摘要
对于GPS区域网坐标序列中普遍存在的共模误差(CME),常规方法是利用堆栈空间滤波去除,通常假设CME是空域不变的;而主成分分析法(PCA)和Karhunen-Loeve(KLE)展开法都是把随时间变化的台站网时间序列分解成时间域的主分量和空间域的特征分量,不限制CME的本征。因此,本文尝试结合PCA和KLE方法对GPS区域网坐标序列进行空间滤波,并通过对美国中加州Carrizo平原的一个连续运行GPS监测网进行分析,表明PCA/KLE法可有效提取共模误差,提高站点坐标精度,增强空间滤波的稳健性。
For Common Mode Error （CME） existed in the regional GPS coordinate time series, it is usually removed by stacking spatial filtering which assumes that CME be invariable in the airspace field. While Principal Component Analysis （PCA） and Karhunen-Loeve Expansion methods （KLE） decompose the network time series into a set of principal components over the time and a characteristic component in the space domain, without limiting the intrinsic characteristic of CME. Therefore, this paper attempted to do the spatial filtering for a regional GPS coordinate time series via PCA and KLE. Some discussion was made for a continuous GPS monitoring network on the Carrizo Plain, California, USA, and the result showed that PCA/KLE method could extract the CME effectively, improve the accuracy of site coordinates and enhance the robustness of the spatial filtering.
出处
《测绘科学》
CSCD
北大核心
2014年第7期113-117,97,共6页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41374007
41174010
41374011)
精密工程与工业测量国家测绘地理信息局重点实验室开放基金项目(PF2013-9)
