Xiaofei Chen Research Group

Department of Earth and Space Sciences, SUSTech, China.

陈晓非课题组 南方科技大学，地球与空间科学系

Python 3 is required.

Anaconda environment is required for installation by make install.

目前仅支持 Python 3，推荐使用Anaconda配置Python环境。请确保您所安装和使用的Python以及编译器是同一个。

Make sure that the CUDA and Python in the same version for you install and run.

Before Installation: Download & Compile fftw Library.

在安装前请安装fftw库，可以在当前目录通过如下命令一键安装。

make fftw

After the installation of the fftw, install the CC-FJpy with the following commands

在安装fftw库之后可以在当前目录通过如下两行命令完成CC-FJpy的安装。

make
make install

If the machine has an NVIDIA graphics card and the nvcc compiler is already in the path, the GPU version of CC-FJpy will be installed. If the above conditions are not met, the CPU version will be installed. If you want to install only the CPU version, you can install it with the following commands.

若本机有英伟达显卡且nvcc编译器已经在路径中时，GPU版本的CC-FJpy将会被安装，如果不满足上述条件则会安装CPU版本。 若想仅安装CPU版本，则可以通过如下两行命令安装。

make cpu
make install

Please import the package before use its functions.

请在使用前先导入。

import ccfj

CCFs=ccfj.CC(
    npts,nsta,nf,fftlen,
    Pairs,startend,data,
    overlaprate=0.0,
    nThreads=8,
    fstride=1,
    ifonebit=0,
    ifspecwhittenning=1)

• npts: The number of points for data of one station read in（每次互相关每个台所使用数据点数，在程序中对于每个台是一致的，没有数据的点可以通过补0处理 ）

• nsta: The number of stations（台站总数）

• nf: The number of points of the frequency domain CC output （输出互相关的频率点数）

• fftlen: The number of points for one CC （单词互相关所用点数）

• Pairs: A numpy array for Station Pairs, the dtype should be np.int32 （要计算的台站对。一维数组,长度为nparis*2。要计算的第i个台站对互相关为Pairs[i*2,i*2+1]） You can use function GetStationPairs to generate Pairs.

• startend: A numpy array records the start point and end point for each station, the dtype should be np.int32. For example, if the npts length is for one day, and station A have all of A day's data. Then for station A the startend is [0,npts] （记录每个台的记录开始时间和结束时间的数组。）

• data: The seismic records, the dtype should be np.float32 （连续地震记录，长度为nsta*npts的一维数组。将每个台的记录（不满npts的点补0）依次排在该数组中。）

• overlaprate: The rate for overlap, do not >= 1

• nThreads: The number of Threads for reading data and CC （互相关计算中omp所使用的核数）

• fstride: The output frequency stride （输出的互相关对应的频率点数相对于原始数据傅里叶变换（fftlen）时频谱点的间隔）

• ifonebit: if perform onebit （是否使用onebit）

• ifspecwhittening: if perform specwhittenning （是否使用谱白化）

• CCFs: The output is the noise cross-correlation functions (CCFs) （输出为频率域互相关函数）

According to our experience, ifonebit and ifspecwhittening, pick one of them is ok.

根据我们的经验，ifonebit 和ifspecwhittening二者选其一即可。

The specific example, please read example_CC.ipynb.

具体例子请参考 example_CC.ipynb。

out = ccfj.fj_noise(uf,r,c,f,fstride=1,itype=1,func=0,num=20)

The F-J Method for CCFs

• uf: the CCFs in the frequency domain （频率域互相关函数,二维数组，大小为[互相关数，nf]）

• r: the list of station distances (unit: m) （对应的台间距数组，单位m，长度为互相关数）

• c: the list of phase velocities you want to calculate （对要计算的相速度数组，单位m/s）

• f: the list of frequencies. The number of points of f should be consistent with the columns of uf (频率域互相关对应的频率)

• fstride: stride of frequency for output （默认1即可）

• itype: 0 for trapezoidal integral; 1 for linear approximate （积分类型0梯形积分，1对格林函数进行线性逼近）

• func: 0 for Bessel function; 1 for Hankel function (使用的积分基底0贝塞尔函数，1汉克尔函数)

• out: the output dispersion spectrum （输出的频散谱）

• num: the number of threads for cpu version (defult 20) and the device number of gpu version (default 0) （若是cpu版本则对应并行cpu核数，若是gpu版本则对应了gpu id）

out = ccfj.fj_earthquake(u,r,c,f,fstride=1,itype=1,func=0,num =20)

• u: the records in time domain （时间域地震记录）

• r: the list of station distances (unit: m) （震中距）

• c: the list of phase velocities you want to calculate （相速度）

• f: the list of frequencies. The number of points of f should be consistent with the columns of uf （频率，这里频率需要对应输入u中时间域点数对应的频率）

• fstride: stride of frequency for output (默认1)

• itype: 0 for trapezoidal integral; 1 for linear approximate （积分类型0梯形积分，1对格林函数进行线性逼近）

• func: 0 for Bessel function; 1 for Hankel function (使用的积分基底0贝塞尔函数，1汉克尔函数)

• out: the output dispersion spectrum （输出的频散谱）

• num: the number of threads for cpu version (defult 20) and the device number of gpu version (default 0) （若是cpu版本则对应并行cpu核数，若是gpu版本则对应了gpu id）

This is mainly for earthquake

该函数主要针对地震或者主动源数据

out = ccfj.MWFJ(u,r,c,f,Fs,nwin,winl,winr,taper=0.9,fstride=1,itype=0,func=0, num=20)

• u: the records in time domain

• r: the list of station distances (unit: m)

• c: the list of phase velocities you want to calculate

• f: the list of frequencies. The number of points of f should be consistent with the columns of uf

• Fs: The sample frequency

• nwin: number of time windows （时间窗个数）

• winl: list of left side of time windows （每个记录时间窗的左端）

• winr: list of right side of time windows （每个记录时间窗的右端）

• fstride: stride of frequency for output

• itype: 0 for trapezoidal integral; 1 for linear approximate

• func: 0 for Bessel function; 1 for Hankel function

• out: the output dispersion spectrum

• num: the number of threads for cpu version (defult 20) and the device number of gpu version (default 0)

make clean
make uninstall

Wang, J., Wu, G., & Chen, X. (2019). Frequency‐Bessel Transform Method for Effective Imaging of Higher‐Mode Rayleigh Dispersion Curves From Ambient Seismic Noise Data. Journal of Geophysical Research: Solid Earth, 124(4), 3708-3723. doi:10.1029/2018jb016595

Wu, G.-x., Pan, L., Wang, J.-n., & Chen, X. (2020). Shear Velocity Inversion Using Multimodal Dispersion Curves From Ambient Seismic Noise Data of USArray Transportable Array. Journal of Geophysical Research: Solid Earth, 125(1), e2019JB018213. doi:10.1029/2019jb018213

Li, Z., & Chen, X., (2020). An Effective Method to Extract Overtones of Surface Wave from Array Seismic Records of Earthquake Events. Journal of Geophysical Research: Solid Earth, 125(3), e2019JB18511. doi:10.1029/2019jb018511

Li, Z., Zhou, J., Wu, G., Wang, J., Zhang, G., Dong, S., Pan, L., Yang, Z., Gao, L., Ma, Q., Ren, H., & Chen, X. (2021). CC-FJpy: A Python Package for seismic ambient noise cross-correlation and the frequency-Bessel transform method. Seismological Research Letters. doi:10.1785/0220210042