def lognormal_object_test():
model_dict = {
'AF': (11.613073, 0.180683),
'CN': (10.328811, 0.100058),
'JP': (11.862534, 0.100779),
'US': (46.155474, 0.434135),
'IR': (9.318099, 0.100001)
}
model1 = LognormalModel('IR', 9.318099, 0.100001, 0.0)
model2 = LognormalModel('US', 46.155474, 0.434135, 0.0)
model3 = LognormalModel('IR', 10.986839, 0.128601, 0.0)
print('Testing that fatality rates calculation is correct...')
rates = model3.getLossRates(np.arange(5, 10))
testrates = np.array([
4.62833172e-10, 1.27558914e-06, 2.28027416e-04, 6.81282956e-03,
6.04383822e-02
])
np.testing.assert_almost_equal(rates, testrates)
print('Fatality rates calculation is correct.')
print('Testing model fatality comparison...')
assert model1 > model2
print('Passed model fatality comparison.')
print('More complete test of model fatality comparison...')
mlist = []
for key, values in model_dict.items():
mlist.append(LognormalModel(key, values[0], values[1], 0.0))
mlist.sort()
names = [m.name for m in mlist]
assert names != ['JP', 'US', 'CN', 'IR', 'AF']
print('Passed more complete test of model fatality comparison.')
print('Sorted list of country models:')
print('%5s %6s %6s %-6s %-14s' %
('Name', 'Theta', 'Beta', 'Area', 'Deaths'))
for model in mlist:
exp_pop = np.array([1e6, 1e6, 1e6, 1e6, 1e6])
mmirange = np.arange(5, 10)
deaths = model.getLosses(exp_pop, mmirange)
print('%5s %6.3f %6.3f %6.4f %14.4f' %
(model.name, model.theta, model.beta, model.getArea(), deaths))
def basic_test():
mmidata = np.array([[7, 8, 8, 8, 7], [8, 9, 9, 9, 8], [8, 9, 10, 9, 8],
[8, 9, 9, 8, 8], [7, 8, 8, 6, 5]],
dtype=np.float32)
popdata = np.ones_like(mmidata) * 1e7
isodata = np.array(
[[4, 4, 4, 4, 4], [4, 4, 4, 4, 4], [4, 4, 156, 156, 156],
[156, 156, 156, 156, 156], [156, 156, 156, 156, 156]],
dtype=np.int32)
shakefile = get_temp_file_name()
popfile = get_temp_file_name()
isofile = get_temp_file_name()
geodict = GeoDict({
'xmin': 0.5,
'xmax': 4.5,
'ymin': 0.5,
'ymax': 4.5,
'dx': 1.0,
'dy': 1.0,
'nx': 5,
'ny': 5
})
layers = OrderedDict([
('mmi', mmidata),
])
event_dict = {
'event_id': 'us12345678',
'magnitude': 7.8,
'depth': 10.0,
'lat': 34.123,
'lon': -118.123,
'event_timestamp': datetime.utcnow(),
'event_description': 'foo',
'event_network': 'us'
}
shake_dict = {
'event_id': 'us12345678',
'shakemap_id': 'us12345678',
'shakemap_version': 1,
'code_version': '4.5',
'process_timestamp': datetime.utcnow(),
'shakemap_originator': 'us',
'map_status': 'RELEASED',
'shakemap_event_type': 'ACTUAL'
}
unc_dict = {'mmi': (1, 1)}
shakegrid = ShakeGrid(layers, geodict, event_dict, shake_dict, unc_dict)
shakegrid.save(shakefile)
popgrid = Grid2D(popdata, geodict.copy())
isogrid = Grid2D(isodata, geodict.copy())
write(popgrid, popfile, 'netcdf')
write(isogrid, isofile, 'netcdf')
ratedict = {
4: {
'start': [2010, 2012, 2014, 2016],
'end': [2012, 2014, 2016, 2018],
'rate': [0.01, 0.02, 0.03, 0.04]
},
156: {
'start': [2010, 2012, 2014, 2016],
'end': [2012, 2014, 2016, 2018],
'rate': [0.02, 0.03, 0.04, 0.05]
}
}
popgrowth = PopulationGrowth(ratedict)
popyear = datetime.utcnow().year
exposure = Exposure(popfile, popyear, isofile, popgrowth=popgrowth)
expdict = exposure.calcExposure(shakefile)
modeldict = [
LognormalModel('AF', 11.613073, 0.180683, 1.0),
LognormalModel('CN', 10.328811, 0.100058, 1.0)
]
fatmodel = EmpiricalLoss(modeldict)
# for the purposes of this test, let's override the rates
# for Afghanistan and China with simpler numbers.
fatmodel.overrideModel(
'AF',
np.array([0, 0, 0, 0, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 0],
dtype=np.float32))
fatmodel.overrideModel(
'CN',
np.array([0, 0, 0, 0, 1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 0],
dtype=np.float32))
print('Testing very basic fatality calculation...')
fatdict = fatmodel.getLosses(expdict)
# strictly speaking, the afghanistant fatalities should be 462,000 but floating point precision dictates otherwise.
testdict = {'CN': 46111, 'AF': 461999, 'TotalFatalities': 508110}
for key, value in fatdict.items():
assert value == testdict[key]
print('Passed very basic fatality calculation...')
print('Testing grid fatality calculations...')
mmidata = exposure.getShakeGrid().getLayer('mmi').getData()
popdata = exposure.getPopulationGrid().getData()
isodata = exposure.getCountryGrid().getData()
fatgrid = fatmodel.getLossGrid(mmidata, popdata, isodata)
assert np.nansum(fatgrid) == 508111
print('Passed grid fatality calculations...')
# Testing modifying rates and stuffing them back in...
chile = LognormalModel('CL', 19.786773, 0.259531, 0.0)
rates = chile.getLossRates(np.arange(5, 10))
modrates = rates * 2 # does this make event twice as deadly?
# roughly the exposures from 2015-9-16 CL event
expo_pop = np.array(
[0, 0, 0, 1047000, 7314000, 1789000, 699000, 158000, 0, 0])
mmirange = np.arange(5, 10)
chile_deaths = chile.getLosses(expo_pop[4:9], mmirange)
chile_double_deaths = chile.getLosses(expo_pop[4:9],
mmirange,
rates=modrates)
print('Chile model fatalities: %f' % chile_deaths)
print('Chile model x2 fatalities: %f' % chile_double_deaths)