def test_read_with_cent(self):
""" Test read_footprints while passing in a Centroids object """
var_names = copy.deepcopy(DEF_VAR_EXCEL)
var_names['sheet_name'] = 'fp_centroids-test'
var_names['col_name']['region_id'] = 'iso_n3'
test_centroids = Centroids()
test_centroids.read_excel(os.path.join(DATA_DIR,
'fp_centroids-test.xls'),
var_names=var_names)
storms = StormEurope()
storms.read_footprints(TEST_NCS, centroids=test_centroids)
self.assertEqual(storms.intensity.shape, (2, 9944))
self.assertEqual(
np.count_nonzero(~np.isnan(storms.centroids.region_id)), 6401)
def test_set_ssi(self):
""" Test set_ssi with both dawkins and wisc_gust methodology. """
storms = StormEurope()
storms.read_footprints(TEST_NCS)
storms.set_ssi(method='dawkins')
ssi_dawg = np.asarray([1.44573572e+09, 6.16173724e+08])
self.assertTrue(np.allclose(storms.ssi, ssi_dawg))
storms.set_ssi(method='wisc_gust')
ssi_gusty = np.asarray([1.42124571e+09, 5.86870673e+08])
self.assertTrue(np.allclose(storms.ssi, ssi_gusty))
storms.set_ssi(threshold=20, on_land=False)
ssi_special = np.asarray([2.96582030e+09, 1.23980294e+09])
self.assertTrue(np.allclose(storms.ssi, ssi_special))
def test_generate_prob_storms(self):
""" Test the probabilistic storm generator; calls _hist2prob as well as
Centroids.set_region_id() """
storms = StormEurope()
storms.read_footprints(TEST_NCS)
storms_prob = storms.generate_prob_storms()
self.assertEqual(
np.count_nonzero(storms.centroids.region_id), 6190
# here, we don't rasterise; we check if the centroids lie in a
# polygon. that is to say, it's not the majority of a raster pixel,
# but the centroid's location that is decisive
)
self.assertEqual(storms_prob.size, 60)
self.assertEqual(np.count_nonzero(storms_prob.orig), 2)
self.assertEqual(storms_prob.centroids.size, 3054)
self.assertIsInstance(storms_prob.intensity, sparse.csr.csr_matrix)
def test_read_with_ref(self):
""" Test read_footprints while passing in a reference raster. """
storms = StormEurope()
storms.read_footprints(TEST_NCS, ref_raster=TEST_NCS[1])
self.assertEqual(storms.tag.haz_type, 'WS')
self.assertEqual(storms.units, 'm/s')
self.assertEqual(storms.event_id.size, 2)
self.assertEqual(storms.date.size, 2)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).year, 1999)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).month, 12)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).day, 26)
self.assertEqual(storms.event_id[0], 1)
self.assertEqual(storms.event_name[0], 'Lothar')
self.assertTrue(isinstance(storms.intensity, sparse.csr.csr_matrix))
self.assertTrue(isinstance(storms.fraction, sparse.csr.csr_matrix))
self.assertEqual(storms.intensity.shape, (2, 9944))
self.assertEqual(storms.fraction.shape, (2, 9944))
def test_read_footprints(self):
""" Test read_footprints function, using two small test files"""
storms = StormEurope()
storms.read_footprints(TEST_NCS, description='test_description')
self.assertEqual(storms.tag.haz_type, 'WS')
self.assertEqual(storms.units, 'm/s')
self.assertEqual(storms.event_id.size, 2)
self.assertEqual(storms.date.size, 2)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).year, 1999)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).month, 12)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).day, 26)
self.assertEqual(storms.event_id[0], 1)
self.assertEqual(storms.event_name[0], 'Lothar')
self.assertIsInstance(storms.intensity, sparse.csr.csr_matrix)
self.assertIsInstance(storms.fraction, sparse.csr.csr_matrix)
self.assertEqual(storms.intensity.shape, (2, 9944))
self.assertEqual(storms.fraction.shape, (2, 9944))
def test_generate_prob_storms(self):
"""Test the probabilistic storm generator; calls _hist2prob as well as
Centroids.set_region_id()"""
storms = StormEurope()
storms.read_footprints(WS_DEMO_NC)
storms_prob = storms.generate_prob_storms()
self.assertEqual(
np.count_nonzero(storms.centroids.region_id), 6402
# here, we don't rasterise; we check if the centroids lie in a
# polygon. that is to say, it's not the majority of a raster pixel,
# but the centroid's location that is decisive
)
self.assertEqual(storms_prob.size, 60)
self.assertTrue(
np.allclose((1 / storms_prob.frequency).astype(int), 330))
self.assertAlmostEqual(storms.frequency.sum(),
storms_prob.frequency.sum())
self.assertEqual(np.count_nonzero(storms_prob.orig), 2)
self.assertEqual(storms_prob.centroids.size, 3054)
self.assertIsInstance(storms_prob.intensity, sparse.csr.csr_matrix)
def test_Forecast_init_raise(self):
"""Test calc and propety functions from the Forecast class"""
#hazard with several event dates
storms = StormEurope()
storms.read_footprints(WS_DEMO_NC, description='test_description')
#exposure
data = {}
data['latitude'] = np.array([1, 2, 3])
data['longitude'] = np.array([1, 2, 3])
data['value'] = np.ones_like(data['latitude']) * 100000
data['deductible'] = np.zeros_like(data['latitude'])
data[INDICATOR_IMPF + 'WS'] = np.ones_like(data['latitude'])
data['region_id'] = np.ones_like(data['latitude'], dtype=int) * 756
expo = Exposures(gpd.GeoDataFrame(data=data))
#vulnerability
#generate vulnerability
impact_function_set = ImpactFuncSet()
#create and calculate Forecast
with self.assertRaises(ValueError):
Forecast({dt.datetime(2018, 1, 1): storms}, expo,
impact_function_set)
def test_Forecast_calc_properties(self):
"""Test calc and propety functions from the Forecast class"""
#hazard
haz = StormEurope()
haz.read_cosmoe_file(
HAZ_DIR.joinpath('storm_europe_cosmoe_forecast_vmax_testfile.nc'),
run_datetime=dt.datetime(2018, 1, 1),
event_date=dt.datetime(2018, 1, 3))
#exposure
data = {}
data['latitude'] = haz.centroids.lat
data['longitude'] = haz.centroids.lon
data['value'] = np.ones_like(data['latitude']) * 100000
data['deductible'] = np.zeros_like(data['latitude'])
data[INDICATOR_IMPF + 'WS'] = np.ones_like(data['latitude'])
data['region_id'] = np.ones_like(data['latitude'], dtype=int) * 756
expo = Exposures(gpd.GeoDataFrame(data=data))
#vulnerability
#generate vulnerability
impact_function = ImpfStormEurope()
impact_function.set_welker()
impact_function_set = ImpactFuncSet()
impact_function_set.append(impact_function)
#create and calculate Forecast
forecast = Forecast({dt.datetime(2018, 1, 1): haz}, expo,
impact_function_set)
forecast.calc()
# test
self.assertEqual(len(forecast.run_datetime), 1)
self.assertEqual(forecast.run_datetime[0], dt.datetime(2018, 1, 1))
self.assertEqual(forecast.event_date, dt.datetime(2018, 1, 3))
self.assertEqual(forecast.lead_time().days, 2)
self.assertEqual(forecast.summary_str(),
'WS_NWP_run2018010100_event20180103_Switzerland')
self.assertAlmostEqual(forecast.ai_agg(), 26.347, places=1)
self.assertAlmostEqual(forecast.ei_exp()[1], 7.941, places=1)
self.assertEqual(len(forecast.hazard), 1)
self.assertIsInstance(forecast.hazard[0], StormEurope)
self.assertIsInstance(forecast.exposure, Exposures)
self.assertIsInstance(forecast.vulnerability, ImpactFuncSet)
def test_icon_read(self):
"""test reading from icon grib"""
haz = StormEurope()
haz.read_icon_grib(dt.datetime(2021, 1, 28),
dt.datetime(2021, 1, 28),
model_name='test',
grib_dir=CONFIG.hazard.test_data.str(),
delete_raw_data=False)
self.assertEqual(haz.tag.haz_type, 'WS')
self.assertEqual(haz.units, 'm/s')
self.assertEqual(haz.event_id.size, 40)
self.assertEqual(haz.date.size, 40)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).year, 2021)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).month, 1)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).day, 28)
self.assertEqual(haz.event_id[-1], 40)
self.assertEqual(haz.event_name[-1], '2021-01-28_ens40')
self.assertIsInstance(haz.intensity,
sparse.csr.csr_matrix)
self.assertIsInstance(haz.fraction,
sparse.csr.csr_matrix)
self.assertEqual(haz.intensity.shape, (40, 49))
self.assertAlmostEqual(haz.intensity.max(), 17.276321,places=3)
self.assertEqual(haz.fraction.shape, (40, 49))
logger = logging.getLogger('climada.hazard.storm_europe')
with mock.patch.object(logger,'warning') as mock_logger:
with self.assertRaises(ValueError):
haz.read_icon_grib(dt.datetime(2021, 1, 28, 6),
dt.datetime(2021, 1, 28),
model_name='test',
grib_dir=CONFIG.hazard.test_data.str(),
delete_raw_data=False)
mock_logger.assert_called_once()
def test_from_footprints(self):
"""Test from_footprints constructor, using one small test files"""
storms = StormEurope.from_footprints(WS_DEMO_NC[0], description='test_description')
self.assertEqual(storms.tag.haz_type, 'WS')
self.assertEqual(storms.units, 'm/s')
self.assertEqual(storms.event_id.size, 1)
self.assertEqual(storms.date.size, 1)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).year, 1999)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).month, 12)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).day, 26)
self.assertEqual(storms.event_id[0], 1)
self.assertEqual(storms.event_name[0], 'Lothar')
self.assertIsInstance(storms.intensity,
sparse.csr.csr_matrix)
self.assertIsInstance(storms.fraction,
sparse.csr.csr_matrix)
self.assertEqual(storms.intensity.shape, (1, 9944))
self.assertEqual(storms.fraction.shape, (1, 9944))
self.assertEqual(storms.frequency[0], 1.0)
"""Test from_footprints constructor, using two small test files"""
storms = StormEurope.from_footprints(WS_DEMO_NC, description='test_description')
self.assertEqual(storms.tag.haz_type, 'WS')
self.assertEqual(storms.units, 'm/s')
self.assertEqual(storms.event_id.size, 2)
self.assertEqual(storms.date.size, 2)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).year, 1999)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).month, 12)
self.assertEqual(dt.datetime.fromordinal(storms.date[0]).day, 26)
self.assertEqual(storms.event_id[0], 1)
self.assertEqual(storms.event_name[0], 'Lothar')
self.assertIsInstance(storms.intensity,
sparse.csr.csr_matrix)
self.assertIsInstance(storms.fraction,
sparse.csr.csr_matrix)
self.assertEqual(storms.intensity.shape, (2, 9944))
self.assertEqual(storms.fraction.shape, (2, 9944))
def test_icon_read(self):
"""test reading from icon grib"""
# for this test the forecast file is supposed to be already downloaded from the dwd
# another download would fail because the files are available for 24h only
# instead, we download it as a test dataset through the climada data api
apiclient = Client()
ds = apiclient.get_dataset_info(name='test_storm_europe_icon_2021012800', status='test_dataset')
dsdir, _ = apiclient.download_dataset(ds)
haz = StormEurope.from_icon_grib(
dt.datetime(2021, 1, 28),
dt.datetime(2021, 1, 28),
model_name='test',
grib_dir=dsdir,
delete_raw_data=False)
self.assertEqual(haz.tag.haz_type, 'WS')
self.assertEqual(haz.units, 'm/s')
self.assertEqual(haz.event_id.size, 40)
self.assertEqual(haz.date.size, 40)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).year, 2021)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).month, 1)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).day, 28)
self.assertEqual(haz.event_id[-1], 40)
self.assertEqual(haz.event_name[-1], '2021-01-28_ens40')
self.assertIsInstance(haz.intensity,
sparse.csr.csr_matrix)
self.assertIsInstance(haz.fraction,
sparse.csr.csr_matrix)
self.assertEqual(haz.intensity.shape, (40, 49))
self.assertAlmostEqual(haz.intensity.max(), 17.276321,places=3)
self.assertEqual(haz.fraction.shape, (40, 49))
with self.assertLogs('climada.hazard.storm_europe', level='WARNING') as cm:
with self.assertRaises(ValueError):
haz = StormEurope.from_icon_grib(
dt.datetime(2021, 1, 28, 6),
dt.datetime(2021, 1, 28),
model_name='test',
grib_dir=CONFIG.hazard.test_data.str(),
delete_raw_data=False)
self.assertEqual(len(cm.output), 1)
self.assertIn('event definition is inaccuratly implemented', cm.output[0])
def test_cosmoe_read(self):
"""test reading from cosmo-e netcdf"""
haz = StormEurope()
haz.read_cosmoe_file(DATA_DIR.joinpath('storm_europe_cosmoe_forecast_vmax_testfile.nc'),
run_datetime=dt.datetime(2018,1,1),
event_date=dt.datetime(2018,1,3))
self.assertEqual(haz.tag.haz_type, 'WS')
self.assertEqual(haz.units, 'm/s')
self.assertEqual(haz.event_id.size, 21)
self.assertEqual(haz.date.size, 21)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).year, 2018)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).month, 1)
self.assertEqual(dt.datetime.fromordinal(haz.date[0]).day, 3)
self.assertEqual(haz.event_id[-1], 21)
self.assertEqual(haz.event_name[-1], '2018-01-03_ens21')
self.assertIsInstance(haz.intensity,
sparse.csr.csr_matrix)
self.assertIsInstance(haz.fraction,
sparse.csr.csr_matrix)
self.assertEqual(haz.intensity.shape, (21, 25))
self.assertAlmostEqual(haz.intensity.max(), 36.426735,places=3)
self.assertEqual(haz.fraction.shape, (21, 25))
def test_centroids_from_nc(self):
""" Test if centroids can be constructed correctly """
cent = StormEurope._centroids_from_nc(TEST_NCS[0])
self.assertTrue(isinstance(cent, Centroids))
self.assertEqual(cent.size, 9944)
def test_Forecast_plot(self):
"""Test cplotting functions from the Forecast class"""
#hazard
haz1 = StormEurope()
haz1.read_cosmoe_file(
HAZ_DIR.joinpath('storm_europe_cosmoe_forecast_vmax_testfile.nc'),
run_datetime=dt.datetime(2018, 1, 1),
event_date=dt.datetime(2018, 1, 3))
haz1.centroids.lat += 0.6
haz1.centroids.lon -= 1.2
haz2 = StormEurope()
haz2.read_cosmoe_file(
HAZ_DIR.joinpath('storm_europe_cosmoe_forecast_vmax_testfile.nc'),
run_datetime=dt.datetime(2018, 1, 1),
event_date=dt.datetime(2018, 1, 3))
haz2.centroids.lat += 0.6
haz2.centroids.lon -= 1.2
#exposure
data = {}
data['latitude'] = haz1.centroids.lat
data['longitude'] = haz1.centroids.lon
data['value'] = np.ones_like(data['latitude']) * 100000
data['deductible'] = np.zeros_like(data['latitude'])
data[INDICATOR_IMPF + 'WS'] = np.ones_like(data['latitude'])
data['region_id'] = np.ones_like(data['latitude'], dtype=int) * 756
expo = Exposures(gpd.GeoDataFrame(data=data))
#vulnerability
#generate vulnerability
impact_function = ImpfStormEurope()
impact_function.set_welker()
impact_function_set = ImpactFuncSet()
impact_function_set.append(impact_function)
#create and calculate Forecast
forecast = Forecast(
{
dt.datetime(2018, 1, 2): haz1,
dt.datetime(2017, 12, 31): haz2
}, expo, impact_function_set)
forecast.calc()
#test plotting functions
forecast.plot_imp_map(run_datetime=dt.datetime(2017, 12, 31),
save_fig=False,
close_fig=True)
forecast.plot_hist(run_datetime=dt.datetime(2017, 12, 31),
save_fig=False,
close_fig=True)
forecast.plot_exceedence_prob(run_datetime=dt.datetime(2017, 12, 31),
threshold=5000,
save_fig=False,
close_fig=True)
#create a file containing the polygons of Swiss cantons using natural earth
cantons_file = CONFIG.local_data.save_dir.dir() / 'CHE_cantons.shp'
adm1_shape_file = shapereader.natural_earth(
resolution='10m',
category='cultural',
name='admin_1_states_provinces')
if not cantons_file.exists():
with fiona.open(adm1_shape_file, 'r') as source:
with fiona.open(cantons_file, 'w', **source.meta) as sink:
for f in source:
if f['properties']['adm0_a3'] == 'CHE':
sink.write(f)
forecast.plot_warn_map(
str(cantons_file),
decision_level='polygon',
thresholds=[100000, 500000, 1000000, 5000000],
probability_aggregation='mean',
area_aggregation='sum',
title="Building damage warning",
explain_text="warn level based on aggregated damages",
save_fig=False,
close_fig=True)
forecast.plot_warn_map(
str(cantons_file),
decision_level='exposure_point',
thresholds=[1, 1000, 5000, 5000000],
probability_aggregation=0.2,
area_aggregation=0.2,
title="Building damage warning",
explain_text="warn level based on aggregated damages",
run_datetime=dt.datetime(2017, 12, 31),
save_fig=False,
close_fig=True)
forecast.plot_hexbin_ei_exposure()
plt.close()
with self.assertRaises(ValueError):
forecast.plot_warn_map(
str(cantons_file),
decision_level='test_fail',
probability_aggregation=0.2,
area_aggregation=0.2,
title="Building damage warning",
explain_text="warn level based on aggregated damages",
save_fig=False,
close_fig=True)
plt.close()
with self.assertRaises(ValueError):
forecast.plot_warn_map(
str(cantons_file),
decision_level='exposure_point',
probability_aggregation='test_fail',
area_aggregation=0.2,
title="Building damage warning",
explain_text="warn level based on aggregated damages",
save_fig=False,
close_fig=True)
plt.close()
with self.assertRaises(ValueError):
forecast.plot_warn_map(
str(cantons_file),
decision_level='exposure_point',
probability_aggregation=0.2,
area_aggregation='test_fail',
title="Building damage warning",
explain_text="warn level based on aggregated damages",
save_fig=False,
close_fig=True)
plt.close()
