Vol. 5, No. 3 - March 1997
You are welcome to send us a list of your favorite images for priority processing. There is a NEWSIPS user request Web page at http://hypatia.gsfc.nasa.gov/IUE/search/IUE_usrreq.html. Your request goes into a semi-automated system, which will send you email reports on the status of the processing. The data may be retrieved from the NDADS system once it has been archived, usually just a few days after it has been processed.
ifitsrd, iue3drow, &
nstwrt were added to the experimental library to allow fields
from a FITS binary table extension to be stored in an IDL structure or an
array of structures. This can simplify the calling sequence when
requesting multiple rows and columns from a FITS binary table extension.
isndads was implemented which
allows archive requests for high dispersion NEWSIPS files using the
program search.
readmx was added to the experimental
library which uses an improved method for deriving splice points when
merging multiple orders. The previous version simply spliced orders at
the mean wavelength within the overlap region, whereas the new version
uses the wavelength at which the difference in the applied ripple
correction within the overlap region is a minimum. The result is a
significant reduction in the apparent noise at the ends of each order.
ripple
iuerip
new_splice
readmx for calculating splice points
between merged spectral orders.
readsi
One of the major differences between the NEWSIPS and IUESIPS high dispersion processing is in how the background is handled. In IUESIPS, a localized background based on the pixels between the orders was used. This led to many well-known problems, since in the area of crowded orders the background is filled in by light from the adjacent orders. The background estimate was therefore too high, and absorption lines were often seen to go well below zero flux. In NEWSIPS, swaths have been chosen across the image to sample the background, and the area affected by interorder light is corrected for that effect. A "global" determination of the background is then made. This produces a more accurate background, and you will see that the absorption line depths are more realistic.
There are limitations to this new technique, of course. If there are local perturbations of the background, such as small cosmic ray hits, they are interpolated over and the background estimate is too low around that small area. If there are large perturbations, such as glancing cosmic ray hits (the ones that look like comets), they may distort the global background fit by producing mild ringing. In some portion of the image the background estimate may then be too high or too low. Flares in late-epoch LWR images, and the lack of any nearby pixels with reliable background fluxes in the short-wavelength corner of the SWP camera, are specialized problems for which there is no good background estimator - by any algorithm.
In addition, images which have only emission lines with no continuum are better handled with a localized background determination. The NEWSIPS software determines whether the image has a continuum spectrum or not, then chooses a "global" or "local" background determination appropriately. This is recorded in the processing history.
You can easily see that NEWSIPS has a more sophisticated background
estimation algorithm, but it isn't perfect. This is to be expected in a
production system designed to process all the widely divergent images in
the IUE archive. A customized background and extraction may be expected
to do a better job for individual images, depending on their
characteristics. In addition, an improved version of readmx
(described above) uses an updated method for
splicing high dispersion orders, thus reducing the apparent noise. A
future issue of the Newsletter will give details on how
customized background extractions might be done easily.
One of the new data products from NEWSIPS is a high dispersion
line-by-line (sihi) file, similar in concept to the
line-by-line files for low dispersion (IUESIPS elbl and
NEWSIPS silo files). These files can be used as a basis for
those users interested in generating customized extractions.
A side note about absorption line depths: First, the noisiness of the data in the depth of an absorption line is much higher than on the wings and nearby continuum, so you may well see flux values both above and below zero in the depth of a dark line. Second, the actual minimum in a dark absorption line is generally not exactly zero, due to grating scattered light. Those of us who measure line depths and equivalent widths in high resolution data normally have to consider that there is a few percent effect from the line being filled in by grating scattered light. In addition, IUE's detectors have a significant amount of halation, so even a perfectly extracted dark absorption line may have a minimum that is a little above zero flux.
If you have any questions, please feel free to contact us.
mxlo) files contain
absolutely-calibrated "sigma" vectors, while
the high dispersion (mxhi) files contain
uncalibrated "noise" vectors. Because no
spectral profile information is available as in low dispersion, the high
dispersion noise spectra are based solely on the camera noise models.
As a reminder, all NEWSIPS data sets use vacuum wavelengths, whereas IUESIPS files use air wavelengths longward of 2000 Ångstroms. This can cause an apparent wavelength differences of 0.65 to 0.9 Ångstroms, when comparing IUESIPS and NEWSIPS long wavelength spectral data.
Complete documentation on the high dispersion processing system is currently being completed by the IUE project and will be available in the near future.
or send a postcard to:
IUE Data Analysis Center (IUEDAC)
Code 684.9, NASA-GSFC
Greenbelt, MD 20771
Back to the IUEDAC Information PageBack to the IUE Project Page