噪声源的时空分布及其对噪声互相关函数的影响——以ChinArray二期数据为例

噪声源的空间分布和季节变化会对噪声互相关函数中的信号产生一定影响.本文选取了ChinArrry二期台阵南部的322个宽频带地震台，利用其2013年9月至2016年6月的垂直分量连续记录计算了台站间的互相关函数，进而通过背景噪声能量流的方法，分析了周期频段4~8 s，8~12 s和12~20 s的噪声能量随时间的演化规律.结果表明，在不同频段，背景噪声的强度及优势来源方向均具明显的季节变化，且不同周期频段的噪声能量变化规律有所差异.总体而言，噪声能量在北半球冬季较强，夏季较弱，与全球海洋活动的季节性变化一致，能量优势来源方向也与全球海浪波高分布相符.同时，在10~20 s频段范围内，噪声互相关函数中存在较强的异常信号.该信号在环形台阵路径上的到时呈现随方位角的规律变化，且冬季较强，夏季较弱.基于走时的分析表明，该信号是由大西洋北部的一个强噪声源激发产生的.在特定路径上，该信号可能对频散提取产生干扰.研究表明，噪声源分布的不均匀性以及季节变化会对噪声互相关函数中信号的细节形态产生影响，进而影响格林函数的收敛程度，相关精细化研究应对噪声源的特性予以关注.

The distribution of noise source both in space and time and its influence on Noise Cross-correlation Functions

The reconstruction of Green's function by long time ambient noise cross-correlation has been widely used in seismology. The distribution of noise sources doesn't meet the theoretical assumption that they are uniformly distributed in both space and time, which can affect signals retrieved from Noise Cross-correlation Functions (NCFs). To further address this issue, using the continuous vertical records of 322 broadband seismic stations which are evenly distributed in a square in southern part of ChinArray Phase II, we calculate the NCFs from September 2013 to June 2016. The seasonal variation and azimuthal anisotropy are analyzed using the normalized background energy flux (NBEF) method for the period bands 4~8 s, 8~12 s and 12~20 s, respectively. Clear seasonal variation and directional anisotropy are observed for all periods and the characteristics of different periods are variable. Overall, the ambient noise energy is stronger in winter and weaker in summer. This is consistent with global oceanic activities which are stronger in winter and become weaker in summer in each local hemisphere. Moreover, strong precursors are observed in NCFs for period 10~20 s, showing clear azimuthal dependency in the NCFs having similar inter-station distance and different azimuths. The precursors are stronger in winter and weaker in summer. Analyzing their seasonality and arrival time suggests that they should be caused by one local noise source located in the northern Atlantic Ocean. The precursors may generate strong signals just before the Rayleigh waves and distort the dispersion measurement. This study suggests that the characteristics of the noise source, such as uneven distribution and seasonal variation, will affect the detail of the signals in NCFs and further the convergence of the Green's function. This phenomenon should be to be considered in noise cross-correlation studies.