A Method for Spatial Deconvolution of Spectra

A method for spatial deconvolution of spectra is presented. It follows thesame fundamental principles as the ``MCS image deconvolution algorithm''(Magain, Courbin, Sohy, 1998) and uses information contained in the spectrum ofa reference Point Spread Function (PSF) to spatially deconvolve spectra of veryblended sources. An improved resolution rather than an infinite one is aimedat, overcoming the well known problem of ``deconvolution artefacts''. As in theMCS algorithm, the data are decomposed into a sum of analytical point sourcesand a numerically deconvolved background, so that the spectrum of extendedsources in the immediate vicinity of bright point sources may be accuratelyextracted and sharpened. The algorithm has been tested on simulated dataincluding seeing variation as a function of wavelength and atmosphericrefraction. It is shown that the spectra of severely blended point sources canbe resolved while fully preserving the spectrophotometric properties of thedata. Extended objects ``hidden'' by bright point sources (up to 4-5 magnitudesbrighter) can be accurately recovered as well, provided the data have asufficiently high total signal-to-noise ratio (200-300 per spectral resolutionelement). Such spectra are relatively easy to obtain, even down to faintmagnitudes, within a few hours of integration time with 10m class telescopes.Comment: 18 pages, 6 postscript figures, in press in Ap