Shallow slow earthquakes to decipher future catastrophic earthquakes in the Guerrero seismic gap

Shallow slow earthquakes have been observed in several subduction zones near the trench, but their mechanisms remain less understood than those of downdip slow events. Observations in the Japan Trench showed episodic tremor and slip preceding the 2011 Mw 9 Tohoku-Oki earthquake, indicating that slow slip and megathrust earthquakes can coexist at shallow depths. The Guerrero seismic gap (GG) along the Pacific coast of Mexico is considered capable of producing a large earthquake (Mw > 8) with severe ground motions in Mexico City and a potentially catastrophic tsunami. The north-west segment of the GG (NW-GG, ~140 km long) has not experienced Mw > 7 since 1911; instead, it hosts near-trench tsunami earthquakes (2002) and aftershocks of the 2014 Mw 7.3 Papanoa event. Guerrero is prone to some of the world’s largest slow slip events (SSEs; Mw > 7.5) recurring ~every 4 years. Geodetic data suggest NW-GG accommodates stress aseismically with coupling ~75% lower than adjacent segments, but offshore shallow plate-interface observations were lacking due to absent instrumentation. Tremor, a proxy for shear slip often accompanying SSEs, and repeating earthquakes (repeaters) can indicate aseismic slip, yet prior Guerrero tremor studies focused downdip (~40–50 km depth) using mainly onshore stations. The study aims to determine whether shallow slow earthquakes occur offshore near the trench, characterize their conditions and properties, and assess implications for future large earthquakes in the Guerrero gap using new offshore observations.

Prior studies document shallow slow earthquakes near trenches in multiple subduction zones, including ETS preceding the 2011 Tohoku-Oki event, linking shallow slow slip to megathrust behavior. In Guerrero, deep tremor (~40–50 km depth) has been identified along a horizontally subducting slab, showing rapid migration and close temporal-spatial association with SSEs and overpressured fluids. Guerrero hosts recurrent large SSEs (~4-year recurrence). Onshore geodetic data indicate low coupling in NW-GG relative to bordering segments, suggesting significant aseismic slip. Offshore, shallow slow earthquakes had not been observed in Mexico’s Middle America Trench due to lack of instrumentation. Subducting relief (e.g., seamounts) can elevate pore pressure, fracture the overriding plate, reduce interplate coupling, and create heterogeneous stress fields that facilitate slow slip, tremor, repeaters, tsunami earthquakes, and creep. Residual gravity and bathymetry have been used elsewhere to infer subducting relief and associations with seismicity and creep. Comparisons to Peru and Japan show low-coupling velocity-strengthening domains linked to subducted ridges/seamounts can act as barriers to rupture propagation.

Offshore network and data: Seven ocean bottom seismometers (OBS) were deployed in the NW-GG for one year starting November 2017 at water depths of ~980–2350 m. Each station had three-component 1 Hz short-period sensors. Station locations were estimated with ~2 m mean uncertainty; timing corrections were applied. Shallow tremor (ST) detection and location: A modified envelope correlation maximum-likelihood method was used. Processing steps: (1) band-pass filter continuous velocity 2–8 Hz; (2) square; (3) low-pass at 0.2 Hz; (4) resample at 1 Hz. Detections used 300 s windows with 150 s step and initial inter-station correlation threshold 0.647. Localization maximized averaged weighted cross-correlations using a 1D Guerrero velocity model and ray-theory travel times. A two-step search was performed: grid search (fixed 10 km depth) considering local maxima, followed by gradient optimization (CCSA). Quality control removed outliers with correlation <0.6 after maximization or template-envelope correlation <0.4; bootstrap estimated location uncertainties and results with >2 km error were excluded. Depth is least constrained due to not modeling a slow seafloor sedimentary layer, likely biasing depths too deep. Tremor source parameters: average seismic energy rate estimated from envelopes with density ρ=3000 kg/m³ and shear velocity β=2.8 km/s; origin time at max energy; duration to quarter-max; energy magnitude Mw computed from log energy. Earthquake detection and location: Continuous OBS data were processed with STA/LTA detection (STA 0.3 s, LTA 10 s; trigger ratio 2) requiring at least three stations. An initial set of 4303 detections was reduced via automatic P/S picking and visual inspection; final manual P/S picks yielded 848 events located using a maximum-likelihood method and a 1D velocity model. Earthquake spatial density was computed on a 0.1° grid. Comparison with the National Seismological Service (SSN) catalogue (1908–; for 2017/11/01–2018/12/01) found 1074 SSN events in the GG region, with 298 common to OBS and 518 OBS-only events; systematic location offsets were attributed to differing velocity models and network geometries. Repeating earthquakes: Analyzed 440,655 waveforms from 13 permanent SSN stations for 75,567 earthquakes (2001–2019), plus the one-year OBS catalogue. For pairs within <100 km, computed correlation coefficient (CC) and spectral coherency (COH) in a 25 s window from P onset; frequency bands: 1–8 Hz (permanent stations) and 4–16 Hz (OBS) to enhance SNR offshore. Repeater pairs required CC and COH >95% at two or more stations; sequences were built via single-linkage hierarchical clustering. Found 51 sequences (2–4 events each; M 3.4–4.5) from onshore networks (2001–2019), and 7 additional sequences from OBS (2017–2018; M 2.7–3.8) with short burst-type recurrence (<days to months). Residual gravity and bathymetry: Used SIO Global Gravity grid v29.1 (1 arc-min; 1985–2019 altimetry) and GMRT bathymetry (7.5 arc-sec). Extracted trench-perpendicular profiles every ~25 km along ~1500 km of the Middle America subduction zone. Stacked and subtracted the average trench-perpendicular profile from original grids to obtain residual gravity anomalies (RGAs) and residual bathymetry/topography (RB). Positive residuals landward of incoming seamounts suggest subducting seamount chains beneath the forearc; negative residuals coincide with basins. Bathymetry and RB indicate seamounts ~100 km offshore Guerrero with basal widths 10–20 km rising several hundred meters above surrounding seafloor.

- First offshore observations in the Guerrero seismic gap identified more than 100 shallow tremors (STs), mostly at 10–16 km depth (depth least constrained), assumed on the plate interface. No clear migration was observed, suggesting mechanically isolated locked patches within a weakly coupled interface. - Spatial separation: Most STs lie within <30 km of the trench, while most regular earthquakes cluster ~60 km from the trench near the coast. A ~20 km-wide trench-perpendicular “silent zone” is devoid of seismicity between these populations. - Repeaters co-locate with STs in the eastern and western regions, indicating aseismic slip in those areas. - ST properties: Weak positive correlation between ST magnitude and downdip distance; larger magnitudes and durations occur closer to the coast where seismicity increases and coupling is expected to be higher. STs form four clusters (S1–S4) with distinct behaviors; S4 (near coast) has the largest tremors with mean Mw ≈ 2.0 ± 0.14. Episodic activity observed: S1 recurs ~every 3 months; S2 ~every 1 month, implying associated nearby short-term SSEs with similar recurrence. - Earthquakes: 4303 STA/LTA detections; 848 events manually picked and located offshore. Comparison with SSN shows 298 common events and 518 OBS-only events, confirming additional offshore seismicity coverage. - Repeaters: 51 sequences (2–4 events; M 3.4–4.5) from 2001–2019 onshore data; 7 additional OBS-era sequences (M 2.7–3.8) with short recurrences (<days to months). - Residual gravity and bathymetry: STs and repeaters are preferentially located over positive residual gravity/bathymetry anomalies; the silent zone coincides with a trench-parallel transition between negative and positive residuals. The 2002 near-trench tsunami earthquake (M 6.7) overlaps a positive residual anomaly. These patterns indicate subducting relief (e.g., seamounts), elevated pore pressure, fracturing, and reduced coupling, fostering diverse slip behaviors (creep, STs, repeaters, tsunami earthquakes). - Plate interface mechanical segmentation: Defined Regions A (trench to ~30 km downdip; frictionally heterogeneous with velocity-weakening asperities within a velocity-strengthening matrix hosting episodic STs, repeaters, possible SSEs and near-trench tsunami earthquakes), A' (~30–50 km; the silent zone; dominantly velocity-strengthening creeping region with low nucleation potential for large earthquakes), B (~50–80 km; heterogeneous with increased small earthquakes, STs, repeaters, and past M 7-class events; penetrated by long-term SSEs), and C (deeper downdip with long-term SSEs and deep tremor). Along-strike, the adjacent Petatlán segment lacks a silent zone and is more velocity-weakening with frequent large earthquakes (~every ~35 years). - Hazard implication: The NW-GG appears mechanically unlocked/weakly coupled offshore, decreasing the likelihood of initiating Mw > 8 within the NW-GG. The silent zone may act as a barrier that halts ruptures from adjacent segments, helping explain why past Mw < 8 ruptures did not propagate into the gap. Nevertheless, dynamic rupture effects could enable a sufficiently large adjacent earthquake to cascade through the gap.

The detection of shallow tremor clusters, co-located repeaters, and the identification of a seismicity-free silent zone together delineate a heterogeneous shallow plate interface offshore Guerrero dominated by weak coupling and aseismic slip. This configuration addresses the central hazard question: why large earthquakes have long recurrence intervals and why ruptures from adjacent segments often stop at the NW-GG. The inferred velocity-strengthening behavior in the silent zone is consistent with a creeping barrier that impedes rupture propagation, analogous to central Peru where low-coupling regions associated with subducted topography act as barriers to M > 8 events. Positive residual gravity/bathymetry anomalies, together with the distribution of slow and fast seismic phenomena, implicate subducting relief (e.g., seamounts) in producing heterogeneous frictional conditions (variable velocity-weakening/strengthening patches), elevated pore pressure, and fracturing that promote slow slip, tremor, repeaters, creeping zones, and occasionally tsunami earthquakes. While the NW-GG is less likely to nucleate a Mw > 8 event, stress concentration around the creeping domain and dynamic effects could allow a large rupture from the adjacent Petatlán segment to propagate through the gap, especially by cascading through locked patches beneath the coast where past M > 7 ruptures have occurred. These insights refine the seismogenic extent offshore Guerrero and argue for continued monitoring and physics-based modeling to improve seismic and tsunami hazard assessments.

This study presents the first offshore evidence of shallow slow earthquakes in the Guerrero seismic gap, documenting over 100 shallow tremors organized in episodic clusters, co-located repeaters, and a ~20 km-wide silent zone indicative of a velocity-strengthening creeping region. Integration of tremor, earthquake, repeater, and residual gravity/bathymetry data reveals heterogeneous frictional conditions likely influenced by subducting relief, which together produce diverse slip behaviors across distinct along-dip regions (A, A', B, C) and along-strike differences with the adjacent Petatlán segment. The offshore NW-GG appears weakly coupled, reducing the likelihood of initiating a Mw > 8 rupture within the segment and explaining past rupture terminations; however, dynamic rupture from neighboring segments could still traverse the gap. Future work should include expanded offshore and onshore instrumentation, long-term seafloor geodesy to directly measure offshore coupling and slow slip, and physics-based source modeling to test the barrier and cascading-rupture scenarios and to refine hazard mitigation strategies.

- Depth uncertainty: Tremor locations used a 1D velocity model without explicitly modeling a slow sedimentary layer beneath OBS, biasing depths too deep and making depth the least constrained parameter. - Observation duration and coverage: Offshore observations span only one year with seven OBS, limiting temporal sampling of episodic behaviors and spatial resolution offshore; the south-east portions of Regions A, A', and B remain poorly constrained. - Statistical strength: The correlation between ST magnitude and downdip distance is weak; further observations are required to confirm trends. - Catalogue and model differences: Systematic location offsets between OBS and SSN catalogues likely arise from differing velocity models and independent network geometries. - Indirect inferences: Slow slip offshore is inferred primarily from tremor and repeater occurrence without direct seafloor geodetic confirmation.