李斐^1,2,,
李文浩¹,
张胜凯^1,,,
雷锦韬¹,
张卿川¹,
袁乐先¹
1. 武汉大学中国南极测绘研究中心, 武汉 430079
2. 武汉大学测绘遥感信息工程国家重点实验室, 武汉 430079

基金项目: 国家重点研发计划项目（2017YFA0603104），国家自然科学基金重点项目（41531069）和武汉大学自主科研学科交叉类项目（2042017kf0209）资助


详细信息

作者简介: 李斐, 男, 1960年生, 博导, 主要从事固体地球物理、物理大地测量学方面的研究.E-mail:fli@whu.edu.cn

通讯作者: 张胜凯, 男, 1977年生, 副教授, 博士, 主要从事极地大地测量学研究.E-mail:zskai@whu.edu.cn

中图分类号: P228

收稿日期:2018-09-02
修回日期:2019-08-21
上线日期:2019-09-05

Spatiotemporal filtering for regional GNSS network in Antarctic Peninsula using independent component analysis

LI Fei^1,2,,
LI WenHao¹,
ZHANG ShengKai^1,,,
LEI JinTao¹,
ZHANG QingChuan¹,
YUAN LeXian¹
1. Chinese Antarctic Center of Surveying and Mapping, Wuhan University, Wuhan 430079, China
2. State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China


More Information

Corresponding author: ZHANG ShengKai,E-mail:zskai@whu.edu.cn

MSC: P228

--> Received Date: 02 September 2018
Revised Date: 21 August 2019
Available Online: 05 September 2019


摘要
摘要:高精度GNSS速度场是研究地壳垂向运动及板块运动的基础，能够为冰川均衡调整（Glacial Isostatic Adjustment，GIA）的建模提供外部检核和新的约束.共性误差（Common Mode Error，CME）是区域连续GNSS时间序列中存在的一种与时空相关的主要误差源，通过空间滤波可有效的降低共性误差的影响，提高坐标时间序列的精度.目前广泛采用的主分量分析法（Principal Component Analysis，PCA），基于二阶统计量（方差和协方差）进行处理，没有充分利用CME高阶统计信息.而独立分量分析ICA（Independent Component Analysis），引入高阶统计量，能够分离出统计独立的非高斯信号.以南极半岛地区的15个GNSS站点为例，由于某些站点存在强烈的局部效应，因此引入了因子分析法首先对异常站进行剔除，然后对比分析了PCA和ICA方法在南极半岛地区区域滤波结果.结果显示，ICA的滤波效果要优于PCA，ICA滤波前后E、N、U三个方向RMS平均降低44.69%、26.94%、34.87%，不确定度分别降低37.43%，44.58%，55.86%，有效的降低了GNSS残差序列的发散性和速度的不确定度.
关键词: GNSS时间序列/
ICA/
PCA/
共性误差/
南极半岛

Abstract:The high-accuracy GNSS velocity field is one of effective approaches to studying the regional crustal displacement, and it can also provide validations and constraints for Glacial Isostatic Adjustment (GIA) models. Common Mode Error (CME) is one of the major errors related to spatiotemporal distribution. Applying a spatiotemporal filter can reduce these effects and improve the precision of GNSS time series. The widely used Principal Component Analysis (PCA) is based on second-order statistics (variance and covariance), which cannot take full advantage of higher order statistics. The so-called Independent Component Analysis (ICA) introduces high-order statistics to help separate statistically independent non-Gaussian signals. For 15 GNSS stations in Antarctic Peninsula, we firstly introduce a factor analysis to remove those anomaly stations influenced by strong local effects. And then we compare the regional filter results derived from PCA and ICA. The result shows that ICA result is better than PCA result. The reduction of mean RMS is 44.69%, 26.94%, 34.87% in E, N, U components after applying ICA filter; the reduction of mean velocity uncertainty can reach up to 37.43%, 44.58%, 55.86%, respectively.
Key words:GNSS time series/
ICA/
PCA/
Common mode error/
Antarctic Peninsula



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=dqwlxb_15147