1. INTRODUCTION

The Extragalactic Background Light (EBL) is the integrated, mean surface brightness of both resolved and unresolved extragalactic sources. In wavelength range of these observations (2500-9500Å), the EBL is dominated by flux from stellar nucleosynthesis at redshifts z 9, and also includes flux from gravitational potential energy such as accreting black holes (Fabian & Iwasawa 1999), brown dwarfs, gravitationally collapsing systems, and possibly decaying particles. As such, the EBL is the fossil record of the star formation history of the universe and a fundamental measure of the luminous energy content of the universe. As we show in Figure 1, upper limits from previous attempts to measure the EBL and lower limits from integrated galaxy counts constrain the EBL to an expected level of roughly 1 × 10^-9 ergs s^-1 cm^-2 sr^-1 Å^-1 at 5500Å, or 28.2 AB mag arcsec^-2. ⁽¹⁾ Because the combined flux from foreground airglow, zodiacal light, and diffuse galactic light (also plotted in Figure 1) is at least 100 times brighter than the EBL, a detection of the EBL requires measurement accuracies of better than 1%.

Previous attempts to measure the optical EBL have employed a variety of different approaches. Mattila (1976) attempted to isolate the EBL by differencing integrations on and off the line of sight to "dark" clouds at high Galactic latitude, under the assumption that the clouds acted as a "blank screens," spatially isolating all foreground contributions from the background. This pioneering work produced upper limits and identified both the rapid temporal variability of terrestrial airglow and extinction and the spatial variability of diffuse Galactic light, which proved to be the primary obstacles to these early efforts to measure the EBL. Toller (1983) later attempted to avoid both atmospheric and zodiacal foregrounds by using data taken with the Pioneer 10 spacecraft at a distance of 3 AU from the Sun, beyond the zodiacal dust cloud. Poor spatial resolution (2°), however, prevented the accurate subtraction of discrete Galactic stars, let alone the diffuse Galactic component from these data. Dube, Wilkes, & Wilkinson (1977, 1979) made the first effort to measure and subtract foreground contributions explicitly based on geometrical modeling of airglow and Galactic foregrounds but including spectroscopic measurement of the zodiacal light (ZL) flux by a technique similar to that which we have adopted. Rapid variability caused uncertainty in their airglow subtraction which dominated the errors in their results.

Figure 1. Relative surface brightnesses of foreground sources, upper limits on the EBL23 (see Section 1), and lower limits based on the integrated flux from resolved galaxies (V555 > 23 AB mag) in the HDF (Williams et al. 1996). The spectral shape and mean flux of zodiacal light and of diffuse galactic light (DGL) are shown at the levels we detect in this work. The airglow spectrum is taken from Broadfoot & Kendall (1968) and is scaled to the flux level we observe at 3800-5100Å (see Section 9). The effective bandpasses for our HST observations are indicated at the bottom of the plot.

In this work, we take advantage of the significant gains in technology and in understanding of the foreground sources which have been achieved since the last attempts to measure the optical EBL (see Mattila 1990 for a review). The most significant technological advance is panoramic, linear CCD detectors. Those on-board HST allowed us to completely avoid bright, time-variable airglow and provide sub-arcsecond spatial resolution. High spatial resolution allowed us to resolve stars to V ~ 27.5 mag and thereby eliminate the possibility of significant contamination from unidentified Galactic stars in the field. Ground-based spectrophotometry with CCDs also made possible much more accurate measurement of the foreground zodiacal light than could be achieved with narrow-band filters and photometers, as were used by Dube, Wickes, & Wilkinson (1977, 1979). Finally, IRAS has provided maps of the thermal emission from dust at high Galactic latitudes. We have use the IRAS maps to select a line of sight for these observations which has a low column density of Galactic dust in order to minimize the DGL contribution caused by dust-scattered starlight, and also to estimate the low-level DGL which cannot be avoided.

Our measurement of the EBL utilizes three independent data sets. Two of these are from HST: (1) images from the Wide Field Planetary Camera 2 (WFPC2) using the F300W, F555W, and F814W filters, each roughly 1000Å wide with central wavelengths of 3000, 5500, and 8000Å, respectively; and (2) low-resolution spectra (300Å per resolution element) from the Faint Object Spectrograph (FOS) covering 3900-7000Å. The FOS data were taken in parallel observing mode with the WFPC2 observations. While flux calibration of WFPC2 images and FOS spectra achieve roughly the same accuracy for point source observations, the increase in spatial resolution, 10⁴ times larger field of view, lower instrumental background, and absolute surface brightness calibration achievable with WFPC2 make it better suited than FOS to an absolute surface brightness measurement of the EBL. Nonetheless, the FOS observations do provide a second, independent measurement of the total background flux of the night sky, also free of terrestrial airglow and extinction, but with greater spectral resolution than the WFPC2 images. The third data set consists of long-slit spectrophotometry of a region of "blank" sky within the WFPC2 field of view. These data were obtained at the 2.5m duPont telescope at Las Campanas Observatory (LCO) using the Boller & Chivens spectrograph simultaneously with one visit of the HST observations (6 of 18 orbits).

The flow chart in Figure 2 summarizes the reduction procedures for each data set used in this measurement, the results obtained from each data set individually, and the coordination of those results to produce a measurement of the EBL. In this paper, we begin by describing the foreground sources in Section 2 and the details of HST scheduling in Section 3. The observations and data reduction of both HST data sets, WFPC2 and FOS, are discussed in detail in this paper. WFPC2 observations and data reduction are discussed in Section 4. In Section 5, we present the first results from the WFPC2 data, which are measurements of the total sky flux (foregrounds plus background) in each bandpass. The FOS observations, data reduction, and results are discussed in Section 6 and Section 7. The modeling of diffuse Galactic light is discussed in Section 8. The LCO data and measurement of ZL are discussed in Bernstein, Freedman & Madore (2001b, henceforth Paper II). A summary of that work is given in Section 9. In Section 10, we present a measurement of the minimum flux of the EBL from resolved sources, which can be made using the WFPC2 data alone. The implications of that result are also discussed in Section 10. Finally, in Section 11 we combine the results of the individual data sets and modeling (horizontal connections shown in the flow chart) to obtain a measurement of the EBL. The implications of these results are discussed in Bernstein, Freedman & Madore (2001c, henceforth Paper III).

Figure 2. Flow chart of data reduction, analysis, and results from all data sets used to measure the EBL. Where appropriate, subscripts to I() indicate the bright magnitude cut-off applied. The thick, horizontal bars divide the pre-reduction and analysis steps. The symbols * and marking steps in data reduction indicate that a systematic or statistical uncertainty is accrued at that step. Normal type-face indicates STScI pipeline data reduction procedures; bold type-face indicates original procedures which we developed for this work. Dashed boxes mark estimates of the ZL color; shaded boxes indicate results derived from one data set alone; thick-lined boxes indicate final measurements of the EBL from combined WFPC2/LCO and FOS/LCO data sets.

¹ All surface brightnesses are specified in ergs s^-1 cm^-2 sr^-1 Å ^-1 unless specifically noted otherwise. AB mag is defined as AB mag = -2.5log F - 48.6, as usual, with F given in ergs s^-1 cm^-2 Hz^-1. Back.