顾及三维方向的GNSS坐标时间序列粗差探测方法

针对GNSS坐标时间序列中粗差为三维位置粗差的特点，提出综合三维坐标的迭代最小二乘粗差探测方法。通过实验模拟数据，分别利用3倍中误差（3σ）准则和四分位距（interquartile range, IQR）准则对单方向和综合N、E、U三方向的粗差进行探测。结果表明，2种判别准则中，综合考虑3个方向的探测效果远好于单方向的探测效果，且探测效率均超过95%；3σ准则的探测效率稍高于IQR准则，但误判率也较高。在实测数据中的应用表明，SOPAC(Scrips Orbit and Permanent Array Center)提供的部分raw类型全球IGS站时间序列中粗差占0.1%~4.7%；运用IQR准则综合3个方向的粗差探测得到的时间序列要比cleaned trend类型更加干净。

Gross Error Detection of GNSS Coordinate Time Series on Three-Dimensional Directions

Because the gross error in the time series of GNSS is a three-dimensional position gross error, this paper proposes an iterative least-squares gross error detection method on 3D coordinates. According to the experimental simulation data, the gross error detection in one direction and in three directions of north, east, and up is compared with the 3σ criterion and the interquartile range (IQR) criterion, respectively. The results show that the effect on three directions is much better than that in one direction. And the detection efficiency of three directions detection is beyond 95%. The 3σ criterion detected more gross errors than the IQR criterion, but the misjudgment rate is higher. Application in real GNSS time-series data shows that gross errors in the raw type of some global IGS stations provided by SOPAC (Scrips Orbit and Permanent Array Center) range from 0.1% to 4.7% of the station data. Time series detected by IQR gross error in three directions is cleaner than its cleaned trend type.