zhangyiss/TomoATT
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
f3f1778f777d0aeb15c65033996bd0144d2aa8c6
TomoATT/examples/utils/functions_for_data.py

2415 lines
87 KiB
Python
Raw Normal View History

initial upload
2025-12-17 10:53:43 +08:00
# %%
# Initialization, class definition, and declaration.
import os
import math
from obspy import UTCDateTime
import numpy as np
import copy
class Event(): # class of earthquake
def __init__(self):
self.name = "nan" # evname1 Earthquake name, recommended as "earthquake".
self.id = -1
self.lat = 0.0
self.lon = 0.0
self.dep = 0.0
self.mag = 0.0
self.ortime = UTCDateTime(1999,1,1,0,0,0)
self.Nt = 0 # Number of the absolute traveltime of earthquake
self.Ncs_dt = 0 # Number of the commmon source differential traveltime of earthquake
self.Ncr_dt = 0 # Number of the commmon receiver differential traveltime of earthquake
self.t = {} # stname1+phase -> (stname1, phase, time, data_weight)
self.cs_dt = {} # stname1 + stname2 + phase -> (stname1, stname2, phase, dif_time, data_weight)
self.cr_dt = {} # stname1 + evname2 + phase -> (stname1, evname2, phase, dif_time, data_weight)
self.azi_gap = 360.0 # the max azimuthal gap of each earthquake
self.misfit = {} # traveltime residual of the data, the difference between real data and synthetic data, used for evaluation. stname or stname1+stname2 or stname1+evname2 -> residual
self.tag = {} # additional tags for the earthquake, e.g., azi_gap, weight. (azimuthal gap, weight of the earthquake)
class Station():
def __init__(self):
self.name = "nan" # stname1, recommend: network.stname
self.id = -1
self.lat = 0.0
self.lon = 0.0
self.ele = 0.0
self.tag = {} # additional tags for the station, e.g., wright
# %% [markdown]
# Functions: some basic auxiliary functions for processing data
# %%
# function： cal_dis(lat1, lon1,lat2, lon2) (in kilometers)， cal_azimuth(lat1, lon1, lat2, lon2) (degree) calculate epicentral distance (km) and azimuth (degree)
def cal_dis(lat1, lon1,lat2, lon2, R = 6371):
latitude1 = (math.pi/180)*lat1
latitude2 = (math.pi/180)*lat2
longitude1 = (math.pi/180)*lon1
longitude2= (math.pi/180)*lon2
# Therefore, the spherical distance between points A and B is:{arccos[sinb*siny+cosb*cosy*cos(a-x)]}*R
# Radius of the earth
if((lat1-lat2)**2+(lon1-lon2)**2<0.000001):
return 0
d = math.acos(math.sin(latitude1)*math.sin(latitude2)+ math.cos(latitude1)*math.cos(latitude2)*math.cos(longitude2-longitude1))/math.pi*180
return d * 2 * math.pi * R / 360
def cal_azimuth(lat1, lon1, lat2, lon2):
lat1_rad = lat1 * math.pi / 180
lon1_rad = lon1 * math.pi / 180
lat2_rad = lat2 * math.pi / 180
lon2_rad = lon2 * math.pi / 180
y = math.sin(lon2_rad - lon1_rad) * math.cos(lat2_rad)
x = math.cos(lat1_rad) * math.sin(lat2_rad) - math.sin(lat1_rad) * math.cos(lat2_rad) * math.cos(lon2_rad - lon1_rad)
brng = math.atan2(y, x) * 180 / math.pi
if((lat1-lat2)**2+(lon1-lon2)**2<0.0001):
return 0
return float((brng + 360.0) % 360.0)
# %%
# Function: Coordinate rotation rotate_src_rec(ev_info, st_info, theta0, phi0, psi): rotate to the new coordinate system, satisfying the center point transformation r0, t0, p0 -> r0, 0, 0 and an anticlockwise rotation angle psi.
# Satisfying the center point transformation r0, t0, p0 -> r0, 0, 0 and an anticlockwise rotation angle psi.
import numpy as np
RAD2DEG = 180/np.pi
DEG2RAD = np.pi/180
R_earth = 6371.0
# Spherical coordinates to Cartesian coordinate
def rtp2xyz(r,theta,phi):
x = r * np.cos(theta*DEG2RAD) * np.cos(phi*DEG2RAD)
y = r * np.cos(theta*DEG2RAD) * np.sin(phi*DEG2RAD)
z = r * np.sin(theta*DEG2RAD)
return (x,y,z)
# Cartesian coordinates to Spherical coordinate
def xyz2rtp(x,y,z):
# theta: -90~90; phi: -180~180
r = np.sqrt(x**2+y**2+z**2)
theta = np.arcsin(z/r)
phi = np.arcsin(y/r/np.cos(theta))
idx = np.where((phi > 0) & (x*y < 0))
phi[idx] = np.pi - phi[idx]
idx = np.where((phi < 0) & (x*y > 0))
phi[idx] = -np.pi - phi[idx]
# for i in range(phi.size):
# if(phi[i] > 0 and x[i]*y[i] < 0):
# phi[i] = np.pi - phi[i]
# if(phi[i] < 0 and x[i]*y[i] > 0):
# phi[i] = -np.pi - phi[i]
return (r,theta*RAD2DEG,phi*RAD2DEG)
# anti-clockwise rotation along x-axis
def rotate_x(x,y,z,theta):
new_x = x
new_y = y * np.cos(theta*DEG2RAD) + z * -np.sin(theta*DEG2RAD)
new_z = y * np.sin(theta*DEG2RAD) + z * np.cos(theta*DEG2RAD)
return (new_x,new_y,new_z)
# anti-clockwise rotation along y-axis
def rotate_y(x,y,z,theta):
new_x = x * np.cos(theta*DEG2RAD) + z * np.sin(theta*DEG2RAD)
new_y = y
new_z = x * -np.sin(theta*DEG2RAD) + z * np.cos(theta*DEG2RAD)
return (new_x,new_y,new_z)
# anti-clockwise rotation along z-axis
def rotate_z(x,y,z,theta):
new_x = x * np.cos(theta*DEG2RAD) + y * -np.sin(theta*DEG2RAD)
new_y = x * np.sin(theta*DEG2RAD) + y * np.cos(theta*DEG2RAD)
new_z = z
return (new_x,new_y,new_z)
# spherical Rotation
# rotate to the new coordinate, satisfying the center r0,t0,p0 -> r0,0,0 and a anticlockwise angle psi
def rtp_rotation(t,p,theta0,phi0,psi):
# step 1: r,t,p -> x,y,z
(x,y,z) = rtp2xyz(1.0,t,p)
# step 2: anti-clockwise rotation with -phi0 along z-axis: r0,t0,p0 -> r0,t0,0
(x,y,z) = rotate_z(x,y,z,-phi0)
# step 3: anti-clockwise rotation with theta0 along y-axis: r0,t0,0 -> r0,0,0
(x,y,z) = rotate_y(x,y,z,theta0)
# # step 4: anti-clockwise rotation with psi along x-axis
(x,y,z) = rotate_x(x,y,z,psi)
# step 5: x,y,z -> r,t,p
(new_r,new_t,new_p) = xyz2rtp(x,y,z)
return (new_t,new_p)
def rtp_rotation_reverse(new_t,new_p,theta0,phi0,psi):
# step 1: r,t,p -> x,y,z
(x,y,z) = rtp2xyz(1.0,new_t,new_p)
# step 2: anti-clockwise rotation with -psi along x-axis
(x,y,z) = rotate_x(x,y,z,-psi)
# step 3: anti-clockwise rotation with -theta0 along y-axis: r0,0,0 -> r0,t0,0
(x,y,z) = rotate_y(x,y,z,-theta0)
# step 4: anti-clockwise rotation with phi0 along z-axis: r0,t0,0 -> r0,t0,p0
(x,y,z) = rotate_z(x,y,z,phi0)
# step 5: x,y,z -> r,t,p
(r,t,p) = xyz2rtp(x,y,z)
return (t,p)
def rotate_src_rec(ev_info,st_info,theta0,phi0,psi):
ev_info_rotate = {}
st_info_rotate = {}
# rotate earthquakes
for key_ev in ev_info:
ev = ev_info[key_ev]
ev_lat = np.array([ev.lat]); ev_lon = np.array([ev.lon])
(ev_lat,ev_lon,) = rtp_rotation(ev_lat,ev_lon,theta0,phi0,psi)
ev.lat = ev_lat[0]; ev.lon = ev_lon[0]
ev_info_rotate[key_ev] = ev
# rotate stations
for key_st in st_info:
st = st_info[key_st]
st_lat = np.array([st.lat]); st_lon = np.array([st.lon])
(st_lat,st_lon) = rtp_rotation(st_lat,st_lon,theta0,phi0,psi)
st.lat = st_lat[0]; st.lon = st_lon[0]
st_info_rotate[key_st] = st
return (ev_info_rotate,st_info_rotate)
def rotate_src_rec_reverse(ev_info_rotate,st_info_rotate,theta0,phi0,psi):
ev_info = {}
st_info = {}
# rotate earthquakes
for key_ev in ev_info_rotate:
ev = ev_info_rotate[key_ev]
ev_lat = np.array([ev.lat]); ev_lon = np.array([ev.lon])
(ev_lat,ev_lon,) = rtp_rotation_reverse(ev_lat,ev_lon,theta0,phi0,psi)
ev.lat = ev_lat[0]; ev.lon = ev_lon[0]
ev_info[key_ev] = ev
# rotate stations
for key_st in st_info_rotate:
st = st_info_rotate[key_st]
st_lat = np.array([st.lat]); st_lon = np.array([st.lon])
(st_lat,st_lon) = rtp_rotation_reverse(st_lat,st_lon,theta0,phi0,psi)
st.lat = st_lat[0]; st.lon = st_lon[0]
st_info[key_st] = st
return (ev_info,st_info)
# %%
# # Function: Coordinate rotation rotate_src_rec(ev_info, st_info, theta0, phi0, psi): rotate to the new coordinate system, satisfying the center point transformation r0, t0, p0 -> r0, 0, 0 and an anticlockwise rotation angle psi.
# # Satisfying the center point transformation r0, t0, p0 -> r0, 0, 0 and an anticlockwise rotation angle psi.
# import numpy as np
# RAD2DEG = 180/np.pi
# DEG2RAD = np.pi/180
# R_earth = 6371.0
# # Spherical coordinates to Cartesian coordinate
# def rtp2xyz(r,theta,phi):
# x = r * np.cos(theta*DEG2RAD) * np.cos(phi*DEG2RAD)
# y = r * np.cos(theta*DEG2RAD) * np.sin(phi*DEG2RAD)
# z = r * np.sin(theta*DEG2RAD)
# return (x,y,z)
# # Cartesian coordinates to Spherical coordinate
# def xyz2rtp(x,y,z):
# # theta: -90~90; phi: -180~180
# r = np.sqrt(x**2+y**2+z**2)
# theta = np.arcsin(z/r)
# phi = np.arcsin(y/r/np.cos(theta))
# if(phi > 0 and x*y < 0):
# phi = np.pi - phi
# if(phi < 0 and x*y > 0):
# phi = -np.pi - phi
# return (r,theta*RAD2DEG,phi*RAD2DEG)
# # anti-clockwise rotation along x-axis
# def rotate_x(x,y,z,theta):
# new_x = x
# new_y = y * np.cos(theta*DEG2RAD) + z * -np.sin(theta*DEG2RAD)
# new_z = y * np.sin(theta*DEG2RAD) + z * np.cos(theta*DEG2RAD)
# return (new_x,new_y,new_z)
# # anti-clockwise rotation along y-axis
# def rotate_y(x,y,z,theta):
# new_x = x * np.cos(theta*DEG2RAD) + z * np.sin(theta*DEG2RAD)
# new_y = y
# new_z = x * -np.sin(theta*DEG2RAD) + z * np.cos(theta*DEG2RAD)
# return (new_x,new_y,new_z)
# # anti-clockwise rotation along z-axis
# def rotate_z(x,y,z,theta):
# new_x = x * np.cos(theta*DEG2RAD) + y * -np.sin(theta*DEG2RAD)
# new_y = x * np.sin(theta*DEG2RAD) + y * np.cos(theta*DEG2RAD)
# new_z = z
# return (new_x,new_y,new_z)
# # spherical Rotation
# # rotate to the new coordinate, satisfying the center r0,t0,p0 -> r0,0,0 and a anticlockwise angle psi
# def rtp_rotation(t,p,theta0,phi0,psi):
# # step 1: r,t,p -> x,y,z
# (x,y,z) = rtp2xyz(1.0,t,p)
# # step 2: anti-clockwise rotation with -phi0 along z-axis: r0,t0,p0 -> r0,t0,0
# (x,y,z) = rotate_z(x,y,z,-phi0)
# # step 3: anti-clockwise rotation with theta0 along y-axis: r0,t0,0 -> r0,0,0
# (x,y,z) = rotate_y(x,y,z,theta0)
# # # step 4: anti-clockwise rotation with psi along x-axis
# (x,y,z) = rotate_x(x,y,z,psi)
# # step 5: x,y,z -> r,t,p
# (new_r,new_t,new_p) = xyz2rtp(x,y,z)
# return (new_t,new_p)
# def rtp_rotation_reverse(new_t,new_p,theta0,phi0,psi):
# # step 1: r,t,p -> x,y,z
# (x,y,z) = rtp2xyz(1.0,new_t,new_p)
# # step 2: anti-clockwise rotation with -psi along x-axis
# (x,y,z) = rotate_x(x,y,z,-psi)
# # step 3: anti-clockwise rotation with -theta0 along y-axis: r0,0,0 -> r0,t0,0
# (x,y,z) = rotate_y(x,y,z,-theta0)
# # step 4: anti-clockwise rotation with phi0 along z-axis: r0,t0,0 -> r0,t0,p0
# (x,y,z) = rotate_z(x,y,z,phi0)
# # step 5: x,y,z -> r,t,p
# (r,t,p) = xyz2rtp(x,y,z)
# return (t,p)
# def rotate_src_rec(ev_info,st_info,theta0,phi0,psi):
# ev_info_rotate = {}
# st_info_rotate = {}
# # rotate earthquakes
# for key_ev in ev_info:
# ev = ev_info[key_ev]
# (ev.lat,ev.lon,) = rtp_rotation(ev.lat,ev.lon,theta0,phi0,psi)
# ev_info_rotate[key_ev] = ev
# # rotate stations
# for key_st in st_info:
# st = st_info[key_st]
# (st.lat,st.lon) = rtp_rotation(st.lat,st.lon,theta0,phi0,psi)
# st_info_rotate[key_st] = st
# return (ev_info_rotate,st_info_rotate)
# def rotate_src_rec_reverse(ev_info_rotate,st_info_rotate,theta0,phi0,psi):
# ev_info = {}
# st_info = {}
# # rotate earthquakes
# for key_ev in ev_info_rotate:
# ev = ev_info_rotate[key_ev]
# (ev.lat,ev.lon) = rtp_rotation_reverse(ev.lat,ev.lon,theta0,phi0,psi)
# ev_info[key_ev] = ev
# # rotate stations
# for key_st in st_info_rotate:
# st = st_info_rotate[key_st]
# (st.lat,st.lon) = rtp_rotation_reverse(st.lat,st.lon,theta0,phi0,psi)
# st_info[key_st] = st
# return (ev_info,st_info)
# %%
# linear_regression(X,Y)
def linear_regression(X,Y):
slope,intercept = np.polyfit(X,Y,deg=1)
fitted_values = slope * X + intercept
residual = Y - fitted_values
SEE = np.std(residual)
return (slope,intercept,SEE)
# %% [markdown]
#
# Functions: obtain target information from ev_info and st_info
# %%
# function: output the [lon,lat,dep,weight] of the earthquake
def data_lon_lat_dep_wt_ev(ev_info):
lat = []
lon = []
dep = []
weight = []
for key in ev_info:
lat.append(ev_info[key].lat)
lon.append(ev_info[key].lon)
dep.append(ev_info[key].dep)
try:
weight.append(ev_info[key].tag["weight"])
except:
weight.append(1.0)
return [np.array(lon),np.array(lat),np.array(dep),np.array(weight)]
# %%
# function: output the [lon, lat, dep, ortime] of the earthquake
def data_ev_loc(ev_info):
lat = []
lon = []
dep = []
ortime = []
for key in ev_info:
lat.append(ev_info[key].lat)
lon.append(ev_info[key].lon)
dep.append(ev_info[key].dep)
ortime.append(ev_info[key].ortime.timestamp)
return [np.array(lon),np.array(lat),np.array(dep),np.array(ortime)]
# %%
# function: output the [lon,lat,dep,weight] of the station
def data_lon_lat_ele_wt_st(ev_info,st_info):
names = {}
lat = []
lon = []
ele = []
weight = []
for key_ev in ev_info:
for key_t in ev_info[key_ev].t: # absolute traveltime data
name_st = ev_info[key_ev].t[key_t][0]
names[name_st] = name_st
for key_t in ev_info[key_ev].cs_dt: # common source differential traveltime data
name_st = ev_info[key_ev].cs_dt[key_t][0]
names[name_st] = name_st
name_st = ev_info[key_ev].cs_dt[key_t][1]
names[name_st] = name_st
for key_t in ev_info[key_ev].cr_dt: # common receiver differential traveltime data
name_st = ev_info[key_ev].cr_dt[key_t][0]
names[name_st] = name_st
for name in names: # only output the station which has data
lat.append(st_info[name].lat)
lon.append(st_info[name].lon)
ele.append(st_info[name].ele)
try:
weight.append(st_info[name].tag["weight"])
except:
weight.append(1.0)
return [np.array(lon),np.array(lat),np.array(ele),np.array(weight)]
# %%
# function: output the [dis,time] of all data
def data_dis_time(ev_info,st_info):
all_dis = []
all_time = []
for key_ev in ev_info:
lat_ev = ev_info[key_ev].lat
lon_ev = ev_info[key_ev].lon
dep_ev = ev_info[key_ev].dep
for key_t in ev_info[key_ev].t:
all_time.append(ev_info[key_ev].t[key_t][2])
lat_st = st_info[ev_info[key_ev].t[key_t][0]].lat
lon_st = st_info[ev_info[key_ev].t[key_t][0]].lon
ele_st = st_info[ev_info[key_ev].t[key_t][0]].ele
dis = math.sqrt(cal_dis(lat_ev,lon_ev,lat_st,lon_st)**2 + (dep_ev+ele_st/1000)**2)
all_dis.append(dis)
return [np.array(all_dis),np.array(all_time)]
# %%
# function: output the [epidis,time] of all data
def data_epidis_time(ev_info,st_info):
all_dis = []
all_time = []
for key_ev in ev_info:
lat_ev = ev_info[key_ev].lat
lon_ev = ev_info[key_ev].lon
for key_t in ev_info[key_ev].t:
all_time.append(ev_info[key_ev].t[key_t][2])
lat_st = st_info[ev_info[key_ev].t[key_t][0]].lat
lon_st = st_info[ev_info[key_ev].t[key_t][0]].lon
dis = cal_dis(lat_ev,lon_ev,lat_st,lon_st)**2
all_dis.append(dis)
return [np.array(all_dis),np.array(all_time)]
# %%
# function: output the [cs_dt] of all data
def data_cs_dt(ev_info):
all_time = []
for key_ev in ev_info:
for key_dt in ev_info[key_ev].cs_dt:
all_time.append(ev_info[key_ev].cs_dt[key_dt][3])
return np.array(all_time)
# %%
# Function: data_dis_time_phase(ev_info, st_info, phase_list) Given a list of seismic phases, output the [epicentral distance, arrival time] for each phase.
def data_dis_time_phase(ev_info,st_info,phase_list):
all_dis = {}
all_time = {}
for phase in phase_list:
all_dis[phase] = []
all_time[phase] = []
for key_ev in ev_info:
lat_ev = ev_info[key_ev].lat
lon_ev = ev_info[key_ev].lon
dep_ev = ev_info[key_ev].dep
for key_t in ev_info[key_ev].t:
phase = key_t.split("+")[1]
if (not phase in phase_list):
continue
all_time[phase].append(ev_info[key_ev].t[key_t][2])
lat_st = st_info[ev_info[key_ev].t[key_t][0]].lat
lon_st = st_info[ev_info[key_ev].t[key_t][0]].lon
ele_st = st_info[ev_info[key_ev].t[key_t][0]].ele
dis = math.sqrt(cal_dis(lat_ev,lon_ev,lat_st,lon_st)**2 + (dep_ev+ele_st/1000)**2)
all_dis[phase].append(dis)
for phase in phase_list:
all_dis[phase] = np.array(all_dis[phase])
all_time[phase] = np.array(all_time[phase])
return [all_dis,all_time]
# %%
# Function: data_lon_lat_dep_wt_ev(ev_info) Outputs the lines connecting station and earthquake for traveltime data as [line_x, line_y].
def data_line(ev_info,st_info):
line_x = []
line_y = []
for key_ev in ev_info:
lat_ev = ev_info[key_ev].lat
lon_ev = ev_info[key_ev].lon
for key_t in ev_info[key_ev].t:
lat_st = st_info[ev_info[key_ev].t[key_t][0]].lat
lon_st = st_info[ev_info[key_ev].t[key_t][0]].lon
line_x.append([lon_ev,lon_st])
line_y.append([lat_ev,lat_st])
return [line_x,line_y]
# %% [markdown]
# Functions: discard some data in ev_info and st_info based on selection criteria
# %%
# Function: limit_ev_region(ev_info, lat1, lat2, lon1, lon2, dep1, dep2) Delete the earthquakes that are out of the specified region.
def limit_ev_region(ev_info,lat_min,lat_max,lon_min,lon_max,dep_min,dep_max):
count_delete = 0
del_key_ev = []
for key_ev in ev_info:
ev = ev_info[key_ev]
lat = ev.lat
lon = ev.lon
dep = ev.dep
if (lat < min(lat_min,lat_max) or lat > max(lat_min,lat_max) \
or lon < min(lon_min,lon_max) or lon > max(lon_min,lon_max) \
or dep < min(dep_min,dep_max) or dep > max(dep_min,dep_max)):
del_key_ev.append(key_ev)
count_delete += 1
del_key_t = []
for key_t in ev_info[key_ev].cr_dt:
name_ev2 = ev_info[key_ev].cr_dt[key_t][1]
lat2 = ev_info[name_ev2].lat
lon2 = ev_info[name_ev2].lon
dep2 = ev_info[name_ev2].dep
if (lat2 < min(lat_min,lat_max) or lat2 > max(lat_min,lat_max) \
or lon2 < min(lon_min,lon_max) or lon2 > max(lon_min,lon_max) \
or dep2 < min(dep_min,dep_max) or dep2 > max(dep_min,dep_max)):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev_info[key_ev].cr_dt[key_t]
ev_info[key_ev].Ncr_dt = len(ev_info[key_ev].cr_dt)
for key_ev in del_key_ev:
del ev_info[key_ev]
print("delete %d events out of the region, now %d earthquakes are retained within the study region"%(count_delete,len(ev_info)))
return ev_info
# %%
# Function: limit_st_region(ev_info, st_info, lat1, lat2, lon1, lon2) Delete the stations that are out of the specified region.
def limit_st_region(ev_info,st_info,lat1,lat2,lon1,lon2):
for key_ev in ev_info:
# delete the station out of the region in the absolute traveltime data
del_key_t = []
for key_t in ev_info[key_ev].t:
name_st = ev_info[key_ev].t[key_t][0]
lat_st = st_info[name_st].lat
lon_st = st_info[name_st].lon
if(lat_st < min(lat1,lat2) or lat_st > max(lat1,lat2) or lon_st < min(lon1,lon2) or lon_st > max(lon1,lon2)):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev_info[key_ev].t[key_t]
ev_info[key_ev].Nt = len(ev_info[key_ev].t)
# delete the station out of the region in the common source differential traveltime data
del_key_t = []
for key_t in ev_info[key_ev].cs_dt:
name_st1 = ev_info[key_ev].cs_dt[key_t][0]
lat_st1 = st_info[name_st1].lat
lon_st1 = st_info[name_st1].lon
name_st2 = ev_info[key_ev].cs_dt[key_t][1]
lat_st2 = st_info[name_st2].lat
lon_st2 = st_info[name_st2].lon
if(lat_st1 < min(lat1,lat2) or lat_st1 > max(lat1,lat2) or lon_st1 < min(lon1,lon2) or lon_st1 > max(lon1,lon2) \
or lat_st2 < min(lat1,lat2) or lat_st2 > max(lat1,lat2) or lon_st2 < min(lon1,lon2) or lon_st2 > max(lon1,lon2)):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev_info[key_ev].cs_dt[key_t]
ev_info[key_ev].Ncs_dt = len(ev_info[key_ev].cs_dt)
# delete the station out of the region in the common receiver differential traveltime data
del_key_st = []
for key_t in ev_info[key_ev].cr_dt:
name_st = ev_info[key_ev].cr_dt[key_t][0]
lat_st = st_info[name_st].lat
lon_st = st_info[name_st].lon
if(lat_st < min(lat1,lat2) or lat_st > max(lat1,lat2) or lon_st < min(lon1,lon2) or lon_st > max(lon1,lon2)):
del_key_st.append(key_t)
for key_t in del_key_st:
del ev_info[key_ev].cr_dt[key_t]
ev_info[key_ev].Ncr_dt = len(ev_info[key_ev].cr_dt)
return ev_info
# %%
# Function: limit_epi_dis(ev_info, st_info, epi_dis1, epi_dis2) Delete the stations with epicentral distance in the range from epi_dis1 to epi_dis2.
def limit_epi_dis(ev_info,st_info,epi_dis1,epi_dis2):
for key_ev in ev_info:
ev = ev_info[key_ev]
lat_ev = ev.lat
lon_ev = ev.lon
# delete the absolute traveltime data
del_key_t = []
for key_t in ev.t:
stname = ev.t[key_t][0]
lat_st = st_info[stname].lat
lon_st = st_info[stname].lon
dis = cal_dis(lat_ev, lon_ev, lat_st, lon_st)
if (dis > epi_dis1 and dis < epi_dis2):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev.t[key_t]
ev.Nt = len(ev.t)
# delete the common source differential traveltime data
del_key_t = []
for key_t in ev.cs_dt:
for i in range(2):
stname = ev.t[key_t][i]
lat_st = st_info[stname].lat
lon_st = st_info[stname].lon
dis = cal_dis(lat_ev, lon_ev, lat_st, lon_st)
if (dis > epi_dis1 and dis < epi_dis2):
del_key_t.append(key_t)
break
for key_t in del_key_t:
del ev.cs_dt[key_t]
ev.Ncs_dt = len(ev.cs_dt)
# delete the common receiver differential traveltime data
del_key_t = []
for key_t in ev.cr_dt:
stname = ev.cr_dt[key_t][0]
lat_st = st_info[stname].lat
lon_st = st_info[stname].lon
dis = cal_dis(lat_ev, lon_ev, lat_st, lon_st)
if (dis > epi_dis1 and dis < epi_dis2):
del_key_t.append(key_t)
lat_ev2 = ev_info[ev.cr_dt[key_t][1]].lat
lon_ev2 = ev_info[ev.cr_dt[key_t][1]].lon
dis = cal_dis(lat_ev2, lon_ev2, lat_st, lon_st)
if (dis > epi_dis1 and dis < epi_dis2):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev.cr_dt[key_t]
ev.Ncr_dt = len(ev.cr_dt)
ev_info[key_ev] = ev
return ev_info
# %%
# Function: limit_data_residual(ev_info, st_info, slope, intercept, up, down) Limit the data within the range defined by the line time = dis * slope + intercept and the bounds up and down.
# remove outliers, only retain data satisfying: slope * dis + intercept + down < time < slope * dis + intercept + up
def limit_data_residual(ev_info,st_info,slope,intercept,up,down):
for key_ev in ev_info:
lat_ev = ev_info[key_ev].lat
lon_ev = ev_info[key_ev].lon
dep_ev = ev_info[key_ev].dep
del_key_t = []
for key_t in ev_info[key_ev].t:
name_st = ev_info[key_ev].t[key_t][0]
lat_st = st_info[name_st].lat
lon_st = st_info[name_st].lon
ele_st = st_info[name_st].ele
dis = math.sqrt(cal_dis(lat_ev,lon_ev,lat_st,lon_st)**2 + (dep_ev+ele_st/1000)**2)
residual = ev_info[key_ev].t[key_t][2] - (slope*dis+intercept)
if (residual < down or residual > up):
del_key_t.append(key_t)
for key_t in del_key_t:
del ev_info[key_ev].t[key_t]
for key_ev in ev_info:
ev_info[key_ev].Nt = len(ev_info[key_ev].t)
return ev_info
# %%
# Function: limit_data_phase(ev_info, phase_list) Retain only the specified seismic phases.
def limit_data_phase(ev_info,phase_list):
for key_ev in ev_info:
# process the absolute traveltime data
new_t = {}
for key_t in ev_info[key_ev].t:
phase = ev_info[key_ev].t[key_t][1]
if phase in phase_list:
new_t[key_t] = ev_info[key_ev].t[key_t]
ev_info[key_ev].t = new_t
ev_info[key_ev].Nt = len(ev_info[key_ev].t)
# process the common source differential traveltime data
new_t = {}
for key_t in ev_info[key_ev].cs_dt:
phase = ev_info[key_ev].cs_dt[key_t][2]
phase = phase.split(",")[0]
if phase in phase_list:
new_t[key_t] = ev_info[key_ev].cs_dt[key_t]
ev_info[key_ev].cs_dt = new_t
ev_info[key_ev].Ncs_dt = len(ev_info[key_ev].cs_dt)
# process the common receiver differential traveltime data
new_t = {}
for key_t in ev_info[key_ev].cr_dt:
phase = ev_info[key_ev].cr_dt[key_t][2]
phase = phase.split(",")[0]
if phase in phase_list:
new_t[key_t] = ev_info[key_ev].cr_dt[key_t]
ev_info[key_ev].cr_dt = new_t
ev_info[key_ev].Ncr_dt = len(ev_info[key_ev].cr_dt)
return ev_info
# %%
# Function: limit_min_Nt(min_Nt_thd, ev_info) Delete the earthquakes with the number of data less than min_Nt_thd.
def limit_min_Nt(min_Nt_thd, ev_info):
Nev = len(ev_info)
del_key_ev = []
for key_ev in ev_info:
if(ev_info[key_ev].Nt < min_Nt_thd):
del_key_ev.append(key_ev)
for key_ev in del_key_ev:
del ev_info[key_ev]
print("Original data set has %d earthquakes, %d earthquakes are deleted, %d earthquakes are retained"%(Nev,len(del_key_ev),len(ev_info)))
return ev_info
# %%
# Function: limit_azi_gap(gap_thd) Calculate the azimuthal gap for all events and delete events with a gap greater than gap_thd.
def limit_azi_gap(gap_thd,ev_info,st_info):
Nev = len(ev_info)
del_key_ev = []
for key_ev in ev_info:
ev = ev_info[key_ev]
gap = cal_azi_gap(ev,st_info)
if (gap > gap_thd):
del_key_ev.append(key_ev)
else:
ev_info[key_ev].tag["azi_gap"] = gap
for key_ev in del_key_ev:
del ev_info[key_ev]
print("Original data set has %d earthquakes, %d earthquakes are deleted, %d earthquakes are retained"%(Nev,len(del_key_ev),len(ev_info)))
return ev_info
# Function: cal_azi_gap(ev, st_info) Calculate the azimuthal gap of a single earthquake.
def cal_azi_gap(ev,st_info):
azi_all = []
lat_ev = ev.lat
lon_ev = ev.lon
stlist = {}
for key in ev.t:
stname = ev.t[key][0]
if (not stname in stlist):
lat_st = st_info[stname].lat
lon_st = st_info[stname].lon
azi = cal_azimuth(lat_ev, lon_ev, lat_st, lon_st)
azi_all.append(azi)
stlist[stname] = 1
azi_all.sort()
if(len(azi_all) < 2):
return 360.0
else:
gap = 0.0
for i in range(len(azi_all)-1):
gap = max(gap,azi_all[i+1] - azi_all[i])
gap = max(gap,azi_all[0] + 360 - azi_all[-1])
return gap
# %%
# Function: limit_earthquake_decluster_Nt(ev_info, dlat, dlon, ddep, Top_N) Divide the region into several subdomains, sort by the number of arrival times, and retain only the top Top_N earthquakes with the most arrival times in each box.
# option 3, declustering. Divide the region into several subdomains, retain the Top N earthquakes in terms of the number of arrival times in each subdomain.
def limit_earthquake_decluster_Nt(ev_info,dlat,dlon,ddep,Top_N):
# subdivide earthquakes into different subdomains
[ev_info,tag2name] = tag_event_cluster(dlat,dlon,ddep,ev_info)
# sort earthquakes in the same subdomain
# Sort the quality of earthquakes within each tag based on the number of arrivals.
tag2name = sort_cluster_Nt(ev_info, tag2name)
# only retain Top_N earthquakes in each subdomain
# Within each tag, prioritize selecting the top Top_N earthquakes.
[ev_info,tag2name] = limit_decluster(ev_info, tag2name,Top_N)
return ev_info
# Function: tag_event_cluster(size_lat, size_lon, size_dep, ev_info) Subdivide the study area, assign each earthquake to a subregion, and place it in a tag.
def tag_event_cluster(size_lat,size_lon,size_dep,ev_info):
tag2name = {}
for key_ev in ev_info:
name = ev_info[key_ev].name
lat = ev_info[key_ev].lat
lon = ev_info[key_ev].lon
dep = ev_info[key_ev].dep
tag = "%d_%d_%d"%(math.floor(lon/size_lon),math.floor(lat/size_lat),math.floor(dep/size_dep))
ev_info[key_ev].tag["cluster"] = tag
if (tag in tag2name):
tag2name[tag].append(name)
else:
tag2name[tag] = []
tag2name[tag].append(name)
return [ev_info,tag2name]
# Function: sort_cluster_Nt(ev_info, tag2name) Sort the quality of earthquakes within each tag based on the number of arrivals.
def sort_cluster_Nt(ev_info, tag2name):
for key_tag in tag2name:
names_ev = tag2name[key_tag]
Nt = []
for key_ev in names_ev:
Nt.append(len(ev_info[key_ev].t))
# Sort the earthquakes within each tag based on the number of arrivals.
sorted_Nt = sorted(enumerate(Nt), key=lambda x: x[1], reverse=True)
tag2name[key_tag] = []
for index, Nt in sorted_Nt:
tag2name[key_tag].append(names_ev[index])
return tag2name
# Function: limit_cluster(ev_info, tag2name, Max) Prioritize selecting the top Max earthquakes within each tag.
def limit_decluster(ev_info, tag2name, Max):
del_key_ev = []
for key_tag in tag2name:
names_ev = tag2name[key_tag]
if(len(names_ev) > Max):
tag2name[key_tag] = names_ev[0:Max]
for i in range(Max,len(names_ev)): # Delete earthquakes that exceed the threshold in the sorted list.
del_key_ev.append(names_ev[i])
for key_ev in del_key_ev:
del ev_info[key_ev]
return [ev_info,tag2name]
# %% [markdown]
# Functions: assign weights to earthquakes, stations, and data
# %%
# Function: box_weighting_ev(ev_info, dlat, dlon, ddep) Assign box-weight to the earthquakes.
def box_weighting_ev(ev_info,dlon,dlat,ddep):
# categorization
distribute = {}
all_tag_wt = {}
for key_ev in ev_info:
lat_id = math.floor((ev_info[key_ev].lat) / dlat)
lon_id = math.floor((ev_info[key_ev].lon) / dlon)
dep_id = math.floor((ev_info[key_ev].dep) / ddep)
tag = '%d_%d_%d'%(lat_id,lon_id,dep_id)
if (tag in distribute):
distribute[tag] += 1
else:
distribute[tag] = 1
max_weight = 0
for tag in distribute:
all_tag_wt[tag] = 1.0/math.sqrt(distribute[tag])
max_weight = max(max_weight,all_tag_wt[tag])
for key_ev in ev_info:
lat_id = math.floor((ev_info[key_ev].lat) / dlat)
lon_id = math.floor((ev_info[key_ev].lon) / dlon)
dep_id = math.floor((ev_info[key_ev].dep) / ddep)
tag = '%d_%d_%d'%(lat_id,lon_id,dep_id)
ev_info[key_ev].tag["weight"] = all_tag_wt[tag]/max_weight
return ev_info
# %%
# Function: geographical_weighting_ev_rough(ev_info, dlat, dlon, ddep) Assign geographical weighting to the earthquakes roughly.
def geographical_weighting_ev_rough(ev_info,dlat,dlon,ddep,coefficient = 0.5):
# categorization
distribute = {}
all_tag_wt = {}
for key_ev in ev_info:
lat_id = int(ev_info[key_ev].lat/dlat)
lon_id = int(ev_info[key_ev].lon/dlat)
dep_id = int(ev_info[key_ev].dep/ddep)
tag = '%d_%d_%d'%(lat_id,lon_id,dep_id)
if (tag in distribute):
distribute[tag] += 1
else:
distribute[tag] = 1
# Calculate the weight of each category.
delta0 = 0
for tag1 in distribute:
tmp1 = tag1.split('_')
# evlat1 = float(tmp1[0])*dlat; evlon1 = float(tmp1[1])*dlon; evdep1 = float(tmp1[2])*ddep
for tag2 in distribute:
tmp2 = tag2.split('_')
# evlat2 = float(tmp2[0])*dlat; evlon2 = float(tmp2[1])*dlon; evdep2 = float(tmp2[2])*ddep
# distance of id
delta_tp = math.sqrt((int(tmp1[0]) - int(tmp2[0]))**2 + (int(tmp1[1]) - int(tmp2[1]))**2 + (int(tmp1[2]) - int(tmp2[2]))**2)
delta0 = delta0 + distribute[tag1] * distribute[tag2] * delta_tp
delta0 = delta0/(len(ev_info)**2) * coefficient
max_weight = 0.0
for tag1 in distribute:
tmp1 = tag1.split('_')
# evlat1 = float(tmp1[0])*dlat; evlon1 = float(tmp1[1])*dlon; evdep1 = float(tmp1[2])*ddep
weight = 0
for tag2 in distribute:
tmp2 = tag2.split('_')
# evlat2 = float(tmp2[0])*dlat; evlon2 = float(tmp2[1])*dlon; evdep2 = float(tmp2[2])*ddep
delta_tp = math.sqrt((int(tmp1[0]) - int(tmp2[0]))**2 + (int(tmp1[1]) - int(tmp2[1]))**2 + (int(tmp1[2]) - int(tmp2[2]))**2)
weight = weight + math.exp(-(delta_tp/delta0)**2) * distribute[tag2]
all_tag_wt[tag1] = (1.0/weight)
max_weight = max(max_weight,1.0/weight)
# Assign weights to each earthquake based on its tag.
for key_ev in ev_info:
lat_id = int(ev_info[key_ev].lat/dlat)
lon_id = int(ev_info[key_ev].lon/dlon)
dep_id = int(ev_info[key_ev].dep/ddep)
tag = '%d_%d_%d'%(lat_id,lon_id,dep_id)
ev_info[key_ev].tag["weight"] = all_tag_wt[tag]/max_weight
return ev_info
# %%
# Function: box_weighting_st(ev_info, st_info, dlat, dlon) Assign geographical weighting to the stations roughly.
def box_weighting_st(ev_info,st_info,dlon,dlat):
[lon_ev,lat_ev,dep_ev,wt_ev] = data_lon_lat_dep_wt_ev(ev_info)
# Integrate all involved stations.
wt_st = {}
name_st = {}
for key_ev in ev_info:
for key_t in ev_info[key_ev].t:
name_rec = ev_info[key_ev].t[key_t][0]
wt_st[name_rec] = -1.0
name_st[name_rec] = 1
# categorization
distribute = {}
all_tag_wt = {}
# Count the number of stations in each subdomain.
for key_st in name_st:
lat_id = math.floor((st_info[key_st].lat) / dlat)
lon_id = math.floor((st_info[key_st].lon) / dlon)
tag = '%d_%d'%(lat_id,lon_id)
if (tag in distribute):
distribute[tag] += 1
else:
distribute[tag] = 1
max_weight = 0
for tag in distribute:
all_tag_wt[tag] = 1.0/math.sqrt(distribute[tag])
max_weight = max(max_weight,all_tag_wt[tag])
# Assign weights to each station based on its tag.
for key_st in name_st:
lat_id = math.floor((st_info[key_st].lat) / dlat)
lon_id = math.floor((st_info[key_st].lon) / dlon)
tag = '%d_%d'%(lat_id,lon_id)
wt_st[key_st] = all_tag_wt[tag]/max_weight
# modify weight tag in st_info
for key_t in wt_st:
st_info[key_t].tag["weight"] = wt_st[key_t]
# modify weight of abs data ev_info
for key_ev in ev_info:
for key_t in ev_info[key_ev].t:
name_rec = ev_info[key_ev].t[key_t][0]
ev_info[key_ev].t[key_t][3] = wt_st[name_rec]
return [ev_info,st_info]
# %%
# Function: geographical_weighting_st(ev_info,st_info) Assign geographical weighting to the stations roughly.
def geographical_weighting_st(ev_info,st_info,coefficient = 0.5):
# Integrate all involved stations.
wt_st = {}
name_st = {}
for key_ev in ev_info:
for key_t in ev_info[key_ev].t:
name_rec = ev_info[key_ev].t[key_t][0]
wt_st[name_rec] = -1.0
name_st[name_rec] = 1
# Calculate the weight of each station.
delta0 = 0
for key_st1 in name_st:
stlat1 = st_info[key_st1].lat
stlon1 = st_info[key_st1].lon
for key_st2 in name_st:
stlat2 = st_info[key_st2].lat
stlon2 = st_info[key_st2].lon
delta_tp = cal_dis(stlat1,stlon1,stlat2,stlon2)
delta0 = delta0 + delta_tp
delta0 = delta0/(len(wt_st)**2)*coefficient
max_weight = 0.0
for key_st1 in name_st:
stlat1 = st_info[key_st1].lat
stlon1 = st_info[key_st1].lon
weight = 0
for key_st2 in name_st:
stlat2 = st_info[key_st2].lat
stlon2 = st_info[key_st2].lon
delta_tp = cal_dis(stlat1,stlon1,stlat2,stlon2)
weight = weight + math.exp(-(delta_tp/delta0)**2)
wt_st[key_st1] = (1.0/weight)
max_weight = max(max_weight,1.0/weight)
for key_st1 in wt_st:
wt_st[key_st1] = wt_st[key_st1]/max_weight
# Add weight to each data point in the earthquakes.
for key_ev in ev_info:
for key_t in ev_info[key_ev].t:
name_rec = ev_info[key_ev].t[key_t][0]
if (not name_rec in wt_st):
ValueError("The station of the data is not in the calculation list")
if (len(ev_info[key_ev].t[key_t])==3):
ev_info[key_ev].t[key_t].append(wt_st[name_rec])
elif (len(ev_info[key_ev].t[key_t])==4):
ev_info[key_ev].t[key_t][3] = wt_st[name_rec]
else:
ValueError("Error in the weight information of the absolute traveltime data")
# modify weight tag in st_info
for key_t in wt_st:
st_info[key_t].tag["weight"] = wt_st[key_t]
# modify weight of abs data ev_info
for key_ev in ev_info:
for key_t in ev_info[key_ev].t:
name_rec = ev_info[key_ev].t[key_t][0]
ev_info[key_ev].t[key_t][3] = wt_st[name_rec]
return [ev_info,st_info]
# %% [markdown]
# Function: add noise into data
# %%
# Function：assign_gaussian_noise():
def assign_gaussian_noise(ev_info,sigma):
# Record which seismic phases correspond to each station.
st2phase = {} # Station name -> [Keys of absolute arrival time data related to this station]
for key_ev in ev_info:
# Absolute arrival time noise
for key_t in ev_info[key_ev].t:
stname = ev_info[key_ev].t[key_t][0]
ev_info[key_ev].t[key_t][2] = ev_info[key_ev].t[key_t][2] + np.random.normal(0,sigma)
if(stname in st2phase):
st2phase[stname].append(key_t)
else:
st2phase[stname] = [key_t]
for key_ev in ev_info:
# Double-difference arrival time noise
for key_dt in ev_info[key_ev].cs_dt:
stname1 = ev_info[key_ev].cs_dt[key_dt][0]
stname2 = ev_info[key_ev].cs_dt[key_dt][1]
t1 = -999
t2 = -999
# Search for the arrival time of the data.
if (stname1 in st2phase):
for key_t in st2phase[stname1]:
if (key_t in ev_info[key_ev].t):
t1 = ev_info[key_ev].t[key_t][2]
break
if (stname2 in st2phase):
for key_t in st2phase[stname2]:
if (key_t in ev_info[key_ev].t):
t2 = ev_info[key_ev].t[key_t][2]
break
if (t1 == -999 or t2 == -999):
# If there is no absolute arrival time data, the double-difference data residuals increase by a factor of sqrt(2) in noise.
ev_info[key_ev].cs_dt[key_dt][3] = ev_info[key_ev].cs_dt[key_dt][3] + np.random.normal(0,sigma*np.sqrt(2))
print('no data: ', key_ev, key_dt)
else:
# If there is absolute arrival time data, the double-difference data is obtained by subtraction.
ev_info[key_ev].cs_dt[key_dt][3] = t1 - t2
# Common station double-difference arrival time
for key_dt in ev_info[key_ev].cr_dt:
stname = ev_info[key_ev].cr_dt[key_dt][0]
key_ev2 = ev_info[key_ev].cr_dt[key_dt][1]
t1 = -999
t2 = -999
# Search for the arrival time of the data.
if (stname in st2phase):
for key_t in st2phase[stname]:
if (key_t in ev_info[key_ev].t):
t1 = ev_info[key_ev].t[key_t][2]
break
else:
print('not found 1: ', key_ev, key_t)
for key_t in st2phase[stname]:
if (key_t in ev_info[key_ev2].t):
t2 = ev_info[key_ev2].t[key_t][2]
break
else:
print('not found 2: ', key_ev, key_t)
if (t1 == -999 or t2 == -999):
# If there is no absolute arrival time data, the double-difference data residuals increase by a factor of sqrt(2) in noise.
ev_info[key_ev].cr_dt[key_dt][3] = ev_info[key_ev].cr_dt[key_dt][3] + np.random.normal(0,sigma*np.sqrt(2))
print('no data: ', key_ev, key_dt)
else:
# If there is absolute arrival time data, the double-difference data is obtained by subtraction.
ev_info[key_ev].cr_dt[key_dt][3] = t1 - t2
return ev_info
# %%
# Function：assign_uniform_noise_to_ev():
def assign_uniform_noise_to_ev(ev_info, range_lat, range_lon, range_dep, range_time):
# Loop through all earthquakes and assign noise to them.
ev_noise = {} # Name of the earthquake -> noise of [lat,lon,dep,ortime]
# loop list of earthquakes
for key_ev in ev_info:
evname = key_ev
if (evname in ev_noise):
print("error: repeated earthquake name")
exit()
else:
# generate noise
ev_noise[evname] = np.random.uniform(-1,1,4) * np.array([range_lat,range_lon,range_dep,range_time])
# Add noise to each data point.
for key_ev in ev_info:
# Absolute arrival time noise
for key_t in ev_info[key_ev].t:
ev_info[key_ev].t[key_t][2] = ev_info[key_ev].t[key_t][2] - ev_noise[key_ev][3]
# Double-difference arrival time noise (double-difference arrival time remains unchanged)
# Common station double-difference arrival time
for key_dt in ev_info[key_ev].cr_dt:
key_ev2 = ev_info[key_ev].cr_dt[key_dt][1]
if (key_ev2 in ev_noise):
ev_info[key_ev].cr_dt[key_dt][3] = ev_info[key_ev].cr_dt[key_dt][3] - ev_noise[key_ev][3] + ev_noise[key_ev2][3]
else:
print("earthquake %s is not included in ev_list"%(key_ev2))
ev_noise[key_ev2] = np.random.uniform(-1,1,4) * np.array([range_lat,range_lon,range_dep,range_time])
ev_info[key_ev].cr_dt[key_dt][3] = ev_info[key_ev].cr_dt[key_dt][3] - ev_noise[key_ev][3] + ev_noise[key_ev2][3]
# Add noise to each earthquake.
for key_ev in ev_noise:
ev_info[key_ev].lat = ev_info[key_ev].lat + ev_noise[key_ev][0]
ev_info[key_ev].lon = ev_info[key_ev].lon + ev_noise[key_ev][1]
ev_info[key_ev].dep = abs(ev_info[key_ev].dep + ev_noise[key_ev][2])
ev_info[key_ev].ortime = ev_info[key_ev].ortime + ev_noise[key_ev][3]
return ev_info
# %% [markdown]
# Functions: generate differential traveltime
# %%
# Function: generate_cs_dif(ev_info, st_info, dis_thd, azi_thd) Generate double-difference arrival times from absolute arrival times, with inter-station distance less than dis_thd and azimuthal difference less than azi_thd.
# function: generate common source differential traveltime data from absolute traveltime data, the stations separation is less than dis_thd, the azimuth difference is less than azi_thd
def generate_cs_dif(ev_info,st_info,dis_thd,azi_thd):
count_t = 0
count_cs_dt = 0
for key_ev in ev_info:
ev = ev_info[key_ev]
lat_ev = ev.lat
lon_ev = ev.lon
# traverse all arrival times
name_st_list = [] # names of stations
t_list = [] # traveltime
wt_list = [] # weight
for key_t in ev.t:
name_st_list.append(ev.t[key_t][0])
t_list.append(ev.t[key_t][2])
wt_list.append(ev.t[key_t][3])
count_t += 1
# search for possible double-difference arrival times
for id_st1 in range(len(name_st_list)-1):
name_st1 = name_st_list[id_st1]
lat_st1 = st_info[name_st1].lat
lon_st1 = st_info[name_st1].lon
t_st1 = t_list[id_st1]
wt_st1 = wt_list[id_st1]
for id_st2 in range(id_st1+1,len(name_st_list)):
name_st2 = name_st_list[id_st2]
lat_st2 = st_info[name_st2].lat
lon_st2 = st_info[name_st2].lon
t_st2 = t_list[id_st2]
wt_st2 = wt_list[id_st2]
dis = cal_dis(lat_st1,lon_st1,lat_st2,lon_st2)
azi_st1 = cal_azimuth(lat_ev,lon_ev,lat_st1,lon_st1)
azi_st2 = cal_azimuth(lat_ev,lon_ev,lat_st2,lon_st2)
azi_dif = abs(azi_st1 - azi_st2)
if(dis < dis_thd and (azi_dif < azi_thd or (360-azi_dif) < azi_thd )):
ev.cs_dt["%s+%s+%s"%(name_st1,name_st2,"P,cs")] = [name_st1,name_st2,"P,cs",t_st1-t_st2,(wt_st1+wt_st2)/2]
count_cs_dt += 1
ev_info[key_ev].Ncs_dt = len(ev.cs_dt)
print('we generate %d common source differential traveltimes from %s absolute traveltimes'%(count_cs_dt,count_t))
return ev_info
# %%
# Function: generate_cr_dif(ev_info, st_info, dis_thd) Generate common station double-difference arrival times from absolute arrival times, with inter-event distance less than dis_thd.
# Function: generate common receiver differential traveltime data from absolute traveltime data, the earthquake separation is less than dis_thd
def generate_cr_dif(ev_info,st_info,dis_thd,azi_thd):
# Construct mapping：rec2src[name_ev] -> {name_st: [name_ev, name_st, t, wt]; name_st: [name_ev, name_st, t, wt]; ...}
rec2src = build_rec_src_map(ev_info,dis_thd)
print("rec to src map generation finished")
# Construct double-difference data association mapping：rec2src_pair[key_t]
rec2src_pair = build_rec_src_pair_map(rec2src)
print("rec to src_pair map generation finished")
for key_t in rec2src_pair:
name_st = key_t.split('+')[0]
lat_st = st_info[name_st].lat
lon_st = st_info[name_st].lon
for ev_tag in rec2src_pair[key_t]:
name_ev1 = rec2src_pair[key_t][ev_tag][0]
lat_ev1 = ev_info[name_ev1].lat
lon_ev1 = ev_info[name_ev1].lon
dep_ev1 = ev_info[name_ev1].dep
name_ev2 = rec2src_pair[key_t][ev_tag][1]
lat_ev2 = ev_info[name_ev2].lat
lon_ev2 = ev_info[name_ev2].lon
dep_ev2 = ev_info[name_ev2].dep
dis_xy = cal_dis(lat_ev1,lon_ev1,lat_ev2,lon_ev2)
dis_z = abs(dep_ev1 - dep_ev2)
dis = math.sqrt(dis_xy**2 + dis_z**2)
if(dis > dis_thd): # limit of the distance between two earthquakes
continue
azi1 = cal_azimuth(lat_ev1,lon_ev1,lat_st,lon_st)
azi2 = cal_azimuth(lat_ev2,lon_ev2,lat_st,lon_st)
azi_dif = abs(azi1 - azi2)
if(azi_dif > azi_thd and (360-azi_dif) > azi_thd): # limit of the azimuth difference between two earthquakes
continue
t_ev1 = ev_info[name_ev1].t[key_t][2]
t_ev2 = ev_info[name_ev2].t[key_t][2]
wt_ev1 = ev_info[name_ev1].t[key_t][3] * ev_info[name_ev1].tag["weight"]
wt_ev2 = ev_info[name_ev2].t[key_t][3] * ev_info[name_ev2].tag["weight"]
# The actual data weight is wt_ev1 + wt_ev2, but in TomoATT calculations, we need to divide it by ev_info[name_ev1].tag["weight"].
wt = (wt_ev1 + wt_ev2)/2/ev_info[name_ev1].tag["weight"]
ev_info[name_ev1].cr_dt["%s+%s+%s"%(name_st,name_ev2,"P,cr")] = [name_st,name_ev2,"P,cr",t_ev1-t_ev2,wt]
# Count the number of double-difference data points.
count_cr_dt = 0
count_t = 0
for key_ev in ev_info:
ev_info[key_ev].Ncr_dt = len(ev_info[key_ev].cr_dt)
count_cr_dt += ev_info[key_ev].Ncr_dt
count_t += ev_info[key_ev].Nt
print('we generate %d common receiver differential traveltimes from %s absolute traveltimes'%(count_cr_dt,count_t))
return ev_info
# Construct mapping: rec2src = {key_t: dict_tag; key_t: dict_tag; ...}
# dict_tag = {tag: list_name_ev; tag: list_name_ev; ...}
# list_name_ev = [name_ev1, name_ev2, ...]
# Assign earthquakes to different subregions based on their locations. The subregion size is dlat * dlon * ddep. When performing common station double-difference calculations, only earthquake pairs within the same subregion or adjacent subregions will be considered.
def build_rec_src_map(ev_info,dis_thd):
rec2src = {}
for key_ev in ev_info:
name_ev = ev_info[key_ev].name
lat = ev_info[key_ev].lat
lon = ev_info[key_ev].lon
dep = ev_info[key_ev].dep
tag_dep = math.floor(dep/dis_thd)
tag_lat = math.floor(lat/180*math.pi*R_earth/dis_thd)
tag_lon = math.floor(lon/180*math.pi*R_earth*math.cos(lat)/dis_thd)
tag = "%d_%d_%d"%(tag_lon,tag_lat,tag_dep)
for key_t in ev_info[key_ev].t:
# create dictionary
if (not key_t in rec2src):
rec2src[key_t] = {tag:[]}
elif (not tag in rec2src[key_t]):
rec2src[key_t][tag] = []
# Add data
rec2src[key_t][tag].append(name_ev)
return rec2src
# Function: generate_adjacent_tag(tag) Generate tags surrounding the given tag.
def generate_adjacent_tag(tag): # Excluding the tag itself.
adjacent_tag_list = []
tmp = tag.split('_')
tag_lon = int(tmp[0])
tag_lat = int(tmp[1])
tag_dep = int(tmp[2])
for i in range(-1,2):
for j in range(-1,2):
for k in range(-1,2):
if(i == 0 and j == 0 and k == 0):
continue
adjacent_tag_list.append("%d_%d_%d"%(tag_lon+i,tag_lat+j,tag_dep+k))
return adjacent_tag_list
# construct mapping：rec2src_pair
def build_rec_src_pair_map(rec2src):
rec2src_pair = {}
for key_t in rec2src:
rec2src_pair[key_t] = {}
for tag in rec2src[key_t]:
name_ev_list1 = rec2src[key_t][tag]
name_ev_list2 = rec2src[key_t][tag]
adjacent_tag_list = generate_adjacent_tag(tag)
for adjacent_tag in adjacent_tag_list:
if (adjacent_tag in rec2src[key_t]): # If the surrounding tag's region has earthquakes, add them to the earthquake list.
name_ev_list2 = name_ev_list2 + rec2src[key_t][adjacent_tag]
# Find possible earthquake pairs.
for id_ev1 in range(len(name_ev_list1)-1):
name_ev1 = name_ev_list1[id_ev1]
for id_ev2 in range(id_ev1+1,len(name_ev_list2)): # Starting from id_ev1 + 1 already excludes duplicate earthquakes within the tag.
name_ev2 = name_ev_list2[id_ev2]
ev_tag1 = "%s+%s"%(name_ev1,name_ev2)
ev_tag2 = "%s+%s"%(name_ev2,name_ev1)
if(ev_tag1 in rec2src_pair[key_t] or ev_tag2 in rec2src_pair[key_t]):
continue
rec2src_pair[key_t][ev_tag1] = [name_ev1,name_ev2]
return rec2src_pair
# %% [markdown]
# Functions: read and write src_rec.dat file
# %%
# Function: reorder_src(ev) Reorder the earthquake IDs. If the earthquake has no data, the ID is -999.
def reorder_src(ev_info):
ev_id = 0
for key_ev in ev_info:
ev = ev_info[key_ev]
if(ev.Nt + ev.Ncs_dt + ev.Ncr_dt == 0):
ev.id = -999
else:
ev_info[key_ev].id = ev_id
ev_id += 1
return ev_info
# %%
# Function: read_src_rec_file(fname) Read the src_rec.dat file.
#
def read_src_rec_file(fname):
ev_info = {}
st_info = {}
tmp_ev_info = {}
doc = open(fname,'r')
doc_input = doc.readlines()
doc.close()
cc = 0
for info in doc_input:
tmp=info.split()
if (cc == 0): # event line
ev = Event()
# 1 2000 1 2 20 28 37.270 38.2843 39.0241 11.00 3.60 8 1725385
# id_ev = int(tmp[0])
ev.id = int(tmp[0])
year = int(tmp[1])
month = int(tmp[2])
day = int(tmp[3])
hour = int(tmp[4])
minute = int(tmp[5])
second = float(tmp[6])
ev.ortime = UTCDateTime(year,month,day,hour,minute,0) + second
ev.lat = float(tmp[7])
ev.lon = float(tmp[8])
ev.dep = float(tmp[9])
ev.mag = float(tmp[10])
ev.Nt = 0
ev.Ncs_dt = 0
ev.Ncr_dt = 0
ev.t = {}
ev.cs_dt = {}
ev.cr_dt = {}
ndata = int(tmp[11])
name_ev = tmp[12]
ev.name = name_ev
cc += 1
try:
ev.tag["weight"] = float(tmp[13])
except:
pass
if (ndata == 0):
cc = 0
ev_info[name_ev] = ev
else: # data line
# 1 1 MYA 38.3261 38.4253 1050.0000 P 52.46 6.630 weight
if(len(tmp) < 10): # absolue traveltime data
name_st = tmp[2]
phase = tmp[6]
if (phase == "PG"):
phase = "Pg"
if (phase == "PB"):
phase = "Pb"
if (phase == "PN"):
phase = "Pn"
if (not name_st in st_info):
st = Station()
st.name = name_st
st.id = float(tmp[1])
st.lat = float(tmp[3])
st.lon = float(tmp[4])
st.ele = float(tmp[5])
st_info[name_st] = st
time = float(tmp[7])
if(len(tmp) == 9):
weight = float(tmp[8])
else:
weight = 1.0
ev.t["%s+%s"%(name_st,phase)] = [name_st,phase,time,weight]
ev.Nt += 1
else: # differential traveltime data
phase = tmp[11]
if (phase.__contains__("cr")): # common receiver differential traveltime
# evid stid1 stname1 lat1 lon1 eve1 evid2 evname2 lat2 lon2 dep2 phase,cr diftime weight
name_st1 = tmp[2]
if (not name_st1 in st_info): # add station to the station list
st = Station()
st.name = name_st1
st.id = float(tmp[1])
st.lat = float(tmp[3])
st.lon = float(tmp[4])
st.ele = float(tmp[5])
st_info[name_st1] = st
name_ev2 = tmp[7]
# add earthquake to the temp earthquake list
ev2 = Event()
ev2.name = name_ev2
ev2.id = float(tmp[6])
ev2.lat = float(tmp[8])
ev2.lon = float(tmp[9])
ev2.dep = float(tmp[10])
tmp_ev_info[name_ev2] = ev2
dif_time = float(tmp[12])
if(len(tmp) == 14):
weight = float(tmp[13])
else:
weight = 1.0
ev.cr_dt["%s+%s+%s"%(name_st1,name_ev2,phase)] = [name_st1,name_ev2,phase,dif_time,weight]
ev.Ncr_dt += 1
else: # common source differential traveltime
# evid stid1 stname1 lat1 lon1 eve1 stid2 stname2 lat2 lon2 ele2 phase,cs diftime weight
name_st1 = tmp[2]
if (not name_st1 in st_info):
st = Station()
st.name = name_st1
st.id = len(st_info)
st.lat = float(tmp[3])
st.lon = float(tmp[4])
st.ele = float(tmp[5])
st_info[name_st1] = st
name_st2 = tmp[7]
if (not name_st2 in st_info):
st = Station()
st.name = name_st2
st.id = float(tmp[6])
st.lat = float(tmp[8])
st.lon = float(tmp[9])
st.ele = float(tmp[10])
st_info[name_st2] = st
dif_time = float(tmp[12])
if(len(tmp) == 14):
weight = float(tmp[13])
else:
weight = 1.0
ev.cs_dt["%s+%s+%s"%(name_st1,name_st2,phase)] = [name_st1,name_st2,phase,dif_time,weight]
ev.Ncs_dt += 1
if (cc == ndata): # end of the event data
cc = 0
ev_info[name_ev] = ev
else:
cc += 1
# Add earthquakes from the temporary earthquake list to the main earthquake list.
for key_ev in tmp_ev_info:
if (not key_ev in ev_info):
ev_info[key_ev] = tmp_ev_info[key_ev]
return [ev_info,st_info]
# %%
# Function: write_src_rec_file(fname, ev_info, st_info) Output the src_rec.dat file.
def write_src_rec_file(fname,ev_info,st_info):
ev_info = reorder_src(ev_info)
doc_src_rec = open(fname,'w')
min_lat = 9999
max_lat = -9999
min_lon = 9999
max_lon = -9999
min_dep = 9999
max_dep = -9999
record_ev = {}
record_st = {}
Nt_total = 0
Ncs_dt_total = 0
Ncr_dt_total = 0
for key_ev in ev_info:
ev = ev_info[key_ev]
evid = ev.id
year = ev.ortime.year
month = ev.ortime.month
day = ev.ortime.day
hour = ev.ortime.hour
minute = ev.ortime.minute
second = ev.ortime.second
msec = ev.ortime.microsecond
lat_ev = ev.lat
lon_ev = ev.lon
dep_ev = ev.dep
mag = ev.mag
ndata = ev.Nt + ev.Ncs_dt + ev.Ncr_dt
name_ev = ev.name
try:
weight_ev = ev.tag["weight"]
except:
weight_ev = 1.0
if(ndata == 0): # if the earthquake has no data, do not output it
continue
doc_src_rec.write('%7d %6d %2d %2d %2d %2d %5.2f %9.4f %9.4f %9.4f %5.2f %7d %s %7.3f\n'%(\
evid,year,month,day,hour,minute,second+msec/1000000,lat_ev,lon_ev,dep_ev,mag,ndata,name_ev,weight_ev))
min_lat = min(min_lat, lat_ev)
max_lat = max(max_lat, lat_ev)
min_lon = min(min_lon, lon_ev)
max_lon = max(max_lon, lon_ev)
min_dep = min(min_dep, dep_ev)
max_dep = max(max_dep, dep_ev)
record_ev[name_ev] = 1 # record this earthquake
Nt_total += ev.Nt
Ncs_dt_total += ev.Ncs_dt
Ncr_dt_total += ev.Ncr_dt
for key_t in ev.t:
data = ev.t[key_t]
st = st_info[data[0]]
stid = st.id
name_st = st.name
lat_st = st.lat
lon_st = st.lon
ele_st = st.ele
phase = data[1]
time = data[2]
try:
weight_data = data[3]
except:
weight_data = 1.0
doc_src_rec.write('%7d %7d %6s %9.4f %9.4f %9.4f %s %8.4f %7.3f \n'%(evid,stid,name_st,lat_st,lon_st,ele_st,phase,time,weight_data))
min_lat = min(min_lat, lat_st)
max_lat = max(max_lat, lat_st)
min_lon = min(min_lon, lon_st)
max_lon = max(max_lon, lon_st)
min_dep = min(min_dep, -ele_st/1000)
max_dep = max(max_dep, -ele_st/1000)
record_st[name_st] = 1 # record this station
for key_t in ev.cs_dt:
data = ev.cs_dt[key_t]
st1 = st_info[data[0]]
stid1 = st1.id
name_st1= st1.name
lat_st1 = st1.lat
lon_st1 = st1.lon
ele_st1 = st1.ele
st2 = st_info[data[1]]
stid2 = st2.id
name_st2= st2.name
lat_st2 = st2.lat
lon_st2 = st2.lon
ele_st2 = st2.ele
phase = data[2]
time = data[3]
try:
weight_data = data[4]
except:
weight_data = 1.0
doc_src_rec.write('%7d %7d %6s %9.4f %9.4f %9.4f %7d %6s %9.4f %9.4f %9.4f %s %8.4f %7.3f \n'%(\
evid,stid1,name_st1,lat_st1,lon_st1,ele_st1,stid2,name_st2,lat_st2,lon_st2,ele_st2,phase,time,weight_data))
min_lat = min(min_lat, lat_st1)
max_lat = max(max_lat, lat_st1)
min_lon = min(min_lon, lon_st1)
max_lon = max(max_lon, lon_st1)
min_dep = min(min_dep, -ele_st1/1000)
max_dep = max(max_dep, -ele_st1/1000)
min_lat = min(min_lat, lat_st2)
max_lat = max(max_lat, lat_st2)
min_lon = min(min_lon, lon_st2)
max_lon = max(max_lon, lon_st2)
min_dep = min(min_dep, -ele_st2/1000)
max_dep = max(max_dep, -ele_st2/1000)
record_st[name_st1] = 1 # record this station
record_st[name_st2] = 1 # record this station
for key_t in ev.cr_dt:
data = ev.cr_dt[key_t]
st = st_info[data[0]]
stid = st.id
name_st = st.name
lat_st = st.lat
lon_st = st.lon
ele_st = st.ele
ev2 = ev_info[data[1]]
evid2 = ev2.id
name_ev2= ev2.name
lat_ev2 = ev2.lat
lon_ev2 = ev2.lon
dep_ev2 = ev2.dep
phase = data[2]
time = data[3]
try:
weight_data = data[4]
except:
weight_data = 1.0
doc_src_rec.write('%7d %7d %6s %9.4f %9.4f %9.4f %7d %6s %9.4f %9.4f %9.4f %s %8.4f %7.3f \n'%(\
evid,stid,name_st,lat_st,lon_st,ele_st,evid2,name_ev2,lat_ev2,lon_ev2,dep_ev2,phase,time,weight_data))
min_lat = min(min_lat, lat_st)
max_lat = max(max_lat, lat_st)
min_lon = min(min_lon, lon_st)
max_lon = max(max_lon, lon_st)
min_dep = min(min_dep, -ele_st/1000)
max_dep = max(max_dep, -ele_st/1000)
min_lat = min(min_lat, lat_ev2)
max_lat = max(max_lat, lat_ev2)
min_lon = min(min_lon, lon_ev2)
max_lon = max(max_lon, lon_ev2)
min_dep = min(min_dep, dep_ev2)
max_dep = max(max_dep, dep_ev2)
record_ev[name_ev2] = 1 # record this station
record_st[name_st] = 1 # record this station
doc_src_rec.close()
print("src_rec.dat has been outputed: %d events, %d stations, %d abs traveltime, %d cs_dif traveltime, %d cr_dif traveltime. " \
%(len(record_ev),len(record_st),Nt_total,Ncs_dt_total,Ncr_dt_total))
print("earthquake and station region, lat: %6.1f - %6.1f, lon: %6.1f - %6.1f, dep: %6.1f - %6.1f"%(min_lat,max_lat,min_lon,max_lon,min_dep,max_dep) )
# %%
# Function: write_src_list_file(fname, ev_info) Output the event list file.
def write_src_list_file(fname,ev_info):
doc_ev_list = open(fname,'w')
for key_ev in ev_info:
ev = ev_info[key_ev]
evid = ev.id
lat_ev = ev.lat
lon_ev = ev.lon
dep_ev = ev.dep
mag = ev.mag
name_ev = ev.name
if (ev.id == -999): # if the earthquake has no data, do not output it
continue
doc_ev_list.write("%7d %s %s %9.4f %9.4f %9.4f %5.2f \n"%(evid,name_ev,ev.ortime,lat_ev,lon_ev,dep_ev,mag))
doc_ev_list.close()
# %%
# Function: write_rec_list_file(fname, ev_info, st_info) Output the station list file.
def write_rec_list_file(fname,ev_info,st_info):
doc_st_list = open(fname,'w')
st_list = {}
for key_ev in ev_info:
ev = ev_info[key_ev]
for key_t in ev.t:
data = ev.t[key_t]
st = st_info[data[0]]
name_st = st.name
lat_st = st.lat
lon_st = st.lon
ele_st = st.ele
if(not name_st in st_list):
doc_st_list.write("%6s %9.4f %9.4f %10.4f \n"%(name_st,lat_st,lon_st,ele_st))
st_list[name_st] = 1
for key_t in ev.cs_dt:
data = ev.cs_dt[key_t]
st1 = st_info[data[0]]
name_st1= st1.name
lat_st1 = st1.lat
lon_st1 = st1.lon
ele_st1 = st1.ele
st2 = st_info[data[0]]
name_st2= st2.name
lat_st2 = st2.lat
lon_st2 = st2.lon
ele_st2 = st2.ele
if(not name_st1 in st_list):
doc_st_list.write("%6s %9.4f %9.4f %10.4f \n"%(name_st1,lat_st1,lon_st1,ele_st1))
st_list[name_st1] = 1
if(not name_st2 in st_list):
doc_st_list.write("%6s %9.4f %9.4f %10.4f \n"%(name_st2,lat_st2,lon_st2,ele_st2))
st_list[name_st2] = 1
for key_t in ev.cr_dt:
data = ev.cr_dt[key_t]
st = st_info[data[0]]
name_st = st.name
lat_st = st.lat
lon_st = st.lon
ele_st = st.ele
if(not name_st in st_list):
doc_st_list.write("%6s %9.4f %9.4f %10.4f \n"%(name_st,lat_st,lon_st,ele_st))
st_list[name_st] = 1
doc_st_list.close()
# %% [markdown]
# Functions: read objective function file
# %%
# Function: read_objective_function_file(path)
def read_objective_function_file(path):
full_curve = []
location_curve = []
model_curve = []
with open('%s/objective_function.txt'%(path)) as f:
for i,line in enumerate(f):
tmp = line.split(',')
if (tmp[0].__contains__("#")):
continue # skip the comment line
iter = int(tmp[0])
tag = tmp[1]
obj = float(tmp[2])
obj_abs = float(tmp[3])
obj_cs = float(tmp[4])
obj_cr = float(tmp[5])
obj_tele = float(tmp[6])
tmp2 = tmp[7].split('/')
mean = float(tmp2[0])
std = float(tmp2[1])
tmp2 = tmp[8].split('/')
mean_abs = float(tmp2[0])
std_abs = float(tmp2[1])
tmp2 = tmp[9].split('/')
mean_cs = float(tmp2[0])
std_cs = float(tmp2[1])
tmp2 = tmp[10].split('/')
mean_cr = float(tmp2[0])
std_cr = float(tmp2[1])
tmp2 = tmp[11].split('/')
mean_tele = float(tmp2[0])
std_tele = float(tmp2[1])
full_curve.append([obj,obj_abs,obj_cs,obj_cr,obj_tele,mean,std,mean_abs,std_abs,mean_cs,std_cs,mean_cr,std_cr,mean_tele,std_tele])
if tag.__contains__("relocation"):
location_curve.append([obj,obj_abs,obj_cs,obj_cr,obj_tele,mean,std,mean_abs,std_abs,mean_cs,std_cs,mean_cr,std_cr,mean_tele,std_tele])
if tag.__contains__("model"):
model_curve.append([obj,obj_abs,obj_cs,obj_cr,obj_tele,mean,std,mean_abs,std_abs,mean_cs,std_cs,mean_cr,std_cr,mean_tele,std_tele])
return np.array(full_curve),np.array(location_curve),np.array(model_curve)
# %% [markdown]
# Functions: read inversion grid file
# %%
# Function: read the inversion grid file
def read_inversion_grid_file(path):
inv_grid_vel = []
inv_grid_ani = []
switch = False
igrid = -1
with open('%s/inversion_grid.txt'%(path)) as f:
tmp_inv_grid = []
for i,line in enumerate(f):
# read the number of inversion grid in dep, lat, lon directions
if(i==0):
tmp = line.split()
ndep = int(tmp[1])
nlines = 3*ndep+1 # The number of rows for each inversion grid is 3*ndep+1
iline = i % nlines
if(iline == 0): # info: number of inversion grid
tmp = line.split()
if (int(tmp[0]) > igrid):
igrid = int(tmp[0])
else: # change from vel to ani
switch = True
igrid = int(tmp[0])
else: # info location of inversion grid
iline_sub = (iline-1) % 3
if(iline_sub == 0): # dep
tmp = line.split()
dep = float(tmp[0])
if(iline_sub == 1): # list of lat
lat_list = line.split()
if(iline_sub == 2): # list of lon
lon_list = line.split()
# add inversion grid
for lat in lat_list:
for lon in lon_list:
tmp_inv_grid.append([float(lon), float(lat), dep])
if(iline == nlines-1): # the last line of inversion grid
if(switch):
inv_grid_ani.append(tmp_inv_grid)
else:
inv_grid_vel.append(tmp_inv_grid)
tmp_inv_grid = []
return [np.array(inv_grid_vel),np.array(inv_grid_ani)]
# %% [markdown]
# Functions: for plotting
# %%
# Function: fig_ev_st_distribution_dep(ev_info, st_info) Plot the distribution of the earthquakes and stations, color-coded by earthquake depth.
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec
def fig_ev_st_distribution_dep(ev_info,st_info):
[lon_ev,lat_ev,dep_ev,wt_ev] = data_lon_lat_dep_wt_ev(ev_info)
[lon_st,lat_st,ele_st,wt_st] = data_lon_lat_ele_wt_st(ev_info,st_info)
min_lon = min(min(lon_ev),min(lon_st))
max_lon = max(max(lon_ev),max(lon_st))
min_lat = min(min(lat_ev),min(lat_st))
max_lat = max(max(lat_ev),max(lat_st))
max_dep = max(dep_ev)
# Insert a value that does not affect the plot to make the colorbar range look better.
lon_ev = np.insert(lon_ev, 0, 9999); lat_ev = np.insert(lat_ev, 0, 9999); dep_ev = np.insert(dep_ev, 0, 0);
fig = plt.figure(figsize=(12,12))
gridspace = GridSpec(12,12,figure = fig)
xrange = max_lon - min_lon + 1.0
yrange = max_lat - min_lat + 1.0
if (xrange > yrange):
fig_x_size = 6
fig_y_size = round(6*yrange/xrange)
else:
fig_x_size = round(6*xrange/yrange)
fig_y_size = 6
ax1 = fig.add_subplot(gridspace[0:fig_y_size,0:fig_x_size])
bar_ev = ax1.scatter(lon_ev,lat_ev,c=dep_ev,cmap="jet",label = "src",s = 3)
# ax1.plot(lon_st,lat_st,'rv',label = "rec",markersize = 6)
bar_st = ax1.scatter(lon_st,lat_st,c="red",label = "rec",s = 100,marker='v',edgecolors='white')
ax1.legend(fontsize = 14)
ax1.tick_params(axis='x',labelsize=18)
ax1.tick_params(axis='y',labelsize=18)
ax1.set_xlabel('Lon',fontsize=18)
ax1.set_ylabel('Lat',fontsize=18)
ax1.set_xlim((min_lon - (max_lon - min_lon)*0.1,max_lon + (max_lon - min_lon)*0.1))
ax1.set_ylim((min_lat - (max_lat - min_lat)*0.1,max_lat + (max_lat - min_lat)*0.1))
ax2 = fig.add_subplot(gridspace[0:fig_y_size, fig_x_size+1 : fig_x_size+3])
ax2.scatter(dep_ev,lat_ev,c=dep_ev,cmap="jet",label = "src",s = 3)
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_xlabel('Dep',fontsize=18)
ax2.set_ylabel('Lat',fontsize=18)
ax2.set_xlim((-max_dep*0.05,max_dep*1.1))
ax2.set_ylim((min_lat - (max_lat - min_lat)*0.1,max_lat + (max_lat - min_lat)*0.1))
ax3 = fig.add_subplot(gridspace[fig_y_size+1:fig_y_size+3,0:fig_x_size])
ax3.scatter(lon_ev,dep_ev,c=dep_ev,cmap="jet",label = "src",s = 3)
ax3.tick_params(axis='x',labelsize=18)
ax3.tick_params(axis='y',labelsize=18)
ax3.set_xlabel('Lon',fontsize=18)
ax3.set_ylabel('Dep',fontsize=18)
ax3.set_xlim((min_lon - (max_lon - min_lon)*0.1,max_lon + (max_lon - min_lon)*0.1))
ax3.set_ylim((-max_dep*0.05,max_dep*1.1))
ax3.invert_yaxis()
# Place the colorbar on a new axis.
ax4 = fig.add_subplot(gridspace[fig_y_size+2:fig_y_size+3,fig_x_size+1:fig_x_size+3])
cbar1 = plt.colorbar(bar_ev, ax=ax4,orientation='horizontal')
cbar1.set_label('Depth of earthquakes',fontsize=16)
cbar1.ax.tick_params(axis='x', labelsize=16) # Colorbar font size.
# Hide the borders of the axes.
ax4.spines['top'].set_visible(False)
ax4.spines['right'].set_visible(False)
ax4.spines['bottom'].set_visible(False)
ax4.spines['left'].set_visible(False)
# Hide the tick values of the axes.
ax4.set_xticks([])
ax4.set_yticks([])
# %%
# Function: fig_ev_st_distribution_wt(ev_info, st_info) Plot the distribution of the earthquakes and stations, color-coded by weight.
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec
from matplotlib.colors import ListedColormap
from mpl_toolkits.axes_grid1 import make_axes_locatable
import math
def fig_ev_st_distribution_wt(ev_info,st_info):
[lon_ev,lat_ev,dep_ev,wt_ev] = data_lon_lat_dep_wt_ev(ev_info)
[lon_st,lat_st,ele_st,wt_st] = data_lon_lat_ele_wt_st(ev_info,st_info)
# Insert a value that does not affect the plot to make the colorbar range look better.
lon_ev = np.insert(lon_ev, 0, lon_ev[0]); lat_ev = np.insert(lat_ev, 0, lat_ev[0]); dep_ev = np.insert(dep_ev, 0, dep_ev[0]); wt_ev = np.insert(wt_ev, 0, 0.0)
lon_ev = np.insert(lon_ev, 0, lon_ev[0]); lat_ev = np.insert(lat_ev, 0, lat_ev[0]); dep_ev = np.insert(dep_ev, 0, dep_ev[0]); wt_ev = np.insert(wt_ev, 0, 1.0)
lon_st = np.insert(lon_st, 0, lon_st[0]); lat_st = np.insert(lat_st, 0, lat_st[0]); ele_st = np.insert(ele_st, 0, ele_st[0]); wt_st = np.insert(wt_st, 0, 0.0)
lon_st = np.insert(lon_st, 0, lon_st[0]); lat_st = np.insert(lat_st, 0, lat_st[0]); ele_st = np.insert(ele_st, 0, ele_st[0]); wt_st = np.insert(wt_st, 0, 1.0)
min_lon = min(min(lon_ev),min(lon_st))
max_lon = max(max(lon_ev),max(lon_st))
min_lat = min(min(lat_ev),min(lat_st))
max_lat = max(max(lat_ev),max(lat_st))
max_dep = max(dep_ev)
fig = plt.figure(figsize=(12,12))
gridspace = GridSpec(12,12,figure = fig)
xrange = max_lon - min_lon + 1.0
yrange = max_lat - min_lat + 1.0
if (xrange > yrange):
fig_x_size = 6
fig_y_size = round(6*yrange/xrange)
else:
fig_x_size = round(6*xrange/yrange)
fig_y_size = 6
ax1 = fig.add_subplot(gridspace[0:fig_y_size,0:fig_x_size])
bar_ev = ax1.scatter(lon_ev,lat_ev,c=wt_ev,cmap="jet",label = "src",s = 3)
bar_st = ax1.scatter(lon_st,lat_st,c=wt_st,cmap="jet",label = "rec",s = 100,marker='^',edgecolors='white')
ax1.legend(fontsize = 14)
ax1.tick_params(axis='x',labelsize=18)
ax1.tick_params(axis='y',labelsize=18)
ax1.set_xlabel('Lon',fontsize=18)
ax1.set_ylabel('Lat',fontsize=18)
ax1.set_xlim((min_lon - (max_lon - min_lon)*0.1,max_lon + (max_lon - min_lon)*0.1))
ax1.set_ylim((min_lat - (max_lat - min_lat)*0.1,max_lat + (max_lat - min_lat)*0.1))
ax2 = fig.add_subplot(gridspace[0:fig_y_size, fig_x_size+1 : fig_x_size+3])
ax2.scatter(dep_ev,lat_ev,c=wt_ev,cmap="jet",label = "src",s = 3)
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_xlabel('Dep',fontsize=18)
ax2.set_ylabel('Lat',fontsize=18)
ax2.set_xlim((-max_dep*0.05,max_dep*1.1))
ax2.set_ylim((min_lat - (max_lat - min_lat)*0.1,max_lat + (max_lat - min_lat)*0.1))
ax3 = fig.add_subplot(gridspace[fig_y_size+1:fig_y_size+3,0:fig_x_size])
ax3.scatter(lon_ev,dep_ev,c=wt_ev,cmap="jet",label = "src",s = 3)
ax3.tick_params(axis='x',labelsize=18)
ax3.tick_params(axis='y',labelsize=18)
ax3.set_xlabel('Lon',fontsize=18)
ax3.set_ylabel('Dep',fontsize=18)
ax3.set_xlim((min_lon - (max_lon - min_lon)*0.1,max_lon + (max_lon - min_lon)*0.1))
ax3.set_ylim((-max_dep*0.05,max_dep*1.1))
ax3.invert_yaxis()
# Place the colorbar on a new axis.
ax4 = fig.add_subplot(gridspace[fig_y_size+2:fig_y_size+3,fig_x_size+1:fig_x_size+3])
cbar1 = plt.colorbar(bar_st, ax=ax4,orientation='horizontal')
cbar1.set_label('Weight of stations',fontsize=16)
cbar1.ax.tick_params(axis='x', labelsize=16) # colorbar font size.
# Hide the borders of the axes.
ax4.spines['top'].set_visible(False)
ax4.spines['right'].set_visible(False)
ax4.spines['bottom'].set_visible(False)
ax4.spines['left'].set_visible(False)
# Hide the tick values of the axes.
ax4.set_xticks([])
ax4.set_yticks([])
# Place the colorbar on a new axis.
ax5 = fig.add_subplot(gridspace[fig_y_size+1:fig_y_size+2,fig_x_size+1:fig_x_size+3])
cbar1 = plt.colorbar(bar_ev, ax=ax5,orientation='horizontal')
cbar1.set_label('Weight of earthquakes',fontsize=16)
cbar1.ax.tick_params(axis='x', labelsize=16) # colorbar font size.
# Hide the borders of the axes.
ax5.spines['top'].set_visible(False)
ax5.spines['right'].set_visible(False)
ax5.spines['bottom'].set_visible(False)
ax5.spines['left'].set_visible(False)
# Hide the tick values of the axes.
ax5.set_xticks([])
ax5.set_yticks([])
# %%
# Plot and function: plot the distance-time scatter plot, remove the outliers.
# Limit the data within the range defined by the line time = dis * slope + intercept and the bounds up and down.
# Remove outliers, only retain data satisfying: slope * dis + intercept + down < time < slope * dis + intercept + up.
def fig_data_plot_remove_outliers(ev_info,st_info,slope,intercept,up,down,dis_min,dis_max):
fig = plt.figure(figsize=(10,10))
gridspace = GridSpec(6,6,figure = fig)
ax2 = fig.add_subplot(gridspace[0:6, 0:6])
# plot original data
[dis_obs,time_obs] = data_dis_time(ev_info,st_info)
ax2.plot(dis_obs,time_obs,'r.',markersize=1.5,label = "discarded")
# remove outliers, only retain data satisfying: slope * dis + intercept + down < time < slope * dis + intercept + up
ev_info = limit_data_residual(ev_info,st_info,slope,intercept,up,down)
[dis_obs,time_obs] = data_dis_time(ev_info,st_info)
ax2.plot(dis_obs,time_obs,'b.',markersize=1.5,label = "retained")
ax2.plot([dis_min,dis_max],[slope*dis_min+intercept+up,slope*dis_max+intercept+up],'b-',linewidth=2)
ax2.plot([dis_min,dis_max],[slope*dis_min+intercept+down,slope*dis_max+intercept+down],'b-',linewidth=2)
ax2.plot([dis_min,dis_max],[slope*dis_min+intercept,slope*dis_max+intercept],'k-',linewidth=2)
ax2.legend(fontsize = 14)
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_xlabel('Distance (km)',fontsize=18)
ax2.set_ylabel('Traveltime',fontsize=18)
ax2.set_xlim((dis_min,dis_max))
ax2.set_ylim((intercept+down-5,slope*dis_max+intercept+up+5))
return ev_info
# %%
# Plot: distance-time scatter plot of given phases.
def fig_data_plot_phase(ev_info,st_info,phase_list,color_list,dis_min,dis_max):
[dis_obs_phase,time_obs_phase] = data_dis_time_phase(ev_info,st_info,phase_list)
regression = {}
# Calculate the least squares y = ax+b
for key_phase in phase_list:
X = dis_obs_phase[key_phase]
Y = time_obs_phase[key_phase]
if(len(X)>20):
regression[key_phase] = linear_regression(X,Y)
else:
print("No enough data: %d, for %s"%(len(X),key_phase))
regression[key_phase] = [0,0,0]
# draw
fig = plt.figure(figsize=(10,10))
gridspace = GridSpec(6,6,figure = fig)
ax2 = fig.add_subplot(gridspace[0:6, 0:6])
y1 = 99999; y2 = -99999
# scatter plot
for iphase in range(len(phase_list)):
phase = phase_list[iphase]
color = color_list[iphase]
ax2.plot(dis_obs_phase[phase],time_obs_phase[phase],'%s.'%(color),markersize=1)
# linear regression plot
for iphase in range(len(phase_list)):
phase = phase_list[iphase]
color = color_list[iphase]
(slope,intercept,SEE)= regression[phase]
ax2.plot([dis_min,dis_max],[dis_min*slope+intercept,dis_max*slope+intercept],'%s-'%(color),linewidth=2,label = "%s: a,b,SEE=%5.2f,%5.2f,%5.2f"%(phase,slope,intercept,SEE))
y1 = min(y1,intercept-5)
y2 = max(y2,dis_max*slope+intercept+5)
ax2.legend(fontsize = 14)
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_xlabel('Distance (km)',fontsize=18)
ax2.set_ylabel('Traveltime (s)',fontsize=18)
ax2.set_xlim((dis_min,dis_max))
for iphase in range(len(phase_list)):
try:
y1 = min(y1,min(time_obs_phase[phase]))
y2 = max(y2,max(time_obs_phase[phase]))
except:
pass
ax2.set_ylim((y1,y2))
title = ""
for phase in dis_obs_phase:
title = title + "%s(%d) "%(phase,len(dis_obs_phase[phase]))
ax2.set_title(title)
print("a is slope, b is intercept, SEE is standard error of estimate")
# %% [markdown]
# Functions: results analysis and evaluation
# %%
# plot: plot the residual histogram of the initial and final model
def fig_residual_histogram(fn_syn_init,fn_syn_final,fn_obs,range_l,range_r,Nbar,tag1 = "initial",tag2 = "final"):
# read synthetic traveltime data in the initial model
[ev_info_syn_init, st_info_syn_init] = read_src_rec_file(fn_syn_init)
time_syn_init = data_dis_time(ev_info_syn_init,st_info_syn_init)[1]
# read synthetic traveltime data in the final model
[ev_info_syn_final, st_info_syn_final] = read_src_rec_file(fn_syn_final)
time_syn_final = data_dis_time(ev_info_syn_final,st_info_syn_final)[1]
# read observed traveltime data
[ev_info_obs, st_info_obs] = read_src_rec_file(fn_obs)
time_obs = data_dis_time(ev_info_obs,st_info_obs)[1]
fig = plt.figure(figsize=(6,6))
gridspace = GridSpec(6,6,figure = fig)
ax2 = fig.add_subplot(gridspace[0:6, 0:6])
bins=np.linspace(range_l,range_r,Nbar)
error_init = time_syn_init - time_obs
error_final = time_syn_final - time_obs
hist_init, _, _ = ax2.hist(error_init,bins=bins,histtype='step', edgecolor = "red", linewidth = 2,
label = "%s: std = %5.3f s, mean = %5.3f s"%(tag1,np.std(error_init),np.mean(error_init)))
hist_final, _, _ = ax2.hist(error_final,bins=bins,alpha = 0.5, color = "blue",
label = "%s: std = %5.3f s, mean = %5.3f s"%(tag2,np.std(error_final),np.mean(error_final)))
print("residual for ",tag1," model is: ","mean: ",np.mean(error_init),"sd: ",np.std(error_init))
print("residual for ",tag2," model is: ","mean: ",np.mean(error_final),"sd: ",np.std(error_final))
ax2.legend(fontsize=14)
ax2.set_xlim(range_l - abs(range_l)*0.1,range_r + abs(range_r)*0.1)
ax2.set_ylim(0,1.3*max(max(hist_init),max(hist_final)))
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_ylabel('Number of data',fontsize=18)
ax2.set_xlabel('Traveltime residuals (s)',fontsize=18)
ax2.set_title("$t_{syn} - t_{obs}$",fontsize=18)
ax2.grid()
# %%
# plot: plot the residual histogram of the initial and final model
def fig_data_difference_histogram(fn_A,fn_B,range_l,range_r,Nbar):
# read data A
[ev_info_A, st_info_A] = read_src_rec_file(fn_A)
# read data B
[ev_info_B, st_info_B] = read_src_rec_file(fn_B)
# absolute traveltime residual
error_t = []
for key_ev in ev_info_A:
for key_t in ev_info_A[key_ev].t:
data_A = ev_info_A[key_ev].t[key_t]
if (key_ev in ev_info_B and key_t in ev_info_B[key_ev].t):
data_B = ev_info_B[key_ev].t[key_t]
error_t.append(data_A[2] - data_B[2])
# common-source differential traveltime residual
error_cs_dt = []
for key_ev in ev_info_A:
for key_dt in ev_info_A[key_ev].cs_dt:
data_A = ev_info_A[key_ev].cs_dt[key_dt]
if (key_ev in ev_info_B and key_dt in ev_info_B[key_ev].cs_dt):
data_B = ev_info_B[key_ev].cs_dt[key_dt]
error_cs_dt.append(data_A[3] - data_B[3])
else:
print(key_ev,key_dt)
# common-receiver differential traveltime residual
error_cr_dt = []
for key_ev in ev_info_A:
for key_dt in ev_info_A[key_ev].cr_dt:
data_A = ev_info_A[key_ev].cr_dt[key_dt]
if (key_ev in ev_info_B and key_dt in ev_info_B[key_ev].cr_dt):
data_B = ev_info_B[key_ev].cr_dt[key_dt]
error_cr_dt.append(data_A[3] - data_B[3])
# plot
fig = plt.figure(figsize=(14,6))
gridspace = GridSpec(6,14,figure = fig)
ax2 = fig.add_subplot(gridspace[0:6, 0:6])
bins=np.linspace(range_l,range_r,Nbar)
# hist_t, _, _ = ax2.hist(error_t,bins=bins,histtype='step', edgecolor = "red", linewidth = 2,
# label = "noise: std = %5.3f s, mean = %5.3f s"%(np.std(error_t),np.mean(error_t)))
hist_t, _, _ = ax2.hist(error_t,bins=bins,alpha = 0.5, color = "blue",
label = "noise: std = %5.3f s, mean = %5.3f s"%(np.std(error_t),np.mean(error_t)))
ax2.legend(fontsize=14)
ax2.set_xlim(range_l - abs(range_l)*0.1,range_r + abs(range_r)*0.1)
try:
ax2.set_ylim(0,1.3*max(hist_t))
except:
ax2.set_ylim(0,1.0)
ax2.tick_params(axis='x',labelsize=18)
ax2.tick_params(axis='y',labelsize=18)
ax2.set_ylabel('Number of data',fontsize=18)
ax2.set_xlabel('Noise (s)',fontsize=18)
ax2.set_title("Noise of traveltime",fontsize=18)
ax3 = fig.add_subplot(gridspace[0:6,8:14])
bins=np.linspace(range_l,range_r,Nbar)
# hist_t, _, _ = ax3.hist(error_t,bins=bins,histtype='step', edgecolor = "red", linewidth = 2,
# label = "noise: std = %5.3f s, mean = %5.3f s"%(np.std(error_t),np.mean(error_t)))
hist_cs_dt, _, _ = ax3.hist(error_cs_dt,bins=bins,alpha = 0.5, color = "blue",
label = "noise: std = %5.3f s, mean = %5.3f s"%(np.std(error_cs_dt),np.mean(error_cs_dt)))
ax3.legend(fontsize=14)
ax3.set_xlim(range_l - abs(range_l)*0.1,range_r + abs(range_r)*0.1)
try:
ax3.set_ylim(0,1.3*max(hist_cs_dt))
except:
ax3.set_ylim(0,1.0)
ax3.tick_params(axis='x',labelsize=18)
ax3.tick_params(axis='y',labelsize=18)
ax3.set_ylabel('Number of data',fontsize=18)
ax3.set_xlabel('Noise (s)',fontsize=18)
ax3.set_title("Noise of differential traveltime",fontsize=18)
Reference in New Issue Copy Permalink