Automated Search for Absorption Systems in Low Resolution QSO Spectra

An automated method to search for metal line absorption systems in low and medium resolution QSO spectra is presented. The method is based on pattern recognition in noisy data: it compares rectified QSO spectra with an input absorption line pattern minimizing X 2 between model and data. The best candidate for a z system is chosen among the minima of the curve x 2=x 2(z) and can be subtracted if weaker systems are to be searched for. A modified version of the X 2 test has been developed in order to compute the probability of finding a system by chance. The method has proved to be successful by detecting 43 metal absorption systems in about 200 low resolution (FWHM≈25A) QSO spectra of the Hamburg-ESO Survey with S/N≈40–50 and emission redshifts 0.33 ≤ zem ≤ 3.06.