GNSS站坐标时间序列噪声估计方法探讨

On the noise estimation method of GNSS station coordinate time series

为了研究GNSS站坐标时间序列噪声估计方法的差异,比较极大似然估计法、两种方差分量估计法(最小二乘方差分量估计法、最小范数二次无偏估计法)对不同时间长度GNSS站坐标时间序列噪声模型的估计效果,确定噪声模型估计的适宜长度为15 a,最佳估计方法为方差分量估计法,并选择最小二乘方差分量估计进行时间序列的噪声模型估计。中国区域15个GNSS站坐标时间序列的噪声模型估计结果表明,闪烁噪声是中国区域GNSS站坐标的主要噪声,且竖直方向的噪声强度远大于水平方向的。在此基础上,利用线性回归法分析水平方向和竖直方向噪声方差的相关性,指出水平方向的噪声能够引起竖直方向32%以上的噪声。

To study the difference of GNSS station coordinate time series noise estimation methods, maximum likelihood estimation(MLE),and two variance component estimation methods(least square variance component estimation(LS-VCE)and minimum norm quadratic unbiased estimation of variance components(MINQUE))were firstly compared and analyzed for the estimation effect of GNSS station coordinate time series noise model with different time length.The suitable length of noise model estimation was determined to be 15 years, the best estimation method was variance component estimation, and the least square variance component estimation was selected for noise model estimation of time series.The noise model estimation results of 15 GNSS stations in China showed that flicker noise was the main noise of GNSS stations in China, and the vertical noise intensity was much higher than the horizontal noise intensity.On this basis, the linear regression method was used to analyze the correlation between the horizontal and vertical noise variances.The analysis showed that the horizontal noise could explain more than 32% of the vertical noise.