Source code for openquake.calculators.base

# -*- coding: utf-8 -*-
# vim: tabstop=4 shiftwidth=4 softtabstop=4
#
# Copyright (C) 2014-2021 GEM Foundation
#
# OpenQuake is free software: you can redistribute it and/or modify it
# under the terms of the GNU Affero General Public License as published
# by the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# OpenQuake is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with OpenQuake. If not, see <http://www.gnu.org/licenses/>.
import os
import sys
import abc
import pdb
import logging
import operator
import traceback
from datetime import datetime
from shapely import wkt
import h5py
import numpy
import pandas

from openquake.baselib import (
    general, hdf5, __version__ as engine_version)
from openquake.baselib import parallel, python3compat
from openquake.baselib.performance import Monitor
from openquake.hazardlib import InvalidFile, site, stats
from openquake.hazardlib.site_amplification import Amplifier
from openquake.hazardlib.site_amplification import AmplFunction
from openquake.hazardlib.calc.filters import SourceFilter, getdefault
from openquake.hazardlib.source import rupture
from openquake.hazardlib.shakemap.maps import get_sitecol_shakemap
from openquake.hazardlib.shakemap.gmfs import to_gmfs
from openquake.risklib import riskinput, riskmodels
from openquake.commonlib import readinput, logictree, datastore
from openquake.calculators.export import export as exp
from openquake.calculators import getters

get_taxonomy = operator.attrgetter('taxonomy')
get_weight = operator.attrgetter('weight')
get_imt = operator.attrgetter('imt')

calculators = general.CallableDict(operator.attrgetter('calculation_mode'))
U8 = numpy.uint8
U16 = numpy.uint16
U32 = numpy.uint32
F32 = numpy.float32
TWO16 = 2 ** 16
TWO32 = 2 ** 32

stats_dt = numpy.dtype([('mean', F32), ('std', F32),
                        ('min', F32), ('max', F32),
                        ('len', U16)])


# this is used for the minimum_intensity dictionaries
[docs]def consistent(dic1, dic2): """ Check if two dictionaries with default are consistent: >>> consistent({'PGA': 0.05, 'SA(0.3)': 0.05}, {'default': 0.05}) True >>> consistent({'SA(0.3)': 0.1, 'SA(0.6)': 0.05}, ... {'default': 0.1, 'SA(0.3)': 0.1, 'SA(0.6)': 0.05}) True """ if dic1 == dic2: return True v1 = set(dic1.values()) v2 = set(dic2.values()) missing = set(dic2) - set(dic1) - {'default'} if len(v1) == 1 and len(v2) == 1 and v1 == v2: # {'PGA': 0.05, 'SA(0.3)': 0.05} is consistent with {'default': 0.05} return True return not missing
[docs]def get_stats(seq): std = numpy.nan if len(seq) == 1 else numpy.std(seq, ddof=1) tup = (numpy.mean(seq), std, numpy.min(seq), numpy.max(seq), len(seq)) return numpy.array(tup, stats_dt)
[docs]class InvalidCalculationID(Exception): """ Raised when running a post-calculation on top of an incompatible pre-calculation """
[docs]def fix_ones(pmap): """ Physically, an extremely small intensity measure level can have an extremely large probability of exceedence, however that probability cannot be exactly 1 unless the level is exactly 0. Numerically, the PoE can be 1 and this give issues when calculating the damage (there is a log(0) in :class:`openquake.risklib.scientific.annual_frequency_of_exceedence`). Here we solve the issue by replacing the unphysical probabilities 1 with .9999999999999999 (the float64 closest to 1). """ for sid in pmap: array = pmap[sid].array array[array == 1.] = .9999999999999999 return pmap
[docs]def build_weights(realizations): """ :returns: an array with the realization weights of shape R """ arr = numpy.array([rlz.weight['default'] for rlz in realizations]) return arr
[docs]def set_array(longarray, shortarray): """ :param longarray: a numpy array of floats of length L >= l :param shortarray: a numpy array of floats of length l Fill `longarray` with the values of `shortarray`, starting from the left. If `shortarry` is shorter than `longarray`, then the remaining elements on the right are filled with `numpy.nan` values. """ longarray[:len(shortarray)] = shortarray longarray[len(shortarray):] = numpy.nan
[docs]class BaseCalculator(metaclass=abc.ABCMeta): """ Abstract base class for all calculators. :param oqparam: OqParam object :param monitor: monitor object :param calc_id: numeric calculation ID """ precalc = None accept_precalc = [] from_engine = False # set by engine.run_calc is_stochastic = False # True for scenario and event based calculators def __init__(self, oqparam, calc_id): self.datastore = datastore.new(calc_id, oqparam) self._monitor = Monitor( '%s.run' % self.__class__.__name__, measuremem=True, h5=self.datastore) # NB: using h5=self.datastore.hdf5 would mean losing the performance # info about Calculator.run since the file will be closed later on self.oqparam = oqparam
[docs] def pre_checks(self): """ Checks to run after the pre_execute but before the execute """
[docs] def monitor(self, operation='', **kw): """ :returns: a new Monitor instance """ mon = self._monitor(operation, h5=self.datastore.hdf5) self._monitor.calc_id = mon.calc_id = self.datastore.calc_id vars(mon).update(kw) return mon
[docs] def save_params(self, **kw): """ Update the current calculation parameters and save engine_version """ if ('hazard_calculation_id' in kw and kw['hazard_calculation_id'] is None): del kw['hazard_calculation_id'] vars(self.oqparam).update(**kw) self.datastore['oqparam'] = self.oqparam attrs = self.datastore['/'].attrs attrs['engine_version'] = engine_version attrs['date'] = datetime.now().isoformat()[:19] if 'checksum32' not in attrs: attrs['checksum32'] = readinput.get_checksum32( self.oqparam, self.datastore.hdf5) logging.info('Checksum of the input files: %(checksum32)s', attrs) self.datastore.flush()
[docs] def check_precalc(self, precalc_mode): """ Defensive programming against users providing an incorrect pre-calculation ID (with ``--hazard-calculation-id``). :param precalc_mode: calculation_mode of the previous calculation """ calc_mode = self.oqparam.calculation_mode ok_mode = self.accept_precalc if calc_mode != precalc_mode and precalc_mode not in ok_mode: raise InvalidCalculationID( 'In order to run a calculation of kind %r, ' 'you need to provide a calculation of kind %r, ' 'but you provided a %r instead' % (calc_mode, ok_mode, precalc_mode))
[docs] def run(self, pre_execute=True, concurrent_tasks=None, remove=True, shutdown=False, **kw): """ Run the calculation and return the exported outputs. :param pre_execute: set it to False to avoid running pre_execute :param concurrent_tasks: set it to 0 to disable parallelization :param remove: set it to False to remove the hdf5cache file (if any) :param shutdown: set it to True to shutdown the ProcessPool """ with self._monitor: self._monitor.username = kw.get('username', '') if concurrent_tasks is None: # use the job.ini parameter ct = self.oqparam.concurrent_tasks else: # used the parameter passed in the command-line ct = concurrent_tasks if ct == 0: # disable distribution temporarily oq_distribute = os.environ.get('OQ_DISTRIBUTE') os.environ['OQ_DISTRIBUTE'] = 'no' if ct != self.oqparam.concurrent_tasks: # save the used concurrent_tasks self.oqparam.concurrent_tasks = ct self.save_params(**kw) try: if pre_execute: self.pre_execute() self.result = self.execute() if self.result is not None: self.post_execute(self.result) self.export(kw.get('exports', '')) except Exception: if kw.get('pdb'): # post-mortem debug tb = sys.exc_info()[2] traceback.print_tb(tb) pdb.post_mortem(tb) else: logging.critical('', exc_info=True) raise finally: if shutdown: parallel.Starmap.shutdown() # cleanup globals if ct == 0: # restore OQ_DISTRIBUTE if oq_distribute is None: # was not set del os.environ['OQ_DISTRIBUTE'] else: os.environ['OQ_DISTRIBUTE'] = oq_distribute readinput.pmap = None readinput.exposure = None readinput.gmfs = None readinput.eids = None readinput.smlt_cache.clear() readinput.gsim_lt_cache.clear() # remove temporary hdf5 file, if any if os.path.exists(self.datastore.tempname) and remove: os.remove(self.datastore.tempname) return getattr(self, 'exported', {})
[docs] def core_task(*args): """ Core routine running on the workers. """ raise NotImplementedError
[docs] @abc.abstractmethod def pre_execute(self): """ Initialization phase. """
[docs] @abc.abstractmethod def execute(self): """ Execution phase. Usually will run in parallel the core function and return a dictionary with the results. """
[docs] @abc.abstractmethod def post_execute(self, result): """ Post-processing phase of the aggregated output. It must be overridden with the export code. It will return a dictionary of output files. """
[docs] def gzip_inputs(self): """ Gzipping the inputs and saving them in the datastore """ logging.info('gzipping the input files') fnames = readinput.get_input_files(self.oqparam) self.datastore.store_files(fnames)
[docs] def export(self, exports=None): """ Export all the outputs in the datastore in the given export formats. Individual outputs are not exported if there are multiple realizations. """ self.exported = getattr(self, 'exported', {}) if isinstance(exports, tuple): fmts = exports elif exports: # is a string fmts = exports.split(',') elif isinstance(self.oqparam.exports, tuple): fmts = self.oqparam.exports else: # is a string fmts = self.oqparam.exports.split(',') keys = set(self.datastore) | {'fullreport'} has_hcurves = ('hcurves-stats' in self.datastore or 'hcurves-rlzs' in self.datastore) if has_hcurves: keys.add('hcurves') for fmt in fmts: if not fmt: continue for key in sorted(keys): # top level keys if 'rlzs' in key and self.R > 1: if (key[:-4] + 'stats') in self.datastore: continue # skip individual curves self._export((key, fmt)) if has_hcurves and self.oqparam.hazard_maps: self._export(('hmaps', fmt)) if has_hcurves and self.oqparam.uniform_hazard_spectra: self._export(('uhs', fmt))
def _export(self, ekey): if ekey not in exp or self.exported.get(ekey): # already exported return with self.monitor('export'): try: self.exported[ekey] = fnames = exp(ekey, self.datastore) except Exception as exc: fnames = [] logging.error('Could not export %s: %s', ekey, exc) if fnames: logging.info('exported %s: %s', ekey[0], fnames) def __repr__(self): return '<%s#%d>' % (self.__class__.__name__, self.datastore.calc_id)
[docs]def check_time_event(oqparam, occupancy_periods): """ Check the `time_event` parameter in the datastore, by comparing with the periods found in the exposure. """ time_event = oqparam.time_event if time_event and time_event not in occupancy_periods: raise ValueError( 'time_event is %s in %s, but the exposure contains %s' % (time_event, oqparam.inputs['job_ini'], ', '.join(occupancy_periods)))
[docs]def check_amplification(ampl_df, sitecol): """ Make sure the amplification codes in the site collection match the ones in the amplification table. :param ampl_df: the amplification table as a pandas DataFrame :param sitecol: the site collection """ codeset = set(ampl_df.index) if len(codeset) == 1: # there is a single amplification function, there is no need to # extend the sitecol with an ampcode field return codes = set(sitecol.ampcode) missing = codes - codeset if missing: raise ValueError('The site collection contains references to missing ' 'amplification functions: %s' % b' '.join(missing). decode('utf8'))
[docs]class HazardCalculator(BaseCalculator): """ Base class for hazard calculators based on source models """
[docs] def src_filter(self): """ :returns: a SourceFilter """ oq = self.oqparam if getattr(self, 'sitecol', None): sitecol = self.sitecol.complete else: # can happen to the ruptures-only calculator sitecol = None return SourceFilter(sitecol, oq.maximum_distance)
@property def E(self): """ :returns: the number of stored events """ try: return len(self.datastore['events']) except KeyError: return 0 @property def N(self): """ :returns: the total number of sites """ if hasattr(self, 'sitecol'): return len(self.sitecol.complete) if self.sitecol else None return len(self.datastore['sitecol']) @property def few_sites(self): """ :returns: True if there are less than max_sites_disagg """ return len(self.sitecol.complete) <= self.oqparam.max_sites_disagg
[docs] def check_overflow(self): """Overridden in event based"""
[docs] def check_floating_spinning(self): f, s = self.csm.get_floating_spinning_factors() if f != 1: logging.info('Rupture floating factor = %s', f) if s != 1: logging.info('Rupture spinning factor = %s', s) if (f * s >= 1.5 and self.oqparam.pointsource_distance is None and 'classical' in self.oqparam.calculation_mode): logging.info( 'You are not using the pointsource_distance approximation:\n' 'https://docs.openquake.org/oq-engine/advanced/common-mistakes.html#pointsource-distance')
[docs] def read_inputs(self): """ Read risk data and sources if any """ oq = self.oqparam self._read_risk_data() self.check_overflow() # check if self.sitecol is too large if ('amplification' in oq.inputs and oq.amplification_method == 'kernel'): logging.info('Reading %s', oq.inputs['amplification']) df = AmplFunction.read_df(oq.inputs['amplification']) check_amplification(df, self.sitecol) self.af = AmplFunction.from_dframe(df) if (oq.calculation_mode == 'disaggregation' and oq.max_sites_disagg < len(self.sitecol)): raise ValueError( 'Please set max_sites_disagg=%d in %s' % ( len(self.sitecol), oq.inputs['job_ini'])) if ('source_model_logic_tree' in oq.inputs and oq.hazard_calculation_id is None): with self.monitor('composite source model', measuremem=True): self.csm = csm = readinput.get_composite_source_model( oq, self.datastore.hdf5) mags_by_trt = csm.get_mags_by_trt() oq.maximum_distance.interp(mags_by_trt) for trt in mags_by_trt: self.datastore['source_mags/' + trt] = numpy.array( mags_by_trt[trt]) it = list(oq.maximum_distance.ddic[trt].items()) if len(it) > 2: md = '%s->%d, ... %s->%d, %s->%d' % ( it[0] + it[-2] + it[-1]) logging.info('max_dist %s: %s', trt, md) self.full_lt = csm.full_lt self.init() # do this at the end of pre-execute self.pre_checks() if (not oq.hazard_calculation_id and oq.calculation_mode != 'preclassical' and not oq.save_disk_space): self.gzip_inputs() # check DEFINED_FOR_REFERENCE_VELOCITY if self.amplifier: gsim_lt = readinput.get_gsim_lt(oq) self.amplifier.check(self.sitecol.vs30, oq.vs30_tolerance, gsim_lt.values)
[docs] def import_perils(self): """Defined in MultiRiskCalculator"""
[docs] def pre_execute(self): """ Check if there is a previous calculation ID. If yes, read the inputs by retrieving the previous calculation; if not, read the inputs directly. """ oq = self.oqparam if 'gmfs' in oq.inputs or 'multi_peril' in oq.inputs: # read hazard from files assert not oq.hazard_calculation_id, ( 'You cannot use --hc together with gmfs_file') with self.monitor('importing inputs', measuremem=True): self.read_inputs() if 'gmfs' in oq.inputs: self.datastore['full_lt'] = logictree.FullLogicTree.fake() if oq.inputs['gmfs'].endswith('.csv'): eids = import_gmfs_csv(self.datastore, oq, self.sitecol.complete.sids) elif oq.inputs['gmfs'].endswith('.hdf5'): eids = import_gmfs_hdf5(self.datastore, oq) else: raise NotImplementedError( 'Importer for %s' % oq.inputs['gmfs']) E = len(eids) if hasattr(oq, 'number_of_ground_motion_fields'): if oq.number_of_ground_motion_fields != E: raise RuntimeError( 'Expected %d ground motion fields, found %d' % (oq.number_of_ground_motion_fields, E)) else: # set the number of GMFs from the file oq.number_of_ground_motion_fields = E else: self.import_perils() self.save_crmodel() elif 'hazard_curves' in oq.inputs: # read hazard from file assert not oq.hazard_calculation_id, ( 'You cannot use --hc together with hazard_curves') haz_sitecol = readinput.get_site_collection(oq) self.load_crmodel() # must be after get_site_collection self.read_exposure(haz_sitecol) # define .assets_by_site poes = fix_ones(readinput.pmap).array(len(haz_sitecol)) self.datastore['_poes'] = poes.transpose(2, 0, 1) # shape GNL self.datastore['assetcol'] = self.assetcol self.datastore['full_lt'] = fake = logictree.FullLogicTree.fake() self.datastore['rlzs_by_g'] = sum( fake.get_rlzs_by_grp().values(), []) with hdf5.File(self.datastore.tempname, 'a') as t: t['oqparam'] = oq self.realizations = fake.get_realizations() self.save_crmodel() self.datastore.swmr_on() elif oq.hazard_calculation_id: parent = datastore.read(oq.hazard_calculation_id) self.check_precalc(parent['oqparam'].calculation_mode) self.datastore.parent = parent # copy missing parameters from the parent if 'concurrent_tasks' not in vars(self.oqparam): self.oqparam.concurrent_tasks = ( self.oqparam.__class__.concurrent_tasks.default) params = {name: value for name, value in vars(parent['oqparam']).items() if name not in vars(self.oqparam)} self.save_params(**params) with self.monitor('importing inputs', measuremem=True): self.read_inputs() oqp = parent['oqparam'] if oqp.investigation_time != oq.investigation_time: raise ValueError( 'The parent calculation was using investigation_time=%s' ' != %s' % (oqp.investigation_time, oq.investigation_time)) hstats, rstats = list(oqp.hazard_stats()), list(oq.hazard_stats()) if hstats != rstats: raise ValueError( 'The parent calculation had stats %s != %s' % (hstats, rstats)) missing_imts = set(oq.risk_imtls) - set(oqp.imtls) if missing_imts: raise ValueError( 'The parent calculation is missing the IMT(s) %s' % ', '.join(missing_imts)) self.save_crmodel() elif self.__class__.precalc: calc = calculators[self.__class__.precalc]( self.oqparam, self.datastore.calc_id) calc.from_engine = self.from_engine calc.pre_checks = lambda: self.__class__.pre_checks(calc) calc.run(remove=False) calc.datastore.close() for name in ('csm param sitecol assetcol crmodel realizations ' 'policy_name policy_dict full_lt exported').split(): if hasattr(calc, name): setattr(self, name, getattr(calc, name)) else: with self.monitor('importing inputs', measuremem=True): self.read_inputs() self.save_crmodel()
[docs] def init(self): """ To be overridden to initialize the datasets needed by the calculation """ oq = self.oqparam if not oq.risk_imtls: if self.datastore.parent: oq.risk_imtls = ( self.datastore.parent['oqparam'].risk_imtls) if 'full_lt' in self.datastore: full_lt = self.datastore['full_lt'] self.realizations = full_lt.get_realizations() if oq.hazard_calculation_id and 'gsim_logic_tree' in oq.inputs: # redefine the realizations by reading the weights from the # gsim_logic_tree_file that could be different from the parent full_lt.gsim_lt = logictree.GsimLogicTree( oq.inputs['gsim_logic_tree'], set(full_lt.trts)) elif hasattr(self, 'csm'): self.check_floating_spinning() self.realizations = self.csm.full_lt.get_realizations() else: # build a fake; used by risk-from-file calculators self.datastore['full_lt'] = fake = logictree.FullLogicTree.fake() self.realizations = fake.get_realizations()
@general.cached_property def R(self): """ :returns: the number of realizations """ if self.oqparam.collect_rlzs: return 1 elif 'weights' in self.datastore: return len(self.datastore['weights']) try: return self.csm.full_lt.get_num_rlzs() except AttributeError: # no self.csm return self.datastore['full_lt'].get_num_rlzs()
[docs] def read_exposure(self, haz_sitecol): # after load_risk_model """ Read the exposure, the risk models and update the attributes .sitecol, .assetcol """ oq = self.oqparam with self.monitor('reading exposure'): self.sitecol, self.assetcol, discarded = ( readinput.get_sitecol_assetcol( oq, haz_sitecol, self.crmodel.loss_types)) self.datastore['sitecol'] = self.sitecol if len(discarded): self.datastore['discarded'] = discarded if 'scenario' in oq.calculation_mode: # this is normal for the case of scenario from rupture logging.info('%d assets were discarded because too far ' 'from the rupture; use `oq show discarded` ' 'to show them and `oq plot_assets` to plot ' 'them' % len(discarded)) elif not oq.discard_assets: # raise an error self.datastore['assetcol'] = self.assetcol raise RuntimeError( '%d assets were discarded; use `oq show discarded` to' ' show them and `oq plot_assets` to plot them' % len(discarded)) if oq.inputs.get('insurance'): k, v = zip(*oq.inputs['insurance'].items()) self.load_insurance_data(k, v) return readinput.exposure
[docs] def load_insurance_data(self, ins_types, ins_files): """ Read the insurance files and populate the policy_dict """ for loss_type, fname in zip(ins_types, ins_files): array = hdf5.read_csv( fname, {'insurance_limit': float, 'deductible': float, None: object}).array policy_name = array.dtype.names[0] policy_idx = getattr(self.assetcol.tagcol, policy_name + '_idx') insurance = numpy.zeros((len(policy_idx), 2)) for pol, ded, lim in array[ [policy_name, 'deductible', 'insurance_limit']]: insurance[policy_idx[pol]] = ded, lim self.policy_dict[loss_type] = insurance if self.policy_name and policy_name != self.policy_name: raise ValueError( 'The file %s contains %s as policy field, but we were ' 'expecting %s' % (fname, policy_name, self.policy_name)) else: self.policy_name = policy_name
[docs] def load_crmodel(self): # to be called before read_exposure # NB: this is called even if there is no risk model """ Read the risk models and set the attribute .crmodel. The crmodel can be empty for hazard calculations. Save the loss ratios (if any) in the datastore. """ oq = self.oqparam logging.info('Reading the risk model if present') self.crmodel = readinput.get_crmodel(oq) if not self.crmodel: parent = self.datastore.parent if 'crm' in parent: self.crmodel = riskmodels.CompositeRiskModel.read(parent, oq) return if oq.ground_motion_fields and not oq.imtls: raise InvalidFile('No intensity_measure_types specified in %s' % self.oqparam.inputs['job_ini']) self.save_params() # re-save oqparam
[docs] def save_crmodel(self): """ Save the risk models in the datastore """ if len(self.crmodel): logging.info('Storing risk model') attrs = self.crmodel.get_attrs() self.datastore.create_df('crm', self.crmodel.to_dframe(), 'gzip', **attrs)
def _read_risk_data(self): # read the risk model (if any), the exposure (if any) and then the # site collection, possibly extracted from the exposure. oq = self.oqparam self.load_crmodel() # must be called first if (not oq.imtls and 'shakemap' not in oq.inputs and oq.ground_motion_fields): raise InvalidFile('There are no intensity measure types in %s' % oq.inputs['job_ini']) if oq.hazard_calculation_id: with datastore.read(oq.hazard_calculation_id) as dstore: haz_sitecol = dstore['sitecol'].complete if ('amplification' in oq.inputs and 'ampcode' not in haz_sitecol.array.dtype.names): haz_sitecol.add_col('ampcode', site.ampcode_dt) else: if 'gmfs' in oq.inputs and oq.inputs['gmfs'].endswith('.hdf5'): with hdf5.File(oq.inputs['gmfs']) as f: haz_sitecol = f['sitecol'] else: haz_sitecol = readinput.get_site_collection(oq, self.datastore) if hasattr(self, 'rup'): # for scenario we reduce the site collection to the sites # within the maximum distance from the rupture haz_sitecol, _dctx = self.cmaker.filter( haz_sitecol, self.rup) haz_sitecol.make_complete() if 'site_model' in oq.inputs: self.datastore['site_model'] = readinput.get_site_model(oq) oq_hazard = (self.datastore.parent['oqparam'] if self.datastore.parent else None) self.policy_name = '' self.policy_dict = {} if 'exposure' in oq.inputs: exposure = self.read_exposure(haz_sitecol) self.datastore['assetcol'] = self.assetcol self.datastore['cost_calculator'] = exposure.cost_calculator if hasattr(readinput.exposure, 'exposures'): self.datastore['assetcol/exposures'] = ( numpy.array(exposure.exposures, hdf5.vstr)) elif 'assetcol' in self.datastore.parent: assetcol = self.datastore.parent['assetcol'] if oq.region: region = wkt.loads(oq.region) self.sitecol = haz_sitecol.within(region) if oq.shakemap_id or 'shakemap' in oq.inputs or oq.shakemap_uri: self.sitecol, self.assetcol = read_shakemap( self, haz_sitecol, assetcol) self.datastore['sitecol'] = self.sitecol self.datastore['assetcol'] = self.assetcol elif hasattr(self, 'sitecol') and general.not_equal( self.sitecol.sids, haz_sitecol.sids): self.assetcol = assetcol.reduce(self.sitecol) self.datastore['assetcol'] = self.assetcol logging.info('Extracted %d/%d assets', len(self.assetcol), len(assetcol)) else: self.assetcol = assetcol else: # no exposure self.sitecol = haz_sitecol if self.sitecol and oq.imtls: logging.info('Read N=%d hazard sites and L=%d hazard levels', len(self.sitecol), oq.imtls.size) if oq_hazard: parent = self.datastore.parent if 'assetcol' in parent: check_time_event(oq, parent['assetcol'].occupancy_periods) elif oq.job_type == 'risk' and 'exposure' not in oq.inputs: raise ValueError('Missing exposure both in hazard and risk!') if oq_hazard.time_event and oq_hazard.time_event != oq.time_event: raise ValueError( 'The risk configuration file has time_event=%s but the ' 'hazard was computed with time_event=%s' % ( oq.time_event, oq_hazard.time_event)) if oq.job_type == 'risk': tmap = readinput.taxonomy_mapping( self.oqparam, self.assetcol.tagcol.taxonomy) self.crmodel.tmap = tmap taxonomies = set() for ln in oq.loss_names: for items in self.crmodel.tmap[ln]: for taxo, weight in items: if taxo != '?': taxonomies.add(taxo) # check that we are covering all the taxonomies in the exposure missing = taxonomies - set(self.crmodel.taxonomies) if self.crmodel and missing: raise RuntimeError('The exposure contains the taxonomies %s ' 'which are not in the risk model' % missing) if len(self.crmodel.taxonomies) > len(taxonomies): logging.info('Reducing risk model from %d to %d taxonomies', len(self.crmodel.taxonomies), len(taxonomies)) self.crmodel = self.crmodel.reduce(taxonomies) self.crmodel.tmap = tmap if hasattr(self, 'sitecol') and self.sitecol: if 'site_model' in oq.inputs: assoc_dist = (oq.region_grid_spacing * 1.414 if oq.region_grid_spacing else 5) # Graeme's 5km sm = readinput.get_site_model(oq) self.sitecol.complete.assoc(sm, assoc_dist) self.datastore['sitecol'] = self.sitecol # store amplification functions if any self.af = None if 'amplification' in oq.inputs: logging.info('Reading %s', oq.inputs['amplification']) df = AmplFunction.read_df(oq.inputs['amplification']) check_amplification(df, self.sitecol) if oq.amplification_method == 'kernel': # TODO: need to add additional checks on the main calculation # methodology since the kernel method is currently tested only # for classical PSHA self.af = AmplFunction.from_dframe(df) self.amplifier = None else: self.amplifier = Amplifier(oq.imtls, df, oq.soil_intensities) else: self.amplifier = None # manage secondary perils sec_perils = oq.get_sec_perils() for sp in sec_perils: sp.prepare(self.sitecol) # add columns as needed mal = {lt: getdefault(oq.minimum_asset_loss, lt) for lt in oq.loss_names} if mal: logging.info('minimum_asset_loss=%s', mal) self.param = dict(individual_curves=oq.individual_curves, ps_grid_spacing=oq.ps_grid_spacing, minimum_distance=oq.minimum_distance, min_iml=oq.min_iml, collapse_level=int(oq.collapse_level), split_sources=oq.split_sources, avg_losses=oq.avg_losses, amplifier=self.amplifier, sec_perils=sec_perils, ses_seed=oq.ses_seed) # compute exposure stats if hasattr(self, 'assetcol'): save_agg_values( self.datastore, self.assetcol, oq.loss_names, oq.aggregate_by)
[docs] def store_rlz_info(self, rel_ruptures): """ Save info about the composite source model inside the full_lt dataset :param rel_ruptures: dictionary TRT -> number of relevant ruptures """ oq = self.oqparam if hasattr(self, 'full_lt'): # no scenario self.realizations = self.full_lt.get_realizations() if not self.realizations: raise RuntimeError('Empty logic tree: too much filtering?') self.datastore['full_lt'] = self.full_lt else: # scenario self.full_lt = self.datastore['full_lt'] R = self.R logging.info('There are %d realization(s)', R) self.datastore['weights'] = arr = build_weights(self.realizations) self.datastore.set_attrs('weights', nbytes=arr.nbytes) if ('event_based' in oq.calculation_mode and R >= TWO16 or R >= TWO32): raise ValueError( 'The logic tree has too many realizations (%d), use sampling ' 'instead' % R) elif R > 10000: logging.warning( 'The logic tree has %d realizations(!), please consider ' 'sampling it', R) if rel_ruptures: self.check_discardable(rel_ruptures)
[docs] def check_discardable(self, rel_ruptures): """ Check if logic tree reduction is possible """ n = len(self.full_lt.sm_rlzs) keep_trts = set() nrups = [] for grp_id, trt_smrs in enumerate(self.datastore['trt_smrs']): trti, smrs = numpy.divmod(trt_smrs, n) trt = self.full_lt.trts[trti[0]] nr = rel_ruptures.get(grp_id, 0) nrups.append(nr) if nr: keep_trts.add(trt) self.datastore['est_rups_by_grp'] = U32(nrups) discard_trts = set(self.full_lt.trts) - keep_trts if discard_trts: msg = ('No sources for some TRTs: you should set\n' 'discard_trts = %s\nin %s') % ( ', '.join(discard_trts), self.oqparam.inputs['job_ini']) logging.warning(msg)
[docs] def store_source_info(self, calc_times, nsites=False): """ Save (eff_ruptures, num_sites, calc_time) inside the source_info """ self.csm.update_source_info(calc_times, nsites) recs = [tuple(row) for row in self.csm.source_info.values()] hdf5.extend(self.datastore['source_info'], numpy.array(recs, readinput.source_info_dt)) return [rec[0] for rec in recs] # return source_ids
[docs] def post_process(self): """For compatibility with the engine"""
[docs]class RiskCalculator(HazardCalculator): """ Base class for all risk calculators. A risk calculator must set the attributes .crmodel, .sitecol, .assetcol, .riskinputs in the pre_execute phase. """
[docs] def build_riskinputs(self, kind): """ :param kind: kind of hazard getter, can be 'poe' or 'gmf' :returns: a list of RiskInputs objects, sorted by IMT. """ logging.info('Building risk inputs from %d realization(s)', self.R) imtset = set(self.oqparam.imtls) | set(self.oqparam.get_sec_imts()) if not set(self.oqparam.risk_imtls) & imtset: rsk = ', '.join(self.oqparam.risk_imtls) haz = ', '.join(imtset) raise ValueError('The IMTs in the risk models (%s) are disjoint ' "from the IMTs in the hazard (%s)" % (rsk, haz)) if not hasattr(self.crmodel, 'tmap'): self.crmodel.tmap = readinput.taxonomy_mapping( self.oqparam, self.assetcol.tagcol.taxonomy) with self.monitor('building riskinputs'): if self.oqparam.hazard_calculation_id: dstore = self.datastore.parent else: dstore = self.datastore riskinputs = getattr(self, '_gen_riskinputs_' + kind)(dstore) assert riskinputs if all(isinstance(ri.hazard_getter, getters.ZeroGetter) for ri in riskinputs): raise RuntimeError(f'the {kind}s are all zeros on the assets') logging.info('Built %d risk inputs', len(riskinputs)) self.acc = None return riskinputs
def _gen_riskinputs_gmf(self, dstore): out = [] if 'gmf_data' not in dstore: # needed for case_shakemap dstore.close() dstore = self.datastore if 'gmf_data' not in dstore: raise InvalidFile('No gmf_data: did you forget gmfs_csv in %s?' % self.oqparam.inputs['job_ini']) rlzs = dstore['events']['rlz_id'] gmf_df = dstore.read_df('gmf_data', 'sid') logging.info('Events per site: ~%d', len(gmf_df) / self.N) logging.info('Grouping the GMFs by site ID') by_sid = dict(list(gmf_df.groupby(gmf_df.index))) asset_df = self.assetcol.to_dframe('site_id') for sid, assets in asset_df.groupby(asset_df.index): try: df = by_sid[sid] except KeyError: getter = getters.ZeroGetter( sid, rlzs, self.R, gmf_df.columns[1:]) # strip eid else: df['rlz'] = rlzs[df.eid.to_numpy()] getter = getters.GmfDataGetter(sid, df, len(rlzs), self.R) if len(dstore['gmf_data/eid']) == 0: raise RuntimeError( 'There are no GMFs available: perhaps you did set ' 'ground_motion_fields=False or a large minimum_intensity') out.append(riskinput.RiskInput(getter, assets)) return out def _gen_riskinputs_poe(self, dstore): out = [] asset_df = self.assetcol.to_dframe('site_id') for sid, assets in asset_df.groupby(asset_df.index): # hcurves, shape (R, N) ws = [rlz.weight for rlz in self.realizations] getter = getters.PmapGetter(dstore, ws, [sid], self.oqparam.imtls) for slc in general.split_in_slices( len(assets), self.oqparam.assets_per_site_limit): out.append(riskinput.RiskInput(getter, assets[slc])) if slc.stop - slc.start >= TWO16: logging.error('There are %d assets on site #%d!', slc.stop - slc.start, sid) return out
[docs] def execute(self): """ Parallelize on the riskinputs and returns a dictionary of results. Require a `.core_task` to be defined with signature (riskinputs, crmodel, param, monitor). """ if not hasattr(self, 'riskinputs'): # in the reportwriter return ct = self.oqparam.concurrent_tasks or 1 maxw = sum(ri.weight for ri in self.riskinputs) / ct self.datastore.swmr_on() smap = parallel.Starmap( self.core_task.__func__, h5=self.datastore.hdf5) smap.monitor.save('crmodel', self.crmodel) for block in general.block_splitter( self.riskinputs, maxw, get_weight, sort=True): for ri in block: # we must use eager reading for performance reasons: # concurrent reading on the workers would be extra-slow; # also, I could not get lazy reading to work with # the SWMR mode for event_based_risk if not isinstance(ri.hazard_getter, getters.PmapGetter): ri.hazard_getter.init() smap.submit((block, self.param)) return smap.reduce(self.combine, self.acc)
[docs] def combine(self, acc, res): """ Combine the outputs assuming acc and res are dictionaries """ if res is None: raise MemoryError('You ran out of memory!') return acc + res
[docs]def import_gmfs_csv(dstore, oqparam, sids): """ Import in the datastore a ground motion field CSV file. :param dstore: the datastore :param oqparam: an OqParam instance :param sids: the complete site IDs :returns: event_ids """ fname = oqparam.inputs['gmfs'] array = hdf5.read_csv(fname, {'sid': U32, 'eid': U32, None: F32}, renamedict=dict(site_id='sid', event_id='eid', rlz_id='rlzi')).array names = array.dtype.names # rlz_id, sid, ... if names[0] == 'rlzi': # backward compatibility names = names[1:] # discard the field rlzi imts = [name.lstrip('gmv_') for name in names[2:]] oqparam.hazard_imtls = {imt: [0] for imt in imts} missing = set(oqparam.imtls) - set(imts) if missing: raise ValueError('The calculation needs %s which is missing from %s' % (', '.join(missing), fname)) imt2idx = {imt: i for i, imt in enumerate(oqparam.imtls)} arr = numpy.zeros(len(array), oqparam.gmf_data_dt()) for name in names: if name.startswith('gmv_'): try: m = imt2idx[name[4:]] except KeyError: # the file contains more than enough IMTs pass else: arr[f'gmv_{m}'][:] = array[name] else: arr[name] = array[name] n = len(numpy.unique(array[['sid', 'eid']])) if n != len(array): raise ValueError('Duplicated site_id, event_id in %s' % fname) # store the events eids = numpy.unique(array['eid']) eids.sort() if eids[0] != 0: raise ValueError('The event_id must start from zero in %s' % fname) E = len(eids) events = numpy.zeros(E, rupture.events_dt) events['id'] = eids logging.info('Storing %d events, all relevant', E) dstore['events'] = events # store the GMFs dic = general.group_array(arr, 'sid') offset = 0 gmvlst = [] for sid in sids: n = len(dic.get(sid, [])) if n: offset += n gmvs = dic[sid] gmvlst.append(gmvs) data = numpy.concatenate(gmvlst) data.sort(order='eid') create_gmf_data(dstore, oqparam.get_primary_imtls(), oqparam.get_sec_imts(), data=data) dstore['weights'] = numpy.ones(1) return eids
def _getset_attrs(oq): # read effective_time, num_events and imts from oq.inputs['gmfs'] # if the format of the file is old (v3.11) also sets the attributes # investigation_time and ses_per_logic_tree_path on `oq` with hdf5.File(oq.inputs['gmfs'], 'r') as f: attrs = f['gmf_data'].attrs etime = attrs.get('effective_time') num_events = attrs.get('num_events') if etime is None: # engine == 3.11 R = len(f['weights']) num_events = len(f['events']) arr = f.getitem('oqparam') it = arr['par_name'] == b'investigation_time' it = float(arr[it]['par_value'][0]) oq.investigation_time = it ses = arr['par_name'] == b'ses_per_logic_tree_path' ses = int(arr[ses]['par_value'][0]) oq.ses_per_logic_tree_path = ses etime = it * ses * R imts = [] for name in arr['par_name']: if name.startswith(b'hazard_imtls.'): imts.append(name[13:].decode('utf8')) else: # engine >= 3.12 imts = attrs['imts'].split() return dict(effective_time=etime, num_events=num_events, imts=imts)
[docs]def import_gmfs_hdf5(dstore, oqparam): """ Import in the datastore a ground motion field HDF5 file. :param dstore: the datastore :param oqparam: an OqParam instance :returns: event_ids """ dstore['gmf_data'] = h5py.ExternalLink(oqparam.inputs['gmfs'], "gmf_data") attrs = _getset_attrs(oqparam) oqparam.hazard_imtls = {imt: [0] for imt in attrs['imts']} # store the events E = attrs['num_events'] events = numpy.zeros(E, rupture.events_dt) events['id'] = numpy.arange(E) rel = numpy.unique(dstore['gmf_data/eid']) logging.info('Storing %d events, %d relevant', E, len(rel)) dstore['events'] = events dstore['weights'] = numpy.ones(1) return events['id']
[docs]def create_gmf_data(dstore, prim_imts, sec_imts=(), data=None): """ Create and possibly populate the datasets in the gmf_data group """ oq = dstore['oqparam'] R = dstore['full_lt'].get_num_rlzs() M = len(prim_imts) n = 0 if data is None else len(data['sid']) items = [('sid', U32 if n == 0 else data['sid']), ('eid', U32 if n == 0 else data['eid'])] for m in range(M): col = f'gmv_{m}' items.append((col, F32 if data is None else data[col])) for imt in sec_imts: items.append((str(imt), F32 if n == 0 else data[imt])) if oq.investigation_time: eff_time = oq.investigation_time * oq.ses_per_logic_tree_path * R else: eff_time = 0 dstore.create_df('gmf_data', items, 'gzip') dstore.set_attrs('gmf_data', num_events=len(dstore['events']), imts=' '.join(map(str, prim_imts)), effective_time=eff_time) if data is not None: df = pandas.DataFrame(dict(items)) avg_gmf = numpy.zeros((2, n, M + len(sec_imts)), F32) for sid, df in df.groupby(df.sid): df.pop('eid') df.pop('sid') avg_gmf[:, sid] = stats.avg_std(df.to_numpy()) dstore['avg_gmf'] = avg_gmf
[docs]def save_agg_values(dstore, assetcol, lossnames, aggby): """ Store agg_keys, agg_values. :returns: the aggkey dictionary key -> tags """ lst = [] aggkey = assetcol.tagcol.get_aggkey(aggby) K = len(aggkey) agg_number = numpy.zeros(K + 1, U32) if aggby: logging.info('Storing %d aggregation keys', len(aggkey)) dt = [(name + '_', U16) for name in aggby] + [ (name, hdf5.vstr) for name in aggby] kvs = [] for key, val in aggkey.items(): val = tuple(python3compat.decode(val)) kvs.append(key + val) lst.append(' '.join(val)) dstore['agg_keys'] = numpy.array(kvs, dt) if aggby == ['id']: kids = assetcol['ordinal'] elif aggby == ['site_id']: kids = assetcol['site_id'] else: key2i = {key: i for i, key in enumerate(aggkey)} kids = [key2i[tuple(t)] for t in assetcol[aggby]] dstore['assetcol/kids'] = U16(kids) agg_number[:K] = general.fast_agg(kids, assetcol['number'], M=K) agg_number[K] = assetcol['number'].sum() dstore['agg_number'] = agg_number lst.append('*total*') if assetcol.get_value_fields(): dstore['agg_values'] = assetcol.get_agg_values(lossnames, aggby) dstore.set_shape_descr( 'agg_values', aggregation=lst, loss_type=lossnames) return aggkey if aggby else {}
[docs]def read_shakemap(calc, haz_sitecol, assetcol): """ Enabled only if there is a shakemap_id parameter in the job.ini. Download, unzip, parse USGS shakemap files and build a corresponding set of GMFs which are then filtered with the hazard site collection and stored in the datastore. """ oq = calc.oqparam E = oq.number_of_ground_motion_fields oq.risk_imtls = oq.imtls or calc.datastore.parent['oqparam'].imtls logging.info('Getting/reducing shakemap') with calc.monitor('getting/reducing shakemap'): # for instance for the test case_shakemap the haz_sitecol # has sids in range(0, 26) while sitecol.sids is # [8, 9, 10, 11, 13, 15, 16, 17, 18]; # the total assetcol has 26 assets on the total sites # and the reduced assetcol has 9 assets on the reduced sites if oq.shakemap_id: uridict = {'kind': 'usgs_id', 'id': oq.shakemap_id} elif 'shakemap' in oq.inputs: uridict = {'kind': 'file_npy', 'fname': oq.inputs['shakemap']} else: uridict = oq.shakemap_uri sitecol, shakemap, discarded = get_sitecol_shakemap( uridict.pop('kind'), uridict, oq.imtls, haz_sitecol, oq.asset_hazard_distance['default']) if len(discarded): calc.datastore['discarded'] = discarded assetcol.reduce_also(sitecol) logging.info('Extracted %d assets', len(assetcol)) # assemble dictionary to decide on the calculation method for the gmfs if 'MMI' in oq.imtls: # calculations with MMI should be executed if len(oq.imtls) == 1: # only MMI intensities if oq.spatial_correlation != 'no' or oq.cross_correlation != 'no': logging.warning('Calculations with MMI intensities do not ' 'support correlation. No correlations ' 'are applied.') gmf_dict = {'kind': 'mmi'} else: # there are also other intensities than MMI raise RuntimeError( 'There are the following intensities in your model: %s ' 'Models mixing MMI and other intensities are not supported. ' % ', '.join(oq.imtls.keys())) else: # no MMI intensities, calculation with or without correlation if oq.spatial_correlation != 'no' or oq.cross_correlation != 'no': # cross correlation and/or spatial correlation after S&H gmf_dict = {'kind': 'Silva&Horspool', 'spatialcorr': oq.spatial_correlation, 'crosscorr': oq.cross_correlation, 'cholesky_limit': oq.cholesky_limit} else: # no correlation required, basic calculation is faster gmf_dict = {'kind': 'basic'} logging.info('Building GMFs') with calc.monitor('building/saving GMFs'): imts, gmfs = to_gmfs(shakemap, gmf_dict, oq.site_effects, oq.truncation_level, E, oq.random_seed, oq.imtls) N, E, M = gmfs.shape events = numpy.zeros(E, rupture.events_dt) events['id'] = numpy.arange(E, dtype=U32) calc.datastore['events'] = events # convert into an array of dtype gmv_data_dt lst = [(sitecol.sids[s], ei) + tuple(gmfs[s, ei]) for ei, event in enumerate(events) for s in numpy.arange(N, dtype=U32)] oq.hazard_imtls = {str(imt): [0] for imt in imts} data = numpy.array(lst, oq.gmf_data_dt()) create_gmf_data(calc.datastore, imts, data=data) return sitecol, assetcol
[docs]def create_risk_by_event(calc): """ Created an empty risk_by_event with keys event_id, agg_id, loss_id and fields for damages, losses and consequences """ oq = calc.oqparam dstore = calc.datastore aggkey = getattr(calc, 'aggkey', {}) # empty if not aggregate_by crmodel = calc.crmodel if 'risk' in oq.calculation_mode: descr = [('event_id', U32), ('agg_id', U32), ('loss_id', U8), ('loss', F32), ('variance', F32)] dstore.create_df('risk_by_event', descr, K=len(aggkey), L=len(oq.loss_names)) else: # damage + consequences dmgs = ' '.join(crmodel.damage_states[1:]) descr = ([('event_id', U32), ('agg_id', U32), ('loss_id', U8)] + [(dc, F32) for dc in crmodel.get_dmg_csq()]) dstore.create_df('risk_by_event', descr, K=len(aggkey), L=len(oq.loss_names), limit_states=dmgs)