Application of Improved Heuristic Segmentation Algorithm to Step Detection of GNSS Coordinate Time Series

Based on the heuristic segmentation algorithm, we introduce z-test and standard normal homogeneity(SNHT) test, proposes a new step detection method, which is applied to 20 CMONOC (Crustal Movement Observation Network of China) and 10 IGS (International GNSS Service) reference stations for nearly 5 years of coordinate time series. For the known step of IGS stations, the detection results show that the average accurate detection rates for three components of E, N, U are 68.5%, 71.3% and 64.8%, respectively, and we use the difference of the initial fitting residuals with average of 25 days before and after the step respectively, to repair the coordinates time series, the results are presented in good continuity. Based on the coordinate time series after eliminating the gross error, for the combination of known and detected step epochs, the results of estimation from coordinate time series analysis software show that:The mean standard deviation of velocity between the estimated velocity and the published velocity of CMONOC's stations for the E, N and U directions of all stations are 2.86 mm/a, 1.14 mm/a, 2.31 mm/a, respectively, which are significantly smaller than the average velocity standard deviation obtained by a combination of known and manually picked step epochs. The above results show that the improved heuristic segmentation algorithm can be applied to detect the step of coordinate time series.