Next Contents Previous

2. SAMPLE SELECTION AND DATA PROCESSING

In constructing the RBGS, the basic methods used to extract candidate bright galaxies from the IRAS catalogs and to compute total fluxes in each of the four IRAS bands using ADDSCAN/SCANPI (hereafter referred to as SCANPI; Helou et al. 1988) were similar to procedures used in earlier compilations of the IRAS BGS (here referred to as BGS1: Soifer et al. 1989, 1987, 1986) and the IRAS BGS-Part II (here referred to as BGS2: Sanders et al. 1995); the reader is referred to these papers for historical perspective as well as for a more thorough description of original sample membership criteria. In this paper we outline the major steps that were followed to construct the new sample, and where appropriate, emphasize the differences between the earlier BGS1 + BGS2 and the more refined RBGS processing.

The IRAS FSC and IRAS PSC were used as starting points for the initial search of the IRAS data archive. It was necessary to use the PSC because the FSC does not include objects in confused regions of the infrared sky with Galactic latitude |b| < 10°. The FSC Rejects (FSCR) 8 were also examined for bright objects at 60 µm. In the end, the only source accepted from the FSCR is NGC 4151 (IRAS Z12080+3940). The reasons for its inclusion in RBGS are: a) the SCANPI signals in all 4 IRAS bands are very strong (see Table 1); b) the FIR signal is positionally coincident with the optical galaxy, and the coadded scan profile from ADDSCAN/SCANPI is consistent with the optical size of the galaxy; c) the IRAS SCANPI flux density values at 12 µm and 25 µm are comparable (slightly higher, as expected given some extended flux) to independent ground-based measurements at 10.6 µm and 21 µm in apertures centered on the nucleus (Lebofsky & Rieke 1979).

A second step involved another search of the IRAS PSC and FSC, this time with a lower 60 µm point source threshold of 4.5 Jy, which was designed primarily for the purpose of capturing sources that might have a reasonable chance of having extended flux sufficient to bring their total 60 µm flux densities above the RBGS threshold. To investigate how many objects with total 60 µmflux density greater than 5.24 Jy may be "hiding" among objects with even fainter point-source components, a few dozen galaxies randomly selected from the FSC with 4.0 Jy < Snu(60µm) < 4.5 Jy were also coadded with SCANPI. These objects were not found to contain sufficient extended emission to bring their total fluxes above the RBGS flux limit. Searches of NED for optical galaxies larger than 4 arcminutes, without regard for their (likely underestimated) flux measurements from the IRAS catalogs, were also performed to obtain additional candidates. However, objects with total Snu(60µm) > 5.42 Jy among these large optical galaxies objects were also recovered in the IRAS catalog searches noted above. This provides confidence that our procedures resulted in a sample with a very high level of completeness at 60 µm.

All candidate IRAS sources were compared with the latest catalog cross-correlations available in the NASA/IPAC Extragalactic Database (NED) 9, which were then inspected using overlays on the DSS1 images. The data were then reprocessed (co-addition of all acceptable IRAS scans) using SCANPI, and the resulting 1-D coadded scan profiles were visually inspected to determine the amount of extended emission, and in the case of blended or confused sources, to determine the best method for computing the total flux density. A computer program was written to determine objectively and consistently whether the total flux density of each source is better represented by the baseline zero-crossing method fnu(z), the value integrated within the nominal IRAS detector size fnu(t), the point source template fit amplitude ("template"), or the peak flux in the profile ("peak"). This was accomplished by checking whether the fnu(t) value is significantly larger than the template fit and peak values (considering also the RMS noise in the coadded scan outside the detected signal range), and if so whether the fnu(z) value is significantly larger than the fnu(t) value. This procedure, combined with comparisons of the coadded scan profile widths at 25% ("W25") and 50% ("W50") of the peak flux with the nominal beam size (point-spread function), was also used to determine whether each IRAS source is resolved, marginally extended or unresolved. All measurements were performed using SCANPI's median (1002) method of coadding the individual IRAS scans, and all sources with 60 µm point source flux greater than 5.24 Jy (the completeness limit of the original BGS) were included in the final sample.

Appendix A gives more details regarding the various methods used for setting thresholds to select between the flux density estimators and for estimating the total flux densities of the extended sources, and provides a tabular listing of all key SCANPI measurements in each IRAS band, as well as the ratio of the new RBGS total flux density measurements compared with the old values reported in the BGS1 + BGS2 where available). Example coadded scan profiles are also plotted to illustrate the extent of IRAS emission compared to the point spread function and to show how different flux measurement methods were objectively selected in the automated SCANPI reprocessing of the data for all potential RBGS sources.

Finally, IRAS flux measurements from the "Catalog of IRAS Observations of Large Optical Galaxies" (Rice et al. 1988) and "An Atlas of High-Resolution IRAS Maps of Nearby Galaxies" (Rice 1993) were examined. Careful comparison of flux densities integrated within the signal range between SCANPI's baseline fit zero-crossing points, fnu(z), to those measured by Rice et al. from IRAS 2-D image products showed generally good agreement for galaxies with optical diameters smaller than about 25 arcminutes. For objects with optical diameters larger than ~ 25 arcminutes it is clear that even SCANPI's fnu(z) method systematically underestimates the total flux density compared to the 2-D method used by Rice. Therefore, to maintain consistency with other objects in the RBGS to the highest degree possible, in the final compilation we adopted Rice et al. measurements over those from SCANPI's fnu(z) estimator only for objects larger than 25 arcminutes; such objects are flagged in Table 1.



8 The FSCR, along with all other IRAS catalogs, is available for queries through the GATOR service of the Infrared Science Archive (IRSA) at IPAC; http://irsa.ipac.caltech.edu/. The FSCR contains objects that have only one IRAS hours-confirmed (HCON) observation; however, IRAS measurements in the FSCR are much more reliable when a high signal-to-noise detection is combined with confirmation via positional cross-identification with a known source detected at another wavelength. Back.

9 NED is available at / Back.

Next Contents Previous