Earthquake source properties of a shallow induced seismic sequence in SE Brazil

Abstract

We study source parameters of a cluster of 21 very shallow (<1 km depth) small‐magnitude (Mw < 2) earthquakes induced by percolation of water by gravity in SE Brazil. Using a multiple empirical Green's functions (meGf) approach, we estimate seismic moments, corner frequencies, and static stress drops of these events by inversion of their spectral ratios. For the studied magnitude range (−0.3 < Mw < 1.9), we found an increase of stress drop with seismic moment. We assess associated uncertainties by considering different signal time windows and by performing a jackknife resampling of the spectral ratios. We also calculate seismic moments by full waveform inversion to independently validate our moments from spectral analysis. We propose repeated rupture on a fault patch at shallow depth, following continuous inflow of water, as the cause for the observed low absolute stress drop values (<1 MPa) and earthquake size dependency. To our knowledge, no other study on earthquake source properties of shallow events induced by water injection with no added pressure is available in the literature. Our study suggests that source parameter characterization may provide additional information of induced seismicity by hydraulic stimulation.

Source Spectral Properties of Small to Moderate Earthquakes in Southern Kansas

“…The observed increase in Δ with hypocentral depth cannot fully account for this discrepancy, which suggests that the nucleation or rupture processes of these events may differ in some more fundamental way. While these results are consistent with several recent studies of likely induced earthquakes (Agurto-Detzel et al, 2017;Boyd et al, 2017;Sumy et al, 2017), others (Huang et al, 2016(Huang et al, , 2017Zhang et al, 2016) have suggested that tectonic and induced events have comparable source parameters. These disparate conclusions may indicate that differences in local faulting conditions, tectonic stress regime, or history of anthropogenic activity may all play an important role.…”

“…The scaling parameter 1 = 0.25 (2 uncertainty ±0.035) is clearly positive and therefore inconsistent with the null hypothesis of self-similar, constant-Δ data, which would have 1 = 0 to within the uncertainties. These results for the southern Kansas events are within the 1 ∼ 0.1-0.4 range of scaling results obtained for California earthquakes by Trugman and Shearer (2017a) and by other studies that have quantified an increase in stress drop or scaled energy with moment (e.g., Agurto-Detzel et al, 2017;, Calderoni et al, 2013;Izutani & Kanamori, 2001;Mayeda and Walter, 1996;Mayeda et al, 2005Mayeda et al, , 2007Mori et al, 2003;Pacor et al, 2016;Takahashi et al, 2005). However, our Kansas data set contains few higher-magnitude events (20 M L > 3.5 events, which is less than 1% of the total count), and as such we have poor resolution of the distribution and scaling of Δ above M ∼ 3.5.…”

“…The depth dependence in both f c and Δ σ (but not M 0 ) implies an increase in rupture velocity v r that outpaces the expected increase in shear wave speed β with depth (i.e., an increase in the ratio v r / β with depth). Because earthquakes in southern Kansas are characterized by a shallower depth distribution (2–8 km) compared to seismicity in the western and eastern United States (which typically extends to a significantly greater maximum depth), it is plausible that the lower median Δ σ values we observe are due in part to the shallowness of the seismicity (Agurto‐Detzel et al, ; Boyd et al, ; Hardebeck & Aron, ; Pacor et al, ; Sumy et al, ). However, the shallower depth distribution of the events in our data set does not fully explain their anomalously low Δ σ , as can be seen quantitatively by comparing median Δ σ for Kansas and California seismicity within a fixed depth range.…”

Spatio‐Temporal Evolution of Aftershock and Repeater Source Properties After the 2016 Pedernales Earthquake (Ecuador)

“…We instead elect to invert the ratios of all suitable earthquake pairs within a cluster together, using only a small minimum magnitude difference of 0.1 (Text S3 and Figure S14 in Supporting Information S1). This approach is similar to that taken by Lengliné et al (2014) and Agurto-detzel et al (2017).…”

“…We calculate uncertainties by performing a jackknife test on our data (Agurto‐detzel et al., 2017; Prieto et al., 2007). For each cluster and each station, we run 100 inversions where we resample the spectral ratio data, removing 20% of data points within spectral ratios.…”