用堆栈(Stack ing)、主分量分析(PCA)和Karhunen-Loeve展开(KLE)等3种区域滤波方法分别处理了南加州12个GPS连续观测站5年的坐标时间序列。结果显示,3种方法都能有效地提取出坐标时间序列中存在的共性误差,使站点坐标的精度显著提高,但优缺点及适用范围各有不同。
Three regional filtering methods:stacking, principal component analysis and Karhunen-Loeve expansion, were respectively applied to analyze the time series of daily coordinate for five years from twelve stations of Southern California integrated GPS network. The results indicate that the common mode errors can be extracted with all three methods and thereby the precision of coordinates will be improved remarkably, but they have respective advantages and disadvantages.