Comments and reply regarding Kotsakis (2005) 'A type of biased estimators for linear models with uniformly biased data'

摘要

As the IAG's official journal, the Journal of Geodesy offers a forum for scientific publications on fundamental or application-oriented research that cover the broad spectrum of geodetic science. As a service to our readers, comments on and corrections to material published in this journal will also be considered for publication. This particular comment and reply concerns some of the alternative thoughts on least squares estimation in the presence of biases, which is published to alert readers to some of the differing thoughts on this problem. Kotsakis (2005) presents, via Eq. (2), a linear model where each observation is disturbed by a systematic error. He gives three applications for this model: (ⅰ) analysis of linearization errors, (ⅱ) analysis of aliasing errors and (ⅲ) outlier detection. Because of the singularity of the resulting normal equations, he introduces the very simple linear model where all measurements are affected by one common and unknown offset.