Datasets:
ArXiv:
DOI:
License:
fix output
Browse files- quakeflow_nc.py +70 -143
quakeflow_nc.py
CHANGED
@@ -120,11 +120,7 @@ class QuakeFlow_NC(datasets.GeneratorBasedBuilder):
|
|
120 |
|
121 |
VERSION = datasets.Version("1.1.0")
|
122 |
|
123 |
-
degree2km = 111.32
|
124 |
nt = 8192
|
125 |
-
feature_nt = 512
|
126 |
-
feature_scale = int(nt / feature_nt)
|
127 |
-
sampling_rate = 100.0
|
128 |
|
129 |
# This is an example of a dataset with multiple configurations.
|
130 |
# If you don't want/need to define several sub-sets in your dataset,
|
@@ -173,30 +169,36 @@ class QuakeFlow_NC(datasets.GeneratorBasedBuilder):
|
|
173 |
or (self.config.name == "station_train")
|
174 |
or (self.config.name == "station_test")
|
175 |
):
|
176 |
-
features=datasets.Features(
|
177 |
{
|
178 |
-
"data": datasets.Array2D(shape=(3, self.nt), dtype=
|
179 |
-
"phase_pick": datasets.Array2D(shape=(3, self.nt), dtype=
|
|
|
|
|
|
|
|
|
|
|
|
|
180 |
"event_location": datasets.Sequence(datasets.Value("float32")),
|
181 |
"station_location": datasets.Sequence(datasets.Value("float32")),
|
182 |
-
}
|
183 |
-
|
184 |
-
elif (
|
185 |
-
|
186 |
-
or (self.config.name == "event_train")
|
187 |
-
or (self.config.name == "event_test")
|
188 |
-
):
|
189 |
-
features=datasets.Features(
|
190 |
{
|
191 |
-
"data": datasets.
|
192 |
-
"phase_pick": datasets.
|
193 |
-
"
|
194 |
-
"
|
195 |
-
"
|
196 |
-
"
|
197 |
-
|
|
|
|
|
|
|
|
|
198 |
)
|
199 |
-
|
200 |
return datasets.DatasetInfo(
|
201 |
# This is the description that will appear on the datasets page.
|
202 |
description=_DESCRIPTION,
|
@@ -268,10 +270,18 @@ class QuakeFlow_NC(datasets.GeneratorBasedBuilder):
|
|
268 |
for file in filepath:
|
269 |
with fsspec.open(file, "rb") as fs:
|
270 |
with h5py.File(fs, "r") as fp:
|
271 |
-
# for event_id in sorted(list(fp.keys())):
|
272 |
event_ids = list(fp.keys())
|
273 |
for event_id in event_ids:
|
274 |
event = fp[event_id]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
275 |
station_ids = list(event.keys())
|
276 |
if (
|
277 |
(self.config.name == "station")
|
@@ -279,143 +289,60 @@ class QuakeFlow_NC(datasets.GeneratorBasedBuilder):
|
|
279 |
or (self.config.name == "station_test")
|
280 |
):
|
281 |
waveforms = np.zeros([3, self.nt], dtype="float32")
|
282 |
-
|
283 |
-
attrs = event.attrs
|
284 |
-
event_location = [
|
285 |
-
attrs["longitude"],
|
286 |
-
attrs["latitude"],
|
287 |
-
attrs["depth_km"],
|
288 |
-
attrs["event_time_index"],
|
289 |
-
]
|
290 |
-
|
291 |
for i, sta_id in enumerate(station_ids):
|
292 |
-
waveforms[:, : self.nt] = event[sta_id][:, :self.nt]
|
293 |
-
# waveforms[:, : self.nt] = event[sta_id][: self.nt, :].T
|
294 |
attrs = event[sta_id].attrs
|
295 |
-
|
296 |
-
|
297 |
-
|
|
|
298 |
station_location = [attrs["longitude"], attrs["latitude"], -attrs["elevation_m"] / 1e3]
|
299 |
|
300 |
yield f"{event_id}/{sta_id}", {
|
301 |
-
"data":
|
302 |
-
"
|
303 |
-
"
|
304 |
-
"
|
|
|
|
|
|
|
|
|
|
|
305 |
}
|
306 |
|
307 |
-
|
308 |
elif (
|
309 |
(self.config.name == "event")
|
310 |
or (self.config.name == "event_train")
|
311 |
or (self.config.name == "event_test")
|
312 |
):
|
313 |
-
event_attrs = event.attrs
|
314 |
|
315 |
-
# avoid stations with P arrival equals S arrival
|
316 |
-
is_sick = False
|
317 |
-
for sta_id in station_ids:
|
318 |
-
attrs = event[sta_id].attrs
|
319 |
-
if attrs["phase_index"][attrs["phase_type"] == "P"] == attrs["phase_index"][attrs["phase_type"] == "S"]:
|
320 |
-
is_sick = True
|
321 |
-
break
|
322 |
-
if is_sick:
|
323 |
-
continue
|
324 |
-
|
325 |
waveforms = np.zeros([len(station_ids), 3, self.nt], dtype="float32")
|
326 |
-
|
327 |
-
|
328 |
-
|
329 |
-
|
330 |
-
station_location =
|
331 |
|
332 |
for i, sta_id in enumerate(station_ids):
|
333 |
-
|
334 |
-
waveforms[i, :, :] = event[sta_id][:, :self.nt]
|
335 |
attrs = event[sta_id].attrs
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
|
341 |
-
|
342 |
-
## assuming only one event with both P and S picks
|
343 |
-
c0 = ((p_picks) + (s_picks)) / 2.0 # phase center
|
344 |
-
c0_width = ((s_picks - p_picks) * self.sampling_rate / 200.0).max() if p_picks!=s_picks else 50
|
345 |
-
dx = round(
|
346 |
-
(event_attrs["longitude"] - attrs["longitude"])
|
347 |
-
* np.cos(np.radians(event_attrs["latitude"]))
|
348 |
-
* self.degree2km,
|
349 |
-
2,
|
350 |
-
)
|
351 |
-
dy = round(
|
352 |
-
(event_attrs["latitude"] - attrs["latitude"])
|
353 |
-
* self.degree2km,
|
354 |
-
2,
|
355 |
)
|
356 |
-
dz = round(
|
357 |
-
event_attrs["depth_km"] + attrs["elevation_m"] / 1e3,
|
358 |
-
2,
|
359 |
-
)
|
360 |
-
|
361 |
-
event_center[i, :] = generate_label(
|
362 |
-
[
|
363 |
-
# [c0 / self.feature_scale],
|
364 |
-
c0,
|
365 |
-
],
|
366 |
-
label_width=[
|
367 |
-
c0_width,
|
368 |
-
],
|
369 |
-
# label_width=[
|
370 |
-
# 10,
|
371 |
-
# ],
|
372 |
-
# nt=self.feature_nt,
|
373 |
-
nt=self.nt,
|
374 |
-
)[1, :]
|
375 |
-
mask = event_center[i, :] >= 0.5
|
376 |
-
event_location[i, 0, :] = (
|
377 |
-
np.arange(self.nt) - event_attrs["event_time_index"]
|
378 |
-
) / self.sampling_rate
|
379 |
-
# event_location[0, :, i] = (np.arange(self.feature_nt) - 3000 / self.feature_scale) / self.sampling_rate
|
380 |
-
# print(event_location[i, 1:, mask].shape, event_location.shape, event_location[i][1:, mask].shape)
|
381 |
-
event_location[i][1:, mask] = np.array([dx, dy, dz])[:, np.newaxis]
|
382 |
-
event_location_mask[i, :] = mask
|
383 |
-
|
384 |
-
## station location
|
385 |
-
station_location[i, 0] = round(
|
386 |
-
attrs["longitude"]
|
387 |
-
* np.cos(np.radians(attrs["latitude"]))
|
388 |
-
* self.degree2km,
|
389 |
-
2,
|
390 |
-
)
|
391 |
-
station_location[i, 1] = round(attrs["latitude"] * self.degree2km, 2)
|
392 |
-
station_location[i, 2] = round(-attrs["elevation_m"]/1e3, 2)
|
393 |
-
|
394 |
-
std = np.std(waveforms, axis=1, keepdims=True)
|
395 |
-
std[std == 0] = 1.0
|
396 |
-
waveforms = (waveforms - np.mean(waveforms, axis=1, keepdims=True)) / std
|
397 |
-
waveforms = waveforms.astype(np.float32)
|
398 |
|
399 |
yield event_id, {
|
400 |
-
"data":
|
401 |
-
"
|
402 |
-
"
|
403 |
-
"
|
404 |
-
"
|
405 |
-
"
|
|
|
|
|
|
|
406 |
}
|
407 |
-
|
408 |
-
|
409 |
-
def generate_label(phase_list, label_width=[150, 150], nt=8192):
|
410 |
-
target = np.zeros([len(phase_list) + 1, nt], dtype=np.float32)
|
411 |
-
|
412 |
-
for i, (picks, w) in enumerate(zip(phase_list, label_width)):
|
413 |
-
for phase_time in picks:
|
414 |
-
t = np.arange(nt) - phase_time
|
415 |
-
gaussian = np.exp(-(t**2) / (2 * (w / 6) ** 2))
|
416 |
-
gaussian[gaussian < 0.1] = 0.0
|
417 |
-
target[i + 1, :] += gaussian
|
418 |
-
|
419 |
-
target[0:1, :] = np.maximum(0, 1 - np.sum(target[1:, :], axis=0, keepdims=True))
|
420 |
-
|
421 |
-
return target
|
|
|
120 |
|
121 |
VERSION = datasets.Version("1.1.0")
|
122 |
|
|
|
123 |
nt = 8192
|
|
|
|
|
|
|
124 |
|
125 |
# This is an example of a dataset with multiple configurations.
|
126 |
# If you don't want/need to define several sub-sets in your dataset,
|
|
|
169 |
or (self.config.name == "station_train")
|
170 |
or (self.config.name == "station_test")
|
171 |
):
|
172 |
+
features = datasets.Features(
|
173 |
{
|
174 |
+
"data": datasets.Array2D(shape=(3, self.nt), dtype="float32"),
|
175 |
+
"phase_pick": datasets.Array2D(shape=(3, self.nt), dtype="float32"),
|
176 |
+
"phase_time": datasets.Sequence(datasets.Value("string")),
|
177 |
+
"phase_index": datasets.Sequence(datasets.Value("int32")),
|
178 |
+
"phase_type": datasets.Sequence(datasets.Value("string")),
|
179 |
+
"phase_polarity": datasets.Sequence(datasets.Value("string")),
|
180 |
+
"begin_time": datasets.Value("string"),
|
181 |
+
"end_time": datasets.Value("string"),
|
182 |
"event_location": datasets.Sequence(datasets.Value("float32")),
|
183 |
"station_location": datasets.Sequence(datasets.Value("float32")),
|
184 |
+
},
|
185 |
+
)
|
186 |
+
elif (self.config.name == "event") or (self.config.name == "event_train") or (self.config.name == "event_test"):
|
187 |
+
features = datasets.Features(
|
|
|
|
|
|
|
|
|
188 |
{
|
189 |
+
"data": datasets.Array2D(shape=(3, self.nt), dtype="float32"),
|
190 |
+
"phase_pick": datasets.Array2D(shape=(3, self.nt), dtype="float32"),
|
191 |
+
"phase_time": datasets.Sequence(datasets.Value("string")),
|
192 |
+
"phase_index": datasets.Sequence(datasets.Value("int32")),
|
193 |
+
"phase_type": datasets.Sequence(datasets.Value("string")),
|
194 |
+
"phase_polarity": datasets.Sequence(datasets.Value("string")),
|
195 |
+
"begin_time": datasets.Value("string"),
|
196 |
+
"end_time": datasets.Value("string"),
|
197 |
+
"event_location": datasets.Sequence(datasets.Value("float32")),
|
198 |
+
"station_location": datasets.Sequence(datasets.Value("float32")),
|
199 |
+
},
|
200 |
)
|
201 |
+
|
202 |
return datasets.DatasetInfo(
|
203 |
# This is the description that will appear on the datasets page.
|
204 |
description=_DESCRIPTION,
|
|
|
270 |
for file in filepath:
|
271 |
with fsspec.open(file, "rb") as fs:
|
272 |
with h5py.File(fs, "r") as fp:
|
|
|
273 |
event_ids = list(fp.keys())
|
274 |
for event_id in event_ids:
|
275 |
event = fp[event_id]
|
276 |
+
event_attrs = event.attrs
|
277 |
+
begin_time = event_attrs["begin_time"]
|
278 |
+
end_time = event_attrs["end_time"]
|
279 |
+
event_location = [
|
280 |
+
event_attrs["longitude"],
|
281 |
+
event_attrs["latitude"],
|
282 |
+
event_attrs["depth_km"],
|
283 |
+
event_attrs["event_time_index"],
|
284 |
+
]
|
285 |
station_ids = list(event.keys())
|
286 |
if (
|
287 |
(self.config.name == "station")
|
|
|
289 |
or (self.config.name == "station_test")
|
290 |
):
|
291 |
waveforms = np.zeros([3, self.nt], dtype="float32")
|
292 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
293 |
for i, sta_id in enumerate(station_ids):
|
294 |
+
waveforms[:, : self.nt] = event[sta_id][:, : self.nt]
|
|
|
295 |
attrs = event[sta_id].attrs
|
296 |
+
phase_type = attrs["phase_type"]
|
297 |
+
phase_time = attrs["phase_time"]
|
298 |
+
phase_index = attrs["phase_index"]
|
299 |
+
phase_polarity = attrs["phase_polarity"]
|
300 |
station_location = [attrs["longitude"], attrs["latitude"], -attrs["elevation_m"] / 1e3]
|
301 |
|
302 |
yield f"{event_id}/{sta_id}", {
|
303 |
+
"data": waveforms,
|
304 |
+
"phase_time": phase_time,
|
305 |
+
"phase_index": phase_index,
|
306 |
+
"phase_type": phase_type,
|
307 |
+
"phase_polarity": phase_polarity,
|
308 |
+
"begin_time": begin_time,
|
309 |
+
"end_time": end_time,
|
310 |
+
"event_location": event_location,
|
311 |
+
"station_location": station_location,
|
312 |
}
|
313 |
|
|
|
314 |
elif (
|
315 |
(self.config.name == "event")
|
316 |
or (self.config.name == "event_train")
|
317 |
or (self.config.name == "event_test")
|
318 |
):
|
|
|
319 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
320 |
waveforms = np.zeros([len(station_ids), 3, self.nt], dtype="float32")
|
321 |
+
phase_type = []
|
322 |
+
phase_time = []
|
323 |
+
phase_index = []
|
324 |
+
phase_polarity = []
|
325 |
+
station_location = []
|
326 |
|
327 |
for i, sta_id in enumerate(station_ids):
|
328 |
+
waveforms[i, :, : self.nt] = event[sta_id][:, : self.nt]
|
|
|
329 |
attrs = event[sta_id].attrs
|
330 |
+
phase_type.append(attrs["phase_type"])
|
331 |
+
phase_time.append(attrs["phase_time"])
|
332 |
+
phase_index.append(attrs["phase_index"])
|
333 |
+
phase_polarity.append(attrs["phase_polarity"])
|
334 |
+
station_location.append(
|
335 |
+
[attrs["longitude"], attrs["latitude"], -attrs["elevation_m"] / 1e3]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
336 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
337 |
|
338 |
yield event_id, {
|
339 |
+
"data": waveforms,
|
340 |
+
"phase_time": phase_time,
|
341 |
+
"phase_index": phase_index,
|
342 |
+
"phase_type": phase_type,
|
343 |
+
"phase_polarity": phase_polarity,
|
344 |
+
"begin_time": begin_time,
|
345 |
+
"end_time": end_time,
|
346 |
+
"event_location": event_location,
|
347 |
+
"station_location": station_location,
|
348 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|