def upgrade_file(path, multipoint):
"""Upgrade to the latest NRML version"""
node0 = nrml.read(path, chatty=False)[0]
shutil.copy(path, path + '.bak') # make a backup of the original file
tag = striptag(node0.tag)
gml = True
if tag == 'vulnerabilityModel':
vf_dict, cat_dict = get_vulnerability_functions_04(path)
# below I am converting into a NRML 0.5 vulnerabilityModel
node0 = Node(
'vulnerabilityModel', cat_dict,
nodes=[obj_to_node(val) for val in vf_dict.values()])
gml = False
elif tag == 'fragilityModel':
node0 = read_nrml.convert_fragility_model_04(
nrml.read(path)[0], path)
gml = False
elif tag == 'sourceModel':
node0 = nrml.read(path)[0]
dic = groupby(node0.nodes, operator.itemgetter('tectonicRegion'))
node0.nodes = [Node('sourceGroup',
dict(tectonicRegion=trt, name="group %s" % i),
nodes=srcs)
for i, (trt, srcs) in enumerate(dic.items(), 1)]
if multipoint:
sourceconverter.update_source_model(node0, path + '.bak')
with open(path, 'wb') as f:
nrml.write([node0], f, gml=gml)
def mfds2multimfd(mfds):
"""
Convert a list of MFD nodes into a single MultiMFD node
"""
_, kind = mfds[0].tag.split('}')
node = Node('multiMFD', dict(kind=kind))
lengths = None
for field in mfd.multi_mfd.ASSOC[kind][1:]:
alias = mfd.multi_mfd.ALIAS.get(field, field)
if field in ('magnitudes', 'occurRates'):
data = [~getattr(m, field) for m in mfds]
lengths = [len(d) for d in data]
data = sum(data, []) # the list has to be flat
else:
try:
data = [m[alias] for m in mfds]
except KeyError:
if alias == 'binWidth':
# missing bindWidth in GR MDFs is ok
continue
else:
raise
node.append(Node(field, text=data))
if lengths: # this is the last field if present
node.append(Node('lengths', text=lengths))
return node
def surface_nodes(self):
"""
A single element list containing a planarSurface node
"""
node = Node('planarSurface')
for name, lon, lat, depth in zip(
'topLeft topRight bottomLeft bottomRight'.split(),
self.corner_lons, self.corner_lats, self.corner_depths):
node.append(Node(name, dict(lon=lon, lat=lat, depth=depth)))
return [node]
def build_slip_list_node(slip_list):
"""
:param slip_list:
an array of shape (N, 2) with columns (slip, weight)
:returns:
a hypoList node containing N slip nodes
"""
sliplist = Node('slipList', {})
for row in slip_list:
sliplist.append(
Node('slip', dict(weight=row[1]), row[0]))
return sliplist
def build_hypo_list_node(hypo_list):
"""
:param hypo_list:
an array of shape (N, 3) with columns (alongStrike, downDip, weight)
:returns:
a hypoList node containing N hypo nodes
"""
hypolist = Node('hypoList', {})
for row in hypo_list:
n = Node(
'hypo', dict(alongStrike=row[0], downDip=row[1], weight=row[2]))
hypolist.append(n)
return hypolist
def serialize(self, data, fmt='%10.7E'):
"""
Serialize a collection of ground motion fields to XML.
:param data:
An iterable of "GMF set" objects.
Each "GMF set" object should:
* have an `investigation_time` attribute
* have an `stochastic_event_set_id` attribute
* be iterable, yielding a sequence of "GMF" objects
Each "GMF" object should:
* have an `imt` attribute
* have an `sa_period` attribute (only if `imt` is 'SA')
* have an `sa_damping` attribute (only if `imt` is 'SA')
* have a `event_id` attribute (to indicate which rupture
contributed to this gmf)
* be iterable, yielding a sequence of "GMF node" objects
Each "GMF node" object should have:
* a `gmv` attribute (to indicate the ground motion value
* `lon` and `lat` attributes (to indicate the geographical location
of the ground motion field)
"""
gmf_set_nodes = []
for gmf_set in data:
gmf_set_node = Node('gmfSet')
if gmf_set.investigation_time:
gmf_set_node['investigationTime'] = str(
gmf_set.investigation_time)
gmf_set_node['stochasticEventSetId'] = str(
gmf_set.stochastic_event_set_id)
gmf_set_node.nodes = gen_gmfs(gmf_set)
gmf_set_nodes.append(gmf_set_node)
gmf_container = Node('gmfCollection')
gmf_container[SM_TREE_PATH] = self.sm_lt_path
gmf_container[GSIM_TREE_PATH] = self.gsim_lt_path
gmf_container.nodes = gmf_set_nodes
with open(self.dest, 'wb') as dest:
nrml.write([gmf_container], dest, fmt)
def gen_gmfs(gmf_set):
"""
Generate GMF nodes from a gmf_set
:param gmf_set: a sequence of GMF objects with attributes
imt, sa_period, sa_damping, event_id and containing a list
of GMF nodes with attributes gmv and location. The nodes
are sorted by lon/lat.
"""
for gmf in gmf_set:
gmf_node = Node('gmf')
gmf_node['IMT'] = gmf.imt
if gmf.imt == 'SA':
gmf_node['saPeriod'] = str(gmf.sa_period)
gmf_node['saDamping'] = str(gmf.sa_damping)
gmf_node['ruptureId'] = gmf.event_id
sorted_nodes = sorted(gmf)
gmf_node.nodes = (
Node('node', dict(gmv=n.gmv, lon=n.location.x, lat=n.location.y))
for n in sorted_nodes)
yield gmf_node
def build_multi_mfd(mfd):
"""
Parses the MultiMFD as a Node
:param mfd:
MFD as instance of :class:
`openquake.hazardlib.mfd.multi_mfd.MultiMFD`
:returns:
Instance of :class:`openquake.baselib.node.Node`
"""
node = Node("multiMFD", dict(kind=mfd.kind))
for name in sorted(mfd.kwargs):
values = mfd.kwargs[name]
if name in ('magnitudes', 'occurRates'):
values = sum(values, [])
node.append(Node(name, text=values))
if 'occurRates' in mfd.kwargs:
lengths = [len(rates) for rates in mfd.kwargs['occurRates']]
node.append(Node('lengths', text=lengths))
return node
def _read_csv(self):
"""
:yields: asset nodes
"""
expected_header = self._csv_header()
for fname in self.datafiles:
with open(fname, encoding='utf-8') as f:
fields = next(csv.reader(f))
header = set(fields)
if len(header) < len(fields):
raise InvalidFile(
'%s: The header %s contains a duplicated field' %
(fname, header))
elif expected_header - header - {'exposure', 'country'}:
raise InvalidFile(
'Unexpected header in %s\nExpected: %s\nGot: %s' %
(fname, sorted(expected_header), sorted(header)))
occupancy_periods = self.occupancy_periods.split()
for fname in self.datafiles:
with open(fname, encoding='utf-8') as f:
for i, dic in enumerate(csv.DictReader(f), 1):
asset = Node('asset', lineno=i)
with context(fname, asset):
asset['id'] = dic['id']
asset['number'] = valid.positivefloat(dic['number'])
asset['taxonomy'] = dic['taxonomy']
if 'area' in dic: # optional attribute
asset['area'] = dic['area']
loc = Node('location',
dict(lon=valid.longitude(dic['lon']),
lat=valid.latitude(dic['lat'])))
costs = Node('costs')
for cost in self.cost_types['name']:
a = dict(type=cost, value=dic[cost])
if 'retrofitted' in dic:
a['retrofitted'] = dic['retrofitted']
costs.append(Node('cost', a))
occupancies = Node('occupancies')
for period in occupancy_periods:
a = dict(occupants=float(dic[period]),
period=period)
occupancies.append(Node('occupancy', a))
tags = Node('tags')
for tagname in self.tagcol.tagnames:
if tagname not in (
'taxonomy', 'exposure', 'country'):
tags.attrib[tagname] = dic[tagname]
asset.nodes.extend([loc, costs, occupancies, tags])
yield asset
def complex_fault_node(edges):
"""
:param edges: a list of lists of points
:returns: a Node of kind complexFaultGeometry
"""
node = Node('complexFaultGeometry')
node.append(edge_node('faultTopEdge', edges[0]))
for edge in edges[1:-1]:
node.append(edge_node('intermediateEdge', edge))
node.append(edge_node('faultBottomEdge', edges[-1]))
return node
def simple_fault_node(fault_trace, dip, upper_depth, lower_depth):
"""
:param fault_trace: an object with an attribute .points
:param dip: dip parameter
:param upper_depth: upper seismogenic depth
:param lower_depth: lower seismogenic depth
:returns: a Node of kind simpleFaultGeometry
"""
node = Node('simpleFaultGeometry')
line = []
for p in fault_trace.points:
line.append(p.longitude)
line.append(p.latitude)
node.append(Node('gml:LineString', nodes=[Node('gml:posList', {}, line)]))
node.append(Node('dip', {}, dip))
node.append(Node('upperSeismoDepth', {}, upper_depth))
node.append(Node('lowerSeismoDepth', {}, lower_depth))
return node
def convert_fragility_model_04(node, fname, fmcounter=itertools.count(1)):
"""
:param node:
an :class:`openquake.commonib.node.Node` in NRML 0.4
:param fname:
path of the fragility file
:returns:
an :class:`openquake.commonib.node.Node` in NRML 0.5
"""
convert_type = {"lognormal": "logncdf"}
new = Node('fragilityModel',
dict(assetCategory='building',
lossCategory='structural',
id='fm_%d_converted_from_NRML_04' %
next(fmcounter)))
with context(fname, node):
fmt = node['format']
descr = ~node.description
limit_states = ~node.limitStates
new.append(Node('description', {}, descr))
new.append((Node('limitStates', {}, ' '.join(limit_states))))
for ffs in node[2:]:
IML = ffs.IML
# NB: noDamageLimit = None is different than zero
nodamage = ffs.attrib.get('noDamageLimit')
ff = Node('fragilityFunction', {'format': fmt})
ff['id'] = ~ffs.taxonomy
ff['shape'] = convert_type[ffs.attrib.get('type', 'lognormal')]
if fmt == 'continuous':
with context(fname, IML):
attr = dict(imt=IML['IMT'],
minIML=IML['minIML'],
maxIML=IML['maxIML'])
if nodamage is not None:
attr['noDamageLimit'] = nodamage
ff.append(Node('imls', attr))
for ffc in ffs[2:]:
with context(fname, ffc):
ls = ffc['ls']
param = ffc.params
with context(fname, param):
m, s = param['mean'], param['stddev']
ff.append(Node('params', dict(ls=ls, mean=m, stddev=s)))
else: # discrete
with context(fname, IML):
imls = ' '.join(map(str, (~IML)[1]))
attr = dict(imt=IML['IMT'])
if nodamage is not None:
attr['noDamageLimit'] = nodamage
ff.append(Node('imls', attr, imls))
for ffd in ffs[2:]:
ls = ffd['ls']
with context(fname, ffd):
poes = ' '.join(map(str, ~ffd.poEs))
ff.append(Node('poes', dict(ls=ls), poes))
new.append(ff)
return new
def _add_asset(self, idx, asset_node, param):
values = {}
deductibles = {}
insurance_limits = {}
retrofitteds = {}
asset_id = asset_node['id'].encode('utf8')
with context(param['fname'], asset_node):
self.asset_refs.append(asset_id)
taxonomy = asset_node['taxonomy']
if 'damage' in param['calculation_mode']:
# calculators of 'damage' kind require the 'number'
# if it is missing a KeyError is raised
number = asset_node['number']
else:
# some calculators ignore the 'number' attribute;
# if it is missing it is considered 1, since we are going
# to multiply by it
try:
number = asset_node['number']
except KeyError:
number = 1
else:
if 'occupants' in param['all_cost_types']:
values['occupants_None'] = number
location = asset_node.location['lon'], asset_node.location['lat']
if param['region'] and not geometry.Point(*location).within(
param['region']):
param['out_of_region'] += 1
return
tagnode = getattr(asset_node, 'tags', None)
dic = {} if tagnode is None else tagnode.attrib.copy()
with context(param['fname'], tagnode):
dic['taxonomy'] = taxonomy
idxs = self.tagcol.add_tags(dic)
try:
costs = asset_node.costs
except AttributeError:
costs = Node('costs', [])
try:
occupancies = asset_node.occupancies
except AttributeError:
occupancies = Node('occupancies', [])
for cost in costs:
with context(param['fname'], cost):
cost_type = cost['type']
if cost_type in param['relevant_cost_types']:
values[cost_type] = cost['value']
retrovalue = cost.get('retrofitted')
if retrovalue is not None:
retrofitteds[cost_type] = retrovalue
if param['insured_losses']:
deductibles[cost_type] = cost['deductible']
insurance_limits[cost_type] = cost['insuranceLimit']
# check we are not missing a cost type
missing = param['relevant_cost_types'] - set(values)
if missing and missing <= param['ignore_missing_costs']:
logging.warn('Ignoring asset %s, missing cost type(s): %s',
asset_id, ', '.join(missing))
for cost_type in missing:
values[cost_type] = None
elif missing and 'damage' not in param['calculation_mode']:
# missing the costs is okay for damage calculators
with context(param['fname'], asset_node):
raise ValueError("Invalid Exposure. "
"Missing cost %s for asset %s" %
(missing, asset_id))
tot_occupants = 0
for occupancy in occupancies:
with context(param['fname'], occupancy):
occupants = 'occupants_%s' % occupancy['period']
values[occupants] = occupancy['occupants']
tot_occupants += values[occupants]
if occupancies: # store average occupants
values['occupants_None'] = tot_occupants / len(occupancies)
area = float(asset_node.get('area', 1))
ass = asset.Asset(idx, idxs, number, location, values, area,
deductibles, insurance_limits, retrofitteds,
self.cost_calculator)
self.assets.append(ass)
def _get_exposure(fname, stop=None):
"""
:param fname:
path of the XML file containing the exposure
:param stop:
node at which to stop parsing (or None)
:returns:
a pair (Exposure instance, list of asset nodes)
"""
[exposure] = nrml.read(fname, stop=stop)
if not exposure.tag.endswith('exposureModel'):
raise InvalidFile('%s: expected exposureModel, got %s' %
(fname, exposure.tag))
description = exposure.description
try:
conversions = exposure.conversions
except AttributeError:
conversions = Node('conversions', nodes=[Node('costTypes', [])])
try:
area = conversions.area
except AttributeError:
# NB: the area type cannot be an empty string because when sending
# around the CostCalculator object we would run into this numpy bug
# about pickling dictionaries with empty strings:
# https://github.com/numpy/numpy/pull/5475
area = Node('area', dict(type='?'))
try:
occupancy_periods = exposure.occupancyPeriods.text or ''
except AttributeError:
occupancy_periods = ''
try:
tagNames = exposure.tagNames
except AttributeError:
tagNames = Node('tagNames', text='')
tagnames = ['id'] + (~tagNames or [])
if set(tagnames) & {'taxonomy', 'exposure', 'country'}:
raise InvalidFile('taxonomy, exposure and country are reserved names '
'you cannot use it in <tagNames>: %s' % fname)
tagnames.insert(0, 'taxonomy')
# read the cost types and make some check
cost_types = []
retrofitted = False
for ct in conversions.costTypes:
with context(fname, ct):
ctname = ct['name']
if ctname == 'structural' and 'retrofittedType' in ct.attrib:
if ct['retrofittedType'] != ct['type']:
raise ValueError(
'The retrofittedType %s is different from the type'
'%s' % (ct['retrofittedType'], ct['type']))
if ct['retrofittedUnit'] != ct['unit']:
raise ValueError(
'The retrofittedUnit %s is different from the unit'
'%s' % (ct['retrofittedUnit'], ct['unit']))
retrofitted = True
cost_types.append(
(ctname, valid.cost_type_type(ct['type']), ct['unit']))
if 'occupants' in cost_types:
cost_types.append(('occupants', 'per_area', 'people'))
cost_types.sort(key=operator.itemgetter(0))
cost_types = numpy.array(cost_types, cost_type_dt)
cc = CostCalculator({}, {}, {},
{name: i
for i, name in enumerate(tagnames)})
for ct in cost_types:
name = ct['name'] # structural, nonstructural, ...
cc.cost_types[name] = ct['type'] # aggregated, per_asset, per_area
cc.area_types[name] = area['type']
cc.units[name] = ct['unit']
exp = Exposure(exposure['id'], exposure['category'], description.text,
cost_types, occupancy_periods, retrofitted, area.attrib, [],
[], cc, TagCollection(tagnames))
assets_text = exposure.assets.text.strip()
if assets_text:
# the <assets> tag contains a list of file names
dirname = os.path.dirname(fname)
exp.datafiles = [os.path.join(dirname, f) for f in assets_text.split()]
else:
exp.datafiles = []
return exp, exposure.assets
def _pointsources2multipoints(srcs, i):
allsources = []
for key, sources in groupby(srcs, get_key).items():
if len(sources) == 1: # there is a single source
allsources.extend(sources)
continue
msr, rar, usd, lsd, hd, npd = key
mfds = [src[3] for src in sources]
points = []
for src in sources:
points.extend(~src.pointGeometry.Point.pos)
geom = Node('multiPointGeometry')
geom.append(Node('gml:posList', text=points))
geom.append(Node('upperSeismoDepth', text=usd))
geom.append(Node('lowerSeismoDepth', text=lsd))
node = Node(
'multiPointSource',
dict(id='mps-%d' % i, name='multiPointSource-%d' % i),
nodes=[geom])
node.append(Node("magScaleRel", text=msr))
node.append(Node("ruptAspectRatio", text=rar))
node.append(mfds2multimfd(mfds))
node.append(Node('nodalPlaneDist', nodes=[
Node('nodalPlane', dict(probability=prob, rake=rake,
strike=strike, dip=dip))
for prob, rake, strike, dip in npd]))
node.append(Node('hypoDepthDist', nodes=[
Node('hypoDepth', dict(depth=depth, probability=prob))
for prob, depth in hd]))
allsources.append(node)
i += 1
return i, allsources
def _read_csv(self, csvnames, dirname):
"""
:param csvnames: names of csv files, space separated
:param dirname: the directory where the csv files are
:yields: asset nodes
"""
expected_header = self._csv_header()
fnames = [os.path.join(dirname, f) for f in csvnames.split()]
for fname in fnames:
with open(fname) as f:
header = set(next(csv.reader(f)))
if expected_header - header:
raise InvalidFile(
'Unexpected header in %s\nExpected: %s\nGot: %s' %
(fname, expected_header, header))
for fname in fnames:
with open(fname) as f:
for i, dic in enumerate(csv.DictReader(f), 1):
asset = Node('asset', lineno=i)
with context(fname, asset):
asset['id'] = dic['id']
asset['number'] = float(dic['number'])
asset['taxonomy'] = dic['taxonomy']
if 'area' in dic: # optional attribute
asset['area'] = dic['area']
loc = Node(
'location',
dict(lon=valid.longitude(dic['lon']),
lat=valid.latitude(dic['lat'])))
costs = Node('costs')
for cost in self.cost_types['name']:
a = dict(type=cost, value=dic[cost])
costs.append(Node('cost', a))
occupancies = Node('occupancies')
for period in self.occupancy_periods:
a = dict(occupants=dic[period], period=period)
occupancies.append(Node('occupancy', a))
tags = Node('tags')
for tagname in self.tagcol.tagnames:
if tagname != 'taxonomy':
tags[tagname] = dic[tagname]
asset.nodes.extend([loc, costs, occupancies, tags])
if i % 100000 == 0:
logging.info('Read %d assets', i)
yield asset
def simple_fault_node(fault_trace, dip, upper_depth, lower_depth):
"""
:param fault_trace: an object with an attribute .points
:param dip: dip parameter
:param upper_depth: upper seismogenic depth
:param lower_depth: lower seismogenic depth
:returns: a Node of kind simpleFaultGeometry
"""
node = Node('simpleFaultGeometry')
line = []
for p in fault_trace.points:
line.append(p.longitude)
line.append(p.latitude)
node.append(Node('gml:LineString', nodes=[Node('gml:posList', {}, line)]))
node.append(Node('dip', {}, dip))
node.append(Node('upperSeismoDepth', {}, upper_depth))
node.append(Node('lowerSeismoDepth', {}, lower_depth))
return node
def combine_ss_models(filedict,
domains_shp,
lt,
outfile,
nrml_version='04',
weight=1.): #, id_base = 'ASS'):
""" Combine smoothed seismicity models based on tectonic region types
:params filedict:
dict of form filedict[trt] = filename specifying input file for that region
:params domains_shp:
shapefile defining tectonic domain regions
:params lt:
LogicTree object containing relevant values and weights for Mmax
:params outfile:
output nrml formatted file
"""
print 'Getting tectonic region type from %s' % domains_shp
driver = ogr.GetDriverByName("ESRI Shapefile")
data_source = driver.Open(domains_shp, 0)
dsf = data_source.GetLayer()
trt_types = []
for feature in dsf:
trt_types.append(feature.GetField('TRT'))
dsf = shapefile.Reader(domains_shp)
dom_shapes = dsf.shapes()
hypo_depth_dist_nc = PMF([(0.5, 10.0), (0.25, 5.0), (0.25, 15.0)])
hypo_depth_dist_c = PMF([(0.5, 5.0), (0.25, 2.5), (0.25, 10.0)])
hypo_depth_dist_ex = hypo_depth_dist_c
hypo_depth_dict = {
'Cratonic': hypo_depth_dist_c,
'Non_cratonic': hypo_depth_dist_nc,
'Extended': hypo_depth_dist_ex
}
nodal_plane_dist = PMF([(0.3, NodalPlane(0, 30, 90)),
(0.2, NodalPlane(90, 30, 90)),
(0.3, NodalPlane(180, 30, 90)),
(0.2, NodalPlane(270, 30, 90))])
merged_pts = []
# Get mmax values and weights
mmaxs = {}
mmaxs_w = {}
for trt, filename in filedict.iteritems():
if trt == 'Cratonic':
mmax_values, mmax_weights = lt.get_weights('Mmax', 'Proterozoic')
else:
mmax_values, mmax_weights = lt.get_weights('Mmax', trt)
mmax_values = [float(i) for i in mmax_values]
mmax_weights = [float(i) for i in mmax_weights]
print mmax_values
print mmax_weights
mmaxs[trt] = mmax_values
mmaxs_w[trt] = mmax_weights
pt_ids = []
for trt, filename in filedict.iteritems():
print trt
print 'Parsing %s' % filename
# Only keep points within domain
pts = read_pt_source(filename)
# shapes = np.where(trt_types
for zone_trt, dom_shape in zip(trt_types, dom_shapes):
print zone_trt
print dom_shape
if zone_trt == trt:
print 'TRT %s, procesing shape %s' % (zone_trt, dom_shape)
dom_poly = Polygon(dom_shape.points)
for pt in pts:
pt_loc = Point(pt.location.x, pt.location.y)
if pt_loc.within(dom_poly):
pt.tectonic_region_type = zone_trt
pt.nodal_plane_distribution = nodal_plane_dist
pt.hypocenter_distribution = hypo_depth_dict[zone_trt]
pt.rupture_aspect_ratio = 2
mfd = pt.mfd
new_mfd = gr2inc_mmax(mfd, mmaxs[trt], mmaxs_w[trt],
weight)
pt.mfd = new_mfd
if pt.source_id in pt_ids:
print 'Point source %s already exists!' % pt.source_id
print 'Skipping this source for trt %s' % zone_trt
else:
merged_pts.append(pt)
pt_ids.append(pt.source_id)
name = outfile.rstrip('.xml')
if nrml_version == '04':
nodes = list(map(obj_to_node, sorted(merged_pts)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(outfile, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns=NAMESPACE)
def _pointsources2multipoints(srcs, i):
# converts pointSources with the same hpdist, npdist and msr into a
# single multiPointSource.
allsources = []
for (hd, npd, msr), sources in groupby(srcs, dists).items():
if len(sources) == 1: # there is a single source
allsources.extend(sources)
continue
mfds = [src[3] for src in sources]
points = []
usd = []
lsd = []
rar = []
for src in sources:
pg = src.pointGeometry
points.extend(~pg.Point.pos)
usd.append(~pg.upperSeismoDepth)
lsd.append(~pg.lowerSeismoDepth)
rar.append(~src.ruptAspectRatio)
geom = Node('multiPointGeometry')
geom.append(Node('gml:posList', text=points))
geom.append(Node('upperSeismoDepth', text=collapse(usd)))
geom.append(Node('lowerSeismoDepth', text=collapse(lsd)))
node = Node(
'multiPointSource',
dict(id='mps-%d' % i, name='multiPointSource-%d' % i),
nodes=[geom])
node.append(Node("magScaleRel", text=collapse(msr)))
node.append(Node("ruptAspectRatio", text=collapse(rar)))
node.append(mfds2multimfd(mfds))
node.append(Node('nodalPlaneDist', nodes=[
Node('nodalPlane', dict(probability=prob, rake=rake,
strike=strike, dip=dip))
for prob, rake, strike, dip in npd]))
node.append(Node('hypoDepthDist', nodes=[
Node('hypoDepth', dict(depth=depth, probability=prob))
for prob, depth in hd]))
allsources.append(node)
i += 1
return i, allsources
def _pointsources2multipoints(srcs, i):
# converts pointSources with the same hpdist, npdist and msr into a
# single multiPointSource.
allsources = []
for (hd, npd, msr), sources in groupby(srcs, dists).items():
if len(sources) == 1: # there is a single source
allsources.extend(sources)
continue
mfds = [src[3] for src in sources]
points = []
usd = []
lsd = []
rar = []
for src in sources:
pg = src.pointGeometry
points.extend(~pg.Point.pos)
usd.append(~pg.upperSeismoDepth)
lsd.append(~pg.lowerSeismoDepth)
rar.append(~src.ruptAspectRatio)
geom = Node('multiPointGeometry')
geom.append(Node('gml:posList', text=points))
geom.append(Node('upperSeismoDepth', text=collapse(usd)))
geom.append(Node('lowerSeismoDepth', text=collapse(lsd)))
node = Node('multiPointSource',
dict(id='mps-%d' % i, name='multiPointSource-%d' % i),
nodes=[geom])
node.append(Node("magScaleRel", text=collapse(msr)))
node.append(Node("ruptAspectRatio", text=collapse(rar)))
node.append(mfds2multimfd(mfds))
node.append(
Node('nodalPlaneDist',
nodes=[
Node(
'nodalPlane',
dict(probability=prob,
rake=rake,
strike=strike,
dip=dip)) for prob, rake, strike, dip in npd
]))
node.append(
Node('hypoDepthDist',
nodes=[
Node('hypoDepth', dict(depth=depth, probability=prob))
for prob, depth in hd
]))
allsources.append(node)
i += 1
return i, allsources
def _read_csv(self):
"""
:yields: asset nodes
"""
expected_header = self._csv_header()
for fname in self.datafiles:
with open(fname, encoding='utf-8') as f:
fields = next(csv.reader(f))
header = set(fields)
if len(header) < len(fields):
raise InvalidFile(
'%s: The header %s contains a duplicated field' %
(fname, header))
elif expected_header - header - {'exposure', 'country'}:
raise InvalidFile(
'Unexpected header in %s\nExpected: %s\nGot: %s' %
(fname, sorted(expected_header), sorted(header)))
occupancy_periods = self.occupancy_periods.split()
for fname in self.datafiles:
with open(fname, encoding='utf-8') as f:
for i, dic in enumerate(csv.DictReader(f), 1):
asset = Node('asset', lineno=i)
with context(fname, asset):
asset['id'] = dic['id']
asset['number'] = valid.positivefloat(dic['number'])
asset['taxonomy'] = dic['taxonomy']
if 'area' in dic: # optional attribute
asset['area'] = dic['area']
loc = Node('location',
dict(lon=valid.longitude(dic['lon']),
lat=valid.latitude(dic['lat'])))
costs = Node('costs')
for cost in self.cost_types['name']:
a = dict(type=cost, value=dic[cost])
if 'retrofitted' in dic:
a['retrofitted'] = dic['retrofitted']
costs.append(Node('cost', a))
occupancies = Node('occupancies')
for period in occupancy_periods:
a = dict(occupants=float(dic[period]),
period=period)
occupancies.append(Node('occupancy', a))
tags = Node('tags')
for tagname in self.tagcol.tagnames:
if tagname not in (
'taxonomy', 'exposure', 'country'):
tags.attrib[tagname] = dic[tagname]
asset.nodes.extend([loc, costs, occupancies, tags])
yield asset
def weighted_pt_source(pt_sources,
weights,
name,
filename=None,
nrml_version='04'):
"""Scales rates by weights for collapsing logic trees
:param pt_sources:
list of PointSource objects
:param weights:
dict contains weights for each tectonic
region type, e.g. weights[trt] = 0.2
:param filenam:
path to output file, if provided will
be written to nrml format as defined by
nrml_version
:param nrml_version:
version of nrml schema to use
:returns weighted_pt_sources:
list of PointSource objects with activity rates
scaled by weights
"""
weighted_point_sources = []
for pt in pt_sources:
new_pt = copy.deepcopy(
pt) # Copy sources to avoid messing with original data
mfd_type = type(pt.mfd).__name__
trt = pt.tectonic_region_type
weight = weights[trt]
if mfd_type == 'TruncatedGRMFD':
b_val = pt.mfd.b_val
# rescale a value in log sapce
a_val = np.log10(np.power(10, pt.mfd.a_val) * weight)
new_pt.mfd.modify_set_ab(a_val, b_val)
elif mfd_type == 'EvenlyDiscretizedMFD':
mag_bins, rates = zip(*pt.mfd.get_annual_occurrence_rates())
mag_bins = np.array(mag_bins)
rates = np.array(rates)
new_rates = rates * weight
new_pt.mfd.modify_set_mfd(new_pt.mfd.min_mag, new_pt.mfd.bin_width,
list(new_rates))
else:
msg = 'Weighting method for mfd type %s not yet defined' % mfd_type
raise (msg)
weighted_point_sources.append(new_pt)
# Now write out
if filename is not None:
source_model_file = filename
print 'Writing to source model file %s' % source_model_file
if nrml_version == '04':
# source_list = []
# for source in weighted_point_sources:
# source_list.append(source)
nodes = list(map(obj_to_node, sorted(weighted_point_sources)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(source_model_file, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns=NAMESPACE)
elif nrml_version == '05':
msg = 'Method not yet implemented for nrml version 0.5'
raise (msg)
# source_group_list = []
# id = 0
# for trt, sources in weighted_sources.iteritems():
# source_group = SourceGroup(trt, sources = sources, id=id)
# id +=1
# source_group_list.append(source_group)
# write_source_model(nrml_pt_file, source_group_list,
# name = name)
else:
print 'Warning: nrml version not specfied, xml not created'
return weighted_point_sources
def write_source_model(dest,
sources_or_groups,
name=None,
investigation_time=None):
"""
Writes a source model to XML.
:param dest:
Destination path
:param sources_or_groups:
Source model in different formats
:param name:
Name of the source model (if missing, extracted from the filename)
"""
if isinstance(sources_or_groups, nrml.SourceModel):
groups = sources_or_groups.src_groups
attrs = dict(name=sources_or_groups.name,
investigation_time=sources_or_groups.investigation_time)
elif isinstance(sources_or_groups[0], sourceconverter.SourceGroup):
groups = sources_or_groups
attrs = dict(investigation_time=investigation_time)
else: # passed a list of sources
srcs_by_trt = groupby(sources_or_groups,
operator.attrgetter('tectonic_region_type'))
groups = [
sourceconverter.SourceGroup(trt, srcs_by_trt[trt])
for trt in srcs_by_trt
]
attrs = dict(investigation_time=investigation_time)
if name or 'name' not in attrs:
attrs['name'] = name or os.path.splitext(os.path.basename(dest))[0]
if attrs['investigation_time'] is None:
del attrs['investigation_time']
nodes = list(map(obj_to_node, groups))
ddict = extract_ddict(groups)
if ddict:
# remove duplicate content from nodes
for grp_node in nodes:
for src_node in grp_node:
if src_node["id"] in ddict:
src_node.nodes = []
# save HDF5 file
dest5 = os.path.splitext(dest)[0] + '.hdf5'
with hdf5.File(dest5, 'w') as h:
for src_id, dic in ddict.items():
for k, v in dic.items():
key = '%s/%s' % (src_id, k)
if isinstance(v, numpy.ndarray):
h.create_dataset(key,
v.shape,
v.dtype,
compression='gzip',
compression_opts=9)
h[key][:] = v
else:
h[key] = v
source_model = Node("sourceModel", attrs, nodes=nodes)
with open(dest, 'wb') as f:
nrml.write([source_model], f, '%s')
if ddict:
return [dest, dest5]
else:
return [dest]
def build_source_model(csm):
nodes = [obj_to_node(sg) for sg in csm.src_groups]
return Node('compositeSourceModel', {}, nodes=nodes)
identifier = 'ASS' + str(j)
name = 'Helmstetter' + str(j)
point = Point(data[j,0],data[j,1],
10)
rate = data[j,2]
# Convert rate to a value
aval = np.log10(rate) + bvalue*config["mmin"]
mfd = TruncatedGRMFD(min_mag, max_mag, 0.1, aval, bvalue)
hypo_depth_dist = PMF([(0.5, 10.0),
(0.25, 5.0),
(0.25, 15.0)])
nodal_plane_dist = PMF([(0.3, NodalPlane(0, 30, 90)),
(0.2, NodalPlane(90, 30, 90)),
(0.3, NodalPlane(180, 30, 90)),
(0.2, NodalPlane(270, 30, 90))])
point_source = PointSource(identifier, name, 'Non_cratonic',
mfd, 2, msr,
2.0, tom, 0.1, 20.0, point,
nodal_plane_dist, hypo_depth_dist)
source_list.append(point_source)
filename = "Australia_Adaptive_K%i_b%.3f_mmin%.1f.xml" % (smoother.config['k'], smoother.config['bvalue'], smoother.config['mmin'])
mod_name = "Australia_Adaptive_K%i_b%.3f" % (smoother.config['k'], smoother.config['bvalue'])
nodes = list(map(obj_to_node, sorted(source_list)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(filename, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns = NAMESPACE)
def build_site_model(site):
return Node('site', get_site_attributes(site))
def area2pt_source(area_source_file,
sources=None,
investigation_time=50,
rupture_mesh_spacing=10.,
width_of_mfd_bin=0.1,
area_source_discretisation=10.,
filename=None,
nrml_version='04',
name=None):
"""Calls OpenQuake parsers to read area source model
from source_mode.xml type file, convert to point sources
and write to a new nrml source model file.
:params area_source_file:
nrml format file of the area source
:params discretisation:
Grid size (km) for the area source discretisation,
which defines the distance between resulting point
sources.
"""
if sources is None:
sources = nrml2sourcelist(
area_source_file,
investigation_time=investigation_time,
rupture_mesh_spacing=rupture_mesh_spacing,
width_of_mfd_bin=width_of_mfd_bin,
area_source_discretisation=area_source_discretisation)
if name is None:
name = '%s_points' % filename
new_pt_sources = {}
for source in sources:
pt_sources = area_to_point_sources(source)
for pt in pt_sources:
pt.source_id = pt.source_id.replace(':', '')
pt.name = pt.name.replace(':', '_')
try:
new_pt_sources[pt.tectonic_region_type].append(pt)
except KeyError:
new_pt_sources[pt.tectonic_region_type] = [pt]
# print [method for method in dir(pt) if callable(getattr(pt, method))]
# print [attribute for attribute in dir(pt)]
nrml_pt_file = area_source_file[:-4] + '_pts.xml'
source_group_list = []
id = 0
for trt, sources in new_pt_sources.iteritems():
source_group = SourceGroup(trt, sources=sources, id=id)
id += 1
source_group_list.append(source_group)
if filename is not None:
if nrml_version == '04':
source_list = []
for trt, sources in new_pt_sources.iteritems():
for source in sources:
source_list.append(source)
nodes = list(map(obj_to_node, sorted(source_list)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(nrml_pt_file, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns=NAMESPACE)
# This will write version 0.5
elif nrml_version == '05':
write_source_model(nrml_pt_file, source_group_list, name=filename)
else:
print 'Warning: nrml version not specfied, xml not created'
return source_group_list
def _pointsources2multipoints(srcs, i):
allsources = []
for key, sources in groupby(srcs, get_key).items():
if len(sources) == 1: # there is a single source
allsources.extend(sources)
continue
msr, rar, usd, lsd, hd, npd = key
mfds = [src[3] for src in sources]
points = []
for src in sources:
points.extend(~src.pointGeometry.Point.pos)
geom = Node('multiPointGeometry')
geom.append(Node('gml:posList', text=points))
geom.append(Node('upperSeismoDepth', text=usd))
geom.append(Node('lowerSeismoDepth', text=lsd))
node = Node(
'multiPointSource',
dict(id='mps-%d' % i, name='multiPointSource-%d' % i),
nodes=[geom])
node.append(Node("magScaleRel", text=msr))
node.append(Node("ruptAspectRatio", text=rar))
node.append(mfds2multimfd(mfds))
node.append(Node('nodalPlaneDist', nodes=[
Node('nodalPlane', dict(probability=prob, rake=rake,
strike=strike, dip=dip))
for prob, rake, strike, dip in npd]))
node.append(Node('hypoDepthDist', nodes=[
Node('hypoDepth', dict(depth=depth, probability=prob))
for prob, depth in hd]))
allsources.append(node)
i += 1
return i, allsources
def get_exposure(oqparam):
"""
Read the full exposure in memory and build a list of
:class:`openquake.risklib.riskmodels.Asset` instances.
:param oqparam:
an :class:`openquake.commonlib.oqvalidation.OqParam` instance
:returns:
an :class:`Exposure` instance
"""
out_of_region = 0
if oqparam.region_constraint:
region = wkt.loads(oqparam.region_constraint)
else:
region = None
all_cost_types = set(oqparam.all_cost_types)
fname = oqparam.inputs['exposure']
exposure, assets_node = _get_exposure(fname, all_cost_types)
relevant_cost_types = all_cost_types - set(['occupants'])
asset_refs = set()
ignore_missing_costs = set(oqparam.ignore_missing_costs)
for idx, asset in enumerate(assets_node):
values = {}
deductibles = {}
insurance_limits = {}
retrofitteds = {}
with context(fname, asset):
asset_id = asset['id'].encode('utf8')
if asset_id in asset_refs:
raise read_nrml.DuplicatedID(asset_id)
asset_refs.add(asset_id)
exposure.asset_refs.append(asset_id)
taxonomy = asset['taxonomy']
if 'damage' in oqparam.calculation_mode:
# calculators of 'damage' kind require the 'number'
# if it is missing a KeyError is raised
number = asset.attrib['number']
else:
# some calculators ignore the 'number' attribute;
# if it is missing it is considered 1, since we are going
# to multiply by it
try:
number = asset['number']
except KeyError:
number = 1
else:
if 'occupants' in all_cost_types:
values['occupants_None'] = number
location = asset.location['lon'], asset.location['lat']
if region and not geometry.Point(*location).within(region):
out_of_region += 1
continue
try:
costs = asset.costs
except AttributeError:
costs = Node('costs', [])
try:
occupancies = asset.occupancies
except AttributeError:
occupancies = Node('occupancies', [])
for cost in costs:
with context(fname, cost):
cost_type = cost['type']
if cost_type in relevant_cost_types:
values[cost_type] = cost['value']
retrovalue = cost.attrib.get('retrofitted')
if retrovalue is not None:
retrofitteds[cost_type] = retrovalue
if oqparam.insured_losses:
deductibles[cost_type] = cost['deductible']
insurance_limits[cost_type] = cost['insuranceLimit']
# check we are not missing a cost type
missing = relevant_cost_types - set(values)
if missing and missing <= ignore_missing_costs:
logging.warn('Ignoring asset %s, missing cost type(s): %s',
asset_id, ', '.join(missing))
for cost_type in missing:
values[cost_type] = None
elif missing and 'damage' not in oqparam.calculation_mode:
# missing the costs is okay for damage calculators
with context(fname, asset):
raise ValueError("Invalid Exposure. "
"Missing cost %s for asset %s" %
(missing, asset_id))
tot_occupants = 0
for occupancy in occupancies:
with context(fname, occupancy):
exposure.time_events.add(occupancy['period'])
occupants = 'occupants_%s' % occupancy['period']
values[occupants] = occupancy['occupants']
tot_occupants += values[occupants]
if occupancies: # store average occupants
values['occupants_None'] = tot_occupants / len(occupancies)
area = float(asset.attrib.get('area', 1))
ass = riskmodels.Asset(idx, taxonomy, number, location, values, area,
deductibles, insurance_limits, retrofitteds,
exposure.cost_calculator)
exposure.assets.append(ass)
exposure.taxonomies.add(taxonomy)
if region:
logging.info(
'Read %d assets within the region_constraint '
'and discarded %d assets outside the region', len(exposure.assets),
out_of_region)
if len(exposure.assets) == 0:
raise RuntimeError('Could not find any asset within the region!')
# sanity check
values = any(len(ass.values) + ass.number for ass in exposure.assets)
assert values, 'Could not find any value??'
return exposure
def combine_ss_models(filename_stem, domains_shp, params,lt, bval_key, output_dir='./',
nrml_version = '04', weight=1.):#, id_base = 'ASS'):
""" Combine smoothed seismicity models based on tectonic region types
:params filename_stem:
String for the start of the xml filename for the source model,
assuming generic components (non generic are inferred,
e.g. bvalue and completeness model)
:params domains_shp:
shapefile defining tectonic domain regions
:params params:
list of dicts containing parameters derivded from the shapefile
:bval_key
key for the dicts in params as we are merging by
bvalues (best, lower, upper)
:params lt:
LogicTree object containing relevant values and weights for Mmax
:params outfile:
output nrml formatted file
"""
dsf = shapefile.Reader(domains_shp)
dom_shapes = dsf.shapes()
# Get indicies of relevant fields
for i, f in enumerate(dsf.fields):
if f[0]=='CODE':
code_index = i-1
if f[0]=='TRT':
trt_index = i-1
hypo_depth_dist_nc = PMF([(0.5, 10.0),
(0.25, 5.0),
(0.25, 15.0)])
hypo_depth_dist_c = PMF([(0.5, 5.0),
(0.25, 2.5),
(0.25, 10.0)])
hypo_depth_dist_ex = hypo_depth_dist_c
hypo_depth_dict = {'Cratonic': hypo_depth_dist_c,
'Non_cratonic': hypo_depth_dist_nc,
'Extended': hypo_depth_dist_ex}
# FIXME! - Temporary solution until nodal plan logic tree
# info can be read directly from shapefile attributes
nodal_plane_dist = PMF([(0.3, NodalPlane(0, 30, 90)),
(0.2, NodalPlane(90, 30, 90)),
(0.3, NodalPlane(180, 30, 90)),
(0.2, NodalPlane(270, 30, 90))])
merged_pts = []
pt_ids = []
# Get mmax values and weights
mmaxs = {}
mmaxs_w = {}
for dom in params:
print 'Processing source %s' % dom['CODE']
print dom['TRT']
if dom['TRT'] == 'NCratonic' or dom['TRT'] == 'Extended':
dom['TRT'] = 'Non_cratonic'
# For the moment, only consider regions within AUstralia
if dom['TRT'] == 'Active' or dom['TRT'] == 'Interface' or \
dom['TRT'] == 'Oceanic' or \
dom['TRT'] == 'Intraslab' or dom['CODE'] == 'NECS' or \
dom['CODE'] == 'NWO':
print 'Source %s not on continental Australia, skipping' % dom['CODE']
continue
elif dom['TRT'] == 'Cratonic':
if dom['DOMAIN'] == 1:
mmax_values, mmax_weights = lt.get_weights('Mmax', 'Archean')
else:
mmax_values, mmax_weights = lt.get_weights('Mmax', 'Proterozoic')
# elif dom['TRT'] == 'Active':
# print 'MMax logic tree not yet defined for active crust, using extended crust'
# mmax_values, mmax_weights = lt.get_weights('Mmax', 'Extended')
else:
mmax_values, mmax_weights = lt.get_weights('Mmax', dom['TRT'])
mmax_values = [float(i) for i in mmax_values]
mmax_weights = [float(i) for i in mmax_weights]
print mmax_values
print mmax_weights
mmaxs[dom['CODE']] = mmax_values
mmaxs_w[dom['CODE']] = mmax_weights
#pt_ids = []
#for trt, filename in filedict.iteritems():
# print trt
completeness_table = np.array([dom['COMPLETENESS'][0]])
completeness_string = 'comp'
for ym in completeness_table:
completeness_string += '_%i_%.1f' % (ym[0], ym[1])
mmin = dom['COMPLETENESS'][0][1]
filename = "%s_b%.3f_mmin_%.1f_0.1%s.xml" % (
filename_stem, dom[bval_key], mmin,
completeness_string)
print 'Parsing %s' % filename
# Only keep points within domain
pts = read_pt_source(filename)
#shapes = np.where(trt_types
for shape in dsf.shapeRecords():
print shape.record[code_index]
if shape.record[code_index] == dom['CODE']:
# Check for undefined depths (-999 values)
if dom['DEP_BEST'] < 0:
print 'Setting best depth to 10 km'
dom['DEP_BEST']=10
if dom['DEP_UPPER'] < 0:
print 'Setting upper depth to 5 km'
dom['DEP_UPPER']=5
if dom['DEP_LOWER'] < 0:
print 'Setting lower depth to 15 km'
dom['DEP_LOWER']=15
hypo_depth_dist = PMF([(0.5, dom['DEP_BEST']),
(0.25, dom['DEP_LOWER']),
(0.25, dom['DEP_UPPER'])])
# Define nodal planes as thrusts except for special cases
str1 = dom['SHMAX'] + 90.
str2 = dom['SHMAX'] + 270.
str3 = dom['SHMAX'] + dom['SHMAX_SIG'] + 90.
str4 = dom['SHMAX']+ dom['SHMAX_SIG'] + 270.
str5 = dom['SHMAX'] - dom['SHMAX_SIG'] + 90.
str6 = dom['SHMAX'] - dom['SHMAX_SIG'] + 270.
strikes = [str1,str2,str3,str4,str5,str6]
for i,strike in enumerate(strikes):
if strike >=360:
strikes[i]=strike-360
nodal_plane_dist = PMF([(0.34, NodalPlane(strikes[0], 30, 90)),
(0.34, NodalPlane(strikes[1], 30, 90)),
(0.08, NodalPlane(strikes[2], 30, 90)),
(0.08, NodalPlane(strikes[3], 30, 90)),
(0.08, NodalPlane(strikes[4], 30, 90)),
(0.08, NodalPlane(strikes[5], 30, 90))])
if dom['CODE'] == 'WARM' or dom['CODE'] == 'WAPM':
print 'Define special case for WARM'
nodal_plane_dist = PMF([(0.75, NodalPlane(45, 90, 0)),
(0.125, NodalPlane(strikes[0], 30, 90)),
(0.125, NodalPlane(strikes[1], 30, 90))])
if dom['CODE'] == 'FMLR':
print 'Define special case for FMLR, 0.5 thrust, 0.5 SS'
nodal_plane_dist = PMF([(0.17, NodalPlane(strikes[0], 30, 90)),
(0.17, NodalPlane(strikes[1], 30, 90)),
(0.04, NodalPlane(strikes[2], 30, 90)),
(0.04, NodalPlane(strikes[3], 30, 90)),
(0.04, NodalPlane(strikes[4], 30, 90)),
(0.04, NodalPlane(strikes[5], 30, 90)),
(0.17, NodalPlane(strikes[0], 90, 0)),
(0.17, NodalPlane(strikes[1], 90, 0)),
(0.04, NodalPlane(strikes[2], 90, 0)),
(0.04, NodalPlane(strikes[3], 90, 0)),
(0.04, NodalPlane(strikes[4], 90, 0)),
(0.04, NodalPlane(strikes[5], 90, 0))])
dom_poly = Polygon(shape.shape.points)
for pt in pts:
pt_loc = Point(pt.location.x, pt.location.y)
if pt_loc.within(dom_poly):
# pt.tectonic_region_type = dom['TRT']
pt.tectonic_region_type = dom['GMM_TRT']
pt.nodal_plane_distribution = nodal_plane_dist # FIXME! update based on data extracted from shapefile
pt.hypocenter_distribution = hypo_depth_dist
pt.rupture_aspect_ratio=2
mfd = pt.mfd
new_mfd = gr2inc_mmax(mfd, mmaxs[dom['CODE']], mmaxs_w[dom['CODE']], weight)
pt.mfd = new_mfd
if pt.source_id in pt_ids:
print 'Point source %s already exists!' % pt.source_id
print 'Skipping this source for trt %s' % dom['TRT']
else:
merged_pts.append(pt)
pt_ids.append(pt.source_id)
outfile = "%s_%s.xml" % (
filename_stem, bval_key)
outfile = os.path.join(output_dir, outfile)
name = outfile.rstrip('.xml')
if nrml_version == '04':
nodes = list(map(obj_to_node, sorted(merged_pts)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(outfile, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns = NAMESPACE)
return outfile
def _get_exposure(fname, ok_cost_types, stop=None):
"""
:param fname:
path of the XML file containing the exposure
:param ok_cost_types:
a set of cost types (as strings)
:param stop:
node at which to stop parsing (or None)
:returns:
a pair (Exposure instance, list of asset nodes)
"""
[exposure] = nrml.read(fname, stop=stop)
if not exposure.tag.endswith('exposureModel'):
raise InvalidFile('%s: expected exposureModel, got %s' %
(fname, exposure.tag))
description = exposure.description
try:
conversions = exposure.conversions
except AttributeError:
conversions = Node('conversions', nodes=[Node('costTypes', [])])
try:
inslimit = conversions.insuranceLimit
except AttributeError:
inslimit = Node('insuranceLimit', text=True)
try:
deductible = conversions.deductible
except AttributeError:
deductible = Node('deductible', text=True)
try:
area = conversions.area
except AttributeError:
# NB: the area type cannot be an empty string because when sending
# around the CostCalculator object we would run into this numpy bug
# about pickling dictionaries with empty strings:
# https://github.com/numpy/numpy/pull/5475
area = Node('area', dict(type='?'))
try:
occupancy_periods = ~exposure.occupancyPeriods or ''
except AttributeError:
occupancy_periods = 'day night transit'
try:
tagNames = exposure.tagNames
except AttributeError:
tagNames = Node('tagNames', text='')
tagnames = ~tagNames or []
tagnames.insert(0, 'taxonomy')
# read the cost types and make some check
cost_types = []
for ct in conversions.costTypes:
if not ok_cost_types or ct['name'] in ok_cost_types:
with context(fname, ct):
cost_types.append(
(ct['name'], valid.cost_type_type(ct['type']), ct['unit']))
if 'occupants' in ok_cost_types:
cost_types.append(('occupants', 'per_area', 'people'))
cost_types.sort(key=operator.itemgetter(0))
cost_types = numpy.array(cost_types, cost_type_dt)
insurance_limit_is_absolute = inslimit.get('isAbsolute', True)
deductible_is_absolute = deductible.get('isAbsolute', True)
tagi = {name: i for i, name in enumerate(tagnames)}
cc = asset.CostCalculator({}, {}, {}, deductible_is_absolute,
insurance_limit_is_absolute, tagi)
for ct in cost_types:
name = ct['name'] # structural, nonstructural, ...
cc.cost_types[name] = ct['type'] # aggregated, per_asset, per_area
cc.area_types[name] = area['type']
cc.units[name] = ct['unit']
assets = []
asset_refs = []
exp = Exposure(exposure['id'],
exposure['category'], ~description, cost_types,
occupancy_periods.split(), insurance_limit_is_absolute,
deductible_is_absolute, area.attrib, assets, asset_refs, cc,
asset.TagCollection(tagnames))
return exp, exposure.assets
def write_combined_faults_points(point_sources, fault_sources,
filename, name, area_sources = None,
nrml_version='04'):
"""Write pts, area and fault sources to file
:param point_sources:
list without trt or dict with trt key of point sources
"""
print 'Writing to source model file %s' % filename
ps_id_index = 1
fs_id_index = 1
if nrml_version == '04':
if type(point_sources) == dict:
source_list = []
for trt, sources in point_sources.iteritems():
for source in sources:
source.source_id = 'PS_%i' % ps_id_index
source_list.append(source)
ps_id_index += 1
# id_index = max(id_index, source.source_id)
elif type(point_sources) == list:
source_list = copy.deepcopy(point_sources)
for source in source_list:
source.source_id = 'PS_%i' % ps_id_index
# source_list.append(source)
ps_id_index += 1
# id_index = max(id_index, source.source_id)
for fault_source in fault_sources:
# id_index += 1
fault_source.source_id = "FS_%i" % fs_id_index
fs_id_index += 1
source_list.append(fault_source)
if area_sources is not None:
for area_source in area_sources:
source_list.append(area_source)
nodes = list(map(obj_to_node, sorted(source_list)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(filename, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns = NAMESPACE)
elif nrml_version == '05':
if type(point_sources) == dict:
source_group_list = []
id = 0
for trt, sources in point_sources.iteritems():
for source in sources:
id_index = max(id_index, source.source_id)
for trt, sources in point_sources.iteritems():
for fault_source in fault_sources:
if fault_source.tectonic_region_type == trt:
id_index += 1
fault_source.source_id = "%i" % id_index
sources.append(fault_source)
if area_sources is not None:
for area_source in area_sources:
if area_source.tectonic_region_type == trt:
sources.append(area_source)
source_group = SourceGroup(trt, sources = sources, id=id)
id +=1
source_group_list.append(source_group)
write_source_model(filename, source_group_list,
name = name)
elif type(point_sources) == list:
msg = 'Method not yet implemented for nrml version 0.5'
raise(msg)
else:
print 'Warning: nrml version not specfied, xml not created'
def convert_fragility_model_04(node, fname, fmcounter=itertools.count(1)):
"""
:param node:
an :class:`openquake.commonib.node.Node` in NRML 0.4
:param fname:
path of the fragility file
:returns:
an :class:`openquake.commonib.node.Node` in NRML 0.5
"""
convert_type = {"lognormal": "logncdf"}
new = Node('fragilityModel',
dict(assetCategory='building',
lossCategory='structural',
id='fm_%d_converted_from_NRML_04' %
next(fmcounter)))
with context(fname, node):
fmt = node['format']
descr = ~node.description
limit_states = ~node.limitStates
new.append(Node('description', {}, descr))
new.append((Node('limitStates', {}, ' '.join(limit_states))))
for ffs in node[2:]:
IML = ffs.IML
# NB: noDamageLimit = None is different than zero
nodamage = ffs.attrib.get('noDamageLimit')
ff = Node('fragilityFunction', {'format': fmt})
ff['id'] = ~ffs.taxonomy
ff['shape'] = convert_type[ffs.attrib.get('type', 'lognormal')]
if fmt == 'continuous':
with context(fname, IML):
attr = dict(imt=IML['IMT'],
minIML=IML['minIML'],
maxIML=IML['maxIML'])
if nodamage is not None:
attr['noDamageLimit'] = nodamage
ff.append(Node('imls', attr))
for ffc in ffs[2:]:
with context(fname, ffc):
ls = ffc['ls']
param = ffc.params
with context(fname, param):
m, s = param['mean'], param['stddev']
ff.append(Node('params', dict(ls=ls, mean=m, stddev=s)))
else: # discrete
with context(fname, IML):
imls = ' '.join(map(str, (~IML)[1]))
attr = dict(imt=IML['IMT'])
if nodamage is not None:
attr['noDamageLimit'] = nodamage
ff.append(Node('imls', attr, imls))
for ffd in ffs[2:]:
ls = ffd['ls']
with context(fname, ffd):
poes = ' '.join(map(str, ~ffd.poEs))
ff.append(Node('poes', dict(ls=ls), poes))
new.append(ff)
return new
def _get_exposure(fname, stop=None):
"""
:param fname:
path of the XML file containing the exposure
:param stop:
node at which to stop parsing (or None)
:returns:
a pair (Exposure instance, list of asset nodes)
"""
[exposure] = nrml.read(fname, stop=stop)
if not exposure.tag.endswith('exposureModel'):
raise InvalidFile('%s: expected exposureModel, got %s' %
(fname, exposure.tag))
description = exposure.description
try:
conversions = exposure.conversions
except AttributeError:
conversions = Node('conversions', nodes=[Node('costTypes', [])])
try:
inslimit = conversions.insuranceLimit
except AttributeError:
inslimit = Node('insuranceLimit', text=True)
try:
deductible = conversions.deductible
except AttributeError:
deductible = Node('deductible', text=True)
try:
area = conversions.area
except AttributeError:
# NB: the area type cannot be an empty string because when sending
# around the CostCalculator object we would run into this numpy bug
# about pickling dictionaries with empty strings:
# https://github.com/numpy/numpy/pull/5475
area = Node('area', dict(type='?'))
try:
occupancy_periods = exposure.occupancyPeriods.text or ''
except AttributeError:
occupancy_periods = ''
try:
tagNames = exposure.tagNames
except AttributeError:
tagNames = Node('tagNames', text='')
tagnames = ~tagNames or []
if set(tagnames) & {'taxonomy', 'exposure', 'country'}:
raise InvalidFile('taxonomy, exposure and country are reserved names '
'you cannot use it in <tagNames>: %s' % fname)
tagnames.insert(0, 'taxonomy')
# read the cost types and make some check
cost_types = []
retrofitted = False
for ct in conversions.costTypes:
with context(fname, ct):
ctname = ct['name']
if ctname == 'structural' and 'retrofittedType' in ct.attrib:
if ct['retrofittedType'] != ct['type']:
raise ValueError(
'The retrofittedType %s is different from the type'
'%s' % (ct['retrofittedType'], ct['type']))
if ct['retrofittedUnit'] != ct['unit']:
raise ValueError(
'The retrofittedUnit %s is different from the unit'
'%s' % (ct['retrofittedUnit'], ct['unit']))
retrofitted = True
cost_types.append(
(ctname, valid.cost_type_type(ct['type']), ct['unit']))
if 'occupants' in cost_types:
cost_types.append(('occupants', 'per_area', 'people'))
cost_types.sort(key=operator.itemgetter(0))
cost_types = numpy.array(cost_types, cost_type_dt)
insurance_limit_is_absolute = il = inslimit.get('isAbsolute')
deductible_is_absolute = de = deductible.get('isAbsolute')
cc = CostCalculator(
{}, {}, {},
True if de is None else de,
True if il is None else il,
{name: i for i, name in enumerate(tagnames)},
)
for ct in cost_types:
name = ct['name'] # structural, nonstructural, ...
cc.cost_types[name] = ct['type'] # aggregated, per_asset, per_area
cc.area_types[name] = area['type']
cc.units[name] = ct['unit']
assets = []
asset_refs = []
exp = Exposure(
exposure['id'], exposure['category'],
description.text, cost_types, occupancy_periods,
insurance_limit_is_absolute, deductible_is_absolute, retrofitted,
area.attrib, assets, asset_refs, cc, TagCollection(tagnames))
assets_text = exposure.assets.text.strip()
if assets_text:
# the <assets> tag contains a list of file names
dirname = os.path.dirname(fname)
exp.datafiles = [os.path.join(dirname, f) for f in assets_text.split()]
else:
exp.datafiles = []
return exp, exposure.assets
def pt2fault_distance(pt_sources, fault_sources, min_distance = 5.0,
filename = 'source_model.xml',
buffer_distance = 100., nrml_version = '04',
name=None):
"""Calculate distances from a pt source rupture plane
to the fault sources to then reduce Mmax on events that are
within a certain distance
:param pt_sources:
list of PointSource objects
:param fault_sources:
List of FaultSource objects
:param min_distance:
Minimum distance (km) within which we want a point source
rupture to be from a fault.
:param filename:
Name of output nrml file for revised pt source model
:param buffer_distance:
Km, initial filter to only process pts within this
distance from the fault
"""
if name is None:
name = filename[:-4] + '_geom_filtered'
id_index = 0 # We need to re-number all sources to avoid duplicate ids
# Extract the points of the fault source mesh
fault_lons = []
fault_lats = []
fault_depths = []
for fault in fault_sources:
whole_fault_surface = SimpleFaultSurface.from_fault_data(
fault.fault_trace, fault.upper_seismogenic_depth,
fault.lower_seismogenic_depth, fault.dip,
fault.rupture_mesh_spacing)
fault_lons.append(whole_fault_surface.mesh.lons.flatten())
fault_lats.append(whole_fault_surface.mesh.lats.flatten())
fault_depths.append(whole_fault_surface.mesh.depths.flatten())
fault_lons = np.concatenate(fault_lons)
fault_lats = np.concatenate(fault_lats)
fault_depths = np.concatenate(fault_depths)
min_fault_lon = np.min(fault_lons)
max_fault_lon = np.max(fault_lons)
min_fault_lat = np.min(fault_lats)
max_fault_lat = np.max(fault_lats)
# Generate ruptures for point sources
minimum_distance_list = []
revised_point_sources = {'Cratonic': [], 'Non_cratonic': [],
'Extended': [], 'Subduction': []}
for pt in pt_sources:
print 'Looping over point sources'
# For speeding things up filter based on initial distances
# to find points very far from or very close to a fault
mfd_type = type(pt.mfd).__name__
pt_depths = []
for probs, depths in pt.hypocenter_distribution.data:
pt_depths.append(depths)
np_probs = []
np_list = []
for prob, nodal_plane in pt.nodal_plane_distribution.data:
np_probs.append(prob)
np_list.append(nodal_plane)
centroid_distances = []
for pt_depth in pt_depths:
centroid_distances.append(distance(pt.location.longitude, pt.location.latitude,
pt_depth, fault_lons, fault_lats, fault_depths))
centroid_distances = np.array(centroid_distances).flatten()
# print 'Minimum distance', min(centroid_distances)
# print 'Maximum distance', max(centroid_distances)
if (min(centroid_distances)) > buffer_distance:
# Keep point as it, not within buffer distance of any faults
revised_point_sources[pt.tectonic_region_type].append(pt)
continue
if (min(centroid_distances)) < min_distance:
# Discard point sources as too close to a fault
print 'Discarding point source, too close to a fault'
continue
rupture_mags = []
rupture_lons = []
rupture_lats = []
rupture_depths = []
rupture_strikes = []
rupture_dips = []
ruptures = pt.iter_ruptures()
for rupture in ruptures:
rupture_mags.append(rupture.mag)
rupture_lons.append(rupture.surface.corner_lons)
rupture_lats.append(rupture.surface.corner_lats)
rupture_depths.append(rupture.surface.corner_depths)
rupture_strikes.append(rupture.surface.strike)
rupture_dips.append(rupture.surface.dip)
rupture_mags = np.array(rupture_mags).flatten()
# make the same length as the corners
rupture_mags = np.repeat(rupture_mags, 4)
rupture_strikes = np.repeat(rupture_strikes, 4)
rupture_dips = np.repeat(rupture_dips, 4)
rupture_lons = np.array(rupture_lons).flatten()
rupture_lats = np.array(rupture_lats).flatten()
rupture_depths = np.array(rupture_depths).flatten()
print 'Doing meshgrid'
lons1,lons2 = np.meshgrid(fault_lons, rupture_lons)
lats1,lats2 = np.meshgrid(fault_lats, rupture_lats)
depths1, depths2 = np.meshgrid(fault_depths, rupture_depths)
# Calculate distance from pt to all fault
print 'Distance calculations'
distances = distance(lons1, lats1, depths1, lons2, lats2, depths2)
closest_distance_to_faults = np.min(distances)
print 'Shortest pt to fault distance is', closest_distance_to_faults
minimum_distance_list.append(closest_distance_to_faults)
# Find where the distance is less than the threshold min_distance
too_close_lons = lons2[np.where(distances < min_distance)]
too_close_lats = lats2[np.where(distances < min_distance)]
if too_close_lons.size > 0:
lon_indices = np.where(np.in1d(rupture_lons, too_close_lons))[0]
lat_indices = np.where(np.in1d(rupture_lats, too_close_lats))[0]
too_close_mags = rupture_mags[np.intersect1d(
lon_indices, lat_indices)]
too_close_strikes = rupture_strikes[np.intersect1d(
lon_indices, lat_indices)]
too_close_dips = rupture_dips[np.intersect1d(
lon_indices, lat_indices)]
# print 'Magnitudes of rupture close to fault', too_close_mags
# print 'Strikes of rupture close to fault', too_close_strikes
# print 'Dips of rupture close to fault', too_close_dips
unique_strikes = np.unique(rupture_strikes)
unique_dips = np.unique(rupture_dips)
src_name_index = 0
for prob, nodal_plane in pt.nodal_plane_distribution.data:
id_index += 1
src_name_index += 1
# We are now splitting the source into many with different
# combinations of Mmaxs and nodal planes
new_pt = copy.deepcopy(pt)
new_pt.source_id ="%i" % id_index
new_pt.name = new_pt.name + ("_%i" % src_name_index)
new_np = NodalPlane(nodal_plane.strike, nodal_plane.dip, nodal_plane.rake)
new_np_distribution = PMF([(1.0, new_np)]) # weight of nodal plane is 1 as making
# a separate source
# Calculate new rates based on probability of original nodal plane
new_pt.nodal_plane_distribution = new_np_distribution
if mfd_type == 'TruncatedGRMFD':
b_val = pt.mfd.b_val
# rescale a value in log sapce
a_val = np.log10(np.power(10, pt.mfd.a_val)*prob)#*area_src_weight))
new_pt.mfd.modify_set_ab(a_val, b_val)
elif mfd_type == 'EvenlyDiscretizedMFD':
mag_bins, rates = zip(*pt.mfd.get_annual_occurrence_rates())
mag_bins = np.array(mag_bins)
rates = np.array(rates)
new_rates = rates*prob#*area_src_weight)
new_pt.mfd.modify_set_mfd(new_pt.mfd.min_mag, new_pt.mfd.bin_width,
list(new_rates))
else:
msg = 'Weighting method for mfd type %s not yet defined' % mfd_type
raise(msg)
pair_index = np.where(np.logical_and(too_close_strikes == nodal_plane.strike,
too_close_dips == nodal_plane.dip))
# Deal with intersecting cases
if len(pair_index[0]) > 0:
intersecting_magnitudes = too_close_mags[pair_index]
minimum_magnitude_intersecting_fault = min(intersecting_magnitudes)
if minimum_magnitude_intersecting_fault >= \
(pt.mfd.min_mag + pt.mfd.bin_width):
new_mmax = minimum_magnitude_intersecting_fault - \
pt.mfd.bin_width
if mfd_type == 'TruncatedGRMFD':
new_pt.mfd.max_mag = new_mmax
if mfd_type == 'EvenlyDiscretizedMFD':
trimmed_rates = new_rates[np.where(mag_bins <= new_mmax)]
else:
print 'Minimum magnitude intersects fault, discarding source'
continue
else:
pass
# Append revised source for given nodal plane distribution to
# list of revised sources
print 'Appending revised source'
revised_point_sources[pt.tectonic_region_type].append(new_pt)
else:
id_index += 1
pt.source_id = "%i" % id_index
'Appending original source'
revised_point_sources[pt.tectonic_region_type].append(pt)
if len(minimum_distance_list) > 0:
print 'Overall minimum distance (km):', min(minimum_distance_list)
# Write pts to source model on their own
source_model_file = filename
print 'Writing to source model file %s' % source_model_file
if nrml_version == '04':
source_list = []
for trt, sources in revised_point_sources.iteritems():
for source in sources:
source_list.append(source)
nodes = list(map(obj_to_node, sorted(source_list)))
source_model = Node("sourceModel", {"name": name}, nodes=nodes)
with open(source_model_file, 'wb') as f:
nrml.write([source_model], f, '%s', xmlns = NAMESPACE)
elif nrml_version == '05':
source_group_list = []
id = 0
for trt, sources in revised_point_sources.iteritems():
source_group = SourceGroup(trt, sources = sources, id=id)
id +=1
source_group_list.append(source_group)
write_source_model(source_model_file, source_group_list,
name = name)
else:
print 'Warning: nrml version not specfied, xml not created'
# Write pts to source model with faults
source_model_file = filename[:-4] +'_inc_faults.xml'
name = name +'_inc_faults'
write_combined_faults_points(revised_point_sources, fault_sources,
source_model_file, name, nrml_version='04')
def build_trunc_gr_from_shp(record):
attribs = {"aValue": float(record["a_val"]),
"bValue": float(record["b_val"]),
"minMag": float(record["min_mag"]),
"maxMag": float(record["max_mag"])}
return Node("truncGutenbergRichterMFD", attribs)
def write_source_model(dest, sources_or_groups, name=None,
investigation_time=None):
"""
Writes a source model to XML.
:param dest:
Destination path
:param sources_or_groups:
Source model in different formats
:param name:
Name of the source model (if missing, extracted from the filename)
:returns:
the list of generated filenames
"""
if isinstance(sources_or_groups, nrml.SourceModel):
groups = sources_or_groups.src_groups
attrs = dict(name=sources_or_groups.name,
investigation_time=sources_or_groups.investigation_time)
elif isinstance(sources_or_groups[0], sourceconverter.SourceGroup):
groups = sources_or_groups
attrs = dict(investigation_time=investigation_time)
else: # passed a list of sources
srcs_by_trt = groupby(
sources_or_groups, operator.attrgetter('tectonic_region_type'))
groups = [sourceconverter.SourceGroup(trt, srcs_by_trt[trt])
for trt in srcs_by_trt]
attrs = dict(investigation_time=investigation_time)
if name or 'name' not in attrs:
attrs['name'] = name or os.path.splitext(os.path.basename(dest))[0]
if attrs['investigation_time'] is None:
del attrs['investigation_time']
nodes = list(map(obj_to_node, groups))
ddict = extract_ddict(groups)
out = [dest]
if ddict:
# remove duplicate content from nodes
for grp_node in nodes:
for src_node in grp_node:
if src_node["id"] in ddict:
src_node.nodes = []
# save HDF5 file
dest5 = os.path.splitext(dest)[0] + '.hdf5'
with hdf5.File(dest5, 'w') as h:
for src_id, dic in ddict.items():
for k, v in dic.items():
key = '%s/%s' % (src_id, k)
if isinstance(v, numpy.ndarray):
h.create_dataset(key, v.shape, v.dtype,
compression='gzip',
compression_opts=9)
h[key][:] = v
else:
h[key] = v
out.append(dest5)
# produce a geometryModel if there are MultiFaultSources
sections = {}
for group in groups:
for src in group:
if hasattr(src, 'sections'):
sections.update(src.sections)
sections = {sid: sections[sid] for sid in sections}
smodel = Node("sourceModel", attrs, nodes=nodes)
with open(dest, 'wb') as f:
nrml.write([smodel], f, '%s')
if sections:
secnodes = [obj_to_node(sec) for sec in sections.values()]
gmodel = Node("geometryModel", attrs, nodes=secnodes)
with open(dest[:-4] + '_sections.xml', 'wb') as f:
nrml.write([gmodel], f, '%s')
out.append(f.name)
return out
