Inversion of Source

  • This is the core of the MCMTpy, and all parameters are written into a JSON file named sample_dc.json. Some brief descriptions of the parameters are included here following the definination.

  • We do not recommend using relative paths because of the possibility of errors. Absolute paths are preferred.

Path

GFs_json_file

  • The path to the Json-file used to calculate Green Function Database.

GFs_file

  • The path to the Green Function Database.

Raw_data_file

  • The path to the Raw data ASDF file.

Source_tag

  • Raw data ASDF file’s source name which setted in the data preprocess function.

Raw_p_n0

  • The number of points before the first arrival. Be consistent with the GFs’ samples_before_first_arrival.

Dt

  • Sampling rate of the raw data.

Output_path

  • Inversion result output file path.

Source Time Function

Trapezoid_stf_mode

  • Whether to use pyfk provided trapezoid shaped source time function. Set to be .true. you need to set the following parameters correctly:

    dura rise delta

dura

  • Duration time (s).

rise

  • Rise time (s).

delta

  • The time interval (s).

Stf_file

  • Stf_file should be an SAC file as the source time function when set Trapezoid_stf_mode to be .false..

Note

To read the data from the Stf_file file, keep the sampling rate of SAC data consistent with the data and GFs.

MPI and Chains_n

MPI_n

  • CPU number

Chains_n

  • A total of n Markov Chains are sampled on each core (CPU).

N

  • Each Markov Chains’ sampling number.

Misfit

Geometrical_spreading_mode

  • Whether the geometric diffusion correction is turned on or not. Waveform normalization is used when it set to be .false., and please set the magnitude M0 not to be inverted in Fixed_FM. Only one of Geometrical_spreading_mode and Amplitude_normalization_mode can be .true..

Distance_0

  • The reference epicentral distance when Geometrical_spreading_mode set to be .true..

Amplitude_normalization_mode

  • Waveform normalization mode.Only one of Geometrical_spreading_mode and Amplitude_normalization_mode can be .true..

Use_file_mode

  • When set it to be .false., don’t need to provide Raw_data_inv_info file. Setting following parameters (Raw_data_inv_info_writed, NET_STA, P_inv_comp, S_inv_comp, Surf_inv_comp, P_win, S_win, Surf_win, Phase_weight, Distance_weight, P_maxlag, S_maxlag, Surf_maxlag, P_filter, S_filter, Surf_filter) correctly, MCMTpy will generate this file automatically, so you can change individual parameter values and use this file for inversion. Raw_data_inv_info_writed is the filename that generate this file automatically. When set it to be .true., Raw_data_inv_info is needed which is the path to inversion file.

Raw_data_inv_info

  • The path to inversion parameters file. It includes the time window length, filtering range, weight and other parameters used for data. When set Use_file_mode to be .true., it must be provided.

  • Raw_data_inv_info.txt format.

 1YN   EYA   
 2Lat_lon_depth	26.108801   99.947502   0.000000   
 3P_inv_comp	yes	yes	no	
 4S_inv_comp	no	no	no	
 5Surf_inv_comp	yes	yes	yes	
 6P_win		-10.00	3.40	
 7S_win		-5.00	25.20	
 8Surf_win		-12.00	50.20	
 9Phase_weight	1.0	1.0	1.0	
10Distance_weight	1.0	
11P_maxlag		5.0000	
12S_maxlag		15.0000	
13Surf_maxlag	10.0000	
14P_filter		0.010	0.20	
15S_filter		0.010	0.100	
16Surf_filter	0.005	0.100	
17
18YN   YUL   
19Lat_lon_depth	25.885401   99.371696   0.000000   
20P_inv_comp	yes	yes	no	
21S_inv_comp	no	no	no	
22Surf_inv_comp	yes	yes	yes	
23P_win		-10.00	6.20	
24S_win		-5.00	25.20	
25Surf_win		-12.00	50.20	
26Phase_weight	1.0	1.0	1.0	
27Distance_weight	1.0	
28P_maxlag		5.0000	
29S_maxlag		15.0000	
30Surf_maxlag	10.0000	
31P_filter		0.010	0.200	
32S_filter		0.010	0.100	
33Surf_filter	0.005	0.100

Raw_data_inv_info_writed

  • Raw_data_inv_info_writed is the filename that generate this file automatically When set Use_file_mode to be .true..

NET_STA

  • The information contained in the two-dimensional list: [network_name] [station_name] [station_lat] [source_lon] [station_depth]

P_inv_comp

  • P wave Z/R/T component. ‘yes’: inversion. ‘no’: no inversion. Select the component you want to invert.

S_inv_comp

  • S wave Z/R/T component. ‘yes’: inversion. ‘no’: no inversion. Select the component you want to invert.

Surf_inv_comp

  • Surf wave Z/R/T component. ‘yes’: inversion. ‘no’: no inversion. Select the component you want to invert.

P_win

  • P wave P_win=[-1,2] unit/s; The P wave fitting band, and ‘-1’ means 1 second before the P wave, ‘2’ means 2 second after the P wave

S_win

  • S wave S_win…

Surf_win

  • Surf wave Surf_win…

Phase_weight

  • P/S/Surface wave inversion weight

Distance_weight

  • The weight of each station in NET_STA

P_maxlag

  • The maximum number of moveable sampling points to fitting waveform of P wave. It is generally half the period of the P phase waveform. unit/s.

S_maxlag

  • The maximum number of moveable sampling points to fitting waveform of S wave.

Surf_maxlag

  • The maximum number of moveable sampling points to fitting waveform of Surf wave.

P_filter

  • P wave filter range, [freqmin, freqmax].

S_filter

  • S wave filter range, [freqmin, freqmax].

Surf_filter

  • Surf wave filter range, [freqmin, freqmax].

Inversion parameters

InvType

  • The type of source, mt(moment tensor) dc(double-couple) sf(single-couple) ep(explosion).

FM0

  • The initial value of the inversion, it’s a 1-d list.

  • dc:[m0, str, dip rake, lat, lon, depth, t0] || mt:[m0, m_xx, m_xy, m_xz, m_yy, m_yz, m_zz, lat, lon, depth, t0] || ep:[m0]

  • The default initial value of t0 is usually 0.

Fixed_FM

  • The length of the list corresponds to FM0, and ‘constant’ means fixed to a constant and unchanged in the inversion process. ‘variable’ is represented as a variable and participates in the inversion.

FM_boundary

  • A 2-d list, FM’s boundary.

Noise_level_waveform

  • Noise level estimation of raw data.

MISFIT0

  • The initial error of the objective function, usually set to a very large value.

alpha_max

  • The initial value of alpha.

alpha_min

  • The final value of alpha.

a

  • The alpha function’s parameter.

b

  • The alpha function’s parameter.

sigma_mw

  • The standard deviation of the new solution of Mw.

sigma_mt

  • The standard deviation of the new solution of Moment Tensor.

sigma_dc

  • The standard deviation of the new solution of Strike Dip Rake.

sigma_sf

  • The standard deviation of the new solution of Single Couple (not test now).

sigma_ep

  • The standard deviation of the new solution of Explosion (not test now).

sigma_lat_lon

  • The standard deviation of the new solution of Latitude and Longitude (degree).

sigma_depth

  • The standard deviation of the new solution of Source Depth (km).

sigma_t

  • The standard deviation of the new solution of T0 (s).

Alpha

  • The alpha function is shown in following script, you can test different parameters alpha_max alpha_min a b and plot thefigure.

 1#!/usr/bin/env python3
 2# -*- coding: utf-8 -*-
 3import math
 4import numpy as np
 5import matplotlib.pyplot as plt
 6
 7# parameters
 8N=5e3
 9a=1000                  # 0.2*N
10b=100                   # 0.04*N
11alpha_max=100
12alpha_min=0
13
14#------------------------------------------------------------#
15# we expect no parameters need to be changed below
16#------------------------------------------------------------#
17tt=np.linspace(0, round(N), num=round(N))
18yy=np.zeros(len(tt))
19for i in range(0,len(tt)):
20    yy[i]=(alpha_max-alpha_min)*(a+math.exp((tt[0]-a)/b))/(a+math.exp((tt[i]-a)/b))+alpha_min
21
22fig, axs = plt.subplots(1,1)
23axs.plot(tt,yy, linestyle='-', color='red', lw=3, alpha=0.6)
24axs.set_xlabel("Sample Number",fontsize=17)
25axs.set_ylabel("Alpha Value",fontsize=17)
26fig.show()

Logging file

  • MCMTpy will generates 1 logging files when it do Inversion, that is Inv_para_info.txt. They record all the parameter information of inversion process. You can find them in Output_path path.

Example Double-Couple Inv

  • The example_path need to de changed to your path, and run this notebook to set sample_dc.json file.

 1#!/usr/bin/env python3
 2# -*- coding: utf-8 -*-
 3
 4import os
 5import sys
 6import json5
 7
 8
 9# The root directory of the project
10example_path = '/Users/yf/3.Project/8.MCMTpy/MCMTpy-master/data/example_yunnan'
11
12
13#------------------------------------------------------------#
14# we expect no parameters need to be changed below
15#------------------------------------------------------------#
16# 1. get notebook path
17notebook_path = sys.path[0]
18os.chdir(notebook_path)
19# change path to Green_Funcs
20os.chdir('../YN.202105212148_Inv/dc_inv')
21
22# 2.show build_GFs.json 
23filename = 'sample_dc.json'
24with open(filename, 'r',encoding='utf-8') as f1:
25    sample_dc_json=json5.load(f1)
26
27# 3.change parameters with absolute path
28sample_dc_json["GFs_json_file"] = os.path.join(example_path,"Green_Funcs/build_GFs_new.json")
29sample_dc_json["GFs_file"] =os.path.join(example_path,"Green_Funcs/GFs/GFs_splits" )          
30sample_dc_json["Raw_data_file"] = os.path.join(example_path,"YN.202105212148_Inv/YN.202105212148_MCMTpy/YN.202105212148.h5")              
31sample_dc_json["Output_path"] = os.path.join(example_path,"YN.202105212148_Inv/dc_inv/Output_YN.202105212148_dc")         
32sample_dc_json["Raw_data_inv_info"] = os.path.join(example_path,"YN.202105212148_Inv/dc_inv/Raw_data_inv_info.txt")     
33sample_dc_json["Raw_data_inv_info_writed"] = os.path.join(example_path,"YN.202105212148_Inv/dc_inv/Raw_data_inv_info_writed.txt") 
34sample_dc_json_new = os.path.join(example_path,'YN.202105212148_Inv/dc_inv/sample_dc_new.json')
35
36sample_dc_json["MPI_n"] = 4                         # CPU num
37sample_dc_json["Chains_n"] = 4                      # Each MK-chains num
38sample_dc_json["N"] = 1e1                           # each chain‘s search number
39
40sample_dc_json["alpha_max"] = 100                   # The initial value of alpha
41sample_dc_json["alpha_min"] = 0                     # The final value of alpha
42sample_dc_json["a"] = 10                     
43sample_dc_json["b"] = 10               
44
45with open(sample_dc_json_new,'w') as f2:
46    json5.dump(sample_dc_json, f2, indent=2)
47f2.close()
48
49# 4.run MCMTpy in shell
50os.chdir(notebook_path)
51os.chdir('../YN.202105212148_Inv/dc_inv')
52# !MCMTpy  sample MH  -c ./sample_dc_new.json #> sample.log

  • Now you can run MCMTpy in bash:

    $ MCMTpy  sample MH  -c ./sample_dc_new.json > sample.log

Note

Please follow the above parameter instructions and set all parameters correctly before running the program. Otherwise, it is easy to report an error!