These directions are for v.1 and are provided until v.2 directions become available. Finder v.2 requires IRAF v2.11beta2 or later. Contact seaman@noao.edu for assistance. _------ Retrieving and using the prototype finder package ------------------------------------------------- The package is contained in a compressed tar file in our anonymous ftp area. % cd [some convenient root directory] % mkdir finder % cd finder ftp noao.edu login: anonymous password: [your email address] ftp> cd misc ftp> binary ftp> get finder.tar.Z ftp> quit % uncompress finder.tar.Z % tar -xpf finder.tar Edit these lines into your loginuser.cl file or into the IRAF hlib$extern.pkg file: set finder = [convenient root directory]/finder/ <- note trailing "/" task finder.pkg = finder$finder.cl The package has been prebuilt for sparc/SunOS. If you need to rebuild for another architecture, contact iraf@noao.edu. The help files are in finder$doc. You can print them out (such as they are) by typing commands like `help finder$doc/tfinder.hlp file+' or `help finder$doc/tfinder.hlp file+ | lprint'. Phil Massey jotted down some notes about using the IRAF finder package, specifically to support the new generation of multi-object spectrographs, but the notes should apply for general purpose use, also. ----------- From massey@tofu Tue Jan 21 10:57:58 1992 From: massey@tofu (Philip Massey x271) To: seaman@ursa Subject: Re: finder intro Here it is! /data/massey/docs/tfinder.doc A GUIDE TO THE PROTOTYPE IRAF/FINDER ROUTINES FOR NESSIE/HYDRA USERS (a rough draft in 12 acts) ---Phil Massey June 18 1991 slightly revised July 2, 1991 completely revised Aug 31, 1991 after Rob's fixes. Hidden away in the "local" package at NOAO is an extremely useful set of routines written by Rob Seaman. These routines allow you to determine accurate positions for objects on CCD frames using the Space Telescope Guide Star Catalogue as reference stars. Although this package is a prototype, and has a few inconvenient features, it is doubtless a zillion times better than whatever you are currently using for producing coordinates for NESSIE/HYDRA. The purpose of this document is to guide the NESSIE/HYDRA user through its intricacies. It assumes you have a good grasp of basic IRAF, and that you are running on some SUN consol that is on good speaking terms with ORION. For each frame you will need a list of accurate x and y positions for your unknowns. You will also need to have some id number in the third column or the package will get hopelessly confused later on. The file needs to be a simple text file, containing x and y in the first two columns and an integer ID in the third column. If you are working from the output of PHOTCAL, say, you will need to (a) edit out the comment lines, and (b) use "fields" in the "proto" package to extract the relevant x and y columns. You will then have to edit in some integer id's. Alternatively, you can use tvmark to generate a "starting" list of x and y positions which you can then center using the "finder" routines. Be warned, however, that sky positions will vanish when you go to center your other objects! Personally, I think it would be better to use "center" in the apphot package to first center on all your objects, use txdump to put the x and y positions into a separate file, and then use tvmark to add sky positions. But remember to add the id's in the third column! 1) Load local and finder. 2) Examine each of your images and determine which way is east and north. Write down the RA, DEC and epoch of field center from the header, or, better yet, write down the x and y of some object on the frame with known coordinates. 3) epar tfinder. Give tfinder the name of the first image you will be working with, as well as a name for the STSDAS "table" that will be used internally. Also provide a name for the "logfile"; this will contain crucial info you'll need for figuing out which plate to restrict yourself to. I would NOT give it a file name for the object x's and y's yet; this can be done later in tpeak when you are doing the final pass. catpars can be left blank. For ra, dec, and epoch you can either give it the coordinates of the frame center (taken from the header) and the epoch of these coordinages, or the ra, dec, and epoch of some object on the frame. Be sure to give it the xref and yref of this position. Give it the best value of the plate scale you know, and set "edge" to some significant number of pixels (200, say) to make sure that it finds all the stars in your field. The default boxsize of 9 for centering seems to work fine. You must specify the orientation for north and east using the terms left, right, top, or bottom. "pangle" is the orientation of the frame. You would expect this to normally be zero, but could be plus or minus a degree or so; our CCDs are not mounted on the telescopes all that accurately. Leave everything else alone. 4) It is necessary to "unlearn" selectpars before the first run of tfinder. 5) Run tfinder. After searching the CDroms containing the ST Guide Star Catalogue tfinder will eventually display the image in your imtool window and draw little red circles where it thinks the guide stars should be. When it is ready for you to do something it will give you a blinking cursor in the imtool window. Examine the marked locations. Hopefully you will see the circles just slighly offset from obviously corresponding stars. Does the scale seem to be exactly right? Is there an angle problem between the marked location and the guide stars? The interactive features in tfinder are great for taking care of linear shifts but can do nothing about scale problems and angle problems. If need be "quit" [using a "q"] out of tfinder and adjust the scale and angle. Rerun tfinder, being sure to first delete the table and logfile It's time to use some of the fantastically NICE features of tfinder. Usually all the red circles (the catalog stars) are all offset from their true locations. hit an "a" and then move the cursor to one of the red circles and hit an "l". Move the cursor to the corresponding star and hit another "l". Miraculously the image will now be redisplayed with all the sources nicely centered (the circles will now be blue). If you want to see what other options are available to you, hit a "?". 6) Repeat 3 & 5 for all the ccd frames for which you want a COMMON astrometric solution, e.g., based upon the SAME ST plate. I will warn you now that it is going to prove necessary to redo all of these runs, but it can't be helped. 7) Now is the time to make sure that your astrometry is all on a common reference system. For each CCD frame you have probably have reference stars from a few different ST "PLATE_ID"'s. This is Very Bad, as the coordinates of these reference stars will TYPICALLY differ by a few arcsecs. So having now seen which stars are CAN be centered on each frame, you are going to have to go back and rerun tfinder but now insisting that all of the reference stars come from the same plate. To do this, you must first ascertain which (if any) plate_id is in common to all the frames you want to observe together. Page through the log files, and figure out what (if any) plate_id is common to all the fields for which you want to have on a common system. If you have only one frame, you still may still find that stars (sometimes the same stars) come from two different plates. Edit in this plate ID into "selectpars". This is just slightly tricky; use "(double quote)'(single quote) plate name '(single quote)"(double quote) i.e., "'004C'" would be a valid entry. It will appear as '004C', (surrounded by single quotes). 8) Run the "tpeak" command, specifying the same table name that you used before. Specify the name of the list that has the xy coordinates of the unknowns. tpeak will do 2 things for you: a) add in the unknowns b) select only the guide stars that come from the plate_id that you specified in selectpars in item 7 above. EXIT tpeak without doing anything: simply type a "q" after it displays your image with the centered guide stars and positions of the unknowns. 9) This item is intensionally left blank. 10) Whew! It is time to do a little astrometry now. The astrometry is actually performed by the AAT program "ASTROM", which seems to do an incredibly nice job...nicer than ASTRO anyway! a) epar tastrom and enter the SDAS table name (which, incidently, now contains your list of unknowns), and some output root name which will be used to generate a bunch of files. Ignore catpars again. For "report" epoch, enter whatever epoch you would like to have the results in. For ra_tan and dec_tan you need to provide the RA and DEC of some point reasonably near the field center of the CCD. However, unlike the input in tfinder you must be VERY fussy about the format. Specify the RA as HH MM SS.S (no colons) and the DEC as SDD MM SS the tepoch is the peoch corresponding to this. b) run tastrom. Nothing will seem to happen, but it has! If you have given tastrom the "root" name of "field3" you will now find that there are 3 files that have been created: "field3.in", which contains the INPUT to astrom, "field3.ast" which contains the full-blown overly chatty analysis perfomed by astrom, and "field3.out", which contains a good summary of the fit and the RA and DEC of your unknowns. 11) Page through the ".out" file and see which standard stars gave high residuals. The first set of residuals listed will be for the "linear interpolation", and these will have gigantic errors if the orientation of the chip is not RA=x and DEC=y, so skip down to the RESIDUALS---6 COEF FIT. For each star the star name and the residuals in arcseconds EW, NS, and on-the-sky are listed, as well as the plate scale in x and y ( these two plate scales should be identical). Write down the star names of the few stars with the worst residuals. Edit the ".in" file and search for these stars. Eliminate the measurement line (x, y, star name) as well as the line above it (which gives the RA, DEC, and gsc region of that object). Rerun astrom by eparing "redo" and running it, setting "overwrite" to "yes". Continue this process until you have a nice, clean solution. I find that from reasonable scale CCD frames (scale <1.5 arcsec/pixel) I can easily get the solution to have an rms of a few tenths of an arcsec, with no star more deviant than 0.5 arcsec. You will find yourself with two useful output files, one name ".coo" that contains the ra's and dec's of all the unknowns, as well as the equinox of the coordinates.