def test_calc_imp_mat_pass(self):
"""Test save imp_mat"""
# Read default entity values
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute the impact over the whole exposures
impact.calc(ent.exposures, ent.impact_funcs, hazard, save_mat=True)
self.assertTrue(isinstance(impact.imp_mat, sparse.csr_matrix))
self.assertEqual(impact.imp_mat.shape,
(hazard.event_id.size, ent.exposures.value.size))
self.assertTrue(
np.allclose(
np.sum(impact.imp_mat, axis=1).reshape(-1), impact.at_event))
self.assertTrue(
np.allclose(
np.array(
np.sum(np.multiply(impact.imp_mat.toarray(),
impact.frequency.reshape(-1, 1)),
axis=0)).reshape(-1), impact.eai_exp))
def test_impact_year_set_empty(self):
"""Test result for empty impact"""
imp = Impact()
iys_all = imp.calc_impact_year_set()
iys = imp.calc_impact_year_set(all_years=False)
self.assertEqual(len(iys), 0)
self.assertEqual(len(iys_all), 0)
def test_calc_imp_mat_pass(self):
"""Test save imp_mat"""
# Read default entity values
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute the impact over the whole exposures
impact.calc(ent.exposures, ent.impact_funcs, hazard, save_mat=True)
self.assertIsInstance(impact.imp_mat, sparse.csr_matrix)
self.assertEqual(impact.imp_mat.shape, (hazard.event_id.size,
ent.exposures.gdf.value.size))
np.testing.assert_array_almost_equal_nulp(
np.array(impact.imp_mat.sum(axis=1)).ravel(), impact.at_event, nulp=5)
np.testing.assert_array_almost_equal_nulp(
np.sum(impact.imp_mat.toarray() * impact.frequency[:, None], axis=0).reshape(-1),
impact.eai_exp)
def test_write_read_excel_pass(self):
"""Test write and read in excel"""
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
imp_write = Impact()
ent.exposures.assign_centroids(hazard)
imp_write.calc(ent.exposures, ent.impact_funcs, hazard)
file_name = os.path.join(DATA_FOLDER, 'test.xlsx')
imp_write.write_excel(file_name)
imp_read = Impact()
imp_read.read_excel(file_name)
self.assertTrue(np.array_equal(imp_write.event_id, imp_read.event_id))
self.assertTrue(np.array_equal(imp_write.date, imp_read.date))
self.assertTrue(np.array_equal(imp_write.coord_exp,
imp_read.coord_exp))
self.assertTrue(np.allclose(imp_write.eai_exp, imp_read.eai_exp))
self.assertTrue(np.allclose(imp_write.at_event, imp_read.at_event))
self.assertTrue(np.array_equal(imp_write.frequency,
imp_read.frequency))
self.assertEqual(imp_write.tot_value, imp_read.tot_value)
self.assertEqual(imp_write.aai_agg, imp_read.aai_agg)
self.assertEqual(imp_write.unit, imp_read.unit)
self.assertEqual(
0,
len([
i for i, j in zip(imp_write.event_name, imp_read.event_name)
if i != j
]))
self.assertIsInstance(imp_read.crs, dict)
def test_impact_pnt_agg(self):
"""Test impact agreggation method"""
gdf_mix = GDF_LINE.append(GDF_POLY).append(GDF_POINT).reset_index(
drop=True)
exp_mix = Exposures(gdf_mix)
exp_pnt = u_lp.exp_geom_to_pnt(exp_mix,
res=1,
to_meters=False,
disagg_met=u_lp.DisaggMethod.DIV,
disagg_val=None)
imp_pnt = Impact()
imp_pnt.calc(exp_pnt, IMPF_SET, HAZ, save_mat=True)
imp_agg = u_lp.impact_pnt_agg(imp_pnt, exp_pnt.gdf, u_lp.AggMethod.SUM)
aai_agg = 1282901.377219451
eai_exp = np.array([
1.73069928e-04, 8.80741357e-04, 4.32240819e-03, 8.62816073e-03,
2.21441154e-02, 1.09329988e-02, 8.58546479e-02, 4.62370081e-02,
8.99584440e-02, 1.27160538e-02, 8.60317575e-02, 2.02440009e-01,
2.32808488e-02, 2.86159458e-02, 4.26205598e-03, 2.40051484e-01,
5.29133033e-03, 2.72705887e-03, 8.87954091e-03, 2.95633263e-02,
6.33106879e-01, 1.33011693e-03, 1.11120718e-01, 7.72573773e-02,
6.12233710e-03, 1.61239410e-02, 1.01492204e-01, 7.45522678e-02,
1.41155415e-01, 1.53820450e-01, 2.27951125e-02, 2.23629697e-02,
8.59651753e-03, 5.98415680e-03, 1.24717770e-02, 1.24717770e-02,
1.48060577e-05, 1.48060577e-05, 5.18270742e-03, 5.18270742e-03,
8.36178802e+03, 7.30704698e+03, 1.20628926e+04, 3.54061498e+04,
1.23524320e+04, 7.78074661e+04, 1.28292995e+05, 2.31231953e+05,
1.31911226e+05, 5.37897306e+05, 8.37016948e+04, 1.65661030e+04
])
check_impact(self, imp_agg, HAZ, exp_mix, aai_agg, eai_exp)
def _cutoff_hazard_damage(self, exposures, if_set, hazard):
"""Cutoff of hazard events which generate damage with a frequency higher
than hazard_freq_cutoff.
Parameters:
exposures (Exposures): exposures instance
imp_set (ImpactFuncSet): impact functions instance
hazard (Hazard): hazard instance
Returns:
ImpactFuncSet
"""
if self.hazard_freq_cutoff == 0:
return hazard
LOGGER.debug('Cutting events whose damage have a frequency > %s.',
self.hazard_freq_cutoff)
from climada.engine.impact import Impact
imp = Impact()
exp_imp = exposures
if self.exp_region_id != 0:
# compute impact only in selected region
exp_imp = exposures[exposures.region_id == self.exp_region_id]
exp_imp = Exposures(exp_imp)
imp.calc(exp_imp, if_set, hazard)
new_haz = copy.deepcopy(hazard)
sort_idxs = np.argsort(imp.at_event)[::-1]
exceed_freq = np.cumsum(imp.frequency[sort_idxs])
cutoff = exceed_freq > self.hazard_freq_cutoff
sel_haz = sort_idxs[cutoff]
new_haz_inten = new_haz.intensity.tolil()
new_haz_inten[sel_haz, :] = np.zeros((sel_haz.size, new_haz.intensity.shape[1]))
new_haz.intensity = new_haz_inten.tocsr()
return new_haz
def test_calc_geom_impact_points(self):
""" test calc_geom_impact() with points"""
imp1 = u_lp.calc_geom_impact(EXP_POINT,
IMPF_SET,
HAZ,
res=0.05,
to_meters=False,
disagg_met=u_lp.DisaggMethod.DIV,
disagg_val=None,
agg_met=u_lp.AggMethod.SUM)
aai_agg1 = 0.0470814
imp11 = Impact()
exp = EXP_POINT.copy()
exp.set_lat_lon()
imp11.calc(exp, IMPF_SET, HAZ)
check_impact(self, imp1, HAZ, EXP_POINT, aai_agg1, imp11.eai_exp)
imp2 = u_lp.calc_geom_impact(EXP_POINT,
IMPF_SET,
HAZ,
res=500,
to_meters=True,
disagg_met=u_lp.DisaggMethod.FIX,
disagg_val=1.0,
agg_met=u_lp.AggMethod.SUM)
exp.gdf['value'] = 1.0
imp22 = Impact()
imp22.calc(exp, IMPF_SET, HAZ)
aai_agg2 = 6.5454249333e-06
check_impact(self, imp2, HAZ, EXP_POINT, aai_agg2, imp22.eai_exp)
def _calc_impact(self, new_exp, new_ifs, new_haz):
"""Compute impact and risk transfer of measure implemented over inputs.
Parameters:
new_exp (Exposures): exposures once measure applied
new_ifs (ImpactFuncSet): impact functions once measure applied
new_haz (Hazard): hazard once measure applied
Returns:
Impact, Impact
"""
from climada.engine.impact import Impact
imp = Impact()
imp.calc(new_exp, new_ifs, new_haz)
return imp.calc_risk_transfer(self.risk_transf_attach, self.risk_transf_cover)
def test_ref_value_pass(self):
"""Test result against reference value"""
# Read default entity values
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute the impact over the whole exposures
impact.calc(ent.exposures, ent.impact_funcs, hazard)
# Check result
num_events = len(hazard.event_id)
num_exp = ent.exposures.shape[0]
# Check relative errors as well when absolute value gt 1.0e-7
# impact.at_event == EDS.damage in MATLAB
self.assertEqual(num_events, len(impact.at_event))
self.assertEqual(0, impact.at_event[0])
self.assertEqual(0, impact.at_event[int(num_events / 2)])
self.assertAlmostEqual(1.472482938320243e+08, impact.at_event[13809])
self.assertEqual(7.076504723057620e+10, impact.at_event[12147])
self.assertEqual(0, impact.at_event[num_events - 1])
# impact.eai_exp == EDS.ED_at_centroid in MATLAB
self.assertEqual(num_exp, len(impact.eai_exp))
self.assertAlmostEqual(1.518553670803242e+08, impact.eai_exp[0])
self.assertAlmostEqual(1.373490457046383e+08,
impact.eai_exp[int(num_exp / 2)], 6)
self.assertTrue(
np.isclose(1.373490457046383e+08,
impact.eai_exp[int(num_exp / 2)]))
self.assertAlmostEqual(1.066837260150042e+08,
impact.eai_exp[num_exp - 1], 6)
self.assertTrue(
np.isclose(1.066837260150042e+08,
impact.eai_exp[int(num_exp - 1)]))
# impact.tot_value == EDS.Value in MATLAB
# impact.aai_agg == EDS.ED in MATLAB
self.assertAlmostEqual(6.570532945599105e+11, impact.tot_value)
self.assertAlmostEqual(6.512201157564421e+09, impact.aai_agg, 5)
self.assertTrue(np.isclose(6.512201157564421e+09, impact.aai_agg))
def _cutoff_hazard_damage(self, exposures, impf_set, hazard):
"""Cutoff of hazard events which generate damage with a frequency higher
than hazard_freq_cutoff.
Parameters
----------
exposures : climada.entity.Exposures
exposures instance
imp_set : climada.entity.ImpactFuncSet
impact function set instance
hazard : climada.hazard.Hazard
hazard instance
Returns
-------
new_haz : climada.hazard.Hazard
Hazard without events which generate damage with a frequency
higher than hazard_freq_cutoff
"""
if self.hazard_freq_cutoff == 0:
return hazard
if self.exp_region_id:
# compute impact only in selected region
in_reg = np.logical_or.reduce([
exposures.gdf.region_id.values == reg
for reg in self.exp_region_id
])
exp_imp = Exposures(exposures.gdf[in_reg], crs=exposures.crs)
else:
exp_imp = exposures
from climada.engine.impact import Impact
imp = Impact()
imp.calc(exp_imp, impf_set, hazard)
LOGGER.debug('Cutting events whose damage have a frequency > %s.',
self.hazard_freq_cutoff)
new_haz = copy.deepcopy(hazard)
sort_idxs = np.argsort(imp.at_event)[::-1]
exceed_freq = np.cumsum(imp.frequency[sort_idxs])
cutoff = exceed_freq > self.hazard_freq_cutoff
sel_haz = sort_idxs[cutoff]
for row in sel_haz:
new_haz.intensity.data[new_haz.intensity.indptr[row]:new_haz.
intensity.indptr[row + 1]] = 0
new_haz.intensity.eliminate_zeros()
return new_haz
def test_impact_pass(self):
"""Plot impact exceedence frequency curves."""
myent = Entity.from_excel(ENT_DEMO_TODAY)
myent.exposures.check()
myhaz = Hazard.from_mat(HAZ_DEMO_MAT)
myimp = Impact()
myimp.calc(myent.exposures, myent.impact_funcs, myhaz)
ifc = myimp.calc_freq_curve()
myax = ifc.plot()
self.assertIn('Exceedance frequency curve', myax.get_title())
ifc2 = ImpactFreqCurve()
ifc2.return_per = ifc.return_per
ifc2.impact = 1.5e11 * np.ones(ifc2.return_per.size)
ifc2.unit = ''
ifc2.label = 'prove'
ifc2.plot(axis=myax)
def test_excel_io(self):
"""Test write and read in excel"""
ent = Entity.from_excel(ENT_DEMO_TODAY)
ent.check()
hazard = Hazard.from_mat(HAZ_TEST_MAT)
imp_write = Impact()
ent.exposures.assign_centroids(hazard)
imp_write.calc(ent.exposures, ent.impact_funcs, hazard)
file_name = DATA_FOLDER.joinpath('test.xlsx')
imp_write.write_excel(file_name)
imp_read = Impact.from_excel(file_name)
np.testing.assert_array_equal(imp_write.event_id, imp_read.event_id)
np.testing.assert_array_equal(imp_write.date, imp_read.date)
np.testing.assert_array_equal(imp_write.coord_exp, imp_read.coord_exp)
np.testing.assert_array_almost_equal_nulp(imp_write.eai_exp,
imp_read.eai_exp,
nulp=5)
np.testing.assert_array_almost_equal_nulp(imp_write.at_event,
imp_read.at_event,
nulp=5)
np.testing.assert_array_equal(imp_write.frequency, imp_read.frequency)
self.assertEqual(imp_write.tot_value, imp_read.tot_value)
self.assertEqual(imp_write.aai_agg, imp_read.aai_agg)
self.assertEqual(imp_write.unit, imp_read.unit)
self.assertEqual(
0,
len([
i for i, j in zip(imp_write.event_name, imp_read.event_name)
if i != j
]))
self.assertIsInstance(imp_read.crs, str)
def _calc_impact(self, new_exp, new_ifs, new_haz):
"""Compute impact and risk transfer of measure implemented over inputs.
Parameters:
new_exp (Exposures): exposures once measure applied
new_ifs (ImpactFuncSet): impact functions once measure applied
new_haz (Hazard): hazard once measure applied
Returns:
Impact, float
"""
from climada.engine.impact import Impact
imp = Impact()
imp.calc(new_exp, new_ifs, new_haz)
risk_transfer = 0
if self.risk_transf_attach + self.risk_transf_cover > 0:
imp_layer = np.minimum(np.maximum(imp.at_event - self.risk_transf_attach, 0),
self.risk_transf_cover)
risk_transfer = np.sum(imp_layer * imp.frequency)
imp.at_event = np.maximum(imp.at_event -imp_layer, 0)
imp.aai_agg = np.sum(imp.at_event * imp.frequency)
# expected annual impact per exposure no longer valid
imp.eai_exp = np.array([])
return imp, risk_transfer
def test_select_event_identity_pass(self):
""" test select same impact with event name, id and date """
# Read default entity values
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
imp = Impact()
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute the impact over the whole exposures
imp.calc(ent.exposures, ent.impact_funcs, hazard, save_mat=True)
sel_imp = imp.select(event_ids=imp.event_id,
event_names=imp.event_name,
dates=(min(imp.date), max(imp.date)))
self.assertEqual(sel_imp.crs, imp.crs)
self.assertEqual(sel_imp.unit, imp.unit)
self.assertTrue(np.array_equal(sel_imp.event_id, imp.event_id))
self.assertEqual(sel_imp.event_name, imp.event_name)
self.assertTrue(np.array_equal(sel_imp.date, imp.date))
self.assertTrue(np.array_equal(sel_imp.frequency, imp.frequency))
self.assertTrue(np.array_equal(sel_imp.at_event, imp.at_event))
self.assertTrue(
np.array_equal(sel_imp.imp_mat.todense(), imp.imp_mat.todense()))
self.assertTrue(np.array_equal(sel_imp.eai_exp, imp.eai_exp))
self.assertAlmostEqual(round(sel_imp.aai_agg, 5),
round(imp.aai_agg, 5))
self.assertEqual(sel_imp.tot_value, imp.tot_value)
self.assertTrue(np.array_equal(sel_imp.coord_exp, imp.coord_exp))
self.assertIsInstance(sel_imp, Impact)
self.assertIsInstance(sel_imp.imp_mat, sparse.csr_matrix)
def test_calc_if_pass(self):
"""Execute when no if_HAZ present, but only if_"""
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.exposures.gdf.rename(columns={'if_TC': 'if_'}, inplace=True)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
impact.calc(ent.exposures, ent.impact_funcs, hazard)
# Check result
num_events = len(hazard.event_id)
num_exp = ent.exposures.gdf.shape[0]
# Check relative errors as well when absolute value gt 1.0e-7
# impact.at_event == EDS.damage in MATLAB
self.assertEqual(num_events, len(impact.at_event))
self.assertEqual(0, impact.at_event[0])
self.assertEqual(0, impact.at_event[int(num_events / 2)])
self.assertAlmostEqual(1.472482938320243e+08, impact.at_event[13809])
self.assertEqual(7.076504723057620e+10, impact.at_event[12147])
self.assertEqual(0, impact.at_event[num_events - 1])
# impact.eai_exp == EDS.ED_at_centroid in MATLAB
self.assertEqual(num_exp, len(impact.eai_exp))
self.assertAlmostEqual(1.518553670803242e+08, impact.eai_exp[0])
self.assertAlmostEqual(1.373490457046383e+08,
impact.eai_exp[int(num_exp / 2)], 6)
self.assertTrue(
np.isclose(1.373490457046383e+08,
impact.eai_exp[int(num_exp / 2)]))
self.assertAlmostEqual(1.066837260150042e+08,
impact.eai_exp[num_exp - 1], 6)
self.assertTrue(
np.isclose(1.066837260150042e+08,
impact.eai_exp[int(num_exp - 1)]))
# impact.tot_value == EDS.Value in MATLAB
# impact.aai_agg == EDS.ED in MATLAB
self.assertAlmostEqual(6.570532945599105e+11, impact.tot_value)
self.assertAlmostEqual(6.512201157564421e+09, impact.aai_agg, 5)
self.assertTrue(np.isclose(6.512201157564421e+09, impact.aai_agg))
def test_ref_value_insure_pass(self):
"""Test result against reference value"""
# Read demo entity values
# Set the entity default file to the demo one
ent = Entity()
ent.read_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard('TC')
hazard.read_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
impact.at_event = np.zeros(hazard.intensity.shape[0])
impact.eai_exp = np.zeros(len(ent.exposures.value))
impact.tot_value = 0
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute impact for 6th exposure
iexp = 5
# Take its impact function
imp_id = ent.exposures.if_TC[iexp]
imp_fun = ent.impact_funcs.get_func(hazard.tag.haz_type, imp_id)
# Compute
insure_flag = True
impact._exp_impact(np.array([iexp]), ent.exposures, hazard, imp_fun,
insure_flag)
self.assertEqual(impact.eai_exp.size, ent.exposures.shape[0])
self.assertEqual(impact.at_event.size, hazard.intensity.shape[0])
events_pos = hazard.intensity[:, ent.exposures.centr_TC[iexp]].nonzero(
)[0]
res_exp = np.zeros((ent.exposures.shape[0]))
res_exp[iexp] = np.sum(impact.at_event[events_pos] *
hazard.frequency[events_pos])
self.assertTrue(np.array_equal(res_exp, impact.eai_exp))
self.assertEqual(0, impact.at_event[12])
# Check first 3 values
self.assertEqual(0, impact.at_event[12])
self.assertEqual(0, impact.at_event[41])
self.assertEqual(1.0626600695059455e+06, impact.at_event[44])
# Check intermediate values
self.assertEqual(0, impact.at_event[6281])
self.assertEqual(0, impact.at_event[4998])
self.assertEqual(0, impact.at_event[9527])
self.assertEqual(1.3318063850487845e+08, impact.at_event[7192])
self.assertEqual(4.667108555054083e+06, impact.at_event[8624])
# Check last 3 values
self.assertEqual(0, impact.at_event[14349])
self.assertEqual(0, impact.at_event[14347])
self.assertEqual(0, impact.at_event[14309])
def test_impact_pass(self):
"""Plot impact exceedence frequency curves."""
myent = Entity()
myent.read_excel(ENT_DEMO_TODAY)
myent.exposures.check()
myhaz = Hazard('TC')
myhaz.read_mat(HAZ_DEMO_MAT)
myimp = Impact()
myimp.calc(myent.exposures, myent.impact_funcs, myhaz)
ifc = myimp.calc_freq_curve()
myfig, _ = ifc.plot()
self.assertIn('Exceedance frequency curve',\
myfig._suptitle.get_text())
ifc2 = ImpactFreqCurve()
ifc2.return_per = ifc.return_per
ifc2.impact = 1.5e11 * np.ones(ifc2.return_per.size)
ifc2.unit = ''
ifc2.label = 'prove'
ifc.plot_compare(ifc2)
def _calc_impact(self, new_exp, new_impfs, new_haz):
"""Compute impact and risk transfer of measure implemented over inputs.
Parameters
----------
new_exp : climada.entity.Exposures
exposures once measure applied
new_ifs : climada.entity.ImpactFuncSet
impact function set once measure applied
new_haz : climada.hazard.Hazard
hazard once measure applied
Returns
-------
: climada.engine.Impact
"""
from climada.engine.impact import Impact
imp = Impact()
imp.calc(new_exp, new_impfs, new_haz)
return imp.calc_risk_transfer(self.risk_transf_attach,
self.risk_transf_cover)
def test_local_exceedance_imp_pass(self):
"""Test calc local impacts per return period"""
# Read default entity values
ent = Entity.from_excel(ENT_DEMO_TODAY)
ent.check()
# Read default hazard file
hazard = Hazard.from_mat(HAZ_TEST_MAT)
# Create impact object
impact = Impact()
# Assign centroids to exposures
ent.exposures.assign_centroids(hazard)
# Compute the impact over the whole exposures
impact.calc(ent.exposures, ent.impact_funcs, hazard, save_mat=True)
# Compute the impact per return period over the whole exposures
impact_rp = impact.local_exceedance_imp(return_periods=(10, 40))
self.assertIsInstance(impact_rp, np.ndarray)
self.assertEqual(impact_rp.size, 2 * ent.exposures.gdf.value.size)
self.assertAlmostEqual(np.max(impact_rp), 2916964966.388219, places=5)
self.assertAlmostEqual(np.min(impact_rp), 444457580.131494, places=5)
def _calc_impact_measures(self,
hazard,
exposures,
meas_set,
imp_fun_set,
when='future',
risk_func=risk_aai_agg,
save_imp=False):
"""Compute impact of each measure and transform it to input risk
measurement. Set reference year from exposures value.
Parameters:
hazard (Hazard): hazard.
exposures (Exposures): exposures.
meas_set (MeasureSet): set of measures.
imp_fun_set (ImpactFuncSet): set of impact functions.
when (str, optional): 'present' or 'future'. The conditions that
are being considered.
risk_func (function, optional): function used to transform impact
to a risk measurement.
save_imp (bool, optional): activate if Impact of each measure is
saved. Default: False.
"""
impact_meas = dict()
# compute impact without measures
LOGGER.debug('%s impact with no measure.', when)
imp_tmp = Impact()
imp_tmp.calc(exposures, imp_fun_set, hazard)
impact_meas[NO_MEASURE] = dict()
impact_meas[NO_MEASURE]['cost'] = (0, 0)
impact_meas[NO_MEASURE]['risk'] = risk_func(imp_tmp)
impact_meas[NO_MEASURE]['risk_transf'] = 0.0
impact_meas[NO_MEASURE]['efc'] = imp_tmp.calc_freq_curve()
if save_imp:
impact_meas[NO_MEASURE]['impact'] = imp_tmp
# compute impact for each measure
for measure in meas_set.get_measure(hazard.tag.haz_type):
LOGGER.debug('%s impact of measure %s.', when, measure.name)
imp_tmp, risk_transf = measure.calc_impact(exposures, imp_fun_set,
hazard)
impact_meas[measure.name] = dict()
impact_meas[measure.name]['cost'] = (
measure.cost, measure.risk_transf_cost_factor)
impact_meas[measure.name]['risk'] = risk_func(imp_tmp)
impact_meas[measure.name]['risk_transf'] = risk_func(risk_transf)
impact_meas[measure.name]['efc'] = imp_tmp.calc_freq_curve()
if save_imp:
impact_meas[measure.name]['impact'] = imp_tmp
# if present reference provided save it
if when == 'future':
self.imp_meas_future = impact_meas
else:
self.imp_meas_present = impact_meas
def _calc_impact_measures(self, hazard, exposures, meas_set, imp_fun_set,
when='future', risk_func=risk_aai_agg, save_imp=False):
"""Compute impact of each measure and transform it to input risk
measurement. Set reference year from exposures value.
Parameters
----------
hazard : climada.Hazard
exposures : climada.entity.Exposures
meas_set : climada.MeasureSet
set of measures.
imp_fun_set : ImpactFuncSet
set of impact functions.
when : str, optional
'present' or 'future'. The conditions that
are being considered.
risk_func : func, optional
function describing risk measure given an Impact.
Default: average annual impact (aggregated).
save_imp : bool, optional
activate if Impact of each measure is
saved. Default: False.
"""
impact_meas = dict()
# compute impact without measures
LOGGER.debug('%s impact with no measure.', when)
imp_tmp = Impact()
imp_tmp.calc(exposures, imp_fun_set, hazard)
impact_meas[NO_MEASURE] = dict()
impact_meas[NO_MEASURE]['cost'] = (0, 0)
impact_meas[NO_MEASURE]['risk'] = risk_func(imp_tmp)
impact_meas[NO_MEASURE]['risk_transf'] = 0.0
impact_meas[NO_MEASURE]['efc'] = imp_tmp.calc_freq_curve()
if save_imp:
impact_meas[NO_MEASURE]['impact'] = imp_tmp
# compute impact for each measure
for measure in meas_set.get_measure(hazard.tag.haz_type):
LOGGER.debug('%s impact of measure %s.', when, measure.name)
imp_tmp, risk_transf = measure.calc_impact(exposures, imp_fun_set, hazard)
impact_meas[measure.name] = dict()
impact_meas[measure.name]['cost'] = (measure.cost, measure.risk_transf_cost_factor)
impact_meas[measure.name]['risk'] = risk_func(imp_tmp)
impact_meas[measure.name]['risk_transf'] = risk_func(risk_transf)
impact_meas[measure.name]['efc'] = imp_tmp.calc_freq_curve()
if save_imp:
impact_meas[measure.name]['impact'] = imp_tmp
# if present reference provided save it
if when == 'future':
self.imp_meas_future = impact_meas
else:
self.imp_meas_present = impact_meas
def plot_waterfall_accumulated(self,
hazard,
entity,
ent_future,
risk_func=risk_aai_agg,
imp_time_depen=1,
axis=None,
**kwargs):
""" Plot waterfall graph with accumulated values from present to future
year. Call after calc() with save_imp=True. Provide same inputs as in calc.
Parameters:
hazard (Hazard): hazard
entity (Entity): entity
ent_future (Entity): entity in the future
risk_func (func, optional): function describing risk measure given
an Impact. Default: average annual impact (aggregated).
imp_time_depen (float, optional): parameter which represent time
evolution of impact. Default: 1 (linear).
axis (matplotlib.axes._subplots.AxesSubplot, optional): axis to use
kwargs (optional): arguments for bar matplotlib function, e.g. alpha=0.5
Returns:
matplotlib.axes._subplots.AxesSubplot
"""
if not self.imp_meas_future or not self.imp_meas_present:
LOGGER.error('Compute CostBenefit.calc() first')
raise ValueError
if ent_future.exposures.ref_year == entity.exposures.ref_year:
LOGGER.error(
'Same reference years for future and present entities.')
raise ValueError
self.present_year = entity.exposures.ref_year
self.future_year = ent_future.exposures.ref_year
# current situation
curr_risk = self.imp_meas_present[NO_MEASURE]['risk']
time_dep = self._time_dependency_array()
risk_curr = self._npv_unaverted_impact(curr_risk, entity.disc_rates,
time_dep)
LOGGER.info('Current total risk at {:d}: {:.3e}'.format(
self.future_year, risk_curr))
# changing future
time_dep = self._time_dependency_array(imp_time_depen)
# socio-economic dev
imp = Impact()
imp.calc(ent_future.exposures, ent_future.impact_funcs, hazard)
risk_dev = self._npv_unaverted_impact(risk_func(imp),
entity.disc_rates, time_dep,
curr_risk)
LOGGER.info('Total risk with development at {:d}: {:.3e}'.format( \
self.future_year, risk_dev))
# socioecon + cc
risk_tot = self._npv_unaverted_impact(self.imp_meas_future[NO_MEASURE]['risk'], \
entity.disc_rates, time_dep, curr_risk)
LOGGER.info('Total risk with development and climate change at {:d}: {:.3e}'.\
format(self.future_year, risk_tot))
# plot
if not axis:
_, axis = plt.subplots(1, 1)
norm_fact, norm_name = _norm_values(curr_risk)
axis.bar(1, risk_curr / norm_fact, **kwargs)
axis.text(1, risk_curr/norm_fact, str(int(round(risk_curr/norm_fact))), \
horizontalalignment='center', verticalalignment='bottom', \
fontsize=12, color='k')
axis.bar(2,
height=(risk_dev - risk_curr) / norm_fact,
bottom=risk_curr / norm_fact,
**kwargs)
axis.text(2, risk_curr/norm_fact + (risk_dev-risk_curr)/norm_fact/2, \
str(int(round((risk_dev-risk_curr)/norm_fact))), \
horizontalalignment='center', verticalalignment='center', fontsize=12, color='k')
axis.bar(3,
height=(risk_tot - risk_dev) / norm_fact,
bottom=risk_dev / norm_fact,
**kwargs)
axis.text(3, risk_dev/norm_fact + (risk_tot-risk_dev)/norm_fact/2, \
str(int(round((risk_tot-risk_dev)/norm_fact))), \
horizontalalignment='center', verticalalignment='center', fontsize=12, color='k')
axis.bar(4, height=risk_tot / norm_fact, **kwargs)
axis.text(4, risk_tot/norm_fact, str(int(round(risk_tot/norm_fact))), \
horizontalalignment='center', verticalalignment='bottom', \
fontsize=12, color='k')
axis.set_xticks(np.arange(4) + 1)
axis.set_xticklabels(['Risk ' + str(self.present_year), \
'Economic \ndevelopment', 'Climate \nchange', 'Risk ' + str(self.future_year)])
axis.set_ylabel('Impact (' + self.unit + ' ' + norm_name + ')')
axis.set_title('Total accumulated impact from {:d} to {:d}'.format( \
self.present_year, self.future_year))
return axis
def plot_waterfall(hazard,
entity,
haz_future,
ent_future,
risk_func=risk_aai_agg,
axis=None,
**kwargs):
""" Plot waterfall graph at future with given risk metric. Can be called
before and after calc().
Parameters:
hazard (Hazard): hazard
entity (Entity): entity
haz_future (Hazard): hazard in the future (future year provided at
ent_future)
ent_future (Entity): entity in the future
risk_func (func, optional): function describing risk measure given
an Impact. Default: average annual impact (aggregated).
axis (matplotlib.axes._subplots.AxesSubplot, optional): axis to use
kwargs (optional): arguments for bar matplotlib function, e.g. alpha=0.5
Returns:
matplotlib.axes._subplots.AxesSubplot
"""
if ent_future.exposures.ref_year == entity.exposures.ref_year:
LOGGER.error(
'Same reference years for future and present entities.')
raise ValueError
present_year = entity.exposures.ref_year
future_year = ent_future.exposures.ref_year
imp = Impact()
imp.calc(entity.exposures, entity.impact_funcs, hazard)
curr_risk = risk_func(imp)
imp = Impact()
imp.calc(ent_future.exposures, ent_future.impact_funcs, haz_future)
fut_risk = risk_func(imp)
if not axis:
_, axis = plt.subplots(1, 1)
norm_fact, norm_name = _norm_values(curr_risk)
# current situation
LOGGER.info('Risk at {:d}: {:.3e}'.format(present_year, curr_risk))
# changing future
# socio-economic dev
imp = Impact()
imp.calc(ent_future.exposures, ent_future.impact_funcs, hazard)
risk_dev = risk_func(imp)
LOGGER.info('Risk with development at {:d}: {:.3e}'.format(
future_year, risk_dev))
# socioecon + cc
LOGGER.info('Risk with development and climate change at {:d}: {:.3e}'.\
format(future_year, fut_risk))
axis.bar(1, curr_risk / norm_fact, **kwargs)
axis.text(1, curr_risk/norm_fact, str(int(round(curr_risk/norm_fact))), \
horizontalalignment='center', verticalalignment='bottom', \
fontsize=12, color='k')
axis.bar(2,
height=(risk_dev - curr_risk) / norm_fact,
bottom=curr_risk / norm_fact,
**kwargs)
axis.text(2, curr_risk/norm_fact + (risk_dev-curr_risk)/norm_fact/2, \
str(int(round((risk_dev-curr_risk)/norm_fact))), \
horizontalalignment='center', verticalalignment='center', fontsize=12, color='k')
axis.bar(3,
height=(fut_risk - risk_dev) / norm_fact,
bottom=risk_dev / norm_fact,
**kwargs)
axis.text(3, risk_dev/norm_fact + (fut_risk-risk_dev)/norm_fact/2, \
str(int(round((fut_risk-risk_dev)/norm_fact))), \
horizontalalignment='center', verticalalignment='center', fontsize=12, color='k')
axis.bar(4, height=fut_risk / norm_fact, **kwargs)
axis.text(4, fut_risk/norm_fact, str(int(round(fut_risk/norm_fact))), \
horizontalalignment='center', verticalalignment='bottom', \
fontsize=12, color='k')
axis.set_xticks(np.arange(4) + 1)
axis.set_xticklabels(['Risk ' + str(present_year), \
'Economic \ndevelopment', 'Climate \nchange', 'Risk ' + str(future_year)])
axis.set_ylabel('Impact (' + imp.unit + ' ' + norm_name + ')')
axis.set_title('Risk at {:d} and {:d}'.format(present_year,
future_year))
return axis
def test_ref_value_rp_pass(self):
"""Test result against reference value with given return periods"""
imp = Impact()
imp.frequency = np.ones(10) * 6.211180124223603e-04
imp.at_event = np.zeros(10)
imp.at_event[0] = 0
imp.at_event[1] = 0.400665463736549e9
imp.at_event[2] = 3.150330960044466e9
imp.at_event[3] = 3.715826406781887e9
imp.at_event[4] = 2.900244271902339e9
imp.at_event[5] = 0.778570745161971e9
imp.at_event[6] = 0.698736262566472e9
imp.at_event[7] = 0.381063674256423e9
imp.at_event[8] = 0.569142464157450e9
imp.at_event[9] = 0.467572545849132e9
imp.unit = 'USD'
ifc = imp.calc_freq_curve(np.array([100, 500, 1000]))
self.assertEqual(3, len(ifc.return_per))
self.assertEqual(100, ifc.return_per[0])
self.assertEqual(500, ifc.return_per[1])
self.assertEqual(1000, ifc.return_per[2])
self.assertEqual(3, len(ifc.impact))
self.assertEqual(0, ifc.impact[0])
self.assertEqual(2320408028.5695677, ifc.impact[1])
self.assertEqual(3287314329.129928, ifc.impact[2])
self.assertEqual('Exceedance frequency curve', ifc.label)
self.assertEqual('USD', ifc.unit)
def test_risk_trans_pass(self):
"""Test calc_risk_transfer"""
# Create impact object
imp = Impact()
imp.event_id = np.arange(10)
imp.event_name = [0, 1, 2, 3, 4, 5, 6, 7, 8, 15]
imp.date = np.arange(10)
imp.coord_exp = np.array([[1, 2], [2, 3]])
imp.crs = DEF_CRS
imp.eai_exp = np.array([1, 2])
imp.at_event = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 15])
imp.frequency = np.ones(10) / 5
imp.tot_value = 10
imp.aai_agg = 100
imp.unit = 'USD'
imp.imp_mat = sparse.csr_matrix(np.empty((0, 0)))
new_imp, imp_rt = imp.calc_risk_transfer(2, 10)
self.assertEqual(new_imp.unit, imp.unit)
self.assertEqual(new_imp.tot_value, imp.tot_value)
self.assertTrue((new_imp.imp_mat == imp.imp_mat).toarray().all())
self.assertEqual(new_imp.event_name, imp.event_name)
self.assertTrue(np.allclose(new_imp.event_id, imp.event_id))
self.assertTrue(np.allclose(new_imp.date, imp.date))
self.assertTrue(np.allclose(new_imp.frequency, imp.frequency))
self.assertTrue(np.allclose(new_imp.coord_exp, np.array([])))
self.assertTrue(np.allclose(new_imp.eai_exp, np.array([])))
self.assertTrue(
np.allclose(new_imp.at_event,
np.array([0, 1, 2, 2, 2, 2, 2, 2, 2, 5])))
self.assertAlmostEqual(new_imp.aai_agg, 4.0)
self.assertEqual(imp_rt.unit, imp.unit)
self.assertEqual(imp_rt.tot_value, imp.tot_value)
self.assertTrue((imp_rt.imp_mat == imp.imp_mat).toarray().all())
self.assertEqual(imp_rt.event_name, imp.event_name)
self.assertTrue(np.allclose(imp_rt.event_id, imp.event_id))
self.assertTrue(np.allclose(imp_rt.date, imp.date))
self.assertTrue(np.allclose(imp_rt.frequency, imp.frequency))
self.assertTrue(np.allclose(imp_rt.coord_exp, np.array([])))
self.assertTrue(np.allclose(imp_rt.eai_exp, np.array([])))
self.assertTrue(
np.allclose(imp_rt.at_event,
np.array([0, 0, 0, 1, 2, 3, 4, 5, 6, 10])))
self.assertAlmostEqual(imp_rt.aai_agg, 6.2)
def test_write_imp_mat(self):
"""Test write_excel_imp_mat function"""
impact = Impact()
impact.imp_mat = np.zeros((5, 4))
impact.imp_mat[0, :] = np.arange(4)
impact.imp_mat[1, :] = np.arange(4) * 2
impact.imp_mat[2, :] = np.arange(4) * 3
impact.imp_mat[3, :] = np.arange(4) * 4
impact.imp_mat[4, :] = np.arange(4) * 5
impact.imp_mat = sparse.csr_matrix(impact.imp_mat)
file_name = os.path.join(DATA_FOLDER, 'test_imp_mat')
impact.write_sparse_csr(file_name)
read_imp_mat = Impact().read_sparse_csr(file_name + '.npz')
for irow in range(5):
self.assertTrue(
np.array_equal(
np.array(read_imp_mat[irow, :].toarray()).reshape(-1),
np.array(impact.imp_mat[irow, :].toarray()).reshape(-1)))
def test_write_read_exp_test(self):
"""Test result against reference value"""
# Create impact object
num_ev = 5
num_exp = 10
imp_write = Impact()
imp_write.tag = {
'exp': Tag('file_exp.p', 'descr exp'),
'haz': TagHaz('TC', 'file_haz.p', 'descr haz'),
'if_set': Tag()
}
imp_write.event_id = np.arange(num_ev)
imp_write.event_name = [
'event_' + str(num) for num in imp_write.event_id
]
imp_write.date = np.ones(num_ev)
imp_write.coord_exp = np.zeros((num_exp, 2))
imp_write.coord_exp[:, 0] = 1.5
imp_write.coord_exp[:, 1] = 2.5
imp_write.eai_exp = np.arange(num_exp) * 100
imp_write.at_event = np.arange(num_ev) * 50
imp_write.frequency = np.ones(num_ev) * 0.1
imp_write.tot_value = 1000
imp_write.aai_agg = 1001
imp_write.unit = 'USD'
file_name = os.path.join(DATA_FOLDER, 'test.csv')
imp_write.write_csv(file_name)
imp_read = Impact()
imp_read.read_csv(file_name)
self.assertTrue(np.array_equal(imp_write.event_id, imp_read.event_id))
self.assertTrue(np.array_equal(imp_write.date, imp_read.date))
self.assertTrue(np.array_equal(imp_write.coord_exp,
imp_read.coord_exp))
self.assertTrue(np.array_equal(imp_write.eai_exp, imp_read.eai_exp))
self.assertTrue(np.array_equal(imp_write.at_event, imp_read.at_event))
self.assertTrue(np.array_equal(imp_write.frequency,
imp_read.frequency))
self.assertEqual(imp_write.tot_value, imp_read.tot_value)
self.assertEqual(imp_write.aai_agg, imp_read.aai_agg)
self.assertEqual(imp_write.unit, imp_read.unit)
self.assertEqual(
0,
len([
i for i, j in zip(imp_write.event_name, imp_read.event_name)
if i != j
]))
self.assertIsInstance(imp_read.crs, dict)
def test_ref_value_pass(self):
"""Test result against reference value"""
imp = Impact()
imp.frequency = np.ones(10) * 6.211180124223603e-04
imp.at_event = np.zeros(10)
imp.at_event[0] = 0
imp.at_event[1] = 0.400665463736549e9
imp.at_event[2] = 3.150330960044466e9
imp.at_event[3] = 3.715826406781887e9
imp.at_event[4] = 2.900244271902339e9
imp.at_event[5] = 0.778570745161971e9
imp.at_event[6] = 0.698736262566472e9
imp.at_event[7] = 0.381063674256423e9
imp.at_event[8] = 0.569142464157450e9
imp.at_event[9] = 0.467572545849132e9
imp.unit = 'USD'
ifc = imp.calc_freq_curve()
self.assertEqual(10, len(ifc.return_per))
self.assertEqual(1610.0000000000000, ifc.return_per[9])
self.assertEqual(805.00000000000000, ifc.return_per[8])
self.assertEqual(536.66666666666663, ifc.return_per[7])
self.assertEqual(402.500000000000, ifc.return_per[6])
self.assertEqual(322.000000000000, ifc.return_per[5])
self.assertEqual(268.33333333333331, ifc.return_per[4])
self.assertEqual(230.000000000000, ifc.return_per[3])
self.assertEqual(201.250000000000, ifc.return_per[2])
self.assertEqual(178.88888888888889, ifc.return_per[1])
self.assertEqual(161.000000000000, ifc.return_per[0])
self.assertEqual(10, len(ifc.impact))
self.assertEqual(3.715826406781887e9, ifc.impact[9])
self.assertEqual(3.150330960044466e9, ifc.impact[8])
self.assertEqual(2.900244271902339e9, ifc.impact[7])
self.assertEqual(0.778570745161971e9, ifc.impact[6])
self.assertEqual(0.698736262566472e9, ifc.impact[5])
self.assertEqual(0.569142464157450e9, ifc.impact[4])
self.assertEqual(0.467572545849132e9, ifc.impact[3])
self.assertEqual(0.400665463736549e9, ifc.impact[2])
self.assertEqual(0.381063674256423e9, ifc.impact[1])
self.assertEqual(0, ifc.impact[0])
self.assertEqual('Exceedance frequency curve', ifc.label)
self.assertEqual('USD', ifc.unit)
def test_impact_year_set_sum(self):
"""Test result against reference value with given events"""
imp = Impact()
imp.frequency = np.ones(10) * 6.211180124223603e-04
imp.at_event = np.zeros(10)
imp.at_event[0] = 0
imp.at_event[1] = 0.400665463736549e9
imp.at_event[2] = 3.150330960044466e9
imp.at_event[3] = 3.715826406781887e9
imp.at_event[4] = 2.900244271902339e9
imp.at_event[5] = 0.778570745161971e9
imp.at_event[6] = 0.698736262566472e9
imp.at_event[7] = 0.381063674256423e9
imp.at_event[8] = 0.569142464157450e9
imp.at_event[9] = 0.467572545849132e9
imp.unit = 'USD'
imp.date = np.array([
732801, 716160, 718313, 712468, 732802, 729285, 732931, 715419,
722404, 718351
])
iys_all = imp.calc_impact_year_set()
iys = imp.calc_impact_year_set(all_years=False)
iys_all_yr = imp.calc_impact_year_set(year_range=(1975, 2000))
iys_yr = imp.calc_impact_year_set(all_years=False,
year_range=[1975, 2000])
iys_all_yr_1940 = imp.calc_impact_year_set(all_years=True,
year_range=[1940, 2000])
self.assertEqual(np.around(sum([iys[year] for year in iys])),
np.around(sum(imp.at_event)))
self.assertEqual(sum([iys[year] for year in iys]),
sum([iys_all[year] for year in iys_all]))
self.assertEqual(len(iys), 7)
self.assertEqual(len(iys_all), 57)
self.assertIn(1951 and 1959 and 2007, iys_all)
self.assertTrue(iys_all[1959] > 0)
self.assertAlmostEqual(3598980534.468811, iys_all[2007])
self.assertEqual(iys[1978], iys_all[1978])
self.assertAlmostEqual(iys[1951], imp.at_event[3])
# year range (yr):
self.assertEqual(len(iys_yr), 2)
self.assertEqual(len(iys_all_yr), 26)
self.assertEqual(sum([iys_yr[year] for year in iys_yr]),
sum([iys_all_yr[year] for year in iys_all_yr]))
self.assertIn(1997 and 1978, iys_yr)
self.assertFalse(2007 in iys_yr)
self.assertFalse(1959 in iys_yr)
self.assertEqual(len(iys_all_yr_1940), 61)
def dummy_impact():
imp = Impact()
imp.event_id = np.arange(6)
imp.event_name = [0, 1, 'two', 'three', 30, 31]
imp.date = np.arange(6)
imp.coord_exp = np.array([[1, 2], [1.5, 2.5]])
imp.crs = DEF_CRS
imp.eai_exp = np.array([7.2, 7.2])
imp.at_event = np.array([0, 2, 4, 6, 60, 62])
imp.frequency = np.array([1 / 6, 1 / 6, 1, 1, 1 / 30, 1 / 30])
imp.tot_value = 7
imp.aai_agg = 14.4
imp.unit = 'USD'
imp.imp_mat = sparse.csr_matrix(
np.array([[0, 0], [1, 1], [2, 2], [3, 3], [30, 30], [31, 31]]))
return imp
