A Method of GNSS Pseudorange Multiple Outlier Prior Detection Based on Observations Domain

It is difficult to distinguish gross error observations when conducting GNSS multiple outlier detection using a statistical hypothesis test method based on least squares.To solve theproblem, a new priordetecting method is proposed in this paper.The detecting measures are gaussian distribution and independent to each other.Thus multiple gross errors can be simultaneous exactly located by testing normality of the detecting measures. An experiment on the method is conducted with IGS data. Results are as follows:First, the detecting measures conform to normal distribution confirmed its good outlier detectable characteristics;Second,the multiple outlier detecting results showed over 92％ detection rate of this method;Atlast,theSPP results influenced by multiple gross errorswere significantly improved byreducing the weights of identified error observables.