def test_jupyter_tutorial_tapers(self):
client = Client("IRIS")
t = UTCDateTime("2018-12-25 12:00:00").timestamp
st = client.get_waveforms("UU",
"SPU",
"*",
"H*",
t,
t + 10 * 60,
attach_response=True)
pre_filt = (0.01, 0.03, 40.0, 45.0)
st = anxcor_utils.remove_response(st,
output='DISP',
pre_filt=pre_filt,
zero_mean=True,
taper=True)
converter = XArrayConverter()
resampler = XArrayResample(target_rate=10.0)
rmmean_trend = XArrayRemoveMeanTrend()
xarray = converter(st)
resampled_array = resampler(xarray)
rmm_array = rmmean_trend(resampled_array)
whitening_op = XArrayWhiten(taper=0.05,
whiten_type='cross_component',
freqmax=3.0,
freqmin=0.01,
smoothing_window_ratio=0.01)
whitened_array = whitening_op(rmm_array)
assert whitened_array.data[0, 0, 0] == pytest.approx(0, abs=1e-2)
def test_jupyter_example(self):
client = Client("IRIS")
t = UTCDateTime("2018-12-25 12:00:00").timestamp
st = client.get_waveforms("UU",
"SPU",
"*",
"H*",
t,
t + 10 * 60,
attach_response=True)
pre_filt = (0.01, 0.03, 40.0, 45.0)
st = anxcor_utils.remove_response(st,
output='DISP',
pre_filt=pre_filt,
zero_mean=True,
taper=True)
converter = XArrayConverter()
resampler = XArrayResample(target_rate=10.0)
rmmean_trend = XArrayRemoveMeanTrend()
temp_normalizer = XArrayTemporalNorm(t_norm_type='reduce_metric',
freqmin=0.001,
freqmax=0.1,
time_window=2.0,
rolling_procedure='mean',
reduce_metric='max')
xarray = converter(st)
resampled_array = resampler(xarray)
rmm_array = rmmean_trend(resampled_array)
t_normed_array = temp_normalizer(rmm_array)
assert True
def test_variable_type(self):
# first, make a noise trace and shift it by tau * sampling rate\
file = source_file
duration = 400
shift = 30
noise_loc_1 = create_random_trace(sampling_rate=40, duration=duration)
noise_loc_1[0].data += create_random_trace(sampling_rate=40,
duration=duration)[0].data
noise_loc_1 = convert(noise_loc_1)
noise_loc_2 = xr.apply_ufunc(shift_trace,
noise_loc_1,
input_core_dims=[['time']],
output_core_dims=[['time']],
kwargs={
'time': shift,
'delta': noise_loc_1.attrs['delta']
},
keep_attrs=True)
# next, add an eq teleseism from file to both noise streams
noise_loc_1_eq = noise_loc_1.copy()
down = XArrayResample(10)
t_norm = XArrayTemporalNorm(time_window=2.0)
noise_loc_2_eq = down(noise_loc_1_eq)
noise_loc_2_eq_tnorm = t_norm(noise_loc_2_eq)
assert noise_loc_2_eq_tnorm.dtype == np.float64
def test_read_resample(self):
anxcor = Anxcor()
anxcor.set_window_length(120.0)
times = anxcor.get_starttimes(starttime_stamp,endtime_stamp, 0.5)
bank = WavebankWrapper(source_dir)
anxcor.add_dataset(bank, 'nodals')
anxcor.add_process(XArrayResample(target_rate=10.0))
anxcor.save_at_process(target_dir,'resample:0')
result = anxcor.process(times)
anxcor = Anxcor()
anxcor.set_window_length(120.0)
anxcor.add_process(XArrayResample(target_rate=10.0))
bank = WavebankWrapper(source_dir)
anxcor.add_dataset(bank, 'nodals')
anxcor.load_at_process(target_dir, 'resample:0')
result = anxcor.process(times)
how_many_nc = _how_many_fmt(target_dir, format='.nc')
_clean_files_in_dir(target_dir)
assert 20 == how_many_nc
def test_increase_earthquake_shift_time_xcorr_func(self):
# first, make a noise trace and shift it by tau * sampling rate\
file = source_file
duration = 400
shift = 3
noise_loc_1 = create_random_trace(sampling_rate=40, duration=duration)
noise_loc_1[0].data+= create_random_trace(sampling_rate=40, duration=duration)[0].data
noise_loc_1 = convert(noise_loc_1,starttime=0,station=0)
noise_loc_2 = xr.apply_ufunc(shift_trace, noise_loc_1, input_core_dims=[['time']],
output_core_dims=[['time']],
kwargs={'time': shift, 'delta': noise_loc_1.attrs['delta']}, keep_attrs=True)
# next, add an eq teleseism from file to both noise streams
noise_loc_1_eq = noise_loc_1.copy()
noise_loc_2_eq = noise_loc_2.copy()
attrs = noise_loc_1.attrs
source_eq = source_earthquake()
noise_loc_1_eq[:, :, :source_eq.data.shape[2]] += source_eq.data[:, :, :]
noise_loc_2_eq[:, :, :source_eq.data.shape[2]] += source_eq.data[:, :, :]
noise_loc_1.attrs = attrs
noise_loc_2.attrs = attrs
noise_loc_2_eq.attrs = attrs
noise_loc_1_eq.attrs = attrs
# downsample both to 10hz sampling rate
down = XArrayResample(10)
t_norm =XArrayTemporalNorm(time_window=2.0)
noise_loc_1_eq = down(noise_loc_1_eq)
noise_loc_2_eq = down(noise_loc_2_eq)
noise_loc_2_eq_tnorm = t_norm(noise_loc_2_eq.copy())
noise_loc_1_eq_tnorm = t_norm(noise_loc_1_eq.copy())
x_corr_eq = self.max_corr_norm(noise_loc_1_eq, noise_loc_2_eq)
x_corr_eq_tnorm= self.max_corr_norm(noise_loc_1_eq_tnorm, noise_loc_2_eq_tnorm)
zero_index = len(x_corr_eq)//2
assert x_corr_eq[zero_index+int(40*shift)] < x_corr_eq_tnorm[zero_index+int(40*shift)]
def test_phase_shift_not_introduced(self):
target_rate = 20
process = XArrayResample(target_rate=target_rate)
trace_initial = converter(
create_sinsoidal_trace(sampling_rate=100, period=0.5,
duration=0.5))
trace_processed = converter(
create_sinsoidal_trace(sampling_rate=100, period=0.5,
duration=0.5))
trace_processed = process(trace_processed)
target = np.argmax(
trace_initial.data.ravel()) * trace_initial.attrs['delta']
source = np.argmax(
trace_processed.data.ravel()) * trace_processed.attrs['delta']
assert round(abs(target - source), int(np.log10(
1 / target_rate))) == 0, "filter introduced phase shift"