Tsunami deposits highlight high-magnitude earthquake potential in the Guerrero seismic gap Mexico

High-magnitude (>8 Mw) earthquakes and megaearthquakes (≥9 Mw) often occur along subduction zones (e.g., Chile 1960, Sumatra 2004, Japan 2011). The Guerrero seismic gap (GSG), an ~200 km-long segment of the Mexican subduction zone (MSZ), is characterized by low seismic activity but could plausibly rupture entirely and generate an Mw ~8.4 earthquake with severe consequences for the Mexico City metropolitan area and coastal communities such as Acapulco. Although recent hypotheses suggest rheology in the GSG favors slow slip over fast slip, this view is based on incomplete assessment of tsunamigenic earthquakes in historical and late Holocene times. In other subduction zones where hazards were initially underestimated, large earthquakes and tsunamis left geologic signatures (e.g., Japan, Chile, Sumatra, Cascadia), demonstrating substantial tsunamigenic potential. Here, the study aims to assess the long-term earthquake and tsunami potential of the GSG by developing geological evidence of past large tsunamis and probable local earthquakes over ~2000 years, and by using numerical modeling to evaluate whether a large Mw >8 event—around ~1300 AD—could explain the observed deposits and coastal deformation.

The paper situates its investigation within global evidence that subduction zones produce the largest tsunamigenic earthquakes (Chile 1960; Sumatra 2004; Japan 2011) and that some regions experienced devastating events despite lacking prior instrumental records of similar magnitude. It references prior work indicating mid- to late-Holocene earthquakes and tsunamis in the Guerrero segment of the MSZ, as well as work documenting seafloor and upper-plate morphologies associated with shallow megathrust seismogenesis. Comparable paleotsunami records from other subduction zones (Japan, Chile, Sumatra, Cascadia) illustrate that coastal stratigraphy can reveal great earthquakes overlooked by short instrumental records. Regionally, the neighboring Oaxaca segment has historical and geologic evidence for a very large event (1787, Mw ~8.6) with >500 km alongshore flooding and inland inundation >6 km, and previous research in the Guerrero coast identified apparent older tsunami events over the last several millennia. Collectively, this literature motivates a geological search for tsunami deposits in Guerrero to reassess the potential for rare, large events.

Field surveys targeted environments favorable for tsunami deposit preservation (salt marshes, estuaries, swales between beach ridges, swamps) along ~55 km of the Guerrero coast. A total of 38 sites were investigated using hang auger, geoslicer, pits, and cores, including a shoreline-normal transect with detailed focus on sites A1 and A2 located ~800 m from the shoreline and ~2 m AMSL. Stratigraphic cross sections were constructed from geoslices and pits; nine sites in the transect and additional alongshore sites were logged. Sedimentological and proxy analyses at A1 and A2 included grain size, geochemistry (elemental composition), diatom microfossils, and magnetic properties (magnetic susceptibility and anisotropy of magnetic susceptibility, AMS). Diatom analyses at A1 used standard acid digestion and slide preparation; 1 g dry sediment per horizon was processed, carbonates removed with 10% HCl, rinsed, and mounted; coarse fractions were removed from sand beds to increase diatom detection. A total of 1420 diatom valves were identified and counted under oil immersion at 100x; environmental preferences were assigned based on literature, and diagrams prepared with Tilia software. AMS sampling at A1 collected contiguous 7 cm³ boxes every 2 cm along a vertical profile, with select natural samples from pit walls to avoid coring deformation. Measurements used a MFK1-FA Kappabridge (200 A/m) with a 3D rotator. Principal susceptibilities K1, K2, K3 were determined; shape parameter T and degree of anisotropy P were calculated. Magnetic fabrics were interpreted to infer depositional energy, flow direction, and tsunami swash processes. Chronology employed multiple methods: 210Pb dating for young sediments (top 16 cm yielding ~40 years), 14C radiocarbon (14 samples measured at Beta Analytic), and OSL dating of quartz (3 samples) performed at Baylor University’s Geoluminescence Research Dating Lab. These complementary methods constrained ages of tsunami sand units and adjacent strata. Tsunami and coseismic deformation modeling used FUNWAVE. Initial sea-surface deformation was computed with TOPICS using Okada elastic half-space solutions. Propagation and inundation were modeled with a fully nonlinear Boussinesq scheme over GEBCO_2022 topo-bathymetry (15 arc-sec grid, ~453 m resolution) and a 5 m LIDAR Digital Terrain Model for land. Simulations ran 120 minutes with 1-minute outputs. Scenarios varied fault distance from coast, magnitude, and rupture area; a best-fit scenario (T5) employed a rupture ~210 km long by 90 km wide at 20 km depth, maximum slip ~20 m concentrated at shallow depths, located ~20 km from the coast and ~40 km from the trench, centered in front of site A1. Modeling outputs included coseismic uplift/subsidence patterns, maximum coastal amplitudes, and inland inundation depths/limits.

- Four laterally correlated sand units (2–5) across multiple sites represent tsunami events over the past ~2000 years. Deposits are preserved up to at least ~800 m inland; this is a conservative minimum inundation estimate due to preservation and source limitations. - Sand unit 2 (observed at A1, A2, and most sites) contains marine diatoms and elevated marine-associated elements (Na, Mg, Ca, Br, Ba). AMS indicates high-energy deposition with basal magnetic susceptibility peak and upward fining. 14C ages at A1 and A2 are 1955–1956 cal AD and 1954–1955 cal AD, respectively. This unit likely corresponds to the 1957 M 7.8 Acapulco-area tsunamigenic event; instrumental records show permanent uplift of 15±3 cm and 7±3 cm for the two tsunamigenic earthquakes of May 11 and 19, 1962 (M7.1 and M7.0), while the 1957 event generated a tsunami but has no instrumental uplift record. - Sand unit 3 (A1, A2, and 20 additional sites) shows sharp basal contact, flame structures, broken shells, and marine/brackish diatoms. AMS indicates higher energy with partial K1 reorientation. Sediments above are clayey silt with predominantly brackish diatoms (>40%), whereas below are fine sands with fewer brackish diatoms (<5%). This abrupt environmental shift indicates coseismic coastal subsidence likely ≥1 m. OSL dating constrains the event to 1240–1370 AD. The evidence implies a local Mw >8 near-trench tsunamigenic earthquake capable of flooding at least ~800 m inland. - Sand unit 4 comprises fine sand with sharp basal contact, shell fragments, and brackish/marine diatoms, indicating marine inundation; environmental transitions across the unit suggest probable uplift. Its age is older than 1240–1370 AD but remains inconclusive. - Sand unit 5 is the lowest, deformed, with sharp basal contact, shell fragments, scarce marine diatoms, and apparent liquefaction, indicating marine inundation related to an earthquake sometime before 590–666 cal AD and after 485–359 cal BC. - AMS fabrics at A1 identify high-energy tsunami deposition for units 2 and 3 and low-energy background deposition in intervening clays. Geochemical and microfossil proxies consistently indicate marine incursions during sand units. - Coseismic deformation modeling shows that fault planes nearshore induce uplift, whereas ruptures >20 km offshore result in coastal subsidence; Mw >8.2 events located beyond ~20 km offshore produce subsidence consistent with the geological inference for unit 3. Best-fit scenario (T5) yields 1–2 m subsidence at the coast. - Tsunami modeling for a hypothesized Mw 8.6 near-trench rupture in front of A1 produces maximum coastal amplitudes >10 m and inundation >1 km inland, consistent with the observed inland extent and deposit distribution. - Ruptures on neighboring segments are unlikely to produce the specific subsidence and inundation observed at A1; historical large events (e.g., Jalisco 1932 Mw8.2; Michoacán 1985 Mw8.1; Oaxaca 1785/1787 large events) did not produce comparable subsidence or inundation there. - The temporal occurrence of large events is highly variable over the last ~2 ka; only the ~1300 AD event shows clear, substantial permanent deformation (subsidence), while the others indicate minor uplift or modest/non-permanent deformation. - A 2002 Mw6.7 near-trench tsunami earthquake in the GSG produced only a limited tsunami response (<10 cm), underscoring the variability of rupture modes and shallow slip behavior.

The multidisciplinary evidence—stratigraphy, grain size, geochemistry, diatoms, AMS fabrics, and multi-method dating—demonstrates that the Guerrero seismic gap has experienced at least four large tsunami inundations over the last ~2000 years, including a significant event around 1240–1370 AD linked to a Mw >8 near-trench earthquake. This long-term geologic record contrasts with the relatively quiescent instrumental period of the past ~110 years and challenges interpretations that emphasize only slow slip or lower-magnitude ruptures in the GSG. The ~1300 AD event’s clear signal of ≥1 m coastal subsidence, coupled with best-fit deformation and tsunami models, indicates a trenchward rupture capable of generating >10 m coastal wave amplitudes and >1 km inundation. These findings directly address the research question by revealing the high-magnitude earthquake and tsunami potential in the GSG that is not evident from instrumental records alone. They highlight variable rupture behavior in the MSZ, including shallow slip and near-trench ruptures that can cause subsidence and produce large tsunamis. The study underscores the necessity of integrating long-term geological archives with instrumental and geophysical observations for robust hazard assessment and preparedness for rare but catastrophic events.

This study establishes a ~2000-year history of large tsunamis in the Guerrero seismic gap and identifies a major event around 1240–1370 AD that likely resulted from an Mw >8 near-trench earthquake, producing substantial coastal subsidence and extensive inundation. Through combined sedimentology, microfossils, magnetic fabrics, and multi-method dating, supported by earthquake rupture and tsunami inundation modeling, the results demonstrate that the GSG is capable of high-magnitude tsunamigenic earthquakes exceeding those recorded instrumentally in the last century. The best-fit modeling scenario (Mw ~8.6) predicts >10 m coastal amplitudes and >1 km inundation at the study area, consistent with the geological evidence. The work emphasizes that subduction zones, including the MSZ, exhibit variable rupture modes and that integrating long-term geologic records with instrumental data is essential for accurate hazard assessment and community preparedness.

- The age of sand unit 4 is unresolved; it predates 1240–1370 AD but remains poorly constrained. - Attribution of sand unit 2 to a specific 1957 or 1962 event is uncertain; while radiocarbon dates constrain timing and uplift was instrumentally recorded in 1962, it is not possible to conclusively link uplift and tsunami to a single event. - Radiocarbon ages show inversions at depth likely due to bioturbation/burrowing; 210Pb chronology is limited to the top ~16 cm (~40 years), constraining only the youngest part of the sequence. - Diatoms in some horizons are scarce and poorly preserved, requiring enrichment procedures; preservation biases and sediment source availability limit the inland extent and completeness of deposits, making the observed ~800 m inland limit a conservative minimum estimate of inundation. - Secondary processes and dynamic coastal settings could influence depositional signals; while multiple proxies support tsunami interpretation, some uncertainties in distinguishing all possible non-tsunami influences remain. - Modeling assumptions (fault geometry, slip distribution, bathymetry/topography resolution) influence deformation and inundation estimates; although the best-fit scenario aligns with observations, alternative parameterizations could yield different amplitudes/distributions within plausible ranges.