Date: Mon, 3 Oct 2005 17:56:14 -0700 (PDT)

Uncertainties in IRS data
-------------------------

While previous versions of the IRS pipeline did not propagate useable
uncertainty estimates along, your S12 data does include an estimate of a
critical part of these uncertainties, namely those associated with fitting
the ramp slopes.  The uncertainties of S12 BCD products, in files
"_func.fits," consist of ramp slope estimation uncertainties. These
uncertainties are typically 0.1% at 5 Jy, 0.5 - 1% at 500 mJy, and about
10% at 20 mJy (in low resolution). At high resolution, they are about 1%
at 10 Jy.  The S12 uncertainties do not include some systematics, such as
those arising from flat fielding or flux calibration. These systematics
will be incorporated in the S14 version of the pipeline (Spring 2006).

The S12 uncertainties are propagated during spectral extractions from
individual BCDs. During extraction, the BCD uncertainties of individual
pixels are summed in quadrature, within a wavelength resolution element,
to get the uncertainties quoted in the extraction tables. These
uncertainties are output in column 4 of the "_spect.tbl" and "_spect2.tbl"
files.  The S12, 2-D coadder does not use the uncertainties from the BCD
products. The 2-D coadder output uncertainties, in files "_c2unc.fits",
are the trimmed standard deviation of the pixel values of individual BCDs.
The output spectral uncertainties extracted from 2-D coadder products
(col. 4 of the "_tune.tbl" files), are based only on the "_c2unc.fits"
statistical uncertainties. A future version of the 2-D coadder will
propagate the uncertainties of BCDs.

It is important that observers examine the repeatability of their spectra,
across multiple cycles and nod positions, to further assess statistical
and systematic uncertainties.

_______________________________________________
IRS-ig mailing list
IRS-ig@lists.ipac.caltech.edu
https://lists.ipac.caltech.edu/mailman/listinfo/irs-ig