Robust Estimation of the Outliers in GPS Baseline Components

Several definitions can be made for the term ‘outlier ’ such as a single observation far away from the rest of the observation set. The outlier is the observation that not includes systematic and gross errors but random errors and differs from the other data with its distribution. It is assumed that geodetic observations free from systematic and gross errors always have random errors which are in normal distribution. If an observation is separated from the other data with respect to its random error which is not in normal distribution, that observation is named ‘outlier’. To date several approaches have been developed to identify the outliers more accurately. The first proposed approaches, conventional methods, used the statistical tests to detect outlying observations and removed them from the data set. The main drawback of such an application is the shape defect of the network. Unlike the conventional methods, the robust estimations do not remove any observation from the network. They reduce the effects of the possible outliers on the adjustment model as making the weights of the outlying observations smaller even zero. In this study, some robust estimations (Danish method, Huber’s, Andrew’s, and IGGIII M-Estimations) are applied to a 17-point GPS network in order to make decision abou