Modeling of point spread functions for astronomical multifiber spectrographs

This thesis presents an improved method for reconstructing the spectra of astronomical objects from two-dimensional (2D) charge-coupled device (CCD) images. We address two issues, namely, estimation of calibration matrix for a CCD, and reconstruction of spectra of astronomical objects, which is referred to as extraction. In the rst part of the thesis we estimate the elements of system calibration matrix by modeling the two-dimensional point-spread functions (PSF) in calibration images. Our PSF model is valid for arbitrarily complicated 2D point-spread functions, as compared to the state-of-art extraction methods which are valid only for class of separable PSFs. We present various models for PSFs and give a quantitative comparison between the models. In the second part of the thesis the system calibration matrix is used to extract the spectra of a particular type of calibration images called arc-images. We also address the issue of resolution and covariance in the extracted spectra, and present a method that establishes optimal properties in both these regards. We also compare quantitatively the performance of our extraction technique with the state-of-art extraction technique. The work presented in this thesis can be deployed for estimation of spectra of faint galaxies in presence of strong night-sky foregrounds.