HHT-EEMD用于IGS站高程时间序列分析

利用经验模态分解（EMD）和整体经验模态分解（EEMD）方法，将BJFS站2000~2015年高程时间序列进行分解，发现其不仅存在1 a、0.5 a、0.25 a、2个月、1个月以及长周期等周期项，还存在以往方法很难探测出来的近似2 a周期信号。与EMD分解结果对比，整体经验模态分解可以明显减弱模式混叠现象。对各分量进行Hilbert 变换（HHT），得到时间-频率-能量的Hilbert 频谱图。结果表明，年周期和2 a周期变化是高程运动的主要贡献项。利用小波变换方法对比验证EEMD的分解结果表明，与小波分析相比，EEMD重构信号与高程序列差异的RMS更小，证明了HHT-EEMD方法在数据资料分析过程中的有效性。对环境负载及GRACE负载造成的测站位移进行功率谱分析得出，环境负载确实会造成IGS站高程时间序列的1 a、0.5 a以及季节性运动，GRACE负载还验证了2 a信号的存在。

Application of the HHT-EEMD Approach in Analysis of GPS Height Time Series

In this paper, we use empirical mode decomposition (EMD) and ensemble empirical mode decomposition (EEMD) to analyze BJFS height time series, spanning from 2000 to 2015. Results show that not only the signals at well-known periods (annual, semi-annual, 3-month, 2-month and 1 month), but also quasi-biennial signals, which were difficult to detect previously, are found. Compared with EMD, it is proven that the EEMD approach can weaken the mode mixing phenomenon significantly. The decomposed intrinsic mode function (IMF) is transformed with a Hilbert algorithm, from which we observe that the annual and quasi-biennial oscillations are the major contributors to the height variations. Compared with wavelet analysis, the RMS of the difference between EEMD reconstruction signal and elevation sequence is smaller, showing the effectiveness of the HHT-EEMD method in data analysis. The annual and semi-annual variations in the height are mainly attributed to the surface mass loading by integrating geophysical models, and the existence of the quasi-biennial signals are verified by GRACE data.