Reading HUT Data

The two HUT missions constituted a link between past (IUE, Copernicus) and future (FUSE) spaceborne spectroscopic facilities for UV-bright astronomical sources. One of several uses made of HUT data was to compile far-UV atlases of early-type stars in the magnitude range m_v = 6 - 12 down to the Lyman limit. Take for example the spectra of HD 93521, a Galactic O9.5Vp star, from the atlas of Buss, Kruk & Ferguson (1995). This atlas was compiled from HUT 2 observations.

Retrieving the Files

Entering the star name, HD 93521, on the HUT Archive Search Web page is sufficient to determine both that 4 data files are available and that of these only the exposures in rows 2 - 4 were long enough to merit coaddition and analysis. We mark the entries, download a tar file containing them by clicking "submit" on the Search page, and finally unpack the resulting file. This step results in the creation of a new "hut" subdirectory, a copy of the original tar file (hut.tar) and a text description of nomenclature of HUT data file types. These files consist of a primary array in nonhomogeneous fits format where the first row represents fluxes and the second row is flux errors. The data consist of postscript plots depicting various "engineering" parameters versus time as well as FITS files of raw, semi-calibrated, and final calibrated data. (The *imcscor* files are not available for some targets, including our example, HD 93521.)

Reading the Files in IRAF

The structure of the data files may be reconnoitered by invoking IRAF and STSDAS and then running the "catfits" task on them. Taking the same example:

st>  catfits hd93521_032_02_a_ph.fits

The output from this command is:

EXT# FITSNAME FILENAME EXTVE DIMENS BITPI OBJECT
0 hd93521_032_0 2048x2 32 HD93521

This tells us that the file has no extensions and that the primary data array is an image containing 2048 entries in two rows. Again, the first row contains the net fluxes of general interest.

A wavelength vector may be built by first giving the command:

cl>   rfits hd93521_032_02_a_ph.fits * hd93521_032_02_a_ph.imh

followed by:

cl>   imhead hd93521_032_02_a_ph.imh l+   .

We may then utilize the W0 and WPC keywords for wavelength-start and wavelength increment per pixels (both in Ångstroms).

The analysis of HUT data files can now proceed with standard IRAF commands in version 2.11. The files may be converted to imh format by the 'rfits' command or (after loading STSDAS) 'strfits'.

One may also plot the fluxes with sgraph in STSDAS:

st>  sgraph hd93521_032_02_a_ph.fits

(Here it is understood that extension [0] designation is implicit at the end of the file name.)

Reading the Calibrated HUT Files in IDL

As described in the page on IDL-based FITS-readers, there are a number of IDL-based tools available for reading of HUT data files, which are simple FITS images containing only a header and primary array. We consider two FITS-readers:

Example 1: (read in with 'ifitsrd')

Most users interested in HUT data have had some familiarity with reading IUE data, so we suggest they download IUEDAC software and familiarize themselves with a few IUE FITS readers such as "iuefhdr" (to read header info) and "ifitsrd" (to read data).

One starts by browsing the important keywords, which for HUT data are W0 and WPC (starting pixel wavelength and wavelength increment, respectively, both in Ångstroms):

IUEDAC>   iuefhdr,'hd93521_032_02_a_ph.fits',params,mhead,exthead'

where params is a parameter array describing the FITS file format (containing record numbers of the main and extension arrays) and the main and extension header (string) arrays; HUT data do not contain an extension array. One prints the mhead array and locates the keywords, W0 and WPC (start wavelength and wavelength increment; both in Ångstroms) and also the array length, which is given by NAXIS1 (= 2048).

IDL users know how to construct the wavelength array using this information:

IUEDAC>   w = w0 + dindgen(2048)*wpc

(note the declaration of double precision). One can also extract the W0, WPC keywords explicitly using the IUEDAC routine stpar, e.g.:

IUEDAC>   stpar,mhead,'W0',w0   , etc.

Next, the flux and flux errors can be read in:

IUEDAC>   ifitsrd,'hd93521_032_02_a_ph.fits',-1,h2,0,p2

where h2 and a2 are the header and primary flux arrays. The flux array, a2, is a 2048 X 2 array in which the first row contains the observed fluxes and the second row the associated errors.

The files for observations 3 and 4 may be read in the same manner, co-added, and passed through a Fourier filter to recover spectra in the form similar to those published by Buss et al.

Example 2: (read in with 'mrdfits')

IUEDAC>  a2=mrdfits('hd93521_032_02_a_ph.fits',0,h2)

Note that 'mrdfits' will not return IDL structures on image files. Note also that some readers (other than 'mrdfits') written especially for FITS tables may not work for these files.

Example 3: (second science example)

HUT data are also useful for analysis of AGN or QSO emission line spectra.

Let's compare a HUT 1 spectrum of the AGN Fairall 9 with an IUE spectrum. After retrieving the files as above and downloading them, we read in the '*imcscor*' files recommended for bright sources:

IUEDAC>   ifitsrd,'fairall9_223_01_a_imcscor_ph.fits',-1,h,e,a

The IDL routines may be used to build the wavelength array. To do so, first determine the pixel-to- wavelength intercept and slope (delta wavelength per pixel) from the keywords W0 and WPC in the header (one must know these kinds of things; it is not explicitly documented). Both are expressed in Ångstroms. Next use the IUEDAC routine 'stpar' to find them:

IUEDAC>   stpar,h,'w0',w0,err
IUEDAC>   if(err le 0.) then print,'w0 not found'
IUEDAC>  stpar,h,'wpc',wpc,err
IUEDAC>   if(err le 0.) then print,'wpc not found'
IUEDAC>  w=fltarr(2048)
IUEDAC>  i=dindgen(2048)
IUEDAC>  w=w0 +i*wpc

Next, obtain an extracted, low-dispersion spectrum from the IUE archives of Fairfall 9.   SWP29545 is one among many examples:

IUEDAC>  readmx,'swp29545.mxlo',h1,w1,f1,e1,s1

where h1,w1,f1,e1,s1 are, respectively, the header, wavelength, calibrated net flux, error flags, and noise vectors, respectively.

When these spectra are overplotted their agreement is impressive!