1. INTROUCTION

1.1. The Need for Reliable Classification of Faint Optical Galaxy Samples

Since the mid 1990's, faint galaxies have been observed with the Hubble Space Telescope (HST). The best statistics, spatial sampling, and areal coverage are currently achieved in the I-band (F814W). Good HST morphological information is available today for ~ 10⁵ galaxies with I 26 mag. These galaxies come from the two Hubble Deep Fields ("HDF"s; Williams et al. 1996, 1998), their flanking fields, the HST Medium-Deep Survey, and other HST parallel surveys (e.g., Griffiths et al. 1994; Driver et al. 1995a, 1995b; Odewahn et al. 1996; Abraham et al. 1996, 1999; Roche et al. 1997; Cohen et al. 2002).

The median half-light radius of faint field galaxies with I ~ 26 mag is r_e 0".2-0".3 (Odewahn et al. 1996). Because of the high spatial resolution required for proper morphological classifications (e.g., using artificial neural networks), HST's 2.4 m aperture limits the useful wavelength for quantitative galaxy classification to 4000 8000Å. NICMOS H-band images (~ 0 16 FWHM) do not necessarily sufficiently resolve these faint galaxies to accurately measure their half-light radii and other important parameters.

For I 23 mag, a few hundred galaxies have their morphology measured at HST resolution and have measured spectroscopic redshifts plus velocity dispersion or rotation curves (e.g., Cohen et al. 2000; Lilly et al. 1998, Simard et al. 1999; Vogt et al. 1997), leading to more quantitative studies of their physical properties such as mass and M/L ratios. However, because morphology can be determined with relative ease with HST for a very large number of galaxies, this ability allows us to explore a part of parameter space that constrains galaxy formation and evolution, and that is not yet accessible to satisfactory spectroscopic study and thus lacks kinematic data for mass measurements.

The most dramatic result from these HST morphological studies in the mid-late 1990's was that at faint fluxes, late-type/irregular galaxies completely dominate the faint blue galaxy counts (see Driver et al. 1995a, 1995b, 1998; Glazebrook et al. 1995; Odewahn et al. 1996; Windhorst et al. 1996, 1998b, 2000). A number of authors interpreted this result as evidence for the existence of a population of star-forming galaxies that underwent substantial evolution since z 3 (Ellis et al. 1996; Pascarelle et al. 1996; Driver et al. 1998; Abraham et al. 1999). Ellipticals and early-type spirals, on the other hand, have evolved much less since z 1 (see Driver et al. 1995a; Lilly et al. 1998; Cohen et al. 2002). The photometric redshift distribution as a function of observed I-band morphological type - N(z_phot, type) - suggests a differential evolution of galaxies as a function of type (Driver et al. 1998), implying a gradual formation of the Hubble sequence with cosmic time.

In the deepest HST fields, where morphological classifications are achievable to I 26 mag (e.g., Odewahn et al. 1996; Driver et al. 1998), the sampled redshift range is z 1 - 3, while the bulk of the galaxies is at z 2. Faint high redshift galaxies observed in the I-band are therefore primarily seen in the rest-frame mid-ultraviolet (mid-UV) wavelength range, or 2000-3200Å.

The reliability of results derived from the faint HST galaxy morphologies is therefore fundamentally limited by the uncertain rest-frame mid-UV morphology of nearby galaxies. For instance, although these faint, late-type/irregular objects resemble some classes of nearby late-type and peculiar galaxies (e.g., Hibbard & Vacca 1997), they need not be physically late-type objects; instead, they may be earlier-type galaxies that look dramatically different in the rest-frame UV. Because of the wavelength dependence of nearby galaxy morphology (especially toward the UV), the outcome of faint galaxy classifications will depend on the rest-frame wavelength sampled. This "morphological K-correction" can be quite significant, even between the B and near-IR bands (e.g., Knapen & Beckman 1996), and must be quantified in order to distinguish genuine evolutionary effects from simple band-pass shifting.

The mid-UV is the optimal wavelength region for comparing the galaxies seen with HST in the I-band at z 2 to nearby samples, and to address the question of whether the numerous late-type objects that dominate the faint blue galaxy counts are truly new classes of objects, or just reflect the redshifted UV morphology of ordinary galaxies as seen nearby. In the mid-UV, one currently attains the highest resolution plus largest field-of-view ("FOV") with the HST Wide Field and Planetary Camera 2 ("WFPC2").

In this paper, we therefore present a systematic WFPC2 mid-UV imaging survey for a representative sample of 37 nearby galaxies. Having observations for both nearby and distant objects at the same rest-frame wavelength will allow us to calibrate the dependence of morphology on rest-frame wavelength and, thus, to improve our ability to interpret the morphology of distant galaxies. Such a survey will also help us better understand the physical drivers of the rest-frame mid-UV emission, i.e., the relation between star formation (SF) and the global physical characteristics of galaxies, their recent star-formation history, and the role of dust absorption and scattering.