As is true of all background measurements, the difficulty in measuring the optical EBL is in differentiating it from the much brighter foregrounds: terrestrial airglow, zodiacal light (ZL), and diffuse Galactic light (DGL). Relative to the EBL flux at ~ 5000Å, airglow and ZL are each more than 100 times brighter than the EBL. Along the most favorable lines of sight, the DGL is roughly equal in flux to the EBL. We have measured the EBL in a field which is out of the ecliptic plane and near the Galactic pole in order to optimally minimize the contributions of zodiacal light, DGL, and nearby stars (see Paper I).
In the EBL measurement presented in Paper I, we have used three simultaneous data sets to isolate the diffuse EBL from the foreground sources: (1) absolute surface photometry taken with WFPC2 aboard HST using the wide-band filters F300W (U300), F555W (V555), and F814W (I814); (2) low resolution (~ 300Å) surface spectrophotometry at 4000--7000Å taken with the FOS, also aboard HST; and (3) moderate resolution (~ 2Å) surface spectrophotometry taken with the Boller and Chivens spectrograph on the 2.5m duPont telescope at LCO. We use the two HST data sets to measure the total mean flux of the night sky, including ZL, DGL and the EBL. We avoid terrestrial airglow all together by using HST for this measurement. We then use the LCO spectra to measure the absolute surface brightness of ZL in the same field and on the same nights as the HST observations. Finally, we estimate the small DGL contribution using a scattering model which is in good agreement with the observations. We then subtract the measured ZL and the modeled DGL from the total flux measured with HST/WFPC2 through each filter and with HST/FOS. These measurements are described in detail in Papers I and II. Below, we summarize the observations, results, and accuracy of the individual measurements which contribute to the EBL detection (see Table 1).
Bright galaxies brighter are not statistically well sampled in the 4.4 arcmin2 WFPC2 field of view. We have, therefore, masked out any sources brighter than V555 = 23 A B mag in the WFPC2 images before we measured the total sky flux. To do so, we used masks which are derived from the F555W images and extend to four times the isophotal radius in those data. We use the abbreviation EBL23 as a reminder of this bright magnitude cut-off. The EBL23 detections can be combined with ground-based counts at V555 < 23 A B mag to obtain the total EBL. The WFPC2 surface brightness measurements have random errors of < 1% and systematic uncertainties of 1-2% of the total background flux. From the HST/WFPC2 data alone, we can also identify a minimum flux from detectable sources. This minimum is given in Table 1, and the method used to obtain this result is summarized in Section 3.
The FOS spectra also provide a measurement of total flux. The random error per resolution element is around 2.1%, and the systematic uncertainty over the full range is 3.5%. The ~ 14 arcsec2 FOS field of view and ~ 4% systematic uncertainty make the FOS less useful than the WFPC2 for measuring the EBL. However, most of systematic uncertainty is due to the poorly constrained solid angle of the aperture and aperture correction. Both of these are wavelength-independent errors, so that the FOS spectra do provide a ±1% measurement of the color of the total background, which is dominated by zodiacal light.
The scattering which produces the ZL is well described by classical Mie theory for the large (> 10µm), rough dust grains which populate the zodiacal dust cloud. The scattering efficiency of the dust is only weakly wavelength dependent, so that the solar spectral features are well preserved in the scattered ZL spectrum. However the broad band spectrum of the zodiacal light is redder than the solar spectrum by about 5% per 1000Å (see Paper II for further discussion) due to surface roughness of the grains, which decreases scattering efficiency at shorter wavelengths. The mean ZL flux in a narrow band can thus be measured from the apparent equivalent width of the solar Fraunhofer lines evident in its spectrum. Small color corrections can then be used to infer the full spectrum relative to that measurement. This requires moderate resolution spectra (~ 2Å) with excellent flux calibration (±1%), which can only be obtained with ground-based observations, and then only at wavelengths where atmospheric emission lines are relatively weak. We have, therefore, measured the ZL in the range 4000-5100Å using spectra taken at LCO. The resulting measurement has a statistical error of < 1% and a systematic uncertainty of ~ 1.2%. This measurement has been extrapolated the 3000Å and 8000Å WFPC2 bandpasses using measurements of the color of the ZL from the FOS and ground-based LCO data.
Within the Galaxy, there is both resolved flux from discrete stars and diffuse light (diffuse Galactic light, DGL) from starlight scattered by interstellar dust. Discrete stars can simply be resolved and subtracted in the WFPC2 images. The intensity of the dust-scattered optical DGL and the 100µm thermal emission from the same dust are both proportional to the dust column density and the strength of the interstellar radiation field. To minimize the optical DGL, our field was selected to have very low 100µm emission. The remaining low-level DGL which does contribute has been estimated using a simple scattering model based on the dust column density and interstellar radiation field along the line of sight and empirical scattering characteristics for interstellar dust. The predictions of this model are in good agreement with observations of the DGL from 2500-9000Å (see Witt et al. 1997 and references therein). Finally, although isotropic line emission from warm interstellar gas is measured at all Galactic latitudes, the strongest line, H, does not lie within any our HST/WFPC2 bandpasses. The next strongest lines, [OIII], are twenty times weaker and contribute negligibly to our results.
The EBL cannot be measured in typical HST data. Our HST observations were scheduled to avoid contaminating scattered light from all anticipated sources: the bright Earth limb, the Moon, and off-axis stars. Also, observations from LCO and HST were strictly simultaneous to guarantee that the ZL measured from the ground is exactly the contribution seen by HST. As an additional safeguard, observations were scheduled in 3 visits, allowing us to look for possible off-axis scattered light with the satellite oriented at different roll angles, to safeguard against unidentified photometric anomalies with the instruments, and to confirm the expected modulation in the ZL with the Earth's orbital position.