IRIS - Improved Reprocessing of the IRAS Survey

Technical

The whole processing and the characteristics of the IRIS data are described in the paper:
"IRIS : A new generation of IRAS maps", M.-A. Miville-Deschênes and G. Lagache, 2005, ApJSS, 157, 302.

Computing

The computing of the IRIS plates has been done with McKenzie, the Beowulf Linux Cluster of the Canadian Institute for Theoretical Astrophysics. The Beowulf cluster is composed of 512 2.4 GHz Xeon processors with a total of 256 GB of RAM. We effectively used only 40 CPUs of the cluster at the same time and the total computation time for the whole processing (deglitching, destriping, responsivity and offset correction) was ~20 hours for the 4893 HCON plates.

Main characteristics of the IRIS data

	12 micron	25 micron	60 micron	100 micron
Angular resolution (arcmin)	3.8 +- 0.2	3.8 +- 0.2	4.0 +- 0.2	4.3 +- 0.2
Noise level (MJy/sr)	0.04 +- 0.01	0.05 +- 0.02	0.03 +- 0.01	0.06 +- 0.02
Gain uncertainty (%)	5.1	15.1	10.4	13.5
Offset uncertainty (MJy/sr)	6.0E-5	8.3E-5	0.027	0.027
The offset uncertainties given in this table do not take into account the uncertainty on the zodiacal light removal (the zodiacal model used is that of Kelsall et al. 1998).

Limitations

The IRIS data photometry is limited mainly by the zodiacal light subtraction of the DIRBE data, based on the model of Kelsall et al., 1998. A second order limitation is due to the fact that the ISSA plates were not constructed completely self-consistently. The overlaping regions between ISSA plates are not exactly the same. Usually this difference is negligeable and ISSA/IRIS plates can be used to make mosaics. Nevertheless, in some cases, especially in bright regions, the difference can be of several percents. Both problems are illustrated in the following image