To measure the ZL contribution to the background flux that we measured with HST, we obtained ground-based spectra at 3900-5100Å within the field of view of our WFPC2 images using the Boller & Chivens spectrograph on the duPont 2.5m Telescope at Las Campanas Observatory in Chile on the nights of 26-29 November 1995, concurrently with the HST observations of that field on 29 November 1995. We have used those spectra to measure the absolute flux of the ZL at 4650Å with a precision of 0.6%, and a systematic uncertainty of 1.0%, using the method outlined in Section 2. We also mesure the color of the zodiacal light to be C(5100, 3900) = 1.05± 0.01. That measurement is discussed in detail in Paper II.
To identify the ZL flux contributing to the FOS and WFPC2 measurements of the total sky flux, we need an absolute, flux-calibrated spectrum of the ZL from roughly 2000-1 µm (see WFPC2 bandpasses plotted in Figure 14). We obtain this spectrum by scaling a solar spectrum to the surface brightness value we measure for the ZL at 4650Å and applying a small reddening correction redward and blueward of 4650Å to compensate for the changing scattering efficiency of the interplanetary dust (see Section 2.2 for discussion). Although our LCO measurement of the ZL color is quite accurate, it only covers a small fraction of the total wavelength range covered by our WFPC2 observations. We therefore use our FOS observations to identify the appropriate reddening correction, which cover the wavelength range 4000-7000Å. Published measurements of the ZL color have absolute uncertainties as large as 10% and are in poor agreement with each other (see Leinert 1998). Our FOS observations, by contrast, are accurate to better than 2% in relative flux as a function of wavelength and are identical in line of sight and epoch to the WFPC2 and LCO observations. Like most measurements of the ZL, however, the FOS observations clearly include the EBL flux. This fact prevents the FOS and LCO results from unambiguously determining the color of the ZL. Fortunately, this is not an insurmountable problem: the WFPC2 observations can be used as a further constraint, as we describe in the following section.