2000ApJ...544..780P

We have conducted a multiwavelength (X-ray, optical, and radio) search for supernova remnants (SNRs) in the nearby Sculptor Group Sd galaxy NGC 300. Our Very Large Array (VLA) radio observations at 6 cm and 20 cm have been combined with previously published optical results, our own optical image, and archived ROSAT X-ray data to search for new SNR candidates. Of the 28 optically identified SNRs found by Blair & Long, three exhibit some combination of X-ray and radio emission, but in general X-ray and radio emission from the optically identified SNRs is undetected.

A radio-selected sample of SNR candidates is constructed by searching for positional coincidences between nonthermal radio continuum sources and sources of Hα emission with the intent of finding SNRs that are typically overlooked using the [S II]/Hα criterion, i.e., those SNRs that are either confused by Hα emission or which are Balmer-dominated. This search has yielded 17 SNR candidates, of which 14 are new and three were already known from the optically selected sample. Four of the radio-selected candidates also possess detectable X-ray emission.

A complementary analysis of the X-ray data has yielded an X-ray-selected sample of candidates consisting of six soft-spectrum sources (kT<1 keV) coincident with regions of Hα emission. Two of these candidate X-ray SNRs are common to the radio-selected sample and two more are common to the optically selected sample. Thus, the X-ray selection has yielded two additional candidates.

Through a multiwavelength campaign, we have added sixteen new candidate SNRs to the 28 previously known SNRs for a total of 44 SNR candidates in NGC 300. The fact that these new candidates were missed in the optical surveys suggests a possible selection effect. SNRs identified through optical methods may represent only the SNRs located in regions with relatively low confusion from Hα emission, well away from star-forming regions. The SNRs in or near star-forming regions are more likely to be confused by emission from H II regions and are therefore more likely to be missed in optical surveys. The radio-selection process, on the other hand, suffers from a separate selection effect: it can only identify candidates if they are associated with H II regions, such that our sample of radio-selected SNRs is biased toward star-forming regions. Finally, the X-ray candidates are selected on the basis of having soft spectra and association with H II regions, so that the sample is biased against X-ray-emitting SNRs with hard spectra and no optical counterparts. The fact that the optically selected, radio-selected, and X-ray-selected data sets of SNR candidates have limited overlap is consistent with these opposing selection effects. We present a simulation to investigate the effects of optical confusion on a hypothetical optical survey for SNRs in a galaxy, using ``artificial'' SNRs that have been placed in a range of confused environments. We find that the detection of SNRs becomes increasingly difficult with distance, and that the most deeply embedded SNRs are not identified even in the nearest distance investigated. We conclude that a multifrequency search for SNRs is required to uniformly sample the SNR population of a galaxy.