Background and scattered-light subtraction in the high-resolution echelle modes of the Space Telescope Imaging Spectrograph

We present a simple, effective approach for estimating the on-order backgrounds of spectra taken with the highest resolution modes of the Space Telescope Imaging Spectrograph (STIS) on board the Hubble Space Telescope. Our scheme for determining the on-order background spectrum for STIS E140H and E230H observations uses moderate-order polynomial fits to the interorder scattered light visible in the two-dimensional STIS MAMA images. We present a suite of high-resolution STIS spectra to demonstrate that our background-subtraction routine produces the correct overall zero point as judged by the small residual flux levels in the centers of strongly saturated interstellar absorption lines. Although there are multiple sources of background light in STIS echelle mode data, this simple approach works very well for wavelengths longward of Ly alpha (lambda greater than or equal to 1215 Angstrom). At shorter wavelengths, the smaller order separation and generally lower signal-to-noise ratios of the data can reduce the effectiveness of our background estimation procedure. Slight artifacts in the background-subtracted spectrum can be seen in some cases, particularly at wavelengths of less than or similar to 1300 Angstrom. Most of these are caused by echelle scattering of strong spectral features into the interorder light. We discuss the limitations of high-resolution STIS data in light of the uncertainties associated with our background-subtraction procedure. We compare our background-subtracted STIS spectra with the Goddard High Resolution Spectrograph (GHRS) Ech-A observations of the DA white dwarf G191-B2B and the GHRS first-order G160M observations of the early-type star HD 218915. We find no significant differences between the GHRS data and the STIS data reduced with our method in either case.