def test_matrices(self):
# distance matrix
lons = numpy.array([84., 84., 84., 85.5, 85.5, 85.5, 87., 87., 87.])
lats = numpy.array([26., 27.5, 29., 26., 27.5, 29., 26., 27.5, 29.])
dmatrix = geo.geodetic.distance_matrix(lons, lats)
aae(dmatrix.sum(), 18539.6131407)
# spatial correlation
sca = spatial_correlation_array(dmatrix, imts, 'yes')
aae(sca.sum(), 36.000370229)
# spatial covariance
std = numpy.array([(0.5, 0.52, 0.64, 0.73)] * 9, imt_dt) # 9 sites
scov = spatial_covariance_array([std[n] for n in imt_dt.names], sca)
aae(scov.sum(), 13.166200147)
# cross correlation
ccor = cross_correlation_matrix(imts, 'yes')
aae(ccor.sum(), 10.49124788)
# cholesky decomposition
L = cholesky(scov, ccor)
self.assertEqual(L.shape, (36, 36))
aae(L.sum(), 30.5121263)
# intensity
val = numpy.array(
[(5.38409665, 3.9383686, 3.55435415, 4.37692394)] * 9, imt_dt)
shakemap = numpy.zeros(9, shakemap_dt) # 9 sites
shakemap['lon'] = lons
shakemap['lat'] = lats
shakemap['vs30'] = numpy.array([301.17] * 9)
shakemap['val'] = val
shakemap['std'] = std
_, gmfs = to_gmfs(
shakemap, 'yes', 'no', site_effects=False, trunclevel=3,
num_gmfs=2, seed=42)
# shape (N, E, M)
aae(gmfs[..., 0].sum(axis=0), [0.3708301, 0.5671011]) # PGA
_, gmfs = to_gmfs(
shakemap, 'yes', 'yes', site_effects=True, trunclevel=3,
num_gmfs=2, seed=42)
aae(gmfs[..., 0].sum(axis=0), [0.5127146, 0.7800232]) # PGA
aae(gmfs[..., 2].sum(axis=0), [0.4932519, 0.6731384]) # SA(1.0)
# disable spatial correlation
_, gmfs = to_gmfs(
shakemap, 'no', 'no', site_effects=False,
trunclevel=3, num_gmfs=2, seed=42)
# shape (N, E, M)
aae(gmfs[..., 0].sum(axis=0), [0.370832, 0.5670994]) # PGA
_, gmfs = to_gmfs(
shakemap, 'no', 'yes', site_effects=True,
trunclevel=3, num_gmfs=2, seed=42)
aae(gmfs[..., 0].sum(axis=0), [0.5127171, 0.7800206]) # PGA
aae(gmfs[..., 2].sum(axis=0), [0.4932519, 0.6731384]) # SA(1.0)
def test_matrices(self):
# distance matrix
lons = numpy.array([84., 84., 84., 85.5, 85.5, 85.5, 87., 87., 87.])
lats = numpy.array([26., 27.5, 29., 26., 27.5, 29., 26., 27.5, 29.])
dmatrix = geo.geodetic.distance_matrix(lons, lats)
aae(dmatrix.sum(), 18539.6131407)
# spatial correlation
sca = spatial_correlation_array(dmatrix, imts, 'spatial')
aae(sca.sum(), 36.000370229)
# spatial covariance
std = numpy.array([(0.5, 0.52, 0.64, 0.73)] * 9, imt_dt) # 9 sites
scov = spatial_covariance_array([std[n] for n in imt_dt.names], sca)
aae(scov.sum(), 13.166200147)
# cross correlation
ccor = cross_correlation_matrix(imts, 'cross')
aae(ccor.sum(), 10.49124788)
# cholesky decomposition
L = cholesky(scov, ccor)
self.assertEqual(L.shape, (36, 36))
aae(L.sum(), 30.5121263)
# intensity
val = numpy.array([(5.38409665, 3.9383686, 3.55435415, 4.37692394)] *
9, imt_dt)
shakemap = numpy.zeros(9, shakemap_dt) # 9 sites
shakemap['lon'] = lons
shakemap['lat'] = lats
shakemap['vs30'] = numpy.array([301.17] * 9)
shakemap['val'] = val
shakemap['std'] = std
gmfs = to_gmfs(shakemap,
crosscorr='corr',
site_effects=False,
trunclevel=3,
num_gmfs=2,
seed=42)
# shape (R, N, E, M)
aae(gmfs[..., 0].sum(axis=1), [[0.3708301, 0.5671011]]) # PGA
gmfs = to_gmfs(shakemap,
crosscorr='cross',
site_effects=True,
trunclevel=3,
num_gmfs=2,
seed=42)
aae(gmfs[..., 0].sum(axis=1), [[0.4101717, 0.6240185]]) # PGA
aae(gmfs[..., 2].sum(axis=1), [[0.3946015, 0.5385107]]) # SA(1.0)
def read_shakemap(self, haz_sitecol, assetcol):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
oq.risk_imtls = oq.imtls or self.datastore.parent['oqparam'].imtls
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
# for instance for the test case_shakemap the haz_sitecol
# has sids in range(0, 26) while sitecol.sids is
# [8, 9, 10, 11, 13, 15, 16, 17, 18];
# the total assetcol has 26 assets on the total sites
# and the reduced assetcol has 9 assets on the reduced sites
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
sitecol, shakemap, discarded = get_sitecol_shakemap(
smap, oq.imtls, haz_sitecol,
oq.asset_hazard_distance['default'],
oq.discard_assets)
if len(discarded):
self.datastore['discarded'] = discarded
assetcol.reduce_also(sitecol)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
imts, gmfs = to_gmfs(
shakemap, oq.spatial_correlation, oq.cross_correlation,
oq.site_effects, oq.truncation_level, E, oq.random_seed,
oq.imtls)
save_gmf_data(self.datastore, sitecol, gmfs, imts)
return sitecol, assetcol
def read_shakemap(self, haz_sitecol, assetcol):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
oq.risk_imtls = oq.imtls or self.datastore.parent['oqparam'].imtls
extra = self.riskmodel.get_extra_imts(oq.risk_imtls)
if extra:
logging.warn('There are risk functions for not available IMTs '
'which will be ignored: %s' % extra)
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
sitecol, shakemap = get_sitecol_shakemap(
smap, oq.imtls, haz_sitecol, oq.asset_hazard_distance
or oq.region_grid_spacing)
assetcol = assetcol.reduce_also(sitecol)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
gmfs = to_gmfs(shakemap, oq.cross_correlation, oq.site_effects,
oq.truncation_level, E, oq.random_seed, oq.imtls)
save_gmf_data(self.datastore, sitecol, gmfs)
events = numpy.zeros(E, readinput.stored_event_dt)
events['eid'] = numpy.arange(E, dtype=U64)
self.datastore['events'] = events
return sitecol, assetcol
def read_shakemap(self):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
haz_sitecol = self.datastore.parent['sitecol']
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
self.sitecol, shakemap = get_sitecol_shakemap(
smap, haz_sitecol, oq.asset_hazard_distance
or oq.region_grid_spacing)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
gmfs = to_gmfs(shakemap, oq.site_effects, oq.truncation_level, E,
oq.random_seed)
save_gmf_data(self.datastore, self.sitecol, gmfs)
events = numpy.zeros(E, readinput.stored_event_dt)
events['eid'] = numpy.arange(E, dtype=U64)
self.datastore['events'] = events
def mean_gmf(shakemap):
gmfs = to_gmfs(shakemap,
site_effects=True,
trunclevel=3,
num_gmfs=10,
seed=42)
return [gmfs[..., i].mean() for i in range(len(imts))]
def read_shakemap(self, haz_sitecol, assetcol):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
oq.risk_imtls = oq.imtls or self.datastore.parent['oqparam'].imtls
extra = self.riskmodel.get_extra_imts(oq.risk_imtls)
if extra:
logging.warning('There are risk functions for not available IMTs '
'which will be ignored: %s' % extra)
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
sitecol, shakemap, discarded = get_sitecol_shakemap(
smap, oq.imtls, haz_sitecol,
oq.asset_hazard_distance['default'],
oq.discard_assets)
if len(discarded):
self.datastore['discarded'] = discarded
assetcol = assetcol.reduce_also(sitecol)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
imts, gmfs = to_gmfs(
shakemap, oq.spatial_correlation, oq.cross_correlation,
oq.site_effects, oq.truncation_level, E, oq.random_seed,
oq.imtls)
save_gmf_data(self.datastore, sitecol, gmfs, imts)
return sitecol, assetcol
def read_shakemap(self, haz_sitecol, assetcol):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
oq.risk_imtls = oq.imtls or self.datastore.parent['oqparam'].imtls
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
# for instance for the test case_shakemap the haz_sitecol
# has sids in range(0, 26) while sitecol.sids is
# [8, 9, 10, 11, 13, 15, 16, 17, 18];
# the total assetcol has 26 assets on the total sites
# and the reduced assetcol has 9 assets on the reduced sites
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
sitecol, shakemap, discarded = get_sitecol_shakemap(
smap, oq.imtls, haz_sitecol,
oq.asset_hazard_distance['default'],
oq.discard_assets)
if len(discarded):
self.datastore['discarded'] = discarded
assetcol.reduce_also(sitecol)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
imts, gmfs = to_gmfs(
shakemap, oq.spatial_correlation, oq.cross_correlation,
oq.site_effects, oq.truncation_level, E, oq.random_seed,
oq.imtls)
save_gmf_data(self.datastore, sitecol, gmfs, imts)
return sitecol, assetcol
def mean_std(shakemap, site_effects):
_, gmfs = to_gmfs(shakemap,
'yes',
'yes',
site_effects,
trunclevel=3,
num_gmfs=1000,
seed=42)
return gmfs.mean(axis=1), numpy.log(gmfs).std(axis=1)
def read_shakemap(self, haz_sitecol, assetcol):
"""
Enabled only if there is a shakemap_id parameter in the job.ini.
Download, unzip, parse USGS shakemap files and build a corresponding
set of GMFs which are then filtered with the hazard site collection
and stored in the datastore.
"""
oq = self.oqparam
E = oq.number_of_ground_motion_fields
oq.risk_imtls = oq.imtls or self.datastore.parent['oqparam'].imtls
logging.info('Getting/reducing shakemap')
with self.monitor('getting/reducing shakemap'):
# for instance for the test case_shakemap the haz_sitecol
# has sids in range(0, 26) while sitecol.sids is
# [8, 9, 10, 11, 13, 15, 16, 17, 18];
# the total assetcol has 26 assets on the total sites
# and the reduced assetcol has 9 assets on the reduced sites
smap = oq.shakemap_id if oq.shakemap_id else numpy.load(
oq.inputs['shakemap'])
sitecol, shakemap, discarded = get_sitecol_shakemap(
smap, oq.imtls, haz_sitecol,
oq.asset_hazard_distance['default'], oq.discard_assets)
if len(discarded):
self.datastore['discarded'] = discarded
assetcol.reduce_also(sitecol)
logging.info('Building GMFs')
with self.monitor('building/saving GMFs'):
imts, gmfs = to_gmfs(shakemap, oq.spatial_correlation,
oq.cross_correlation, oq.site_effects,
oq.truncation_level, E, oq.random_seed,
oq.imtls)
N, E, M = gmfs.shape
events = numpy.zeros(E, rupture.events_dt)
events['id'] = numpy.arange(E, dtype=U32)
self.datastore['events'] = events
# convert into an array of dtype gmv_data_dt
lst = [(sitecol.sids[s], ei) + tuple(gmfs[s, ei])
for s in numpy.arange(N, dtype=U32)
for ei, event in enumerate(events)]
oq.hazard_imtls = {imt: [0] for imt in imts}
data = numpy.array(lst, oq.gmf_data_dt())
create_gmf_data(self.datastore, len(imts), data=data)
return sitecol, assetcol
def mean_gmf(shakemap):
_, gmfs = to_gmfs(
shakemap, 'yes', 'yes', site_effects=True, trunclevel=3,
num_gmfs=10, seed=42)
return [gmfs[..., i].mean() for i in range(len(imts))]
def test_matrices(self):
# distance matrix
lons = numpy.array([84., 84., 84., 85.5, 85.5, 85.5, 87., 87., 87.])
lats = numpy.array([26., 27.5, 29., 26., 27.5, 29., 26., 27.5, 29.])
dmatrix = geo.geodetic.distance_matrix(lons, lats)
aae(dmatrix.sum(), 18539.6131407)
# spatial correlation
sca = spatial_correlation_array(dmatrix, imts, 'yes')
aae(sca.sum(), 36.000370229)
# spatial covariance
std = numpy.array([(0.5, 0.52, 0.64, 0.73)] * 9, imt_dt) # 9 sites
scov = spatial_covariance_array([std[n] for n in imt_dt.names], sca)
aae(scov.sum(), 13.166200147)
# cross correlation
ccor = cross_correlation_matrix(imts, 'yes')
aae(ccor.sum(), 10.49124788)
# cholesky decomposition
L = cholesky(scov, ccor)
self.assertEqual(L.shape, (36, 36))
aae(L.sum(), 30.5121263)
# intensity
val = numpy.array([(5.38409665, 3.9383686, 3.55435415, 4.37692394)] *
9, imt_dt)
shakemap = numpy.zeros(9, shakemap_dt) # 9 sites
shakemap['lon'] = lons
shakemap['lat'] = lats
shakemap['vs30'] = numpy.array([301.17] * 9)
shakemap['val'] = val
shakemap['std'] = std
_, gmfs = to_gmfs(shakemap,
'yes',
'no',
site_effects=False,
trunclevel=3,
num_gmfs=2,
seed=42)
# shape (N, E, M)
aae(gmfs[..., 0].sum(axis=0), [0.3708301, 0.5671011]) # PGA
_, gmfs = to_gmfs(shakemap,
'yes',
'yes',
site_effects=True,
trunclevel=3,
num_gmfs=2,
seed=42)
aae(gmfs[..., 0].sum(axis=0), [0.5127146, 0.7800232]) # PGA
aae(gmfs[..., 2].sum(axis=0), [0.4932519, 0.6731384]) # SA(1.0)
# disable spatial correlation
_, gmfs = to_gmfs(shakemap,
'no',
'no',
site_effects=False,
trunclevel=3,
num_gmfs=2,
seed=42)
# shape (N, E, M)
aae(gmfs[..., 0].sum(axis=0), [0.370832, 0.5670994]) # PGA
_, gmfs = to_gmfs(shakemap,
'no',
'yes',
site_effects=True,
trunclevel=3,
num_gmfs=2,
seed=42)
aae(gmfs[..., 0].sum(axis=0), [0.5127171, 0.7800206]) # PGA
aae(gmfs[..., 2].sum(axis=0), [0.4932519, 0.6731384]) # SA(1.0)
# set stddev to zero
shakemap['std'] = 0
with self.assertRaises(ValueError) as ctx:
to_gmfs(shakemap,
'no',
'yes',
site_effects=True,
trunclevel=3,
num_gmfs=2,
seed=42)
self.assertIn('stddev==0 for IMT=PGA', str(ctx.exception))
