Abstract:The algorithm of eliminating gross errors in robust estimation and gross error detection IQR criteria, and the orthogonal polynominal interpolation algorithm were studied. The experimental object is GPS singleday height time series of CORS fiducial stations in China from March 1, 1999 to December 31, 2009 lasting nearly 11 years. Through the gross error detection and interpolation tests of three selected base stations, the result shows that robust estimation is more effective than the IQR criterion to detect and remove gross error. When the interpolation being carried out by use of cubic spline function if the number of epoch is continuously short of over 5~10,then the result of interpolation will distorted easily.However,the result from fitting by use of loworder orthogonal polynomial will be stable,as usuall,the ideal result can be achieved with fitting two or three times.When the epoches are continuously short of three months the curve after fitting with orthgonal polynomial will still be basically consistent with the trend of the sequence,i.g. the result of intergolation is good.