4.4. Resolved Objects
We used the image analysis package SExtractor (Bertin & Arnouts 1996) for object identification and photometry. The requirement for detecting an object in our data is a core surface brightness, µcore, of at least 2sky within the core 4 pixels. The isophotal detection threshold, µiso was defined as 1.5sky over 6 pixels, after smoothing with a 3 × 3 boxcar filter. This corresponds to isophotal detection thresholds of 24.7, 25.8, and 25.3 ST mag arcsec-2, at F300W, F555W, and F814W, respectively. The shape of the smoothing kernel and the detection threshold have little affect on detection of objects with V555 < 27.5 A B mag, and photometry is not strongly affected for objects with V555 < 27 A B mag. Total magnitudes are defined as the flux within an area at least three times larger than the ellipse defined by the first moment radius (see Kron 1980) and isophotal elongation. For well-detected sources, the detection and photometry parameters we employ are similar to those used by the HDF team in producing the HDF catalogs (see the photometry discussion in Williams et al. 1996). Isophotal detection limits are roughly 1.5 mag brighter than the HDF, in keeping with the difference in exposure time, which is a factor of 15 - 21 in the various bandpasses. Differential number counts for our field are plotted in Figure 8. Our detection limits are illustrated by the turn-over in the counts in each band. The N-error bars show that V ~ 23 A B mag is reasonable bright-magnitude cut-off for observations in a field of this size. The galaxy counts from the HDF are also plotted in Figure 8 to demonstrate that the field we have observed contains a typical field-galaxy population in both color and number density. The HDF is a convenient benchmark for this comparison simply because the data are publicly available, widely studied, and of a similarly "blank" field. The HDF counts show twice as many galaxies at magnitudes brighter than 23 A B mag in each of the three band, while the difference in number density fainter than 23 A B mag is modest. Such differences at bright magnitudes are well within the typical field-to-field fluctuations for galaxy counts and illustrate the need for a bright magnitude cut-off.
Figure 8. Differential galaxy counts with N error bars from WFPC2 EBL images (filled circles), and the HDF images at F300W, F606W, and F814W (× 's). HDF counts are taken directly from the published catalogs (Williams et al. 1996). The difference in completeness limits between the two data sets reflects the difference in total exposure times, which are roughly × 21, × 16, and × 17 at U, V, and I, respectively.
Four stars are detected in the WF chips, three in WF2 with V555 = 19.0, 20.5 and 20.8 A B mag, respectively, and one in WF4 with V555 = 22.0 A B mag. Star-galaxy separation poses no difficulty; stars brighter than V555 ~ 23.0 A B mag are masked out, regardless, and the flux from stars beyond that limit is < 10% of that from detected galaxies at the same apparent magnitudes (Infante 1997).