+The user can check the results of \fBphot\fR in several 
+ways.  The following command will mark  all
+stars in the \fBphot\fR output file on the display in red.
+
+.YS
+da> display test 1
+da> pdump test.mag.1 xcenter,ycenter yes | tvmark 1 STDIN col=204
+.YE
+
+The following command will mark all the stars whose magnitudes are INDEF on
+the screen in green.
+
+.YS
+da> pdump test.mag.1 xcenter,ycenter "mag == INDEF" | tvmark \\ 
+    1 STDIN col=205
+.YE
+
+The following command will plot magnitude error versus magnitude for all
+the stars in the photometry file.
+
+.YS
+da> pdump test.mag.1 mag,merr yes | graph STDIN point+
+.YE
+
+The following command will plot a histogram of the magnitude distribution.
+
+.YS
+da> pdump test.mag.1 mag,merr yes | phist STDIN plot_type=box
+.YE
+
+The photometry file can be examined interactively with the \fBpexamine\fR
+task as shown below.
+
+.YS
+da> pexamine test.mag.1 "" test
+.YE
+.IP ...
+A vector plot of magnitude error versus magnitude appears on the screen.
+.IP ...
+To examine individual stars in the vector plot move the graphics cursor to a
+star and  type \fBo\fR to get a record listing for the star,
+followed by \fBr\fR, \fBc\fR, or \fBs\fR to see a radial profile plot,
+contour plot, or surface plot respectively, of the star.
+.IP ...
+To activate the image cursor type \fBi\fR, move the cursor to a star
+and type \fBo\fR to get a record listing
+for the star, followed by \fBr\fR, \fBc\fR or \fBs\fR to draw the desired plot.
+To reactivate the graphics cursor type \fBg\fR.
+.IP ...
+To plot magnitude error versus x coordinate for all the stars in the file, type
+\fB:xcolumn xcenter\fR and \fB:ycolumn merr\fR followed by \fBp\fR to
+redraw the plot.
+.IP ...
+To plot a histogram of the magnitudes of the objects type \fBh\fR.
+.IP ...
+Type \fBq\fR to quit.
+
+.NH 2
+Creating a Psf Star List with Pstselect
+
+.PP
+The psf model fitting routines require a list of bright isolated stars
+well distributed over the image to use as psf model templates.
+The \fBpstselect\fR 
+task is used to select suitable candidate stars from the photometry file
+for input to the psf modeling task \fBpsf\fR.
+
+.NH 3
+The Pstselect Algorithm
+
+.PP
+By default the \fBpstselect\fR task performs the following functions: 
+.IP[1]
+reads the task parameters including the input image name, input 
+photometry file, and output psf star list, reads the \fBdatapars\fR
+and \fBdaopars\fR algorithm parameters, and determines whether the
+task will be run interactively or non-interactively
+.IP [2]
+reads the dimensions of the input image from the input image header, and 
+the ids, x and y coordinates, magnitudes, and sky values of up to
+\fImaxnstar\fR stars from the input photometry file
+.IP [2]
+assigns a large negative number to the magnitudes of all stars whose
+measured magnitudes are INDEF in the input photometry file
+.IP [3]
+sorts the stars in order of increasing magnitude so that
+the saturated and brightest stars are at the beginning of the list
+.IP [4]
+selects the
+brightest \fImaxnpsf\fR stars (where maxnpsf is a number chosen by the user)
+which are, not saturated, more than \fIfitrad\fR pixels away from the edge
+of the input image, have no bad data within \fIfitrad\fR pixels,
+and have no brighter neighbor stars within
+(\fIpsfrad\fR + \fIfitrad\fR + 2) pixels
+.IP [5]
+writes the ids, x and y coordinates, magnitudes, and sky values 
+of the selected stars as read from the input photometry list
+to the output psf star list
+
+.NH 3
+The Pstselect Algorithm Parameters
+
+.PP
+The critical \fBpstselect\fR algorithm parameters are \fIpsfrad\fR,
+\fIfitrad\fR, \fIdatamin\fR, and \fIdatamax\fR.
+.PP
+\fIPsfrad\fR and \fIfitrad\fR are used by \fBpstselect\fR to eliminate
+potential psf stars which have bright neighbors.
+For the test image these parameters are currently set
+to 4 * \fIfwhmpsf\fR + 1 and 1 * \fIfwhmpsf\fR or 11 and 3 pixels respectively.
+However as \fBpstselect\fR is the first task to actually use the values
+of these parameters, the user should
+check them here one more time before running \fBpstselect\fR.
+\fIFitrad\fR should be ~ 1 * \fIfwhmpsf\fR
+for optimal psf model computation and fitting so the user leaves it
+at its current value of 3.0. \fIPsfrad\fR should be set to the radius at which
+the profile of the brightest stars of interest disappear into the
+noise. Normally 4 * \fIfwhmpsf\fR + 1 pixels is a good starting value for
+this quantity.
+If \fIpsfrad\fR is too small the fitted stars will not subtract completely
+from the input image, if it is too big DAOPHOT
+will consume cpu time doing unnecessary data extractions
+and subtractions.
+One way to check the value of the \fIpsfrad\fR parameter is to
+use the \fBdaoedit\fR task to examine
+radial profiles of isolated stars in the input image as shown below.
+
+.YS
+da> display test 1 fi+
+
+da> daoedit test
+.YE
+	    
+.IP ...
+Move cursor to star at 36,42 and press the \fBr\fR key.
+.IP ...
+Examine the resulting radial profile and note that the stellar profile
+disappears into the noise at ~4 pixels.
+.IP ...
+Move the cursor to the star at 8,23 and press the \fBr\fR key.
+.IP ...
+Examine the radial profile and note that this stellar profile
+also disappears into the noise at ~4 pixels.
+.IP ...
+Set \fIpsfrad\fR to 5.0 pixels by typing the command \fB:psfrad 5.0\fR.
+.IP ...
+Type \fBq\fR to quit the \fBdaoedit\fR task.
+.YE
+
+.PP
+The \fBpexamine\fR task and the input photometry file can also be
+used to examine the radial profiles of isolated stars in the
+photometry file. 
+
+.YS
+da> display test 1 fi+
+
+da> pexamine test.mag.1 "" test
+.YE
+
+.IP ...
+A plot of magnitude error versus magnitude appears on the screen.
+.IP ...
+Type \fBi\fR to activate the image cursor.
+.IP ...
+Move the cursor to the star at 36,42 and type \fBr\fR, adjust the
+outer radius of the plot with the command \fB:router\fR if necessary,
+e.g., \fB:router 10\fR.
+.IP ...
+Examine the radial profile and note that it disappears into the noise at a
+radius of ~4 pixels.
+.IP ...
+Move the cursor to the star at 8,23 and type \fBr\fR.
+.IP ...
+Examine the radial profile and note that that it
+also disappears into the noise at a radius of ~4 pixels.
+.IP ...
+Type \fBq\fR to quit the pexamine task.
+.LP
+
+The new value of \fIpsfrad\fR can be stored by editing the \fBdaopars\fR
+parameter set with \fBepar\fR in the usual manner or on the command
+line as shown below.
+
+.YS
+da> daopars.psfrad = 5.0
+.YE
+
+.PP
+Why is the value of 5.0 pixels for \fBpsfrad\fR so different from the
+original estimate of 11.0 ?
+There are two reasons. Firstly the stars in artificial image
+test are quite faint,
+with the brightest peaking at ~400 counts above background. Their stellar
+profiles disappear into the noise quite quickly. Secondly the artificial
+stars are gaussian in shape with a  sigma \(~= 1.0 pixels.
+Unlike real stars they have almost all their light
+in the core and none in the wings. For realistic optical images
+11.0 pixels rather than 5.0 would be a more reasonable choice
+for \fIpsfrad\fR than 5.0.
+.PP
+The \fIdatamin\fR and \fIdatamax\fR parameters are used to reject
+psf stars with bad data within \fIfitrad\fR pixels.
+If \fIdatamin\fR and \fIdatamax\fR  are set correctly before
+the \fBphot\fR task is run, these parameters are redundant as stars
+with bad data inside the photometry aperture will have INDEF magnitudes.
+.PP
+At this point the user should check that the current value of the
+\fImaxnstar\fR parameter is larger than the total number of stars
+in the photometry file written by the \fBphot\fR task. If \fImaxnstar\fR
+is too small, \fBpstselect\fR cannot read the entire input photometry
+file into memory and potential psf stars may be missed.
+
+.NH 3
+How Many Psf Stars Should Be Selected ?
+
+.PP
+How many stars should the user select to create the psf model ?
+An absolute minimum
+set by the mathematics is 1 star for a constant psf model,
+3 stars for a linearly variable psf model, and
+6 stars for a quadratically variable psf model. A more reasonable minimum
+suggested by Stetson (1992) is 3 stars
+per degree of freedom or,
+3 stars for a constant psf model, 9 stars for a linearly variable psf model,
+and 18 stars for a quadratically variable psf model. If a variable psf model
+is required, it is vitally  important that the psf star list
+sample the region of interest in the input image completely and
+reasonably uniformly.
+As the contribution of each psf star to the psf model is weighted by
+its signal-to-noise, the psf stars may cover a range in magnitude without
+compromising the resulting psf model.
+
+.NH 3
+Running Pstselect Non-interactively
+
+.PP
+The following example shows how to run the \fBpstselect\fR task
+in non-interactive mode.
+
+.YS
+da> pstselect image default default 3
+
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Star 5 has been added to the PSF star list
+ X:   35.98 Y:   22.00  Mag:  16.372  Dmin: 82.96088  Dmax: 535.1335
+Star 9 has been added to the PSF star list
+ X:   35.98 Y:   42.03  Mag:  16.594  Dmin: 80.25255  Dmax: 489.9732
+Star 6 has been added to the PSF star list
+ X:    8.02 Y:   22.97  Mag:  16.603  Dmin: 71.00896  Dmax: 436.3393
+
+Total of 3 PSF stars selected\fR
+.YE
+
+By default \fBpstselect \fR looks for an input photometry file called
+"test.mag.?" and writes an output psf star list called
+"test.pst.?".
+
+.NH 3
+Running Pstselect Interactively
+
+.PP
+\fBPstselect\fR may also be run interactively. In this mode of operation
+the stars selected
+by \fBpstselect\fR are examined by the user and accepted or rejected on
+the basis of the appearance of their mesh, contour or radial profile plots
+until a total of \fImaxnpsf\fR psf stars is reached.
+Stars from the input photometry file which do not meet the
+\fBpstselect\fR task selection criteria,
+can be added  to the psf star list by the user with the image cursor
+until a total of \fImaxnpsf\fR psf stars have been selected.
+.PP
+The following example shows how to run \fBpstselect\fR in interactive
+mode.
+
+.YS
+da> pstselect image default default 3 inter+ plottype=radial
+
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+.YE
+
+.IP ...
+The image cursor appears on the screen  ready to accept user input.
+.IP ...
+Type \fBn\fR to display the first potential psf star found by
+\fBpstselect\fR, \fBa\fR to select the star, or \fBd\fR to delete it.
+.IP ...
+Repeat the previous step for 2 more stars.
+.IP ...
+Type \fBl\fR to list the selected psf stars.
+.IP ...
+Type \fBq\fR to quit the task.
+
+.PP
+By default \fBpstselect \fR looks for an input photometry file called
+"test.mag.?" and writes an output psf star list called
+"test.pst.?" as before.
+
+.NH 3
+The Pstselect Output
+
+.PP
+The output psf star list consists of the ids, x and y coordinates,
+magnitudes, and sky values of the selected psf stars copied from the
+input photometry file without change.
+
+.NH 3
+Examining and/or Editing the Results of Pstselect
+
+.PP
+The \fBpdump\fR and \fBtvmark\fR commands can be used to mark
+and label the selected psf stars on the image display as shown in the
+following example.
+
+.YS
+da> display test 1 fi+
+da> pdump test.pst.1 xcenter,ycenter,id yes | tvmark 1 STDIN \\ 
+    col=204 label+\fR
+.YE
+
+Bad stars can be removed from the psf star list using the displayed and
+labeled image and
+the text editor, the \fBpselect\fR task, or
+the \fBpexamine\fR task.
+.PP
+The following command shows how to create a new psf star list
+without the psf star whose id is "5" using the \fBpselect\fR task.
+
+.YS
+da> pselect test.pst.1 test.pst.2 "id != 5"\fR
+.YE
+
+.PP
+The same operation can be accomplished using the \fBpexamine\fR task
+as shown below.
+
+.YS
+da> pexamine test.pst.1 test.pst.2 test 
+.YE
+
+.IP ...
+A message appears on the screen to the affect that
+pexamine cannot plot x versus y since the default y
+column merr is not in the input file.
+.IP ...
+The user types \fBh\fR to plot a histogram of the magnitudes
+and notes that there are three stars in the histogram.
+.IP ...
+The user decides that the star with id number 5 marked on the
+display should be deleted because it is too crowded.
+.IP ...
+The user types the \fBi\fR key to bring up the image cursor, moves
+it to star number 5, types the \fBd\fR key to delete the star, and
+the \fBp\fR key to replot the data.
+.IP ...
+The user types the \fBf\fR key to make the deletions permanent and the \fBe\fR
+key to exit the task, and update the output catalog.
+.LP
+
+Finally the user marks the new list on the display image using a
+different marking color.
+
+.YS
+da> pdump test.pst.2 xcenter,ycenter,id yes | tvmark 1 STDIN \\
+    col=205 label+
+.YE
+.LP
+
+.NH 2
+Computing the Psf Model with Psf
+
+.PP
+The \fBpsf\fR task computes the psf model used by the
+\fBpeak\fR, \fBnstar\fR, and \fBallstar\fR tasks to do psf fitting
+photometry, by the \fBgroup\fR task to estimate magnitudes for stars
+whose initial magnitudes are INDEF, and by the \fBaddstar\fR and
+\fBsubstar\fR tasks to add stars to and subtract stars from an image.
+
+.NH 3
+The Psf Algorithm
+
+.PP
+By default the \fBpsf\fR task performs the following functions:
+.IP [1]
+reads the \fBpsf\fR task parameters including, the input image name,
+the input photometry file name, the input psf star list name,
+the output psf image name,
+the output psf star list name, the output psf star group file name, and
+the \fBdatapars\fR and \fBdaopars\fR algorithm parameters
+.IP [2]
+reads the ids, x and y coordinates, magnitudes, and sky values
+of the first \fImaxnstar\fR stars in the input photometry file
+.IP [3]
+reads the ids, x and y coordinates, magnitudes, and sky values
+of the candidate psf stars from the input psf star list and/or
+the image cursor, rejecting stars
+which are not in the input photometry file, are within \fIfitrad\fR pixels
+of the edge of the image, are saturated (if 
+the parameter \fIsaturated\fR is "no"),
+or have bad data within \fIfitrad\fR pixels
+.IP [4]
+computes the analytic component of the psf model 
+specified by the parameter \fIfunction\fR using, data within \fIfitrad\fR
+pixels of each psf star, weights proportional to the signal-to-noise
+ratio in each psf star, and non-linear least-squares fitting techniques
+.IP [5]
+computes the residuals of each psf star from the best fit analytic function
+within a radius of \fIpsfrad\fR + 2 pixels
+.IP [6]
+scales the residuals for each psf star to match the intensity of the first
+psf star, subsamples the scaled residuals in x and y  by a factor of 2,
+weights the residuals by the signal-to-noise ratio of the psf star, 
+and combines the scaled, subsampled, and weighted residuals
+to  create 0, 1, 3, or 6, depending on the \fIvarorder\fR parameter,
+psf model look-up tables
+.IP [7]
+repeats steps [5] and [6] \fInclean\fR times, down-weighting the contributions
+to the psf model look-up table(s) of pixels with particularly large
+residuals each time through the loop
+.IP [8]
+estimates magnitudes for the saturated psf stars (if any exist and
+if the parameter \fIsaturated\fR is "yes"),
+by fitting the current psf model to the wings of the saturated stars
+using the \fBpeak\fR task fitting algorithm,
+.IP [9]
+computes the residuals of each saturated psf star (if any exist and they
+were successfully fit) from the best fit
+analytic function within a radius of \fIpsfrad\fR + 2 pixels, weights
+the residuals by a factor of 0.5, and adds
+the contribution of the scaled, subsampled, and weighted residuals 
+to the psf model look-up table(s)
+.IP [10]
+writes the computed analytic function parameters and 
+look-up tables to the output psf image
+.IP [11]
+identifies all stars within  (psfrad + 2 * fitrad + 1) pixels of
+a psf star as psf star neighbors, and stars within (2 * fitrad) pixels of
+the psf star neighbors as friends of the neighbors
+.IP [12]
+writes the ids, x and y coordinates, magnitudes, and sky values of
+the final list of psf stars to the output psf star list, and the group
+and star ids,
+x and y coordinates, magnitudes, and sky values of the psf stars,
+psf star neighbors, and friends of the psf star neighbors
+to the output psf star group file
+
+.NH 3
+Choosing the Appropriate Analytic Function
+
+.PP
+DAOPHOT offers several choices for the functional form of the analytic
+component of the psf model (see Appendix 8.2 for details).
+To achieve the best fits and to minimize interpolation errors in
+the psf model look-up tables,
+users should choose the analytic function that most closely
+approximates the stellar psf. The options are:
+.IP [1]
+\fBgauss\fR (2 parameters),
+a 2D elliptical gaussian function aligned along the x and y axes of the image.
+Gauss is generally the best choice for well-sampled, fwhmpsf >= 2.5 pixels,
+ground-based images because the interpolation errors are small
+and evaluation is efficient as the function is separable in x and y.
+.IP [2]
+\fBmoffat25\fR and \fBmoffat15\fR (3 parameters),
+elliptical Moffat functions of beta 2.5 and 1.5 respectively which can
+be aligned along an arbitrary position angle. The Moffat functions
+are good choices for under-sampled ground-based data.
+.IP [3]
+\fBlorentz\fR (3 parameters),
+an elliptical Lorentz function which can be aligned along an arbitrary position
+angle. The Lorenz function is a good choice for old ST data since it has
+extended wings.
+.IP [4]
+\fBpenny1\fR (4 parameters),
+a two component model consisting of an elliptical gaussian core 
+which can be aligned along an arbitrary position angle and
+lorentzian wings aligned along the x and y axes of the image.
+The Penny1 function is a good choice for a purely analytic psf model.
+.IP [5]
+\fBpenny2\fR (5 parameters),
+a two component model consisting of an elliptical gaussian core 
+aligned along an arbitrary position angle and lorentzian wings aligned
+along an arbitrary position angle which may be different from that of the
+core.  The Penny2 function is a good choice for a purely analytic psf model.
+.IP [6]
+\fBauto\fR (2, 3, 4 or 5 parameters),
+try each of the 6 analytic psf functions in turn and select the one which
+yields the smallest scatter in the fit. Users should use
+this option with caution 
+since the greater number of free parameters in some models may
+artificially produce a fit with less scatter without significantly
+improving the resulting photometry.
+.IP [7]
+\fBlist\fR (2, 3, 4 or 5 parameters),
+check only those functions in a user specified list, e.g.
+"gauss,moffat25,lorentz" and select the one that gives the smallest
+scatter.
+.PP
+Users uncertain of which analytic function to choose should leave
+\fIfunction\fR set to "gauss"
+and only if the results prove unsatisfactory experiment with one of the
+more complicated analytic functions.
+
+.NH 3
+The Analytic Psf Model
+
+.PP
+A purely analytic psf model may be computed
+by setting the \fBdaopars\fR parameter \fIvarorder\fR = -1.
+Analytic psf models are
+constant, i.e. they have the same shape everywhere in the
+image.
+In the majority
+of cases this is NOT the best modeling option,
+as a better representation of the true psf is almost always obtained by
+computing an empirical psf model composed of an
+analytic function plus one look-up table.
+.PP
+An analytic psf model may be required to model severely undersampled
+data because interpolation errors can produce large uncertainties
+in the computed look-up tables and the resulting fits.
+.PP
+Fields which are so crowded that 
+isolated psf stars are non-existent, may also require psf modeling and 
+psf star neighbor subtraction with an analytic psf model, 
+before a more accurate higher order model free of ghosts produced
+by the psf star neighbors can be computed.
+This step is particularly important if the field is very crowded AND the
+psf is known to be variable.
+
+.NH 3
+The Empirical Constant Psf Model
+
+.PP
+Most users with typical ground-based optical
+data choose to compute an empirical constant psf
+model composed of an analytic component and a single look-up table,
+by setting the \fBdaopars\fR parameter \fIvarorder\fR = 0.
+This type of model is constant, i.e. the psf model has
+the same shape everywhere in the image.
+.PP
+Because of interpolation errors, severely undersampled data may be better
+fit with a purely analytic psf model as described in the previous section.
+.PP
+Fields which are so crowded that 
+isolated psf stars are non-existent may require psf modeling and 
+psf neighbor star subtraction with an analytic psf model, 
+before an accurate look-up table free of ghosts caused by the bright
+psf star neighbors can be computed.
+
+.NH 3
+The Empirical Variable Psf Model
+
+.PP
+Psf models which vary linearly or quadratically
+with position in the image can be computed by setting
+the \fIvarorder\fR parameter to 1 or 2 respectively. 
+In the first case a total of 3 look-up tables will be computed;
+in the second case 6 look-up tables will be computed.
+Users should always begin their analysis with \fIvarorder\fR = -1 or
+0 if their data is from a telescope/instrument combination that
+is unfamiliar to them. Only if the patterns of the residuals around stars
+fit and subtracted with a constant psf model show systematic variations
+with position
+in the image, should the user proceed to experiment with the variable
+psf models.
+.PP
+In very crowded regions it may be necessary to compute a good
+variable psf model iteratively, starting with \fIvarorder\fR = -1
+and proceeding to \fIvarorder\fR = 2 by, computing the psf model,
+fitting the psf model to the psf stars and their neighbors, subtracting
+the psf star neighbors but not the psf stars from the original image,
+increasing \fIvarorder\fR by 1, and recomputing the
+psf model using the subtracted image, until all the psf stars and their
+neighbors subtract out cleanly.
+
+.NH 3
+Rejecting Bad Data from the Psf Model
+
+.PP
+The \fBpsf\fR task uses the \fBdatapars\fR parameters \fIdatamin\fR
+and \fIdatamax\fR to flag bad data.
+If the \fBdaopars\fR parameter \fIsaturated\fR is "no", a prospective
+psf star
+will be rejected outright if it has high or low bad data inside the fitting
+radius; if \fIsaturated\fR is "yes" a star with low bad data will
+be rejected outright but one with high bad data will be flagged
+as saturated and accepted. Except in rare cases (see below) users should leave
+\fIsaturated\fR set to "no". Stars with bad data outside the fitting radius
+but inside the psf radius are flagged, and the user warned, but are still
+accepted as psf stars.
+.PP
+All data within one fitting radius of the unsaturated psf stars is weighted by
+the signal-to-noise ratio of the psf star and used to compute the 
+analytic component of the psf model.
+Pixels which deviate strongly from the current best fit 
+analytic function are down-weighted during the course of the fit.
+.PP
+After the analytic function is fit, the residuals of the psf star data
+from the best fit analytic function are computed, scaled to the magnitude
+of the first psf star, weighted by the signal-to-noise in the psf star,
+subsampled for a factor
+of 2 in x and y, and added into the look-up table(s).
+If there are too few psf stars with
+good data to compute a particular element of the look-up table(s),
+\fBpsf\fR will quit with an error. If the \fBdaopars\fR parameter
+\fInclean\fR > 0, deviant pixels contributing to the psf model look-up tables
+are down-weighted and the look-up table(s) are recomputed \fInclean\fR
+times.
+.PP
+For images where all the bright candidate psf stars are saturated and all the
+remaining
+candidate psf stars are faint, it may be necessary to use the faint stars to
+compute the analytic component of the psf model and bright saturated stars
+to compute the look-up tables(s).
+In this circumstance the user must set the parameter \fIsaturated\fR 
+to "yes" and include several saturated stars in the psf star list.
+After the analytic function and an initial set of look-up tables(s)
+is computed without using the saturated psf stars, the \fBpeak\fR task
+fitting algorithm is used to compute accurate magnitudes for the
+saturated psf stars by fitting the wings of the saturated stars to
+the current psf model. New look-up table(s) are computed
+which include the contributions weighted by 0.5 of the saturated psf stars.
+
+.NH 3
+The Model Psf Psfrad and Fitrad
+
+.PP
+The \fBdaopars\fR parameter \fIpsfrad\fR defines the region over which 
+the psf model is defined. This radius should equal the radius at which the
+radial profile of the brightest star of interest disappears into the noise,
+e.g. \(~= 5 pixels for the test image as determined
+in the section describing the
+\fBpstselect\fR task. The fitting radius defines the region of data around
+each psf star used to compute the analytic component of the psf model
+and should be the larger of the numbers 3 and  1 * \fIfwhmpsf\fR pixels.
+
+.NH 3
+Modeling the Psf Interactively Without a Psf Star List
+
+.PP
+The psf can be modeled interactively without an initial list of candidate
+psf stars by displaying the image and selecting candidate psf stars
+with the image
+cursor. Good candidate psf stars must be in the input photometry file,
+have no neighbors within \fIfitrad\fR
+pixels, and be free of cosmetic blemishes. 
+.PP
+The following example shows how to model the psf interactively without
+using an initial psf star list.
+
+.YS
+da> display test 1 fi+
+
+da> psf test default "" default default default
+
+Analytic psf function(s) (gauss):
+        Analytic psf function(s): gauss
+Order of variable psf (0):
+        Order of variable psf: 0
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Warning: Graphics overlay not available for display device.
+
+Computing PSF for image: test
+9 stars read from test.mag.1
+.YE
+
+.IP ...
+A message appears on the screen telling the user how
+many stars have been read from the photometry file
+(users should make sure that this is the entire star list)
+and the image cursor begins blinking.
+.IP ...
+The user types the \fBa\fR keystroke at pixel 36,42
+followed by another \fBa\fR keystroke after the default
+plot appears,
+to add star 9 psf star list. Star 6 at pixel 8,23 is added
+to the psf star list in the identical manner.
+.LP
+
+.YS
+Star 9 has been added to the PSF star list
+ X:   35.98 Y:   42.03  Mag:  16.594  Dmin: 80.25255  Dmax: 489.9732
+Star 6 has been added to the PSF star list
+ X:    8.02 Y:   22.97  Mag:  16.603  Dmin: 71.00896  Dmax: 436.3393
+.YE
+
+.IP ...
+The user types the \fBl\fR keystroke command to list the selected psf stars.
+.LP
+
+.YS
+Current PSF star list
+    Star:    9  X:   35.98 Y:   42.03  Mag:   16.59  Sky:      101.0
+    Star:    6  X:    8.02 Y:   22.97  Mag:   16.60  Sky:       98.9
+.YE
+.LP
+
+.IP ...
+The user types the \fBf\fR keystroke command to compute the psf model.
+.LP
+
+.YS
+Fitting function gauss    norm scatter: 0.03422394
+
+Analytic PSF fit
+    Function: gauss  X: 25.  Y: 25.  Height: 523.8066  Psfmag: 16.594
+    Par1: 1.082032  Par2: 1.162063
+
+Computed 1 lookup table(s)
+.YE
+
+.IP ...
+The user reviews the model fit with the \fBr\fR keystroke command and
+decides to keep both psf stars.
+.LP
+
+.YS
+PSF star 9 saved by user
+PSF star 6 saved by user
+.YE
+
+.IP ...
+The user types the \fBf\fR keystroke command to remodel the psf.
+.LP
+
+.YS
+Fitting function gauss    norm scatter: 0.03422394
+
+Analytic PSF fit
+    Function: gauss  X: 25.  Y: 25.  Height: 523.8066  Psfmag: 16.594
+    Par1: 1.082032  Par2: 1.162063
+
+Computed 1 lookup table(s)
+.YE
+
+.IP ...
+The user types the \fBw\fR keystroke command  to save the psf model
+followed by the \fBq\fR keystroke command, executed twice, to quit the task.
+.LP
+
+.YS
+Writing PSF image test.psf.1.imh
+Writing output PSF star list test.pst.1
+Writing output PSF star group file test.psg.1
+.YE
+
+.PP
+At this point the user has created an initial psf model in the image
+test.psf.1, a list of the psf stars in test.pst.1, and a list of the
+psf stars and their neighbors in the file
+test.psg.1 respectively.
+.PP
+Users may occasionally see "Star not found" messages when 
+selecting psf stars with the image cursor. This may mean: 1) that the star
+is truly not in the input photometry file (this can be checked
+with the \fBtvmark\fR task), 2) that the image cursor 
+is more than \fImatchrad\fR pixels from the position of the star
+in the input photometry file (either position the image cursor more
+carefully by hand or increase the value of the \fImatchrad\fR parameter), or,
+3) that the input photometry file contains
+more than \fImaxnstar\fR stars (increase the value of the parameter
+\fImaxnstar\fR so that it is greater than the number of stars in
+the photometry file).
+
+.NH 3
+Fitting the Psf Model Interactively Using an Initial Psf Star List
+
+.PP
+The \fBpsf\fR task can also be run interactively using an initial list of
+psf stars chosen by the user with the \fBpstselect\fR task.
+If the \fBpsf\fR task parameter \fIshowpsf\fR is "yes" (the default),
+the psf stars are read from the psf star list one at a time,
+a mesh, contour, or radial profile plot is displayed in the graphics window,
+and the user can accept or delete the star with the \fBa\fR or \fBd\fR
+keystroke commands. If
+\fIshowplots\fR is "no", the psf star list is read without intervention
+by the user. In both cases new stars can be added to the end of the psf
+star list with the image cursor in the usual manner.
+.PP
+A sample run is shown below.
+
+.YS
+da> display test 1 fi+
+
+da> pdump test.pst.1 xcenter,ycenter,id yes | tvmark 1 STDIN \\ 
+    col=205 label+
+.YE
+
+.IP ...
+The user marks and labels the initial list of psf stars on the image display.
+.LP
+
+.YS
+da> psf test default test.pst.1 default default default
+
+Analytic psf function(s) (gauss):
+        Analytic psf function(s): gauss
+Order of variable psf (0):
+        Order of variable psf: 0
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Warning: Graphics overlay not available for display device.
+
+Computing PSF for image: test
+9 stars read from test.mag.1
+.YE
+
+.IP ...
+The user rejects or accepts the stars in the .pst file by typing the
+\fBd\fR or \fBa\fR keystroke commands respectively after the
+default plot appears.
+.LP
+
+.YS
+Star 5 rejected by user
+Star 9 has been added to the PSF star list
+ X:   35.98 Y:   42.03  Mag:  16.594  Dmin: 80.25255  Dmax: 489.9732
+Star 6 has been added to the PSF star list
+ X:    8.02 Y:   22.97  Mag:  16.603  Dmin: 71.00896  Dmax: 436.3393
+
+2 PSF stars read from test.pst.1
+.YE
+
+.IP ...
+The user types the \fBl\fR keystroke command to view the psf star
+list one more time.
+.LP
+
+.YS
+Current PSF star list
+    Star:    9  X:   35.98 Y:   42.03  Mag:   16.59  Sky:      101.0
+    Star:    6  X:    8.02 Y:   22.97  Mag:   16.60  Sky:       98.9
+.YE
+
+.IP ...
+The user computes the psf model with the \fBf\fR keystroke command.
+.LP
+
+.YS
+Fitting function gauss    norm scatter: 0.03422394
+
+Analytic PSF fit
+    Function: gauss  X: 25.  Y: 25.  Height: 523.8066  Psfmag: 16.594
+    Par1: 1.082032  Par2: 1.162063
+
+Computed 1 lookup table(s)
+.YE
+
+.IP ...
+The user saves the psf model with the \fBw\fR keystroke command.
+.LP
+
+.YS
+Writing PSF image test.psf.1.imh
+Writing output PSF star list test.pst.2
+Writing output PSF star group file test.psg.1
+.YE
+
+.IP ...
+The user types the \fBq\fR keystroke command to quit the task.
+.LP
+
+The user notes that the output psf star list is given a version number
+of 2 in this example, since version 1 was written by the \fBpstselect\fR task.
+
+.NH 3
+Fitting the Psf Model Interactively Without an Image Display
+
+.PP
+Users without access to an image display, may still run \fBpsf\fR
+interactively by redirecting
+the image cursor commands to the terminal as shown below.
+
+.YS
+da> set stdimcur = text
+
+da> psf test default test.pst.1 default default default
+
+Analytic psf function(s) (gauss):
+        Analytic psf function(s): gauss
+Order of variable psf (0):
+        Order of variable psf: 0
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Warning: Graphics overlay not available for display device.
+
+Computing PSF for image: test
+9 stars read from test.mag.1
+.YE
+
+.IP ...
+The user rejects or accepts the stars in the .pst file by typing 
+the \fBd\fR or \fBa\fR keystroke commands respectively at the prompt.
+.LP
+
+.YS
+Star 5 rejected by user
+Star 9 has been added to the PSF star list
+ X:   35.98 Y:   42.03  Mag:  16.594  Dmin: 80.25255  Dmax: 489.9732
+Star 6 has been added to the PSF star list
+ X:    8.02 Y:   22.97  Mag:  16.603  Dmin: 71.00896  Dmax: 436.3393
+
+2 PSF stars read from test.pst.1
+.YE
+
+.IP ...
+The user types the \fBl\fR keystroke command at the prompt to
+view the psf star list one more time.
+.LP
+
+.YS
+Current PSF star list
+    Star:    9  X:   35.98 Y:   42.03  Mag:   16.59  Sky:      101.0
+    Star:    6  X:    8.02 Y:   22.97  Mag:   16.60  Sky:       98.9
+.YE
+
+.IP ...
+The user computes the psf model by typing the \fBf\fR keystroke
+command at the prompt.
+.LP
+
+.YS
+Fitting function gauss    norm scatter: 0.03422394
+
+Analytic PSF fit
+    Function: gauss  X: 25.  Y: 25.  Height: 523.8066  Psfmag: 16.594
+    Par1: 1.082032  Par2: 1.162063
+
+Computed 1 lookup table(s)
+.YE
+
+.IP ...
+The user saves the psf model by typing the \fBw\fR keystroke command
+at the prompt.
+.LP
+
+.YS
+Writing PSF image test.psf.1.imh
+Writing PSF output star list test.pst.2
+Writing PSF output star group file test.psg.1
+.YE
+
+.IP ...
+The user types the \fBq\fR keystroke command at the prompt to quit the task.
+.LP
+
+Additional stars can be added to the psf star list by commands of
+the form \fB":a id#"\fR or \fB"100.2 305.6 1 a"\fR typed in at the
+terminal prompt.
+The user should remember to reset the image cursor to the logical
+image cursor with the command \fB"reset stdimcur = stdimage"\fR
+after running the \fBpsf\fR task in "no image display" mode.
+
+.NH 3
+Fitting the Psf Model Non-interactively
+
+.PP
+Finally the psf model can be fit non-interactively by setting the
+\fIinteractive\fR parameter to "no", and using the list of psf stars produced by
+the \fBpstselect\fR task as input. This is the preferred method for computing
+the psf model when the number of psf stars is large (e.g.  the psf
+model to be computed is variable).
+
+.YS
+da> psf test default test.pst.1 default default default inter-
+.YE
+
+.NH 3
+The Output of Psf
+
+.PP
+\fBPsf\fR writes an output psf star list test.pst.# containing the ids, x and
+y coordinates, magnitudes, and sky values, copied from the
+input photometry file,
+of the psf stars actually used to compute the final psf model.
+Because of the ability
+to add and subtract stars within \fBpsf\fR itself, this list may be different
+from the input psf star list if any. A sample output psf star list is
+shown below.
+
+.YS
+#K IRAF       = NOAO/IRAFV2.10EXPORT    version    %-23s
+#K USER       = davis                   name       %-23s
+#K HOST       = tucana                  computer   %-23s
+#K DATE       = 05-28-93                mm-dd-yr   %-23s
+#K TIME       = 14:34:31                hh:mm:ss   %-23s
+#K PACKAGE    = daophot                 name       %-23s
+#K TASK       = psf                     name       %-23s
+#K IMAGE      = test                    imagename  %-23s
+#K PHOTFILE   = test.mag.1              filename   %-23s
+#K PSTFILE    = test.pst.1              filename   %-23s
+#K PSFIMAGE   = test.psf.1              imagename  %-23s
+#K GRPSFILE   = test.psg.1              filename   %-23s
+#K OPSTFILE   = test.pst.2              filename   %-23s
+#K SCALE      = 1.                      units/pix  %-23.7g
+#K OTIME      = 00:07:59.0              timeunit   %-23s
+#K IFILTER    = V                       filter     %-23s
+#K XAIRMASS   = 1.238106                number     %-23.7g
+#K PSFRAD     = 5.                      scaleunit  %-23.7g
+#K FITRAD     = 3.                      scaleunit  %-23.7g
+#
+#N ID    XCENTER   YCENTER   MAG         MSKY                                  \\
+#U ##    pixels    pixels    magnitudes  counts                                \\
+#F %-9d  %-10.3f   %-10.3f   %-12.3f     %-12.3f
+#
+9        35.980    42.029    16.594      101.043
+6        8.022     22.970    16.603      98.891
+.YE
+
+.PP
+\fBPsf\fR also writes an output psf star group photometry file 
+test.psg.? containing the group ids, and the star ids, 
+x and y coordinates, magnitudes, and sky values
+copied from the input photometry file,
+for the psf stars and their
+neighbors. A sample psf star group file is shown below.
+
+.YS
+#K IRAF       = NOAO/IRAFV2.10EXPORT    version    %-23s
+#K USER       = davis                   name       %-23s
+#K HOST       = tucana                  computer   %-23s
+#K DATE       = 05-26-93                mm-dd-yr   %-23s
+#K TIME       = 16:10:48                hh:mm:ss   %-23s
+#K PACKAGE    = daophot                 name       %-23s
+#K TASK       = psf                     name       %-23s
+#K IMAGE      = test                    imagename  %-23s
+#K PHOTFILE   = test.mag.2              filename   %-23s
+#K PSTFILE    = test.pst.2              filename   %-23s
+#K PSFIMAGE   = test.psf.2              imagename  %-23s
+#K GRPSFILE   = test.psg.2              filename   %-23s
+#K SCALE      = 1.                      units/pix  %-23.7g
+#K OTIME      = 00:07:59.0              timeunit   %-23s
+#K IFILTER    = V                       filter     %-23s
+#K XAIRMASS   = 1.238106                number     %-23.7g
+#K PSFRAD     = 5.                      scaleunit  %-23.7g
+#K FITRAD     = 3.                      scaleunit  %-23.7g
+#
+#N ID    GROUP XCENTER   YCENTER   MAG         MSKY                            \\
+#U ##    ##    pixels    pixels    magnitudes  counts                          \\
+#F %-9d  %-6d  %-10.3f   %-10.3f   %-12.3f     %-14.3f
+#
+9        1     35.980    42.029    16.594      101.043
+8        1     28.958    33.924    17.781      100.619
+6        2     8.022     22.970    16.603      98.891
+.YE
+
+There are two stellar groups, one group per psf star,
+in this file. The first psf star
+has a single neighbor so there are two stars in the first group.
+The first star in a group is always the psf star. The header parameters
+record the input and output image and file names, the name of the
+computed psf model, and the values
+of the parameters
+\fIpsfrad\fR and \fIfitrad\fR used to define the psf star groups.
+.PP
+The psf image contains, in the image header, the values of the
+parameters that were used
+to compute the psf model, the best fit values of the parameters of
+the chosen analytic function, and a record of all the
+psf stars used to compute the psf, and  in the image pixels, the best fit
+look-up table(s) of the residuals from the analytic function
+subsampled by a factor of 2.
+The psf image look-up table(s) can be plotted and examined just like any
+other IRAF image.
+.PP
+A sample psf image header is shown below.
+
+.YS
+test.psf.2[23,23][real]: PSF for image: test
+    No bad pixels, no histogram, min=unknown, max=unknown
+    Line storage mode, physdim [23,23], length of user area 1540 s.u.
+    Created Wed 16:10:47 26-May-93, Last modified Wed 16:10:47 26-May-93
+    Pixel file 'tucana!/d0/iraf/davis/test.psf.2.pix' [ok]
+    IRAF    = 'NOAO/IRAFV2.10EXPORT'
+    HOST    = 'tucana  '
+    USER    = 'davis   '
+    DATE    = '05-26-93'
+    TIME    = '16:10:48'
+    PACKAGE = 'daophot '
+    TASK    = 'psf     '
+    IMAGE   = 'test    '
+    PHOTFILE= 'test.mag.2'
+    PSTFILE = 'test.pst.2'
+    PSFIMAGE= 'test.psf.2'
+    GRPSFILE= 'test.psg.2'
+    SCALE   =                   1.
+    PSFRAD  =                   5.
+    FITRAD  =                   3.
+    DATAMIN =                  50.
+    DATAMAX =               24500.
+    NCLEAN  =                    0
+    USESAT  =                    F
+    FUNCTION= 'gauss   '
+    PSFX    =                  25.
+    PSFY    =                  25.
+    PSFHEIGH=             523.8066
+    PSFMAG  =               16.594
+    NPARS   =                    2
+    PAR1    =             1.082032
+    PAR2    =             1.162063
+    VARORDER=                    0
+    FEXPAND =                    F
+    NPSFSTAR=                    2
+    ID1     =                    9
+    X1      =                35.98
+    Y1      =               42.029
+    MAG1    =               16.594
+    ID2     =                    6
+    X2      =                8.022
+    Y2      =                22.97
+    MAG2    =               16.603
+.YE
+
+This psf image header records that: the psf model
+was computed with a gaussian
+analytic function (function = gauss), the analytic
+function has two parameters (npars=2)
+whose values are 1.08 and 1.16 (par1 and par2 are the fwhm of the function
+in x and y respectively in this case), the psf is constant but there is
+1 look-up
+table (varorder = 0),  no saturated stars were used to compute the psf
+(usesat=no), and no cleaning of bad pixels was done to compute the
+lookup table (nclean=0). The number of psf stars and their positions
+and magnitudes
+are also listed. The psf model is defined over a radius of 5
+pixels (psfrad = 5.0), resulting in a square look-up table with dimensions
+of 2 * (nint (2 * psfrad) + 1) + 1 pixels in x and y, and a fitting radius
+of 3 (fitrad = 3.0) was used to compute the analytic portion of the psf
+model.
+.PP
+The height of the best fit gaussian psfheigh is 523.81 counts.
+The psf model has been assigned a magnitude of 16.594 which is the
+magnitude of the first psf star in the input photometry file.
+All fits to the psf model are scaled with respect to this
+magnitude. Therefore a star which is twice as bright as the psf model will
+have a fitted magnitude of ~15.841.
+.PP
+Psfx and psfy define the distance from the center of the input image to the 
+center of the edge pixels in x and y respectively, e.g psfx =
+(ncols - 1.0) / 2.0 and psfy = (nlines - 1) / 2.0. These numbers are used to
+evaluate the psf model only if the model is variable, \fIvarorder\fR > 0.
+.PP
+If the value of \fIvarorder\fR  in this example had been 1 or 2 the
+psf model image would have been 3-dimensional with 3 and 6 23 by 23
+pixel look-up
+tables in planes 1-3 and 1-6 of the image respectively.
+In both cases planes 1-3 would contain the 0th, 1st order in x,
+and 1st order in y Taylor expansion coefficients around the analytic function.
+In the latter case planes 4-6 would contain the
+2nd order in x, 2nd order in xy, and 2nd order in y Taylor expansion
+coefficients.  If the value of \fIvarorder\fR
+had been -1 no look-up tables would have been computed and the
+psf model image would consist of an image header but no pixel file.
+
+.NH 3
+Checking the Psf Model
+
+.PP
+To check the accuracy of the psf model, the user must fit each psf
+star and its neighbors and friends
+as a group using the \fBnstar\fR task, and the psf model and psf star group
+photometry file
+produced by \fBpsf\fR as shown below. In the following example
+the user has chosen to set the rejections
+file to "", so that all stars, even those too faint to be properly
+fit, will be placed in the same output file.
+
+.YS
+da> nstar test test.psg.1 default default ""
+
+Recenter the stars (yes):
+        Recenter the stars: yes
+Refit the sky (no):
+        Refit the sky: no
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Maximum group size in number of stars (60):
+        New maximum group size: 60 stars
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Group:    1 contains  2 stars
+        ID:     9  XCEN:    35.98  YCEN:    42.01  MAG:    16.60
+        ID:     8  XCEN:    28.96  YCEN:    33.91  MAG:    17.73
+Group:    2 contains  1 stars
+        ID:     6  XCEN:     8.02  YCEN:    22.97  MAG:    16.63
+.YE
+
+The results of the fits will appear in test.nst.? as shown below.
+
+.YS
+da> page test.nst.1
+
+#K IRAF       = NOAO/IRAFV2.10EXPORT    version    %-23s
+#K USER       = davis                   name       %-23s
+#K HOST       = tucana                  computer   %-23s
+#K DATE       = 05-28-93                mm-dd-yy   %-23s
+#K TIME       = 14:46:13                hh:mm:ss   %-23s
+#K PACKAGE    = daophot                 name       %-23s
+#K TASK       = nstar                   name       %-23s
+#K IMAGE      = test                    imagename  %-23s
+#K GRPFILE    = test.psg.1              filename   %-23s
+#K PSFIMAGE   = test.psf.1              imagename  %-23s
+#K NSTARFILE  = test.nst.1              filename   %-23s
+#K REJFILE    = ""                      filename   %-23s
+#K SCALE      = 1.                      units/pix  %-23.7g
+#K DATAMIN    = 50.                     counts     %-23.7g
+#K DATAMAX    = 24500.                  counts     %-23.7g
+#K GAIN       = 1.                      number     %-23.7g
+#K READNOISE  = 0.                      electrons  %-23.7g
+#K OTIME      = 00:07:59.0              timeunit   %-23s
+#K XAIRMASS   = 1.238106                number     %-23.7g
+#K IFILTER    = V                       filter     %-23s
+#K RECENTER   = yes                     switch     %-23b
+#K FITSKY     = no                      switch     %-23b
+#K PSFMAG     = 16.594                  magnitude  %-23.7g
+#K PSFRAD     = 5.                      scaleunit  %-23.7g
+#K FITRAD     = 3.                      scaleunit  %-23.7g
+#K MAXITER    = 50                      number     %-23d
+#K MAXGROUP   = 60                      number     %-23d
+#K FLATERROR  = 0.75                    percentage %-23.7g
+#K PROFERROR  = 5.                      percentage %-23.7g
+#K CLIPEXP    = 6                       number     %-23d
+#K CLIPRANGE  = 2.5                     sigma      %-23.7g
+#
+#N ID    GROUP XCENTER   YCENTER   MAG         MERR          MSKY              \\
+#U ##    ##    pixels    pixels    magnitudes  magnitudes    counts            \\
+#F %-9d  %-6d  %-10.3f   %-10.3f   %-12.3f     %-14.3f       %-12.3f
+#
+#N         NITER SHARPNESS   CHI         PIER  PERROR                          \\
+#U         ##    ##          ##          ##    perrors                         \\
+#F         %-17d %-12.3f     %-12.3f     %-6d  %-13s
+#
+9        1     35.982    42.006    16.601      0.019         101.043           \\
+           4     -0.019      0.512       0     No_error
+8        1     28.962    33.912    17.730      0.074         100.619           \\
+           4     0.026       1.093       0     No_error
+6        2     8.017     22.968    16.628      0.021         98.891            \\
+           3     0.022       0.558       0     No_error
+.YE
+
+In this example the chi values computed by \fBnstar\fR for the two
+psf stars are lower than expected, ~ 0.5 instead of ~ 1.0, meaning that the
+observed errors are less than the predicted errors.
+This occurs because there are only 2 psf stars,
+and therefore the model psf and the fitted psf stars are not totally
+independent.
+For example, if only one psf star is used to compute
+the psf model, the chi
+computed by \fBnstar\fR for that star would be ~ 0.0 and for any others
+such as its neighbors ~ 1.0.
+.PP
+After checking that the chi values look reasonable, the user subtracts the
+fitted psf stars and their neighbors from the input image
+with the \fBsubstar\fR task, and examines the residuals
+of the fit around the psf stars as shown below. After \fBsubstar\fR is
+run the subtracted
+image is displayed and the psf stars are marked in  green and the
+psf neighbor stars are marked in red.
+
+.YS
+da> substar test default "" default default
+
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+SUBTRACTING - Star:    6 X =    8.02 Y =   22.97 Mag =   16.63
+SUBTRACTING - Star:    8 X =   28.96 Y =   33.91 Mag =   17.73
+SUBTRACTING - Star:    9 X =   35.98 Y =   42.01 Mag =   16.60
+
+A total of 3 stars were subtracted out of a possible 3
+
+da> display test.sub.1 1 fi+
+da> pdump test.nst.1 xc,yc,id yes | tvmark 1 STDIN col=204 label+
+da> pdump test.pst.1 xc,yc,id yes | tvmark 1 STDIN col=205 label+
+.YE
+
+.PP
+The psf stars and their neighbors should subtract out cleanly
+with no systematic patterns in the residuals as a function of distance
+from the star (note this may not be the case if the psf is purely analytic
+or so severely undersampled that the interpolation errors near the
+center are very large), with magnitude, or with position in the image.
+There should be no hidden
+underlying neighbors revealed after the subtraction (these psf stars
+should be rejected) or neighbors
+that are not in the photometry file (this can be fixed up later).
+The amplitudes of the fit residuals
+should be consistent with the noise if sufficient stars are
+used to determine the psf model.
+.PP
+The displayed and marked subtracted image and the output of \fBnstar\fR can
+be examined in more detail with the \fBpexamine\fR task as shown below.
+
+.YS
+da> pexamine test.nst.1 "" test.sub.1
+.YE
+.IP ...
+A plot of magnitude versus magnitude error for the psf stars and their
+neighbors will appear.
+.IP ...
+Change the default plot to mag versus chi with the command \fB:ycolumn chi\fR
+followed by the \fBp\fR keystroke command.
+.IP ...
+Activate the image cursor with the \fBi\fR keystroke command.
+.IP ...
+Move to psf star number 9 and type \fBr\fR to examine the radial profile
+of the subtracted star, followed by \fBo\fR to get a listing of its position,
+magnitude, magnitude error, sky value, etc.
+.IP ...
+Examine the radial profiles of the other subtracted psf stars and their
+neighbors.
+.IP ...
+Type \fBq\fR to quit the task.
+
+.PP
+For the test image "test", examination of the radial profiles of the
+subtracted stars shows that the residuals are
+consistent with the noise in the image and have no unusual features
+leading to the conclusion that the current psf model is a good representation
+of the true psf.
+
+.NH 3
+Removing Bad Stars from the Psf Model
+
+.PP
+Bad psf stars detected after the psf star and neighbor subtraction,
+for example those with
+a cosmetic blemish, a close double, or an underlying neighbor star,
+should be removed altogether from
+the psf star list. This can be done by editing the psf star list written
+by \fBpsf\fR with the text editor, with the \fBpselect\fR task and an
+expression, e.g. "id != 5", specifying the star to be deleted, or
+with the interactive \fBpexamine\fR task using the delete and update keys,
+and rerunning \fBpsf\fR with the new psf star list.
+.PP
+This step is not required for the test image as both psf stars and their
+neighbors subtracted cleanly from the image.
+
+.NH 3
+Adding New Stars to a Psf Star Group
+
+.PP
+Occasionally stars that are too faint to be included
+in initial star list produced by \fBdaofind\fR and measured with \fBphot\fR, are
+nevertheless sufficiently bright and close to a psf star
+to affect the computation of the psf model.
+Ideally the psf stars should have no such companions and/or the look-up
+table cleaning option in the \fBpsf\fR task should minimize the problem of
+undetected neighbors.  However in some cases it is necessary for
+the user to intervene and add faint stars to the photometry list.
+.PP
+The easiest way to accomplish this
+is to run \fBphot\fR interactively, selecting the
+missing neighbor stars with the image cursor and appending the results
+for the new neighbor stars to the photometry file produced by the first
+run of \fBnstar\fR.
+
+.YS
+da> phot test "" psf.mag inter+ calgorithm=centroid
+.YE
+
+.IP ...
+Point cursor to undetected psf star neighbors and hit spacebar.
+.IP ...
+Type \fBq\fR to quit.
+.LP
+
+.YS
+da> prenumber psf.mag idoffset=5000
+.YE
+
+.IP ...
+Renumber the stars in the new file starting with a number greater
+than the number of stars in the original photometry file in order
+to insure that all the stars have unique ids.
+.LP
+
+.YS
+da> pfmerge test.psg.1,psf.mag test.psf.psg
+.YE
+
+.IP ...
+Combine the psf star group photometry file produced by the \fBpsf\fR task
+with the photometry file for the new psf star neighbors produced by
+the \fBphot\fR task.
+.LP
+
+Users should note that there is no input coordinate list to \fBphot\fR in
+this case.  Therefore the coordinate list is set to "", interactive mode
+is on, and centering is turned on.
+The remaining photometry parameters should be set exactly as they were in the
+first run of \fBphot\fR.
+.PP
+This step is not required for the test image as all the significant
+psf star neighbors were detected by the first run of the \fBdaofind\fR task.
+
+.NH 3
+Refitting the Psf Model With the New Psf Star Groups
+
+.PP
+After the \fBpsf \fR and \fBphot\fR results have been merged, the
+user must regroup the psf stars and their neighbors with
+the \fBgroup\fR task, and refit the
+new groups with the \fBnstar\fR task.
+
+.YS
+da> group test test.psf.psg default test.grp  critov=.2 
+.YE
+
+.IP ...
+Regroup the stars using a very small value for the critical overlap
+parameter.
+.LP
+
+.YS
+da> nstar test test.grp default default ""
+.YE
+
+.IP ...
+Rerun nstar on the new psf star groups.
+.LP
+
+.YS
+da> substar test test.nst.2 ""  default default
+.YE
+
+.IP ...
+Check that the new psf star groups subtract cleanly from the original
+image.
+.PP
+This step is not required for the test image as all the significant
+psf star neighbors were detected by the first run of the \fBdaofind\fR task.
+
+
+.NH 3
+Computing the Final Psf Model
+
+.PP
+Once the psf star and psf star neighbors subtract out cleanly
+from the input image with the current psf model,
+a final psf model should be computed using an image from which all 
+the psf star neighbors but not the psf stars have been subtracted.
+To do this the user runs the \fBsubstar\fR task, setting the input photometry
+file to the final output of \fBnstar\fR, and the exclude file to the final
+psf star list written by the \fBpsf\fR task, and reruns \fBpsf\fR.
+An example of this procedure is shown below.
+
+.YS
+da> substar test test.nst.2 test.pst.2 default default
+da> psf test.sub.3 test.grp test.pst.2 test.psf.2 test.pst.3 \\ 
+    test.psg.3 inter-
+.YE
+
+After this step the user should have a good psf model and can proceed to do
+psf fitting photometry.
+.PP
+This step is not required for the test image as the single psf star
+neighbor is sufficiently far from the psf star to have a negligible
+effect on the computation of the psf model.
+
+.NH 3
+Visualizing the Psf Model with the Seepsf Task
+
+.PP
+The psf analytic function parameters are stored in the psf image header
+and the look-up table(s) in the psf image pixels. The look-up table(s) are
+subsampled by a factor of 2 with respect to the image, and cannot
+be used directly to visualize what the psf model looks like
+at the scale of the image. The task \fBseepsf\fR can  be used to
+do this conversion as shown below.
+
+.YS
+da> seepsf test.psf.3 psf3
+.YE
+
+The output image will contain a picture of what an ideal star of magnitude
+equal to the magnitude of the psf (see the psfmag keyword in the psf
+image header) should look like at the center of the image. 
+.PP
+In the case of a variable psf the appearance of the psf model can be examined at
+various 
+places in the image by specifying a position at which to compute the psf
+model.
+
+.YS
+da> seepsf test.psf.3 psf.13.8 xpsf=13 ypsf=8
+.YE
+
+.PP
+The total power in a variable psf should be constant
+as a function of position in the image even though its shape is
+different. The variable psf fitting code
+in DAOPHOT does perform flux conservation. Users can check this by using
+the \fBimstatistics\fR task to check that there is no net power in the
+look-up tables 2-3 or
+2-6 if the psf order is 1 or 2. Similarly they can use \fBseepsf\fR to compute
+the psf at various positions in the input image and \fBimstat\fR to check
+that the net power in the psf is constant over the image
+even though the shape of the
+psf is variable.
+
+.NH 3
+Problems Computing the Psf Model
+
+.PP
+Computing the psf model is the most crucial step in DAOPHOT.
+The \fBdaofind\fR and \fBphot\fR
+steps are usually straight-forward, and the principal fitting task
+\fBallstar\fR runs entirely in batch once started. However computing
+a good psf model requires user input.
+.PP
+This section suggests a few
+things to check if the computed psf model is not doing a good
+job of fitting and subtracting the psf or program stars. The user 
+should check:
+.IP [1]
+that the sky annulus chosen in the \fBphot\fR step is neither too
+close or too far from the stars. If the sky annulus is too close
+the computed skies will tend to be too high and the psf model
+will have too small an amplitude,
+producing false halos around the fitted and subtracted stars. If the sky annulus
+is too far away the computed sky value will not represent the sky under the
+psf star well, adding scatter to the photometry.
+.IP [2]
+the psf radius.
+If the stars appear to be well fit in the cores but have residual halos with a
+sharp inner edge around
+them then the psf radius may be too small. The psf radius needs to
+be big enough to give good subtractions for the brightest stars of interest.
+.IP [3]
+that the analytic component of the psf function is appropriate for
+the data.
+If the data is somewhat undersampled, fwhmpsf < 2.5 pixels,
+one of the Moffat functions may give a better fit to the data than
+the Gauss function. If that data is extremely undersampled an analytic
+function may do better than one involving look-up tables.
+.IP [4]
+the psf stars.
+One or more of the psf stars may not be stars, may be doubles, or contain
+bad data. Although psf does try to detect and down-weight bad data it may
+not be completely successful. Users need to examine the subtracted image
+for objects with bad residuals and for stars with large fitted chi values
+and eliminate them.
+.IP [5]
+for psf variability with position in the image. The true psf may be variable
+and inadequately fit by a constant psf model. The user should examine the
+residuals around the subtracted psf stars to see if there are patterns
+with position in the image and
+increase the order of the psf model if these are detected.
+Large fitted chi values may also be an indication of a poor psf model. 
+.IP [6]
+the distribution of the psf stars.
+If the psf is variable the user must ensure that the psf stars adequately
+cover the region of interest in the image. For example if there are no
+psf stars in a certain portion of the image the psf may not
+be well represented there.
+.IP [7]
+the data. If the psf is a higher order than quadratic \fBpsf\fR may
+not be able to model it adequately. The user should check the image data
+reduction history, and investigate any image combining, bad pixel and cosmic
+ray removal
+operations, etc., that may have fundamentally altered the data.
+The data should also be checked for linearity.
+.IP [8]
+the noise model. If the chi values for the fitted psf stars are unusually
+large or small, the effective readout noise and gain for the image
+may not be correct. The user should check that these values are
+being read from the image header correctly and that they are appropriate
+for the data.
+
+
+.NH 2
+Doing Psf Fitting Photometry with Peak, Nstar, or Allstar 
+
+.PP
+There are three psf fitting
+photometry tasks in DAOPHOT. The \fBpeak\fR task fits the current psf
+model to stars individually; the \fBnstar\fR task fits the psf model to
+stars in fixed stellar groups simultaneously; the \fBallstar\fR task
+groups and fits the psf model to stellar groups dynamically
+and subtracts the fitted stars from the input image.
+\fBAllstar\fR is the
+task of choice for the majority of users, but all three options are
+discussed in the following sections.
+
+.NH 3
+Fitting Single Stars with Peak
+
+.PP
+\fBPeak\fR is the simplest psf fitting task.
+\fBPeak\fR fits the psf model to the stars in the
+input photometry list individually. Because \fBpeak\fR cannot fit stars in
+groups as the \fBnstar\fR and \fBallstar\fR tasks do, and in
+uncrowded frames aperture photometry is often simpler and just as accurate,
+\fBpeak\fR has few unique functions.
+However \fBpeak\fR can be useful in cases where the user wishes to: 1) improve
+the signal to noise of faint stars by taking advantage of
+\fBpeak's\fR optimal weighting scheme, 2) do astrometry of single
+stars, 3) fit and remove single stars from the frame in order to 
+examine the underlying light distribution.
+
+.NH 4
+The Peak Algorithm
+
+.PP
+By default the \fBpeak\fR task performs the following functions:
+.IP [1]
+reads the task parameters, including the name of
+the input image, the input photometry file, the psf model, the 
+output photometry and rejections files, and the
+\fBdatapars\fR and \fBdaopars\fR algorithm parameter sets
+.IP [2]
+reads the id, x and y coordinates, magnitude, and sky value
+of a star from the input photometry file
+.IP [3]
+rejects the star if it has an undefined sky value, too
+few good data pixels to obtain a fit,  or is too faint
+.IP [4]
+extracts the image data within one fitting radius of each star and
+performs an optimally weighted non-linear least-squares fit of the psf model
+to the extracted data 
+.IP [5]
+rejects the star if its signal-to-noise ratio is too low
+or a unique solution cannot be found
+.IP [6]
+computes the best fit x, y, and magnitude for the star
+.IP [7]
+writes the id, new x and y coordinates, sky value, new magnitude, magnitude
+error, number of iterations required to fit the star, chi statistic,
+and sharpness
+statistic for the fitted star to the
+output photometry file, and the id, last computed x and y position,
+and sky value of the rejected star, to
+the rejections file
+.IP [8]
+repeats steps [2]-[7] for each star in the input photometry list
+
+.NH 4
+Running Peak 
+
+.PP
+A sample run of the \fBpeak\fR task is shown below. The user is prompted
+for all the input and output file names and asked to verify the
+critical parameters \fIrecenter\fR, \fIfitsky\fR, \fIpsfrad\fR, \fIfitrad\fR,
+\fIdatamin\fR, and \fIdatamax\fR.
+
+.YS
+da> peak test default default default default
+
+Recenter the stars (yes):
+        Recenter the stars: yes
+Refit the sky (no):
+        Refit the sky: no
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Star:     1 X:    40.97 Y:     4.01 Mag:    17.22 Sky:   100.74
+ FIT:  Star:     1 X:    41.01 Y:     4.03 Mag:    17.22 Sky =  100.74
+Star:     2 X:    23.06 Y:     7.03 Mag:    17.65 Sky:   100.33
+ FIT:  Star:     2 X:    23.02 Y:     7.05 Mag:    17.75 Sky =  100.33
+Star:     3 X:    18.02 Y:     7.96 Mag:    17.48 Sky:    99.40
+ FIT:  Star:     3 X:    17.98 Y:     8.00 Mag:    17.57 Sky =   99.40
+Star:     4 X:    25.99 Y:    22.01 Mag:    16.80 Sky:   101.34
+ FIT:  Star:     4 X:    25.99 Y:    22.01 Mag:    16.81 Sky =  101.34
+Star:     5 X:    35.98 Y:    22.00 Mag:    16.37 Sky:   101.12
+ FIT:  Star:     5 X:    35.97 Y:    22.02 Mag:    16.39 Sky =  101.12
+Star:     6 X:     8.02 Y:    22.97 Mag:    16.60 Sky:    98.89
+ FIT:  Star:     6 X:     8.02 Y:    22.97 Mag:    16.63 Sky =   98.89
+Star:     7 X:    30.97 Y:    25.01 Mag:    17.05 Sky:   101.29
+ FIT:  Star:     7 X:    30.96 Y:    25.04 Mag:    17.05 Sky =  101.29
+Star:     8 X:    28.96 Y:    33.92 Mag:    17.78 Sky:   100.62
+ FIT:  Star:     8 X:    28.96 Y:    33.93 Mag:    17.74 Sky =  100.62
+Star:     9 X:    35.98 Y:    42.03 Mag:    16.59 Sky:   101.04
+ FIT:  Star:     9 X:    35.98 Y:    42.01 Mag:    16.60 Sky =  101.04\fR
+.YE
+
+In this example the user chose to recenter the stars (almost always
+the best choice), and not to refit the sky (usually the best choice).
+Recentering should only be turned off if the initial centers in the input
+photometry file are known to be very accurate, e.g. they
+are derived from a better seeing
+image or one that has gone through some image restoration program.
+Users who elect to refit the sky,
+should realize that they will almost certainly need to increase
+the fitting radius to
+obtain a reasonable fitted sky value.  Increasing the fitting radius
+however may also increase the scatter caused by neighboring stars.
+.PP
+The fitted stars can be subtracted from the input image with the \fBsubstar\fR
+task as shown below.
+
+.YS
+da> substar test test.pk.1 "" default default\fR
+.YE
+
+.PP
+
+.NH 4
+The Peak Output
+
+.PP
+Peak writes the quantities: id, xcenter, ycenter, mag, merr, msky,
+niter, chi, sharp, pier, and perror to the output photometry file and the
+rejections file.
+.IP [1]
+Id is the id number of the star as read from the input photometry file.
+.IP [2]
+Xcenter and ycenter are the best fit position of the star.
+If the star was rejected xcenter and ycenter will be the computed
+values of x and
+y at the time it was rejected. If \fIrecenter\fR is "no", xcenter and ycenter
+will be the position of the star in the input photometry file.
+.IP [3]
+Mag and merr are the best fit magnitude and magnitude error respectively.
+The instrumental magnitude is computed relative to the
+magnitude assigned to the
+psf model.
+Mag and merr are set to INDEF if the star cannot be fit to the psf model.
+.IP [4]
+Msky is the sky value in the input photometry file if fitsky = "no",
+otherwise it is the
+fitted sky value. If the star is not fit for some reason,
+msky is the computed sky value
+at the time the star was rejected.
+.IP [5]
+Niter is the number of iterations it took to fit the star. If this
+number if equal to the \fBdaopars\fR parameter \fImaxiter\fR the user should be
+suspicious of the computed positions, magnitudes, and sky values.
+However as the convergence criteria are
+conservative the star may still be reasonably well fit.
+Niter is set to 0 if the star cannot be fit  to the psf model.
+.IP [6]
+Chi and sharp are measures of the goodness of fit and the shape
+of the star respectively. Chi should be ~ 1.0. If it is not then,
+either the object is not a single star, the noise model including
+one or more of the gain, readout noise, flat-fielding error, and
+interpolation error parameters for the image are incorrect, the
+psf model is a poor representation of the true psf, or the input
+image does not conform to the requirements of the DAOPHOT package.
+Sharp is a measure of the difference between the observed width  
+of the object and the width of the psf model. Stars should have a sharpness
+value ~ 0.0, resolved objects a sharpness of > 0.0, and cosmic rays and similar
+blemishes a sharpness of < 0. Chi and sharp are set to INDEF if the star
+cannot be fit to the psf model.
+.IP [7]
+Pier and perror are an integer error code and error string respectively.
+If no error was encountered during the fit  pier is 0 and perror is
+"No_error". Stars are rejected by the \fBpeak\fR task if 1) the sky value of
+the star is INDEF 2) there are too few good data pixels to fit the star
+3) the fitting matrix is singular meaning a unique solution could not
+be found 4) the star is too faint, i.e. its signal / noise < 2.0. A fifth
+condition, the solution did not converge by \fImaxiter\fR iterations, is not
+used to reject the star, although users should be suspicious of a star
+for which niter = \fImaxiter\fR.
+
+
+.NH 3 
+Fitting Stars with Group, Grpselect, Nstar and Substar
+
+.PP
+Stars can be fit simultaneously in fixed groups using the
+\fBnstar\fR task.
+This psf fitting technique requires grouping the stars
+into physically meaningful
+associations with the \fBgroup\fR and/or the \fBgrpselect\fR tasks,
+fitting the stars in each group simultaneously with the \fBnstar\fR task,
+and subtracting
+the fitted stars from the image with the \fBsubstar\fR task.
+\fBNstar\fR is the task of choice when the user wishes to explicitly 
+control the grouping process or fit stars in a small number of 
+widely separated groups efficiently.
+\fBNstar\fR is most commonly used to fit the psf model to the psf stars and
+their neighbors.
+
+.NH 4
+The Group and Nstar Algorithms
+
+.PP
+By default the \fBgroup\fR task performs the following steps:
+
+.IP [1]
+reads the task parameters, including the name of the input image, the input
+photometry file, the psf model, the output photometry
+file, and the \fBdatapars\fR and \fBdaopars\fR 
+algorithm parameter sets
+.IP [2]
+reads the ids, x and y coordinates, magnitudes, and sky values
+of up to \fImaxnstar\fR stars in the input photometry file, computes an
+approximate magnitude for the stars with INDEF magnitudes, and sorts
+the stars in increasing order of y 
+.IP [3]
+finds all the stars which are within \fIpsfrad\fR + \fIfitrad\fR + 1 pixels
+of a given star, evaluates the psf of the brighter star at a distance
+of \fIfitrad\fR pixels from
+the fainter, and if this value is larger than \fIcritovlap\fR
+times the expected
+error per pixel,  or the two stars are within  \fIfitrad\fR + 1 pixels of
+each other, adds the star to the group
+.IP [4]
+writes the group and star ids, x and y coordinates, magnitudes and sky values
+for all the groups, to the output group photometry file.
+
+.PP
+By default the \fBnstar\fR task performs the following steps:
+
+.IP [1]
+reads the task parameters including the name of
+the input image, the input group file, the psf image, the 
+output group photometry and rejections files and the \fBdatapars\fR and
+\fBdaopars\fR algorithm parameter sets
+.IP [2]
+reads the group and star ids, x and y coordinates, magnitudes, and
+sky values for all the stars in a group from the input group photometry
+file
+.IP [3]
+extracts the data within psfrad + fitrad pixels
+around the group and
+performs a weighted least-squares fit of the psf model to the extracted
+data
+.IP [4]
+rejects stars which have an undefined sky value, which are too faint (more
+than 12.5 magnitudes fainter than the psf), which are
+too noisy (faintest star in the group less than a 1.0, 1.5, or 2.0
+sigma detection after 5, 10, and 15 iterations or convergence respectively),
+for which there are too few good
+pixels to compute a fit, for which a unique solution cannot
+be found, which have merged with another star (fainter star < 0.37 *
+fwhmpsf from a brighter star in the group), which are both too
+noisy and too
+close to a brighter star (a star is between .37 and 1.0 fwhm of
+a brighter star and is a 1.0, 1.5, or 2.0 sigma detection before
+iterations 5, 10, and 15 respectively), or which are in a group
+too large (> than the value of the \fImaxgroup\fR parameter) to be reduced.
+.IP [5]
+estimates new x and y coordinates and magnitudes for each star in
+the group
+.IP [6]
+iterates until all the stars in the group satisfy the convergence criteria
+backing up the iteration counter by 1 each time a star is rejected from
+the group to allow the remaining stars time to settle into a new fit
+.IP [7]
+writes the star and group ids, new x and y coordinates, sky values,
+new magnitudes and magnitude errors, chi and sharpness statistics
+for the fitted stars to the
+output group photometry and rejections files
+.IP [8]
+repeats steps [2]-[7] for each group in the input group photometry file.
+
+.NH 4
+Running Group, Grpselect, and Nstar
+
+.PP
+Before \fBnstar\fR can be run the stars must be grouped with the
+\fBgroup\fR task as shown below.
+
+.YS
+da> group test default default default
+
+Psf radius in scale units (5.):
+       	New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+       	New fitting radius: 3. scale units 3. pixels
+Critical overlap in stdevs per pixel (1.): .2
+       	New critical overlap: 0.2 stdevs per pixel
+Minimum good data value (50.) (CR or value):
+       	New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+       	New maximum good data value: 24500. counts
+
+Size of     Number of
+group       groups
+
+1            4
+2            1
+3            1
+
+Total of 9 stars in 6 groups\fR
+.YE
+
+The critical overlap parameter \fIcritovlap\fR determines the degree
+to which crowding and/or random photometric errors are expected/allowed
+to influence the photometry. If the default value of 1 is required to group
+all the stars into associations of <= the current value of 
+\fImaxgroup\fR stars, then unavoidable random photometric errors and
+crowding errors will affect the photometry about equally.
+If a critical
+overlap much greater than 1 is required, then crowding errors will
+dominate the random photometric errors. If a critical overlap
+much less than 1 does the
+job then
+unavoidable random photometric errors will dominate, and crowding errors
+are relatively insignificant.
+In 
+the previous example the user chose to set \fIcritovlap\fR to a value
+much smaller
+than 1 to test whether random photometric rather than crowding errors will
+dominate the photometry.
+.PP
+If the first run of \fBgroup\fR separates all the stars into groups of less than
+60 all is well and the user can proceed to the \fBnstar\fR task. Otherwise
+the \fBgrpselect\fR task must be used to select out the larger groups
+and subdivide them as shown in the following example.
+
+.YS
+da> grpselect test.grp.1 small.grp 1 60
+.YE
+
+.IP ...
+First separate out the small groups.
+.LP
+
+.YS
+da> grpselect test.grp.1 big.grp.1 61 10000
+.YE
+
+.IP ...
+Next separate out the large groups.
+.LP
+
+.YS
+da> group test big.grp.1 default big.grp.2 critovlap=1.0
+.YE
+
+.IP ...
+Rerun the group task on the large group file with a bigger value
+of critovlap.
+.LP
+
+.YS
+da> pconcat small.grp,big.grp.2 all.grp
+.YE
+
+.IP ...
+Finally concatenate all the new group files together.
+.LP
+This step is not required for the test image since there are only a
+few stars and the field is not very crowded.
+
+.PP
+Run \fBnstar\fR on the grouped photometry file and \fBsubstar\fR on the
+fitted photometry file.
+
+.YS
+da> nstar test default default default default
+
+Recenter the stars (yes):
+        Recenter the stars: yes
+Refit the sky (no):
+        Refit the sky: no
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Maximum group size in number of stars (60):
+        New maximum group size: 60 stars
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+Group:    1 contains  1 stars
+        ID:     1  XCEN:    41.01  YCEN:     4.03  MAG:    17.22
+Group:    2 contains  2 stars
+        ID:     2  XCEN:    23.04  YCEN:     7.07  MAG:    17.75
+        ID:     3  XCEN:    17.99  YCEN:     8.00  MAG:    17.57
+Group:    3 contains  3 stars
+        ID:     5  XCEN:    35.98  YCEN:    22.01  MAG:    16.39
+        ID:     7  XCEN:    30.99  YCEN:    25.04  MAG:    17.04
+        ID:     4  XCEN:    26.00  YCEN:    22.01  MAG:    16.80
+Group:    4 contains  1 stars
+        ID:     6  XCEN:     8.02  YCEN:    22.97  MAG:    16.63
+Group:    5 contains  1 stars
+        ID:     8  XCEN:    28.96  YCEN:    33.93  MAG:    17.74
+Group:    6 contains  1 stars
+        ID:     9  XCEN:    35.98  YCEN:    42.01  MAG:    16.60
+
+da> substar test default "" default default
+.YE
+
+The parameter \fImaxgroup\fR specifies the
+maximum number of stars that \fBnstar\fR will
+fit simultaneously. The default value of 60
+is a conservative number based
+on the observed numerical behavior of the matrix inversion routines.
+.PP
+For most crowded field photometry applications it is simpler and easier
+to use the automated \fBallstar\fR task.
+
+.NH 4
+The Nstar Output
+
+.PP
+By default \fBnstar\fR writes the quantities: id, group, xcenter, ycenter,
+mag, merr, msky, niter, chi, sharp, pier, and perror to the output
+photometry and rejections files.
+.IP [1]
+Id and group are the star and group id numbers in the input group
+photometry file.
+.IP [2]
+Xcenter and ycenter are the best fit coordinates of the star.
+If the star was rejected xcenter and ycenter will be the best fit
+values of x and y at the time it was rejected. If \fIrecenter\fR is "no"
+xcenter and ycenter will be the position of the star in the input
+group photometry file.
+.IP [3]
+Mag and merr are the best fit magnitude and magnitude error respectively.
+The instrumental magnitude is computed relative to the magnitude
+of the psf model.
+Mag and merr are set to INDEF if the star cannot be fit to the
+psf model.
+.IP [4]
+Msky is always the individual sky for the star in the input photometry
+file regardless of whether  fitsky is "no" or "yes". In the former
+case the actual value of the sky used in \fBnstar\fR is
+the mean of all the sky values for all the stars in the group. In the
+latter case it is a fitted parameter.
+.IP [5]
+Niter is the number of iterations it took to fit the star. If \fBniter\fR
+is equal to the parameter \fImaxiter\fR the user should be
+suspicious of the result. However since the convergence criteria are
+quite tightly constrained the result may still be reasonable.
+Niter is set to 0 if the star cannot be fit to the psf model.
+.IP [6]
+Chi and sharp are measures of the goodness of fit and the star's shape
+respectively. Chi should be ~ 1.0. If it is not then
+either the object is not a single star,
+the noise model including the ccd gain and readout noise, the flat
+fielding error and the interpolation error parameters assumed for the image are
+not correct, the
+psf model is a poor representation of the true psf, or the input
+image does not conform to the requirements of the DAOPHOT package.
+Sharp is a measure of the difference between the observed width  
+of the object and the width of the psf model. Stars should have sharpness
+values \(~= 0.0, resolved objects sharpness values > 0.0, and cosmic rays
+and similar blemishes sharpnesses values < 0.0. Chi and sharp are set to INDEF
+if the star cannot be fit to the psf model.
+.IP [7]
+Pier and perror are an integer error code and error string respectively.
+If no error was encountered during the fit, pier is 0 and perror is
+"No_error".
+
+.NH 3
+Fitting Stars With Allstar
+
+.PP
+\fBAllstar\fR groups, fits
+and subtracts the fitted stars from the input image without intervention
+by the user.
+Because the grouping process is dynamic and the best
+fit stars are fit and subtracted first, fewer weak stars
+and noise spikes migrate to the position of stronger stars in \fBallstar\fR
+than is the case with \fBnstar\fR. \fBAllstar\fR replaces the functionality
+of the tasks \fBgroup\fR, \fBgrpselect\fR, \fBnstar\fR, and \fBallstar\fR.
+\fBAllstar\fR is the task of choice for doing crowded
+field photometry with DAOPHOT.
+
+.NH 4
+The Allstar Algorithm
+
+.PP
+By default the \fBallstar\fR task performs the following steps:
+.IP [1]
+reads the task parameters including the name of the input image, the input
+photometry file, the psf
+model, the output photometry and rejections files, the output
+subtracted image, and the \fBdatapars\fR and \fBdaopars\fR algorithm
+parameter sets
+sets
+.IP [2]
+reads the ids, x and y coordinates, magnitudes, and sky values for
+the first \fImaxnstars\fR stars in the input photometry file, rejecting
+at the start stars which have undefined sky values or which
+are too close to another star
+.IP [3]
+reads the original image into a working array and initializes
+two scratch arrays containing 1) the noise model and 2) the residuals from
+the current best fit for all the stars
+.IP [4]
+at the beginning of each iteration:
+1) groups the stars into physical associations that contain fewer
+than \fImaxgroup\fR
+stars, regrouping as necessary until all the groups are less than
+\fImaxgroup\fR or until the group is too dense to reduce, 2)
+subtracts the current best fit for all the stars that are still unfit from
+the working copy of the image and stores the results in the residuals
+array 3) initializes the weight array for all the unfitted stars
+.IP [5]
+during each iteration: 1) extracts the data within fitrad pixels
+around each star in each group from the residuals image, 2)
+performs a weighted non-linear least-squares fit of the psf model to
+the extracted 
+data, ignoring bad pixels and down-weighting pixels that deviate too
+far from the psf model, and 3) computes new x and y coordinates and
+magnitudes for each star in each group
+.IP [6]
+after the fourth iteration 1) writes the id, new x and y coordinates,
+sky value, new magnitude and magnitude error, number of iterations
+required to fit the star, and the chi and sharpness statistic of stars
+which meet the convergence criteria, to
+the output photometry file, 2) subtracts the fitted star permanently from 
+the working copy of the image, 3) updates the
+noise model in the weight array, 4) and eliminates the star from
+the active star list
+.IP [7]
+after the fourth iteration rejects stars which: 1) are too faint
+(more than 12.5 magnitudes fainter
+than the psf model), 2) have too low a signal-to-
+noise ratio (1.0, 1.5 and 2.0 sigma detection after 5, 10, and 15 iterations
+respectively), 3) have too few good
+pixels to compute a fit, 4) do not permit a unique solution, 
+5) have merged with another star (star is
+< 0.37 * fwhmpsf from a brighter star), 6) are both too noisy
+and too close to a neighbor star (star is between 0.37 and 1.0 * fwhmpsf from
+a brighter star in the group and is a 1.0, 1.5, or 2.0 sigma
+detection before iterations 5, 10, and 15 respectively),
+or 7) are part of a group too dense to be reduced.
+.IP [8]
+writes out the final version of the work array into the output subtracted
+image
+
+.NH 4
+Running Allstar
+
+.PP
+\fBAllstar\fR is run as shown below.
+
+.YS
+da> allstar test default default default default default
+
+Recenter the stars (yes):
+        Recenter the stars: yes
+Refit the sky (no):
+        Refit the sky: no
+Psf radius in scale units (5.):
+        New psf radius: 5. scale units 5. pixels
+Fitting radius in scale units (3.):
+        New fitting radius: 3. scale units 3. pixels
+Maximum group size in number of stars (60):
+        New maximum group size: 60 stars
+Minimum good data value (50.) (CR or value):
+        New minimum good data value: 50. counts
+Maximum good data value (24500.) (CR or value):
+        New maximum good data value: 24500. counts
+
+NITER = 1
+NITER = 2
+NITER = 3
+NITER = 4
+FITTING:   ID:     1  XCEN:    41.01  YCEN:     4.03  MAG:    17.22
+FITTING:   ID:     4  XCEN:    26.00  YCEN:    22.01  MAG:    16.80
+FITTING:   ID:     7  XCEN:    30.99  YCEN:    25.04  MAG:    17.04
+FITTING:   ID:     6  XCEN:     8.02  YCEN:    22.97  MAG:    16.63
+FITTING:   ID:     8  XCEN:    28.96  YCEN:    33.93  MAG:    17.74
+FITTING:   ID:     9  XCEN:    35.98  YCEN:    42.01  MAG:    16.60
+NITER = 5
+FITTING:   ID:     2  XCEN:    23.04  YCEN:     7.05  MAG:    17.75
+FITTING:   ID:     3  XCEN:    18.00  YCEN:     7.99  MAG:    17.56
+FITTING:   ID:     5  XCEN:    35.98  YCEN:    22.01  MAG:    16.39
+.YE
+
+.PP
+Users can easily run \fBallstar\fR as a background job as shown below.
+
+.YS
+da> allstar test default default default default default verify-  \\ 
+    >& allstar.out &
+.YE
+
+.NH 4
+The Allstar Output
+
+.PP
+\fBAllstar\fR writes the following
+quantities: id, xcenter, ycenter, mag, merr, msky, niter, chi, sharp,
+pier, and perror to the output photometry and 
+rejections files.
+.IP [1]
+Id is the id number of the star as read from the input photometry file.
+.IP [2]
+Xcenter and ycenter are the best fit position of the star.
+If the star was rejected xcenter and ycenter will be the computed
+values of x and
+y at the time it was rejected. If \fIrecenter\fR is "no", xcenter and ycenter
+will be the position of the star in the input photometry file.
+.IP [3]
+Mag and merr are the best fit magnitude and magnitude error respectively.
+The instrumental magnitude is computed relative to the magnitude of the
+psf model.
+Mag and merr are set to INDEF if the star cannot be fit to the psf model.
+.IP [4]
+Msky is the sky value in the input photometry file if \fIfitsky\fR = "no",
+otherwise it is the
+recomputed sky value. If \fIfitsky\fR is "yes" the sky is recomputed
+every third iteration after the current best fit for the star is
+subtracted from the image data. The new sky value is set to the 
+average of 40% of the sky pixels, centered on the median sky value,
+which are inside the sky
+annulus defined by the parameters \fIsannulus\fR and
+\fIwsannulus\fR. The sky value is not recomputed
+if there are fewer than 100 sky pixels in the specified sky annulus
+even if \fIfitsky\fR is "yes".
+If the star is not fit for some reason,
+msky is the sky value at the time the star was rejected.
+.IP [5]
+Niter is the number of iterations it took to fit the star. If this
+number is equal to \fImaxiter\fR the user should be
+suspicious of the result. However as the convergence criteria are
+conservative the star may still be reasonably well fit.
+Niter is set to 0 if the star cannot be fit to the psf model.
+.IP [6]
+Chi and sharp are measures of the goodness of fit and the shape
+respectively. Chi should be ~ 1.0. If it is not, then
+either the object is not a single star, the noise model including
+one or more of the gain, readout noise, flat-fielding error, and
+interpolation error parameters for the image are incorrect, the
+psf model is a poor representation of the true psf, or the input
+image does not conform to the requirements of the DAOPHOT package.
+Sharpness is a measure of the difference between the observed width  
+of the object and the width of the psf model. Stars should have a sharpness
+value ~ 0.0, resolved objects a sharpness of > 0.0, and cosmic rays and similar
+blemishes a sharpness of < 0. Chi and sharp are set to INDEF if the star
+cannot be fit for some reason.
+.IP [7]
+Pier and perror are an integer error code and error string respectively.
+If no error was encountered during the fit,  pier is 0 and perror is
+"No_error".
+
+.NH 2
+Examining the Output Photometry Files
+
+.PP
+The identical tools can be used to examine the output of the \fBpeak\fR,
+\fBnstar\fR, and \fBallstar\fR tasks.  Some examples using the output of
+\fBallstar\fR are shown below.
+.PP
+The following command produces a plot of magnitude error versus magnitude.
+
+.YS
+da> pdump test.als.1 mag,merr yes | graph point+
+.YE
+
+The following command produces a plot of chi versus magnitude.
+
+.YS
+da> pdump test.als.1 mag,chi yes | graph STDIN point+
+.YE
+
+The following command produces a plot of chi versus sharpness.
+
+.YS
+da> pdump test.als.1 sharp,chi yes | graph STDIN point+
+.YE
+
+The output photometry file can also be examined interactively
+with the \fBpexamine\fR task and the displayed subtracted image.
+Note that the fitted stars are marked in green and the rejected stars
+are marked in red on the display.
+
+.YS
+da> display test.sub.1 1 fi+
+
+da> pdump test.als.1 xcenter,ycenter yes | tvmark 1 STDIN col=205
+
+da> pdump test.arj.1 xcenter,ycenter yes | tvmark 1 STDIN col=204
+
+da> pexamine test.als.1 "" test.sub.1
+.YE
+
+.IP ...
+A plot of magnitude error versus magnitude appears on the screen.
+.IP ...
+The user moves to a discrepant point in the graph and types o to get a
+listing of the results for the star, r to get a radial profile plot
+around the subtracted star, and concludes on the basis of the plots
+that the bad chi value is due to the star being a close double.
+.IP ...
+The user types i to switch to image cursor mode,
+moves to several other stars with poor subtractions and types
+s to see a surface plot of the residuals.
+.IP ...
+The user types q to quit.
+
+.NH 2
+Problems with the Photometry
+
+.PP
+Bad chi values in, and poor
+subtractions of, \fBpeak\fR, \fBnstar\fR or \fBallstar\fR
+photometry can usually be traced to: 1) a psf model which was
+poorly determined in the first place, e.g. poor choice of
+parameters, bad choice of
+psf stars or too few stars used for determining a good variable psf model,
+2) data reduction problems e.g. the mean sky value was subtracted from the
+image, the image statistics have been altered,  or cosmic ray removal
+clipped the tops of the stars, to give a few of many examples,
+or 3) the properties of the image, e.g. non-linearity, a psf which has
+very high order variability or very undersampled data, make computation
+of a good psf model difficult or impossible.
+.PP
+Bad chi values can also be caused by incorrect
+values of gain and readout noise or by a data reduction operation
+which has significantly affected the image statistics.
+.PP
+Poor sky fitting can also cause scatter
+in the photometry. Users should carefully check the position of the 
+sky annulus used in \fBphot\fR if they are seeing poor subtractions.
+If the images have a rapidly varying
+background due,  due for example to nebulosity, it might be useful to
+check out the alternate sky fitting routines, median or centroid, in the
+\fBphot\fR task. The refit sky option in \fBpeak\fR and \fBnstar\fR
+should be exercised with caution
+since a larger fitting radius is often required
+to get a reasonable sky fit, than is required to get good positions
+and magnitudes, and this in turn can cause more scatter
+in the photometry due to the influence of neighbors. On the other
+hand the refit sky option in \fBallstar\fR can often significantly
+reduce scatter in very crowded regions since it can use data closer to
+or even underneath (!) the star to improve the sky estimate.
+Users who use this option  must remember to set the inner
+radius of the \fBallstar\fR sky annulus to avoid the inner stellar
+core region where there is a lot of noise in the subtraction.
+.PP
+After running \fBsubstar\fR on a file produced by the \fBpeak\fR
+task, users will sometimes see large holes in the data at the
+position of some subtracted
+stars. This is usually caused by fainter stars (which are fit individually)
+migrating to the position of a brighter nearby star and then being
+subtracted out twice by \fBsubstar\fR. Keeping the fitting radius small
+will help minimize this problem, but if it is frequent and the frame
+is somewhat crowded, the user should run \fBnstar\fR or \fBallstar\fR
+instead of \fBpeak\fR. 
+.PP
+A similar problem can be caused by users running \fBdaofind\fR
+with a very low threshold and detecting a lot of noise spikes, which
+then migrate to the positions of brighter stars
+and cause scatter and holes in the subtracted \fBpeak\fR photometry,
+or attach themselves to noise spikes in the stellar profiles and cause
+scatter and holes in the subtracted \fBnstar\fR photometry. 
+Similar problems can affect \fBallstar\fR photometry but to a much
+lesser degree since the
+stars are grouped dynamically and subtracted from the input data as
+they are fit. For all three photometry tasks spurious detections 
+can consume a lot of excess computer time because the stellar groups
+become much larger.
+
+.NH 2
+Detecting Stars Missed By Daofind
+
+.PP
+In very crowded fields many new stars, missed by the first run of \fBdaofind\fR,
+will be detected after
+the first run of \fBpeak\fR+\fBsubstar\fR, \fBnstar\fR+\fBsubstar\fR,
+or \fBallstar\fR. If there
+are many "missed" stars
+\fBdaofind\fR should be run on the subtracted image after increasing the
+\fIthreshold\fR parameter to avoid detecting the residuals
+of previously subtracted stars. If there
+are only a few such stars they can be "detected" by creating a coordinate
+file using the subtracted image and \fBtvmark\fR in interactive mode.
+Examples of both techniques are shown below.
+
+.YS
+da> daofind test.sub.1 newstars.coo threshold=5.0
+.YE
+
+or
+
+.YS
+da> display test.sub.1 1 fi+
+
+da> pdump test.als.1 xcen,ycen yes | tvmark 1 STDIN col=204
+
+da> tvmark 1 newstars.coo inter+
+.YE
+
+.IP ...
+Move cursor to missing stars and tap the \fBa\fR key to append them to the
+output coordinate file.
+
+.NH 2
+Initializing the Missing Star Photometry with Phot
+
+.PP
+The next step is to get initial photometry for the "missing"
+stars. The simplest way is
+to run \fBphot\fR on the original image using the coordinate list created
+by \fBdaofind\fR or \fBtvmark\fR, and the same algorithm parameters as
+were used
+in the first run of \fBphot\fR. It is also possible to use \fBphot\fR directly
+in interactive mode to create a photometry file of missed stars. Both
+options are shown below.
+
+.YS
+da> phot test newstars.coo newstars.mag
+.YE
+
+or
+
+.YS
+da> display test.sub.1 1 fi+
+
+da> pdump test.als.1 xcen,ycen yes | tvmark 1 STDIN col=204
+
+da> phot test "" newstars.mag centroid=calgorithm inter+
+.YE
+
+.IP ...
+Point the cursor to the missing stars and tap \fBspacebar\fR.
+.LP
+
+Note that if the stars are or were marked with the cursor, the user must
+turn centroiding on in order to center them correctly.
+
+.NH 2
+Merging Photometry Files with Pfmerge
+
+.PP
+The photometry file containing the aperture photometry for the new stars
+can be combined with the best psf fitting photometry already computed
+by the \fBnstar\fR or \fBallstar\fR tasks 
+for the original star list, using the task \fBpfmerge\fR as shown below.
+The \fBprenumber\fR task ensures that the new stars all have unique ids.
+
+.YS
+da> pfmerge test.als.1,newstars.mag newstars.als.1
+da> prenumber newstars.als.1\fR
+.YE
+
+
+.NH 2
+Refitting the Stars with Allstar
+
+.PP
+After the photometry  files have been merged a final run of \fBallstar\fR or
+\fBgroup\fR+\fBnstar\fR+\fBsubstar\fR on the combined file in order
+to compute accurate magnitudes for the new stars should be made
+as shown below.
+
+.YS
+da> allstar test newstars.als.1 default default default default
+.YE
+
+.NH 2
+Examining the Subtracted Image
+
+.PP
+The user should search the subtracted image for any remaining unfit
+stars and perform another iteration of \fBdaofind\fR, \fBphot\fR,
+\fBpfmerge\fR and \fBallstar\fR to computed fitted magnitudes for
+the new objects.
+
+.NH 2
+Computing an Aperture Correction
+
+.PP
+The aperture correction is the number which must be added to the
+fitted instrumental magnitudes computed by the \fBpeak\fR,
+\fBnstar\fR, or \fBallstar\fR tasks to produce the total instrumental
+magnitude.
+.PP
+In order to compute aperture corrections for an image with a constant psf model
+the user must:
+.IP [1]
+identify several bright isolated stars in the input image or subtract
+all the neighbors from around several bright stars such as the
+psf stars using the current psf model and the \fBsubstar\fR task
+.IP [2]
+using a minimum aperture radius equal to the one used in \fBphot\fR to compute
+initial aperture photometry for all the crowded field stars,
+and a maximum aperture radius equal to the one through which
+the instrumental magnitudes of the standard stars were or will be measured,
+use the \fBphot\fR task to
+do multi-aperture photometry of the stars identified in [1] through at
+least five apertures
+.IP [3]
+run the \fBmkapfile\fR task in the PHOTCAL package on the aperture
+photometry file produced in 2, to determine the aperture
+correction for the image as shown below
+.LP
+
+.YS
+da> phot test "" test.apmags calg=centroid aperture="3,3.5,4.0,4.5 5.0"
+.YE
+
+.IP ...
+Do multi-aperture photometry of the selected stars.
+.LP
+
+.YS
+da> mkapfile test.apmags 5 test.apcors  
+.YE
+
+.IP ...
+Compute the aperture correction between apertures 1 and 5.
+
+.PP
+To compute compute aperture corrections for an image with a variable psf model
+the user must:
+.IP [1]
+identify several bright isolated stars in the input image or subtract
+all the neighbors from around several bright stars such as the
+psf stars using the current psf model and the \fBsubstar\fR task
+.IP [2]
+using a photometry aperture equal to the one through which
+the magnitudes of the standard stars were or will be measured,
+use the \fBphot\fR task to
+do aperture photometry of the stars identified in [1]
+.IP [3]
+extract the fitted magnitudes for these stars from existing \fBnstar\fR
+or \fBallstar\fR photometry or recompute them using the
+\fBnstar\fR or \fBallstar\fR tasks and the current psf model
+.IP [4]
+set the aperture correction to the mean difference between the fitted
+magnitudes computed in [3] and the aperture photometry magnitudes
+computed through the large aperture in [2]
+
+.NH
+References
+
+.LP
+.nf
+Stetson, P. B. 1987 Pub .A.S.P., \fB99\fR, 191
+Stetson, P. B., Davis, L.E. and Crabtree, D.B. 1989, in
+    \fICCDs in Astronomy\fR, G.H. Jacoby, San Francisco: Astronomical
+    Society of the Pacific, 289
+Stetson, P. B. 19 Pub .A.S.P., \fB102\fR, 932
+Stetson, P.B, 1992, \fIUser's Manual for DAOPHOT II\fR
+Stetson, P. B. 1992  in \fIAstronomical Data Analysis Software and Systems I\fR,
+    D.M. Worall, C. Biemesderfer, and J. Barnes, San Francisco: Astronomical
+    Society of the Pacific, 297
+.fi
+
+.NH
+Appendices
+
+.NH 2
+The Instrumental Magnitude Scale
+
+.PP
+The instrumental magnitude scale is set by the magnitude assigned
+to the psf model, the quantity \fIpsfmag\fR stored in the psf image header.
+Psfmag is the magnitude of the first psf star in the input photometry
+file, usually but not always the file written by the \fBphot\fR task.
+If magnitudes were measured through more than one aperture
+in \fBphot\fR, the magnitude used will be the
+magnitude through the smallest aperture. 
+
+.NH 2
+The Analytic Psf Models
+
+.PP
+The functional forms of the currently supported analytic psf models
+are listed below.
+The quantity A is a normalization factor. The  Pn  are
+the parameters which are fit during the psf modeling process.
+
+.nf
+	z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2
+	gauss = A * exp (-0.5 * z)
+
+	z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p3
+	moffat15 = A / (1 + z) ** 1.5
+	moffat25 = A / (1 + z) ** 2.5
+
+	z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p3
+	lorentz = A / (1.0 + z)
+
+	z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2
+	e = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p4
+	penny1 = A * ((1 - p3) / (1.0 + z) + p3 * exp (-0.693*e))
+
+	z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + p5 * x * y
+	e = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p4
+	penny2 = A * ((1 - p3) / (1.0 + z) + p3 * exp (-0.693*e))
+.fi
+
+.NH 2
+The Error Model
+
+.PP
+The  predicted errors in the the DAOPHOT photometry are computed per
+pixel as shown below, where terms 1, 2, 3, and 4 represent  the  readout
+noise,   the   poisson  noise,  the  flat-fielding  error,  and  the
+interpolation error respectively. The quantities  readnoise,  epadu,
+I,  M,  p1,  and  p2 are the effective readout noise in electrons, the
+effective gain in
+electrons per ADU,  the  pixel  intensity  in  ADU,  the  PSF  model
+intensity in ADU, the FWHM in x in pixels, and the FWHM in y in pixels.
+
+.nf
+	error = sqrt (term1 + term2 + term3 + term4)  (ADU)
+	term1 = (readnoise / epadu) ** 2
+	term2 = I / epadu
+	term3 = (.01 * flaterr * I) ** 2
+	term4 = (.01 * proferr * M / p1 / p2) ** 2
+.fi
+
+
+.NH 2
+The Radial Weighting Function
+
+.PP
+The  radial weighting function employed by all the psf fitting tasks
+is shown below, where dx and dy are  the  distance  of  the  pixel
+in question from the centroid of the star being fit.
+
+.nf
+	wtr = 5.0 / (5.0 + rsq / (1.0 - rsq))
+	rsq = (dx ** 2 + dy ** 2) / fitrad ** 2
+.fi
+
+.NH 2
+Total Weights
+
+.PP
+The  total weight  assigned  each  pixel  in  the  fit  is the
+following.
+
+.nf
+	wtp = wtr / error ** 2
+.fi
+
+.NH 2
+Bad Data Detection
+
+.PP
+Pixels less than the good data minimum \fIdatamax\fR or greater than
+the good data maximum  \fIdatamax\fR are rejected immediately from the
+fit.
+.PP
+After a few iterations and if clipexp > 0, a clipping scheme to
+reject bad data is enabled. The weights of the pixels are
+recomputed  as follows. Pixels having a residual of cliprange sigma
+will have their weight reduced by half.
+
+.nf
+	wt = wtp / (1.0 + (residual / error / chiold /
+             cliprange) ** clipexp)
+.fi
+
+.NH 2
+Stellar Mergers
+
+.PP
+In order for two stars to merge during the course of the psf fitting
+process either their separation must be < 0.37 * FWHM of the psf model,
+or their separation must be > 0.37 * FWHM but < 1.0 * FWHM of the
+psf model and the signal-to-noise ratio of the fainter  is less than 1.0, 1.5,
+or 2.0 after iterations 4, 9, and 14 respectively.
+
+.NH 2
+Faint Stars
+
+.PP
+Stars are considered to be too faint if they are more than 12.5
+magnitudes fainter than the psf, or if after a certain number of iterations,
+they have a signal-to-noise ratio less than 2.0.