GPS基准站坐标分量噪声的时间序列与分类特征

Time Sequences of Noise of the Coordinate Component of GPS Reference Stations and Its Classified Characteristics

根据地震预测对异常信息(长期、中期和短期异常)与噪声分离(白噪声、闪烁噪声和随机漫步噪声)的需要,文中应用小波方法对中国大陆的有关GPS基准站坐标分量和测站间坐标分量进行了分离,并将它们归纳为四类高频、中频、较低频和低频。通过对高频、中频和较低频的时间序列特征进行分类分析,获得如下看法:①高频类可能主要是白噪声造成的,其离散的程度较中频类略大,或许其中在某时刻包含临震前兆和震时形变信息;②中频类主要突出闪烁噪声的特征,或许其中在某时刻包含着短临前兆信息;③较低频类若剔除年和半年周期成分则显示随机漫步噪声的特点,或许其中包含着中期前兆信息;④低频类主要体现趋势性形变信息;⑤高频与中频类噪声的大小一般与观测季节明显有关,噪声最大的时段大约是8月份,噪声最小的时段大约是2月份;⑥判定不同频域信息的显著性时,其噪声要进行相应的综合或分离;⑦一定的空间域内的测站间坐标增量噪声平均小于测站相应坐标的噪声,而且噪声大小随时间的变化更加突出,甚至相差达一倍以上。这些可能说明,虽然这些噪声是GPS观测所固有的,但并非完全是独立的,测站间越近,共性的成分越突出。所有这些将为我们变形分析时信息显著性的判断、方法的选择、资料处理时权的确定、观测的最佳时段选择、跟踪监测方案制定和参考基准站的确定等提供重要的参考。

According to the needs for abnormal information （long-term, medium-term and short-term） and noise separation （white noise, flickering noise and random walking noise） in earthquake prediction, the wavelet method to separate the coordinate components of relational GPS reference stations and coordinate components between monitoring stations in the Chinese mainland has been applied, and it can be divided into 4 categories-high frequency, medium frequency, lower frequency and low frequency. By classified analyzing of the characteristics of time sequence of high frequency, medium frequency and lower frequency, we obtain some ideas as follows： ① The high frequency category is mainly caused by white noise, and its deviation degree is greater than the medium frequency category, maybe on certain time point there contains imminent precursory information before earthquake and co-seismic deformation information. ② In medium frequency category, the characteristic of flickering noise is mostly outstanding , maybe on certain time point there contains short-term precursory information. ③ If the periodic components in year and semi-year are eliminated, the lower frequency category exhibits the characteristic of random walking noise, maybe there contains medium-term precursory information. ④ The low frequency category mainly embodies the information of trending deformation.⑤ The magnitude of high and medium frequency noises are obviously related to the observation season generally, the period of time with the maximal noise is approximately in August, and the period of time with the minimal noise is approximately in February.⑥ In determining the information prominence of different frequency, the noise should be integrated or separated accordingly. ⑦ In certain special domain, the noises of coordinate increments between stations are averagelv smaller than the noises of corresponding coordinate of stations, and the variation of magnitude of noises is much more clear with the time. These may suggest that although the noises are inherent in GPS observation, they are not completely independent, and the more close between stations, the more common features of components. All these will provide us an important references in judging information prominence and selecting the method for analyzing deformation, also in determining of weight, selecting the best observation period of time, establishing the monitoring project as well as the reference stations in processing data.