Comparative study of tectonic tremor locations: Characterization of slow earthquakes in Guerrero, Mexico

Abstract

Deep tectonic tremor in Guerrero, Mexico, has been observed using dense temporal seismic networks (i.e., the Meso‐American Subduction Experiment and Guerrero Gap Experiment (G‐GAP) arrays) during two different time periods. We apply a set of seismic waveform analysis methods to these data sets to constrain the locations of tremors and determine the associated moment tensors. First we detect and locate the tremors. Next, very low frequency (VLF) signals are identified by stacking waveform data during tremor bursts, and their moment tensors are determined. Finally, to better investigate the link between tremors and VLF earthquakes, we detect VLF events using a matched filtering algorithm to search continuous seismic records. None of the 11 VLF events detected by this method occurred in the absence of tremor bursts suggesting they are indeed part of the same phenomena. Unlike previous investigations, our results for the G‐GAP period reveal that downdip tremor activity (i.e., in the so‐called “sweet spot”) is segmented into two patches separated by 40 km in the along‐trench direction, indicating possible variations in the geometry of the plate interface and/or slab effective pressure. Moment tensors of VLF signals are consistent with shear slip on the near‐horizontal plate interface, but source depths are about 5 km deeper than the established plate interface. The slip directions of the VLF events are slightly (~10°) counterclockwise of the plate convergence direction, indicating that strain energy promoting left‐lateral strike‐slip motion may accumulate in the continental crust during the interseismic period.

Insights into the causal relationship between slow slip and tectonic tremor in Guerrero, Mexico

“…Using this numerical model we obtained the stress tensor and the volumetric changes (i.e., the trace of the stress tensor) at every point of our 3‐D domain associated with every slip increment of the SSE. We thus solved for the normal and shear stresses acting on fault planes parallel to the plate interface that we have assumed horizontal in the tremor region according to its geometry [ Pérez‐Campos et al ., ] and recent observations of TTs [ Cruz‐Atienza et al ., ], LFEs [ Frank et al ., ], and very low frequency earthquakes (VLFEs) [ Maury et al ., ] in Guerrero, which have focal mechanisms with zero or very low dipping angles. We then estimated pore pressure changes from the volumetric strain field assuming undrained conditions as where Δ σ kk is the change of dilatancy and B represents the Skempton coefficient ranging from 0 to 1, where the rock is fluid‐saturated for B close to 1.…”

“…The TREP method determines TT hypocentral locations from the spatial distribution of the tremor energy, its spatial derivative, and the azimuth of the particle motion polarization ellipsoid. According to the source mechanisms reported for LFEs and VLFEs in the region [ Frank et al ., ; Maury et al ., ], the method assumes horizontal fault planes and determines both the slip direction (i.e., the rake angle) and the source location that best explain the three above mentioned observables. This technique performs a grid search in a 3‐D lattice with possible hypocenters beneath the study region.…”

“…Different studies of TTs and LFEs in Guerrero [ Payero et al ., ; Kostoglodov et al ., ; Husker et al ., ; Cruz‐Atienza et al ., ; Frank et al ., ; Maury et al ., ] show that these events concentrate in two main source regions of the horizontal plate interface: the so‐called sweet spot, downdip between 200–240 km from the trench, and the transient zone, located updip and close to the slab kink where it becomes horizontal, between 150 and 175 km from the trench (Figure ) [ Husker et al ., ; Frank et al ., ]. Unlike the transient zone, TTs and LFEs take place persistently in the sweet spot even in the absence of long‐term events [ Kostoglodov et al ., ; Husker et al ., ; Frank et al ., ].…”

Seismic Moment, Seismic Energy, and Source Duration of Slow Earthquakes: Application of Brownian slow earthquake model to three major subduction zones

“…This model predicts that tremor signals are often accompanied by VLF components. The VLF signals hidden in a broadband seismogram can be detected by stacking the broadband records relative to the tremor timing, as demonstrated for Nankai, Japan (Ide & Tanaka, 2014;, Taiwan , Cascadia (Ide, 2016), and Mexico (Maury et al, 2016(Maury et al, , 2018. The model also explains the ratios between the tremor energy and VLF seismic moment, which are between 10 À10 and 10 À9 , and shows little change among regions (e.g., Maury et al, 2018).…”

Detection of Low‐Frequency Earthquakes in Broadband Random Time Sequences: Are They Independent Events?

“…In the Nankai subduction zone, Takeo et al (2010) and Ide and Yabe (2014) demonstrated that stacking the broadband seismograms relative to the tremor timing recovers seismic signals at the typical frequencies of VLF events, which can then be utilized to determine their focal mechanisms. The same method has also been successfully applied in Taiwan (Ide et al, 2015), Cascadia (Ide, 2016), and Mexico (Maury et al, 2016. A similar strategy has been employed for tremor bursts and SSEs.…”