Brazilian “Strong-Motion” Data Stéphane Drouet, ON, Rio de Janeiro Marlon Pirchiner, IAG/USP, São Paulo Marcelo Assumpção, IAG/USP, São Paulo Luis Carlos Ribotta, IPT, São Paulo GEM-SARA Topic 6 Workshop 12-14 November 2014
Primary source of the records Initial selection criteria Mmin~3 (whatever M scale) Period 1993-2010 Records collected by Marcelo Assumpção (IAG-USP) from the largest recorded events (11 earthquakes) Additional records from Luis Carlos Ribotta (ITP), 11 events (from which 9 are recorded at a single station) From 2011 Records from the Rede Sismografica Brasileira (www.rsbr.org.br), 17 events
Map Map of the events (red stars) and stations (blue triangles)
Data type and pre-processing Only velocimetric digital records Most of the recordings are sample at 100 sps (some of the oldest have lower sampling rate) For all the records noise, P-wave and S-wave windows have been selected by hand
Processing Taper 5% Padding with 0.0 up to next power of 2 of record length Integration and derivation to get displacement and acceleration Records are cut between tp-10 sec and ts+duration (duration defined by hand but based on 95% of energy after S and 10% of maximum amplitude) Mean removed using pre-event if tp>5 sec or using the mean of the whole record fmin and fmax are selected based on the theoretical shape of the Fourier acceleration spectra Zero-phase bandpass Butterworth filter between fmin and fmax Data files written including pads
Example of time series (original)
Example of time series (processed)
Magnitude-distance scatter mw from 2.0 to 5.2 Repi from 4 to 1800 km No “strong-motion” data
Ground-Motion amplitude
Metadata Very little information Event information Localization, mR from the IAG/USP catalogue, Mw from spectral analysis 3 focal mechanisms for the largest events from specific studies (to be updated) Site information Kappa and H/V analysis
Spectral analysis Far-field S-waves Fourier spectra = Source x Path X Site 𝐴 𝑖𝑗𝑘 = 2R 𝜃𝜑 4 𝜋𝜌𝑣 𝑆 3 𝑀 0 𝑖 × 2 𝜋𝑓 𝑘 2 1+ 𝑓 𝑘 𝑓 𝑐 𝑖 2 1 𝑅 𝑖𝑗 𝛾 exp − 𝜋𝑅 𝑖𝑗 𝑓 𝑘 𝑄 0 𝑓 𝛼 𝑣 𝑆 𝑆 𝑗𝑘
Magnitudes
High-frequency attenuation Get station information K (hard-rock/rock) Get anelastic attenuation Q-values (regionalization) Fig. 4 – Example of high-frequency fit to estimate K. S-waves noise Fig. 5 – Computed K values against distance. The resulting Q factor is indicated on top. Fig. 5 – Computed K values against distance. The resulting Q factor is indicated on top.
H/V ratios from S-waves Flat site response, with amplitude close to one => good rock sites, consistent with the very low kappa values
To be done Check geology of the sites Search for additional focal mechanisms and eventually fault information Produce a stochastic GMPE using the results of the spectral analysis (stress drop, Q) Include other data: 2013-2014 from RSBR (data available) Search additional data from collaborators in Brazil