4.5. On-Axis Scattered Light
Scattered light from the "bright" (19 < V < 23 AB mag) objects imaged in the field is one possible source of error in a measurement of the background from fainter objects. To quantify this effect, we begin by considering how much light these sources produce relative to the total flux in an image. In Figure 9, we plot the ratio of the flux from bright sources to the total flux detected in each of the three WF CCDs during each of four exposures. The total flux is just the sum of all counts in the calibrated images. The flux from bright sources is a background-subtracted flux within four times the isophotal radius (4riso) for galaxies (see Section 10.1) and within a 5 arcsec radius for stars. For each bandpass, the ratio of flux in bright sources to the total flux varies from chip to chip. In the WF2 or WF3 chips, bright objects contribute only ~ 5% of the total integrated flux. As several of the brightest galaxies in the field are imaged on the WF4 chip, almost ~ 15% of the flux in F555W and F814W in that field comes from objects with V < 23 ST mag.
Figure 9. The flux from bright (V555 < 23 AB mag) sources relative to the total flux in each WFPC2, EBL image is plotted for four exposures through three filters (twelve images total). The total flux is just the sum of all counts in the calibrated images. The flux from bright sources is a background-subtracted flux within four times the isophotal radius (4riso) for galaxies (see Section 10.1) and within a 5 arcsec radius for stars. The F555W, F814W, and F300W flux ratios are marked by open circles (overlapping), triangles, and crosses, respectively.
In order exclude these V < 23 AB mag objects from the measured background, we masked regions around each resolved galaxy which extend to four times the isophotal detection radius of the galaxy, and masked regions with radii of 5 arcsec for stars. Based on the encircled energy curves in H95a and growth curves in our own images (see Section 10 and Appendix B), we estimate that less than 2% of the light from these objects is imaged beyond the extent of the masked regions. As an upper limit on their scattered light, 2% of the flux from these objects constitutes only 0.1 × 10-9ergs s-1 cm-2 sr-1 Å-1, less than 3% of the total flux at 5500Å in our estimate of the EBL. As discussed previously in Section 3, the scattered light from off-axis sources is also negligible.
At the wavelength of the F300W filter, the flux from bright objects on the WF4 chip is 15-20% of the total flux, and variation in the percentage flux contributed by objects in the F300W images is greater than in the other two bands. These characteristics are explained by two facts: (1) errors in dark glow subtraction cause noticeable variation in the background level between exposures in this band; and (2) the ZL falls off rapidly below 4000Å, so that the ZL contribution to this band is a factor of three smaller than in the F555W or F814W and variations in source flux from chip to chip are fractionally larger in this band than in the other two. In contrast, ZL contributes roughly 95% of the diffuse background at 5500-9000Å (see Section 9 and Paper II). Nonetheless, the contribution to the background from detected objects in the F300W images, again assuming 2% scattered light, is still a minor uncertainty in our results, being 0.3 × 10-9ergs s-1 cm-2 sr-1 Å-1, or 7% of the EBL we detect at 3000Å.