Installing
and Using IRAF
to prepare images for use with HOU's Image Processing software
by Jeff Adkins
This page details the technical problems and solutions encountered
while trying to prepare images from the TLRBSE project's Teacher
Observing Program, which yielded images from a 0.9 meter telescope
at Kitt Peak National Observatory.
=====
Background
The information in this section was published as an article at
the web site Low End Mac.
Recently two of my students were selected to participate in the Teacher
Observing Program at Kitt Peak National Observatory, through a program
sponsored by the National Optical Astronomy Observatory in Tucson, Arizona.
This involved making a proposal for the use of a 36" telescope (for
you non-astronomers, that means bigger than what most amateurs own, but on
the
small end of professional-class scopes). My students suggested observing
BL Lac, a blazar or active galactic nucleus, and various globular clusters
to
observe RR Lyrae variables. A brief overview of our experience is posted
here (http://www.noao.edu/outreach/tlrbse/top-200410.html).
The purpose of this article is to describe the technical problems we encountered
and solved along the way, providing assistance to others who may face similar
problems and to make comments about the state of advanced scientific software
for the Macintosh.
The task
When using professional-class telescopes, astronomers are interested
in isolating the true signal from an object being imaged from
the noise or interference in the telescope and camera which can
obscure
the thing you're trying to measure. Telescopes use CCD cameras
similar to those used in digital cameras, but they require careful
calibration to remove these sources of unwanted signal. Calibration
of the camera involves taking "bias frames" to show how
electrical signals used to read the camera's picture may provide
false spots in the image, and "flat frames" which are
pictures of blank white spots on the observatory wall to show
how dust in the optical path may cast shadows on the image making
parts
of the image artificially brighter or dimmer than it would otherwise
appear. In practice a user must apply these corrections to each
picture taken to remove these unwanted signals so the remaining
image is pure signal with no static, so to speak. Then measurements
of brightness and other characteristics can be taken which are
scientifically meaningful.
An excellent tutorial on the details of this process is located here:
http://www.astro.ufl.edu/~oliver/ast3722/lectures/CCDImaging/CCDimaging.htm.
Using the camera, then, provided us with a collection of dark frames, bias
frames, and "raw" images of our various target objects. Our task
was to combine these into a single calibrated image, and then take measurements.
All of the images generated by this program are in FITS format. FITS is a file
format which preserves the original data as read out on the chip in the camera.
Unlike jpegs or other formats, altering the appearance of a FITS file does
not alter the original data, just how it is displayed. This is pretty much
a universal standard among professional astronomers.
More about FITS format here: http://fits.gsfc.nasa.gov/fits_home.html
All of our computers were Mac laptops. We had access to the following pieces
of software. Links are included to the sources users can get to these. The
limitations of each piece of software is noted.
NIH Image, This
freeware program for Mac OS 9 can open FITS files, but requires
customized Pascal-like macros to do basic astronomical functions
such as brightness counts (adding up the counts detected in
each pixel of a star's image). The macros available for us
to use (from the Teacher Leaders in Research Based Education
project--not available for download) would not open the images
provided by the telescope we used.
Image processing: This program from the Hands-On
Universe project could open FITS files we'd collected in other places (if
you ignored some user warnings, etc.) Unfortunately it would not open images
we took ourselves at the Kitt Peak telescope.
Image J: This Java-based successor to
NIH Image opens everything just fine, but doesn't yet have the specialized
macros needed by amateur and professional astronomers to perform measurements.
I got a copy of a beta set of macros under development, but it isn't functional
yet. Like NIH Image, most plugins and macros available on the web are aimed
at medical applications.
IRAF: This UNIX-based program is what is used by most professional astronomers
to analyze images. It is very powerful, difficult to use, and uses many arcane
text-based commands similar to UNIX commands. Installing it is not a simple
task, but it is doable. An experienced power user who encounters no problems
can finish the installation in about two hours. I encountered problems, so
it took perhaps three days of off-and-on work to solve all the issues before
IRAF became functional. This was only possible at all because of the very thorough
and well-written set of instructions at Marcos
Huerta's web site.
This points out a common problem with science software for the Mac. Many professional
astronomers use Macs, because they need easy access to UNIX. I've met several
myself who use them for exactly that reason. These programs represent the low-end
of the spectrum (such as NIH Image) where programs are available, but in some
limited way or for specialized educational purposes. At the other end is the
high-powered IRAF, which does anything, but has a steep learning curve. There
isn't anything in the middle for Mac users equivalent to a program like, for
example, Maxim DL for the PC, which is what most amateur astronomers are using
these days for this kind of work.
Mac users wanting to work with Maxim DL have no choice but to use Virtual PC,
which is pretty much an advertisement for getting a real PC--just functional
enough to run real PC software, just slow enough to be annoying to the point
of wanting a real machine. This is why I say Mac astronomical image processing
software is polarized. There's the low-end educational versions (great at what
they are designed for, but limited) and then there's the highest end there
is (IRAF--which, by the way, doesn't run in Windows at all), but there's nothing
in between.
There are a number of programs which show promise but are not complete packages.
The most useful of these is Keith's
Image Stacker which can be used to combine bias and dark frames as
well as "stack" images to enhance contrast. You cannot take measurements
such as brightness counts with KIS, however, and my experience was it could
not open my FITS files for technical reasons I'll explain in a separate
article. For what it does, it is really useful, but it doesn't do exactly
what I need,
which is sort of in a gray area between amateur and professional needs.
There has been an ongoing discussion thread about these and other astronomical
issues sponsored by the Applelust site (http://lists.topica.com/lists/hubble_boy).
If you are interested in astronomy and use a Mac, you should subscribe, and
read the astronomy related articles posted at Applelust.
In the remainder of this document I detail how I installed IRAF
using Huerta's instructions and help files, and then how I managed
to convert the images from the format we got them into a format
our older software at school could read, plus how we generated
some preliminary data from the images.
Installing IRAF
Image Reduction for Astronomy Facility (IRAF) is a professional-level
astronomy program used by most professional astronomers for data
reduction. It will install in OS X, but not in Windows. I used
Marcos Huerta's excellent instructions and installer packages from
this site:
http://www.owlnet.rice.edu/~marcosh/iraf/
I am not going to attempt to improve on his excellent instructions,
but just provide some notes about how it went when I tried it.
Everything went pretty well, more or less, with just a couple
of exceptions. As his instructions indicated, you must go into
the
System Preferences for Accounts, and create a user named "iraf."
In step 2 (http://www.owlnet.rice.edu/~marcosh/iraf/iraf.html)
he suggests that you download the installer while in your normal
account but then put it in /Users/You/Public so the "iraf" user
can get to it. This did not work for me because of permission
issues, and I had to download and install it at the root level
of the hard
drive. This is the directory you get when you double-click on
the Macintosh HD icon on your desktop. It is not inside of any
particular
user's folder.
Later on Huerta notes that in Mac OS X 10.1 and 10.3, you may
get some weird error. I did, and because he said to ignore it,
I did, and as predicted, it worked anyway.
When you are instructed to use the Terminal, you will be typing
some UNIX commands such as cd (change directory) and sudo. These
are extremely sensitive to syntax. If a command ends in a slash,
don't leave it off. If there is a space between a character and
a slash, don't leave it out. Once the installer runs, don't move
any files around.
When performing the mkiraf command, a script is run which tries
to install everything in subdirectories preset within directories
which all refer to each other later, so don't try to install
it in a "convenient" location. Mine didn't work; as it turned
out, I had permissions issues. If your install says a series of "command
not recognized" after each step, the script didn't work.
I eventually resorted to using the FAQ (question # 8) http://www.owlnet.rice.edu/~marcosh/iraf/faq.html#q8.
Following these directions seemed to repair the permissions flaw,
which, for whatever reason, caused the installer to run incorrectly.
In the next major step, The X11Iraf package is installed. This
installed correctly, but I did have to follow the instructions
for editing and creating a file called .cshrc for this. This
file must exist in the directory you use to start IRAF. The instructions
on the site tell you how to tell if this is set incorrectly,
and
refer you to a "Unix for Newbies" page for instructions
on how to fix the problem. I did, so my advice is : don't skip
the "You might have to" sections. Look for the part of
the instructions on this page which says "You have to edit
your .cshrc or .tcshrc file to fix this." I had to.
Everything else went pretty well, but as some of the astronomers
I worked with told me, using IRAF is fine when it works. Fixing
things is part finding good instructions, part intuition, and
part reading between the lines. Knowing just a little UNIX doesn't
hurt
either. While running in the terminal, I used the following commands
frequently and independently of the "just type it like I said
to" instructions on the web page.
ls (list files in the current directory)
cd .. (move one level up in the directory heirarchy)
cd directory name (move to the directory named, if it exists within
the current directory. sd
The UNIX for Newbies page at Huerta's site tells you many other
commands you might need.
I also recommend installing Huerta's "IRAF Button" application,
which autolaunches not only the IRAF application and the X11
environment, but other support software such as the graphic display
software
called DS9. I had an issue involved with the placement of the
.cshrc file not being in the user level directory (/Users/yourname/)
where
I would usually be when running IRAF. This took a little while
to figure out, but eventually, after erasing everything, deleting
the iraf user, and starting over, I managed to run the installer
packages exactly as described on Huerta's site (when he says
to be in a certain account when installing, believe him) to make
everything
function correctly.
Make sure you install the xgterm application as IRAF is really
not useful without it. I also installed the "popular graphic
package" on the download page. There is a lot of talk about
OroborosX and ximtool; these are only options. I used X11 which
is from Apple, and it worked fine.
If you've done everything correctly, when you run "The IRAF
button" then you should see a screen that looks more or
less like this:
Click on the image for a magnified view.
The IRAF Button application is in the lower left corner. IRAF
itself is running in the upper left within xgterm. IRAF is actually
a text-only program and requires an external viewer such as DS9
(shown on the right) to display images. The two applications talk
to each other and can feed data into subroutines and functions
as you work.
Bit depth of FITS images
Like most other graphic formats, FITS is not a single one-size
fits all protocol. It can handle a variety of bit depths. Bit
depth refers to the number of bits in the number variables used
to store
data from each pixel. An image with a bit depth of 1 is black
and white. Each pixel can either be a 1 or a 0, which requires
only
1 bit. Mac Classic computers (the kind with the old 9" screen)
had a bit depth of 1. Most amateur class cameras generate images
with a bit depth of 8 or 16, corresponding to 256 or thousands
of colors. Many of the common programs available to educators
open one or the other of these formats, sometimes not both
(depending
on what the developer of the software usually uses). Professional
level cameras can generate images with millions of shades of
grey (too many for monitors to display or your eyes to discern
even
if your monitor could display them.) This is one of the reasons
false color is used in image processing, because it brings
out detailt to subtle to recognize with the more realistic
looking
greyscale images.
IRAF can be used for any kind of image processing. Ostensibly
I should have used it to take our bias images and flats taken by
my students Robert Johnson and Tri Nguyen at the 0.9 meter telescope
(details are here) and used IRAF to calibrate or reduce the images
to a form which could be measured. While I struggled with installing
IRAF, NOAO astronomer Katy Garmany calibrated the images for us
using IRAF, so I left that task to be learned another day. Once
she generated calibrated images, however, we discovered they would
not open in the Hands On Universe program Image Processing which
we usually use for collecting data from images. I eventually determined
this was because the HOU software requires a 16 bit unsigned image
and the FITS images were 32 bit images.
Signed images remove one bit from the color depth data and use
it to signify positive or negative values. There is a thorough
discussion of this issue at this web site: http://www.rwc.uc.edu/koehler/comath/13.html .
My task therefore became to convert the calibrated images from
Dr. Garmany into a lower bit depth where our software could open
and read the information. We could have used IRAF for the measurements,
but as I only have one or two computers capable of running IRAF,
and I have 15 which can run the older software, I opted to convert
the data so more students could work simultaneously.
If you want to know how to persuade IRAF to do data reduction
of raw images, there is a tutorial written for another project
posted here: http://www.nuro.nau.edu/nuro/info/iraf.html. I
haven't tried it yet but it looks like it is pretty straightforward,
written for brave beginners.
Conversion of bit depth using IRAF
Here are the steps I used to prepare my images for student use.
To begin with:
The original images are in a directory I could find easily.
IRAF is installed and running properly.
In the following instructions type exactly what is between the
brackets [ ] but don't type the brackets themselves.
Using UNIX commands at the cl> prompt, navigate to the the
directory with the images. Type [ls] to list the directory,
[cd ..] to go one level higher in the heirarchy, and [cd foldername]
to move into a folder. For example you might type [ls], press
return,
[cd ..] and press return, [cd ..] and press return again to
go two levels higher in the directory tree, then [cd images] to
move
into the directory images. The exact sequence of commands you
type will depend on precisely where you have installed IRAF and
where
the images are stored on your computer. It is very similar
to DOS commands or Apple II file commands if you're old enough
to remember
that.
Verify you can see the files to be converted by typing [ls]. There
shouldn't be anything else in the directory except the images you
wish to convert.
1. In the IRAF window, type [noao].
Note, due to the way X11 and xgterm work your cursor must be in the window
before typing letters will appear on the screen.
2. Then type [imred].
3. Next type [ccdred].
4. Now type [epar ccdlist]. This means you will EDit the PARameters
of the function. Using arrow keys to move up and down and return
to make changes, change the filename to [*.fits] which will use
a wildcard to select all of the images in the directory.
5. Set the ccdtype to [object].
6. Type a colon -q [:q] to exit the epar subroutine.
7. Enter [ccdlist > filename] where filename is the name of
a text file to which all the files you've selected will be
listed. This shows the real power of IRAF--batch processing. If
a professional
has hundreds of images they can all be manipulated at once
like this. I expect the calibration steps are done in a similar
way.
8. Now a new file will appear in your image directory containing
a list of all the images you wish to convert.
9. Next do [epar wfits].
10. Set fits source files to [@filename] where [filename] is the
same exact name of the list you just made.
11. Set fits file to new name such as [16bitconverted]. This phrase
will be appended to the new files as they are written. The old
filename will still be there so you can tell which is which. Each
file will be numbered in the sequence it is done.
12. Set bi = [16]. (This is the whole reason we went through this.
14. Type [:q] to exit epar, then execute the wfits command by
typing [wfits]. If everything is setup OK, all the files will convert.
Large files take a few seconds each on a 1 GHz TiBook.
You're done with this step!
Opening files with HOU Image Processing.
This is for users of the HOU program Image Processing. You can
now open 16 bit files generated by IRAF from professional-grade
cameras. All you have to do is:
1. Always open from the file menu, not by double-clicking.
2. Don't use the "recently opened" menu to open files.
Doesn't work.
3. Don't panic when the software says "Unrecognized format.
Opening as FITS". It'll work.
4. Then you can do brightness counts on stars, draw slices to
get distances, etc. You can even read the FITS header if necessary.
When I learn how to reduce images using IRAF, I'll write that
up as well, but it may take a while...
If you found any of this useful, I'd appreciate a note. Enjoy!
P.S. Don't move or rename files in the directory after making
the list, or the process will abort in the middle.
