-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdatareader.py
230 lines (194 loc) · 7.98 KB
/
datareader.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
import os
import numpy as np
import h5py
import pandas as pd
import obspy
from obspy import UTCDateTime
from scipy import signal as s
from tqdm import tqdm
from tensorflow.keras.utils import to_categorical
class DataReader:
def __init__(self, format="numpy", sampling_rate=100, **kwargs):
self.buffer = {}
self.format = format
self.sampling_rate = sampling_rate
if format == "numpy":
self.data_dir = kwargs["data_dir"]
try:
csv = pd.read_csv(kwargs["data_list"], header=0,sep="[,|\s+]", engine="python")
except:
csv = pd.read_csv(kwargs["data_list"], header=0, sep="\t")
self.data_list = csv["fname"]
self.num_data = len(self.data_list)
#elif format == "hdf5":
# self.h5 = h5py.File(kwargs["hdf5_file"], "r", libver="latest")
# self.h5_data
elif format == "mseed":
self.data_dir = kwargs["data_dir"]
try:
csv = pd.read_csv(kwargs["data_list"], header=0,sep="[,|\s+]", engine="python")
except:
csv = pd.read_csv(kwargs["data_list"], header=0)
self.evids = csv["evid"]
self.stas = csv["sta"]
self.tphases = csv["UTCDateTime"]
else:
raise (f"{format} not supported")
def read_numpy(self,fname):
if fname not in self.buffer:
npz = np.load(fname)
meta = {}
meta["data"] = npz["data"][:,2] # Z-comp
meta["p_idx"] = npz["p_idx"]
else:
meta = self.buffer[fname]
return meta
def read_mseed(self,fname,tphase):
if fname not in self.buffer:
meta = {}
msd = obspy.read(fname)
if len(msd)<3:
meta["inrange"] = False
return meta
meta["data"] = msd[2].data
p_cnt = int((UTCDateTime(tphase) - UTCDateTime(msd[2].stats.starttime))/msd[2].stats.delta)
if p_cnt > msd[2].stats.npts:
#raise ValueError(f"Calculated phase index for {fname} is greater than record length")
meta["inrange"] = False
else:
meta["inrange"] = True
meta["p_idx"] = p_cnt
else:
meta = self.buffer[fname]
return meta
class DataReader_train(DataReader):
def __init__(self, format="numpy",**kwargs):
super().__init__(format=format,**kwargs)
def get_numpy_data(self):
#def __getitem__(self):
base_name = self.data_list
Xarr = []
yarr = []
pbar = tqdm(total=len(base_name))
for rec in base_name:
pbar.update()
if self.format == "numpy":
meta = self.read_numpy(os.path.join(self.data_dir,rec))
data = meta["data"]
p_idx = meta["p_idx"]
Xarr.append(data[p_idx-32:p_idx+32])
#yarr.append() # need to read in polarity values somewhere
#return (np.array(Xarr), to_categorical((np.array(yarr) == 'positive').astype(int)))
return (np.array(Xarr))
def get_mseed_data(self):
evids = self.evids
stas = self.stas
tphases = self.tphases
rec_list = np.zeros(len(evids),dtype='<U50')
Xarr = np.zeros((len(evids),64))
SNRarr = np.zeros(len(evids))
rec_bool = np.zeros(len(evids),dtype=bool)
pbar = tqdm(total=len(evids))
for i in range(len(evids)):
pbar.update()
rec = evids[i] + "_" + stas[i] + ".mseed"
if os.path.isfile(os.path.join(self.data_dir,rec)):
tphase = tphases[i]
if self.format == "mseed":
meta = self.read_mseed(os.path.join(self.data_dir,rec),tphase)
if meta["inrange"] == False:
continue
data = meta["data"]
p_idx = meta["p_idx"]
if(len(data[p_idx-32:p_idx+32])<64):
continue
signal = data[p_idx:p_idx+25].copy()
noise = data[p_idx-50:p_idx].copy()
if(np.max(signal)-np.min(signal)==0.):
continue
if((len(signal)<25)|(len(noise)<50)):
SNR = 0.1
else:
signal -= np.arange(len(signal))/len(signal)*(signal[-1]-signal[0])/0.25+signal[0]
noise -= np.arange(len(noise))/len(noise)*(noise[-1]-noise[0])/0.5+noise[0]
signal_std = np.std(signal)
noise_std = np.std(noise)
SNR = signal_std/noise_std
SNRarr[i] = SNR
rec_list[i] = rec
Xarr[i] = data[p_idx-32:p_idx+32]
rec_bool = True
Xarr = Xarr[rec_bool,:]
rec_list = rec_list[rec_bool]
SNRarr = SNRarr[rec_bool]
return(Xarr,list(rec_list),SNRarr)
def get_mseed_data_filtered(self,fmin,fmax):
dt = self.sampling_rate
evids = self.evids
stas = self.stas
tphases = self.tphases
rec_list = []
Xarr = []
SNRarr = []
sost = s.butter(2,[fmin,fmax],'bandpass',fs=dt,output='sos')
pbar = tqdm(total=len(evids))
for i in range(len(evids)):
pbar.update()
rec = evids[i] + "_" + stas[i] + ".mseed"
if os.path.isfile(os.path.join(self.data_dir,rec)):
tphase = tphases[i]
if self.format == "mseed":
meta = self.read_mseed(os.path.join(self.data_dir,rec),tphase)
if meta["inrange"] == False:
continue
data = meta["data"]
p_idx = meta["p_idx"]
if(len(data[p_idx-32:p_idx+32])<64):
continue
data = s.sosfiltfilt(sost,data)
signal = data[p_idx:p_idx+100].copy()
noise = data[p_idx-200:p_idx].copy()
if((len(signal)<100)|(len(noise)<200)):
SNR = 0.5
else:
signal -= np.arange(len(signal))/len(signal)*(signal[-1]-signal[0])/1.+signal[0]
noise -= np.arange(len(noise))/len(noise)*(noise[-1]-noise[0])/2.+noise[0]
signal_std = np.std(signal)
noise_std = np.std(noise)
SNR = signal_std/noise_std
SNRarr.append(SNR)
rec_list.append(rec)
Xarr.append(data[p_idx-32:p_idx+32])
return(np.array(Xarr),rec_list,np.array(SNRarr))
def calc_mseed_SNR(self):
evids = self.evids
stas = self.stas
tphases = self.tphases
SNR_arr = []
pbar = tqdm(total=len(evids))
for i in range(len(evids)):
pbar.update()
rec = evids[i] + "_" + stas[i] + ".mseed"
# employs the same discrimination metric as get_mseed_data()
if os.path.isfile(os.path.join(self.data_dir,rec)):
tphase = tphases[i]
if self.format == "mseed":
meta = self.read_mseed(os.path.join(self.data_dir,rec),tphase)
if meta["inrange"] == False:
continue
data = meta["data"]
p_idx = meta["p_idx"]
if(len(data[p_idx-32:p_idx+32])<64):
continue
signal = data[p_idx:p_idx+100]
noise = data[p_idx-200:p_idx]
if((len(signal)<100)|(len(noise)<200)):
SNR = 0.5
else:
signal -= np.arange(len(signal))/len(signal)*(signal[-1]-signal[0])/1.+signal[0]
noise -= np.arange(len(noise))/len(noise)*(noise[-1]-noise[0])/2.+noise[0]
signal_std = np.std(signal)
noise_std = np.std(noise)
SNR = signal_std/noise_std
SNR_arr.append(SNR)
return(np.array(SNR_arr))