采用变系数回归模型提取GPS坐标序列季节性信号  

EXTRACTING SEASONAL SIGNALS FROM CONTINUOUS GPS TIME SERIES BASED ON VARYING-COEFFICIENT REGRESSION MODELS

作　　者：卢辰龙[1] 匡翠林[1] 戴吾蛟[1,2] 卢骏[1] 

机构地区：[1]中南大学地球科学与信息物理学院,长沙410083 [2]湖南省精密工程测量与形变灾害监测重点实验室,长沙410083

出　　处：《大地测量与地球动力学》2014年第5期94-100,共7页Journal of Geodesy and Geodynamics

基　　金：国家973计划项目(2013CB73303);湖南省国土资源厅科技项目(2012-41)

摘　　要：提出采用变系数回归模型提取连续GPS坐标序列中包含的振幅时变季节性信号。对模拟数据及实际GPS坐标序列两组数据的分析结果表明,变系数回归模型在提取GPS坐标序列季节信号方面比传统模型更有效。经处理,GPS坐标序列能获得更合理的速度及噪声估计。Continuous GPS coordinate series contains modulated amplitude seasonal signals, and using the least-squares fitting the traditional model will introduce model error. According to the time-varying characteristics, a method to extract the modulated seasonal signals using the varying-coefficient regression model was proposed. Two groups of data simulated and the real series of GPS coordinate were analyzed with the method,the results show that the varying-coefficient regression model in extracting the seasonal signals of GPS coordinate series is more eftective than traditional model, and the GPS time series processed by the method can get a more reasonable estimation of speed and noise.

关 键 词：GPS坐标序列 时变季节性信号 变系数回归模型 最小二乘拟合 最大似然估计 

分 类 号：P228.41[天文地球—大地测量学与测量工程]