2.2. Sample Selection

2.2.1. Available Ground- and Space-based Observations

In selecting our galaxy sample we aimed to take advantage of as much pre-existing ground- and space-based UV, optical and near-IR images as possible. In particular, high priority was given to galaxies that have been observed with UIT in the far-UV (1500Å) and obeyed the SB and size criteria (Section 2.1.2 and Section 2.1.3). Some of these galaxies are close to the UIT detection limit, as they are minor members in galaxy groups targeted by UIT.

Also, nearby galaxies already observed with HST/WFPC2 in the mid-UV F300W and/or F255W filter that match our criteria will be included in our final sample. When the final Cycle 9 target list was compiled in early 2000, the HST Archive contained 15 such galaxies with F300W images obtained and two more imaged in F255W. These Archival observations came primarily from Cycle 4 program GO 5381 (PI: Koratkar) and Cycle 6 program GO 6355 (PI: Giavalisco). We will analyze these 17 Archival mid-UV images in a separate paper (Taylor et al. 2002, in preparation). The current Cycle 9 sample is designed and must be understood to be complementary to this existing Archival sample.

Over the last decade, several of us have systematically imaged from the ground a total of about 750 nearby galaxies covering all Hubble types and inclinations. The majority of these galaxies - which includes most of the galaxies imaged in the far-UV with UIT (primarily at 1500Å) - were observed in UBVRI, but a good fraction was observed in the near-IR JHK filters as well (Eskridge et al. 2002b). The data base thus compiled consists of seven individual galaxy samples:

(1) 86 face-on spiral galaxies of de Jong & van der Kruit (1994);

(2) 220 galaxies with Hubble types later than S0 from the OSU BVRJHK survey (Frogel et al. 1996; Eskridge et al. 2000, 2002b) (all inclinations);

(3) 113 galaxies from the Frei et al. (1996) survey in B and R, to which we added U and V-band images (good number of early-type galaxies; all inclinations; Taylor et al. 2002; in prep.);

(4) 100 galaxies with 1500Å and 40 with 2500Å images from the Astro/UIT mission (Kuchinski et al. 2000; Marcum et al. 2001) (all inclinations);

(5) 48 edge-on galaxies (de Grijs & Peletier 1997; de Grijs, Peletier, & van der Kruit, 1997);

(6) 150 late-type UGC galaxies selected to be morphologically irregular, peculiar, or merging (Hibbard & Vacca 1997; Hibbard et al. 2002). These have a range of impact parameters (i.e., the distance between the two interacting or colliding objects); and

(7) 49 late-type dwarf spiral galaxies (Matthews & Gallagher 1997) and compact, high-SB luminous blue galaxies (Gallagher et al. 1989, 2001) (all inclinations).

with some overlap among the samples. The last two sets were specifically added to provide a good number of likely local counterparts of the peculiar and irregular galaxies seen with HST in large numbers at high redshifts. The UBVRI + JHK images from these ground-based surveys, where available, are included in the data presented in Section 3 (Fig. 4).

2.2.2. Required Statistics

To date, the HST Archive contains over 10,000 orbits of data on a multitude of distant galaxy projects (including parallels). To match these numerous observations of distant galaxies, we need a statistically significant and representative comparison sample of nearby galaxies that includes all morphological types. To cover a broad range of nearby galaxies, we will use three broad bins of Hubble types: (1) early-type galaxies (E-S0), (2) early-mid type spiral galaxies (Sa-Sc), and (3) late-type galaxies, irregulars (Sd-Irr), plus peculiars and mergers. For statistical studies as a function of photometric parameters, we will cover a range in inclination angles in three bins of sini: edge-on, face-on, and intermediate angles. Thus, we need about 54 galaxies in total, to populate each of the nine type-inclination bins with 6 galaxies to begin to average over galaxy properties.

Of the 20 galaxies with WFPC2 F300W images in the Archive as of early 2000 (mostly E/S0/Sabc's), about 17 are small enough to be contained within the WFPC2 FOV. To this available mid-UV sample we add here a complementary set of 37 galaxies, which we selected from the ground-based samples of about 750 galaxies described in section 2.2.1. These objects comprise 24 new irregulars, peculiars, and merging/interacting galaxies, and 13 new E/S0/Sabc's to complement the 17 usable galaxies available in the HST Archive.

2.2.3. Selection of the HST/WFPC2 Mid-UV Sample

The selection criteria for inclusion in the HST/WFPC2 sample of 37 galaxies are:

(a) the B-band half-light radius should fall within the range 0.'1 r_e 1.'0;

(b) the predicted average SB in F300W out to r r_e should be in the range 18 µ_F300W 22.5-23.0 mag arcsec ^{- 2};

(c) the distribution of Hubble types should be representative for nearby galaxy samples when combined with the set of archival galaxies;

(d) the distribution over apparent axis ratio (and hence inclination) should resemble that of nearby galaxy samples;

(e) priority should be given, first, to the 15 galaxies with available UIT 1500Å images that obey the other selection criteria above, and then to galaxies with multi-band ground-based images available;

(f) as much as possible, select galaxies that are schedulable in the HST CVZ, while satisfying (a) - (e). Replace candidates that are difficult to schedule by alternates.

In Table 1 we list the properties of the selected sample. Most of the table entries are taken or derived from values listed in the RC3 catalogue (de Vaucouleurs et al. 1991). Cols. 2 and 3 list the coordinates of the galaxy centers as measured on the Digital Sky Survey (DSS). In a few cases of complex galaxies, merging/interacting galaxies, or galaxies in small groups, the listed coordinates refer to a "center of mass" position, allowing optimal centering inside the WFPC2 FOV. Col. 4 shows the apparent axis ratio of the minor and major axis isophotal diameter measured at the µ_B = 25.0 mag arcsec ^{- 2} isophote. The half-light radius, r_e and the radius at the µ_B = 25.0 mag arcsec ^{- 2} isophote, r₂₅, are listed in Cols. 5 and 6, respectively; the predicted average SB in F300W out to r_e and r₂₅, and the predicted total F300W magnitude (see section 2.2) are tabulated in Cols. 7, 8, and 12. Col. 13 lists the revised T-type (numeric morphological type on the 16-step de Vaucouleurs scale). The redshifts in Col. 14 are given with respect to the Galactic Standard of Rest. Col. 15 lists for each galaxy the physical image scale in pc/(") at the distance of the galaxy, assuming a Hubble constant H₀ = 67 km s ^{- 1} Mpc ^{- 1}. And Col. 16 gives the physical size of a single WFC CCD-frame at that redshift. For the typical redshift range, z 0.004-0.02 (i.e., Virgo-Coma cluster like distances), a single WFPC2 CCD covers 6.8-34 kpc.

The SB-selection of the sample is illustrated in Fig. 1, which shows the normalized distributions of the predicted average µ_F300W out to r_e for the full (54 galaxy) sample and for the 37 galaxies in the Cycle 9 WFPC2 mid-UV sample, and compares these with the distribution over SB derived for the 3009 galaxies in the RC3 with measured B_T, (U - B) and r_e. Clearly, we sample predominantly galaxies with higher SB than the median SB-value in the RC3, which is µ_F300W 22.3 mag arcsec ^{- 2}.

Figure 1. The normalized distribution of the predicted average mid-UV surface brightness (SB) out to re (see Section 2.1.2) for the 37 galaxies observed during HST Cycle 9 (hashed histogram) and for the full galaxy sample (solid histogram), which includes 17 galaxies with mid-UV data taken prior to Cycle 9. For comparison, we also show the SB distribution for the 3009 galaxies in the RC3 with measured BT, (U - B) and re (open histogram). The galaxies were selected to have µF300W 23 mag arcsec - 2, allowing us to detect each object in F300W in no more than a single HST orbit. For our purpose of comparing nearby and distant galaxies this imposed SB-bias is justified, since the strong cosmological SB-dimming acts similarly in hiding lower SB objects at high redshifts from deep surveys.

As nearby galaxy samples like the RC3 are also limited by significant SB-selection (Disney 1976, Driver 1999), this thus means that our Cycle 9 HST sample represents only the tip of the local galaxy SB-distribution. Fig. 1 shows that our sample covers only about the brightest 50% SB-percentile seen in the RC3, while it over-represents the highest SB galaxies seen in the RC3. If one wanted to use our sample to represent all galaxies in the RC3 with µ_F300W 23 mag arcsec ^{- 2}, one would have to apply weights to each of our galaxies in a given observed SB-bin that could be derived from dividing the two distributions over SB in Fig. 1. This will obviously not work for galaxies with µ_F300W 23 mag arcsec ^{- 2}, since almost no galaxies at average SB-levels that faint were observed with HST.

As aforementioned, this SB-selection is justified when the primary goal is to construct a sample for comparison of nearby and high-redshift galaxies, since the strong cosmological SB-dimming limits our view of the high-redshift universe in a similar way (or worse). For example, simulations by Windhorst et al. (2002), who investigate how the galaxies in the present sample would appear to NGST under realistic observing conditions, show that at very high redshifts (z ~ 7-15) only the highest SB mid-UV structures (with average rest-frame µ_F300W out to r_e brighter than 20-21 mag arcsec ^{- 2}) will be visible to NGST.

Figures 2a-d show the distributions of morphological type, of apparent axis ratio b/a, of the estimated average (U - B) color within the effective radius, and of the effective radius r_e, within the selected sample. These figures will help the reader to evaluate the usefulness or the weight to apply to individual galaxies in the present sample when using these data for his or her particular purpose. We stress that the present sample is not formally complete and, therefore is not suitable for studies that rely on complete statistics (i.e., studies that cover the complete local galaxy luminosity function or the complete range in local galaxy SB). Nonetheless, important trends are visible in Figs. 2a-2d that will help the reader to make our sample useful for high-redshift comparisons. These trends are:

• Fig. 2a shows that our selected mid-UV galaxies overemphasize late-types a bit more than the RC3. This was done on purpose, since late-types are the dominant galaxy population at high redshifts (see Driver et al. 1995a).

• Fig. 2b shows that our selected mid-UV galaxies sample the b/a distribution of the RC3 fairly, given the small number statistics. Since Odewahn, Burstein, & Windhorst (1997) found no significant trend in the galaxy b/a distribution from the RC3 level (B 15 mag) all the way down to the HDF limit (B 28 mag), our nearby comparison sample is thus a fair one in terms of galaxy ellipticities at all redshifts.

• Fig. 2c shows that our selected mid-UV galaxies overemphasize the bluest object in (U - B) a bit more than the RC3. Like Fig. 2a, this was done on purpose, since faint blue galaxies are the dominant population at faint magnitudes.

• Fig. 2d shows that our selected mid-UV galaxies overemphasize the intrinsically smaller galaxies in r_e more than the RC3, which was again done on purpose, since faint blue galaxies are on average significantly smaller than the ones seen nearby (see Odewahn et al. 1996, Pascarelle et al. 1996, Cohen et al. 2002).

Figure 2. Other properties of the selected galaxy sample. As in Fig. 1, we present normalized distributions for the Cycle 9 galaxies (hashed), the full sample (solid), and the RC3 (open) of (a) morphological type; (b) apparent axis ratio, b/a; (c) average (U - B) color out to re; and (d) effective radius re. Taking deviations due to small number statistics into account, our sample approximates the RC3 distribution, except that for the purpose of comparison with high redshift objects, we placed extra emphasis on the very latest types, bluest optical galaxy colors, and smallest angular sizes (to fit the FOV of the WFPC2). We also somewhat over-represent highly inclined systems.

Again, any non-representativeness of our galaxy sample compared to the RC3, or to any other galaxy catalog for that matter, can be addressed by applying the appropriate weights to individual galaxies in our sample when comparing it to any other samples. These weights can be determined from the ratios between the observed distributions in Fig. 2a-2d, or between our sample and any other sample than the RC3.