SYSTEMATIC ERRORS IN SPECTRAL FEATURES INTRODUCED BY RANDOM NOISE

Abstract

ABSTRACT A common IUE computer routine for measuring equivalent widths is shown to introduce systematic errors in data containing large amounts of noise. During statistical tests using simulations of synthetic spectral features with noise characteristics similar to those of the IUE, the routine was found to overestimate the true strength of an absorption line 177 out of 256 attempts, about six standard deviations more frequently than is predicted by random chance. The average of the measured equivalent widths was shown to be larger than the true equivalent width by 1.4sigma of the rms noise of the continuum.