Characteristic Analysis of 3D Outlier Detection Method for GNSS Network Adjustments

摘要

In this study, characteristics of the outlier detection method for global navigation satellite systems (GNSS) network adjustments are discussed and alternative hypothesis models are proposed to handle an outlier vector at one observed site. First, we explain that the classical outlier detection method, on either a baseline-component level or a baseline-vector level, can identify the location of outlying observations but cannot estimate the value of the outlier accurately because the outlier and the corresponding random observation errors are inseparable. Then we describe the alternative hypothesis models, specifically designed by considering the spatial correlation between multiple baseline vectors, so that one outlier vector imposing to multiple baseline observations can be modeled. Numerical examples are used to compare the detection accuracy and reliability when different stochastic models are used and also to validate the proposed method, which contrasts with the less precise classical three-dimensional (3D) outlier detection method.