iraf.net

hi,
how can I create a 3-dimensional spectral cube in IRAF?
I have several individual 1-dim spectra and I would like to stack these into a single file with three dimensions, e.g., spectrum[108,10,20] where the first dimension is the wavelength.With the task 'scopy' I seem to be able to create only 2-dim. spectral cubes.thank you in advance for your help,
Hermine

Frank would be the best person to answer but is away for the holiday. The IMSTACK task can certainly create a 3-D image from your 2-D images, but this wouldn't be a true spectral cube since the WCS wouldn't be what you probably need without some tweaking. The GEMINI package has some tasks for handling cube data but may be too instrument specific for your needs. ANyway, I hope this can get you started or somebody else can post a solution, some more details about exactly what you're trying to do might help.Cheers,
-Mike

thanks Mike. 'imstack' is exactly what I needed.

Hi Hermine & Mike,Sounds like you're happy with imstack, but I'll just comment on the alternatives in the gemini package that Mike mentioned. There are a few tasks specifically for creating spectral datacubes, although most of them are indeed fairly specific to Gemini instruments. These tasks are needed for cases where your spectra are not on a square grid to begin with (otherwise, imstack is probably fine, if potentially slow).First, Frank recently wrote a task called gemcube that can map pixels from a 2D image to a 3D cube using a version of the Drizzle algorithm. This task is actually quite generic, but it is currently distributed as part of the gemini package (under gemtools) because it was needed to support the Gemini instruments. You can grab the gemini package and do "help gemcube" if you'd like to read more about it. Before it will work, you might have to hack your image WCS headers around to describe the 2D-3D mapping you want.Second, the gemini package includes a task called gfcube (under gmos) that will take a set of extracted 1D spectra and resample them onto a 3D cube using a smooth spatial interpolant. This task was designed specifically for GMOS and requires a multi-extension FITS file as input, including a binary table extension that describes the co-ordinates of each 1D spectrum on the sky. There is also a hard limit of 1500 spectra (which is how many GMOS has, although I could make a version without that limitation if you need it).Third, there is also a task called nfcube in the GNIRS package for stacking 2D image slice spectra into a 3D cube, but that's probably not what you need since you have 1D spectra. Likewise, there is a task called nifcube in the nifs package, but it is just an instrument-specific front end to gemcube.The practical difference between gemcube and gfcube is mainly in the interpolation methods they use (currently Drizzle in the first case and a smooth 2D piecewise spline in the second). I would use gemcube for undersampled or marginally sampled data (where the seeing is less than twice the spatial pixel size) and gfcube for well-sampled data. If you don't need to interpolate because your spectra are already on a square grid (eg. from a square microlens IFU), imstack should be fine as long as the I/O is fast enough.There are also some other options around, such as the Euro3D tools, that will let you visualize or analyze your 1D spectra in 3D.Cheers,James.