Source code for openquake.hazardlib.gsim.skarlatoudis_2013

# -*- coding: utf-8 -*-
# vim: tabstop=4 shiftwidth=4 softtabstop=4
#
# Copyright (C) 2014-2020 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/>.

"""
Module exports :class:`SkarlatoudisEtAlSSlab2013`.
"""
import numpy as np
from scipy.constants import g

from openquake.hazardlib.gsim.base import GMPE, CoeffsTable
from openquake.hazardlib import const
from openquake.hazardlib.imt import PGA, PGV, SA


[docs]class SkarlatoudisEtAlSSlab2013(GMPE): """ Implements GMPEs developed by A.A.Skarlatoudis, C.B.Papazachos, B.N.Margaris, C.Ventouzi, I.Kalogeras and EGELADOS group published as "Ground-Motion Prediction Equations of Intermediate-Depth Earthquakes in the Hellenic Arc, Southern Aegean Subduction Area“, Bull Seism Soc Am, DOI 10.1785/0120120265 SA are given up to 4 s. The regressions are developed considering the RotD50 (Boore, 2010) of the as-recorded horizontal components """ #: Supported tectonic region type is ‘subduction intraslab’ because the #: equations have been derived from data from Hellenic Arc events, as #: explained in the 'Introduction'. DEFINED_FOR_TECTONIC_REGION_TYPE = const.TRT.SUBDUCTION_INTRASLAB #: Set of :mod:`intensity measure types <openquake.hazardlib.imt>` #: this GSIM can calculate. A set should contain classes from module #: :mod:`openquake.hazardlib.imt`. DEFINED_FOR_INTENSITY_MEASURE_TYPES = set([ PGA, PGV, SA ]) #: Supported intensity measure component is the RotD50 of two #: horizontal components DEFINED_FOR_INTENSITY_MEASURE_COMPONENT = const.IMC.RotD50 #: Supported standard deviation types are inter-event, intra-event #: and total, page 1961 DEFINED_FOR_STANDARD_DEVIATION_TYPES = set([ const.StdDev.TOTAL, const.StdDev.INTER_EVENT, const.StdDev.INTRA_EVENT ]) #: Required site parameter is Vs30 and backarc flag REQUIRES_SITES_PARAMETERS = {'vs30', 'backarc'} #: Required rupture parameters are magnitude and hypocentral depth REQUIRES_RUPTURE_PARAMETERS = {'mag', 'hypo_depth'} #: Required distance measure is Rhypo. REQUIRES_DISTANCES = {'rhypo'}
[docs] def get_mean_and_stddevs(self, sites, rup, dists, imt, stddev_types): """ See :meth:`superclass method <.base.GroundShakingIntensityModel.get_mean_and_stddevs>` for spec of input and result values. """ # extracting dictionary of coefficients specific to required # intensity measure type. C = self.COEFFS[imt] imean = (self._compute_magnitude(rup, C) + self._compute_distance(rup, dists, C) + self._get_site_amplification(sites, C) + self._compute_forearc_backarc_term(C, sites, dists, rup)) istddevs = self._get_stddevs(C, stddev_types, num_sites=len(sites.vs30)) # Convert units to g, # but only for PGA and SA (not PGV): if imt.name in "SA PGA": mean = np.log((10.0 ** (imean - 2.0)) / g) else: # PGV: mean = np.log(10.0 ** imean) # Return stddevs in terms of natural log scaling stddevs = np.log(10.0 ** np.array(istddevs)) # mean_LogNaturale = np.log((10 ** mean) * 1e-2 / g) return mean, stddevs
def _get_stddevs(self, C, stddev_types, num_sites): """ Return standard deviations as defined in table 1. """ stddevs = [] for stddev_type in stddev_types: assert stddev_type in self.DEFINED_FOR_STANDARD_DEVIATION_TYPES if stddev_type == const.StdDev.TOTAL: stddevs.append(C['epsilon'] + np.zeros(num_sites)) elif stddev_type == const.StdDev.INTRA_EVENT: stddevs.append(C['sigma'] + np.zeros(num_sites)) elif stddev_type == const.StdDev.INTER_EVENT: stddevs.append(C['tau'] + np.zeros(num_sites)) return stddevs def _compute_distance(self, rup, dists, C): """ equation 3 pag 1960: ``c31 * logR + c32 * (R-Rref)`` """ rref = 1.0 c31 = -1.7 return (c31 * np.log10(dists.rhypo) + C['c32'] * (dists.rhypo - rref)) def _compute_magnitude(self, rup, C): """ equation 3 pag 1960: c1 + c2(M-5.5) """ m_h = 5.5 return C['c1'] + (C['c2'] * (rup.mag - m_h)) def _get_site_amplification(self, sites, C): """ Compute the fourth term of the equation 3: The functional form Fs in Eq. (1) represents the site amplification and it is given by FS = c61*S + c62*SS , where c61 and c62 are the coefficients to be determined through the regression analysis, while S and SS are dummy variables used to denote NEHRP site category C and D respectively Coefficents for categories A and B are set to zero """ S, SS = self._get_site_type_dummy_variables(sites) return (C['c61'] * S) + (C['c62'] * SS) def _get_site_type_dummy_variables(self, sites): """ Get site type dummy variables, three different site classes, based on the shear wave velocity intervals in the uppermost 30 m, Vs30, according to the NEHRP: class A-B: Vs30 > 760 m/s class C: Vs30 = 360 − 760 m/s class D: Vs30 < 360 m/s """ S = np.zeros(len(sites.vs30)) SS = np.zeros(len(sites.vs30)) # Class C; 180 m/s <= Vs30 <= 360 m/s. idx = (sites.vs30 < 360.0) SS[idx] = 1.0 # Class B; 360 m/s <= Vs30 <= 760 m/s. (NEHRP) idx = (sites.vs30 >= 360.0) & (sites.vs30 < 760) S[idx] = 1.0 return S, SS def _compute_forearc_backarc_term(self, C, sites, dists, rup): """ Compute back-arc term of Equation 3 """ # flag 1 (R < 335 & R >= 205) flag1 = np.zeros(len(dists.rhypo)) ind1 = np.logical_and((dists.rhypo < 335), (dists.rhypo >= 205)) flag1[ind1] = 1.0 # flag 2 (R >= 335) flag2 = np.zeros(len(dists.rhypo)) ind2 = (dists.rhypo >= 335) flag2[ind2] = 1.0 # flag 3 (R < 240 & R >= 140) flag3 = np.zeros(len(dists.rhypo)) ind3 = np.logical_and((dists.rhypo < 240), (dists.rhypo >= 140)) flag3[ind3] = 1.0 # flag 4 (R >= 240) flag4 = np.zeros(len(dists.rhypo)) ind4 = (dists.rhypo >= 240) flag4[ind4] = 1.0 A = flag1 * ((205 - dists.rhypo)/150) + flag2 B = flag3 * ((140 - dists.rhypo)/100) + flag4 if (rup.hypo_depth < 80): FHR = A else: FHR = B H0 = 100 # Heaviside function if (rup.hypo_depth >= H0): H = 1 else: H = 0 # ARC = 0 for back-arc - ARC = 1 for forearc ARC = np.zeros(len(sites.backarc)) idxarc = (sites.backarc == 1) ARC[idxarc] = 1.0 return ((C['c41'] * (1 - ARC) * H) + (C['c42'] * (1 - ARC) * H * FHR) + (C['c51'] * ARC * H) + (C['c52'] * ARC * H * FHR)) #: Coefficients from SA from Table 1 #: Coefficients from PGA e PGV from Table 5 COEFFS = CoeffsTable(sa_damping=5, table=""" IMT c1 c2 c32 c41 c42 c51 c52 c61 c62 sigma tau epsilon pga 4.229 0.877 -0.00206 -0.481 -0.152 0.425 0.303 0.267 0.491 0.352 0.112 0.369 pgv 2.965 1.069 -0.00178 -0.264 0.018 0.390 0.333 0.408 0.599 0.315 0.144 0.346 0.010 4.235 0.876 -0.00206 -0.482 -0.153 0.425 0.304 0.265 0.488 0.353 0.111 0.370 0.025 4.119 0.877 -0.00202 -0.490 -0.140 0.415 0.326 0.301 0.511 0.352 0.103 0.367 0.050 4.320 0.863 -0.00212 -0.483 -0.178 0.410 0.286 0.245 0.475 0.376 0.095 0.388 0.100 4.565 0.867 -0.00244 -0.515 -0.185 0.452 0.371 0.234 0.442 0.404 0.066 0.410 0.200 4.613 0.842 -0.00199 -0.596 -0.221 0.396 0.291 0.289 0.469 0.379 0.154 0.409 0.400 4.463 0.926 -0.00190 -0.427 -0.110 0.459 0.295 0.298 0.516 0.322 0.141 0.351 1.000 3.952 1.102 -0.00178 -0.199 0.112 0.316 0.442 0.371 0.512 0.305 0.201 0.365 2.000 3.281 1.260 -0.00106 -0.136 0.055 0.196 0.352 0.408 0.578 0.277 0.203 0.343 4.000 2.588 1.384 -0.00039 -0.179 -0.046 0.113 0.189 0.264 0.475 0.278 0.176 0.329 """)