JGR–Solid Earth

The b-value in earthquake magnitude-frequency distribution quantifies the relative frequency of large versus small earthquakes. Monitoring its evolution could provide fundamental insights into temporal variations of stress on different fault patches. However, genuine b-value changes are often difficult to distinguish from artificial ones induced by temporal variations of the detection threshold. A highly innovative and effective solution to this issue has recently been proposed by van der Elst (2021, https://doi.org/10.1029/2020jb021027) by means of the b-positive estimator, which is based on analyzing only the positive differences in magnitude between successive earthquakes. Here, we demonstrate the robustness of the estimator, which remains largely unaffected by detection issues due to the properties of conditional probability. We illustrate that this robustness can be further improved by considering positive differences in magnitude, not only between successive earthquakes but also between different pairs of earthquakes. This generalized approach, defined as the “b-more-positive estimator,” enhances efficiency by providing a precise estimate of the b-value while including a larger number of earthquakes from an incomplete catalog. However, our analysis reveals that the accuracy of the b estimators diminishes when earthquakes below the completeness threshold are included in the catalog. This leads to the paradoxical observation that greater efficiency is achieved when the catalog is more incomplete. To address this, we introduce the “b-more-incomplete estimator,” where the b-more-positive estimator is applied only after artificially filtering the instrumental catalog to make it more incomplete. Our findings show the superior efficiency of the b-more-incomplete method.

Induced earthquakes are still highly unpredictable, and often caused by variations in pore fluid pressure. Monitoring and understanding the mechanisms of fluid-induced fault slip is essential for seismic risk mitigation and seismicity forecasting. Fluid-induced slip experiments were performed on critically stressed faulted sandstone samples, and the evolution of the actively sent ultrasonic waves throughout the experiment was measured. Two different fault types were used: smooth saw-cut fault samples at a 35° angle, and a rough fault created by in situ faulting of the samples. Variations in the seismic slip velocity and friction along the fault plane were identified by the coda of the ultrasonic waves. Additionally, ultrasonic amplitudes show precursory signals to laboratory fault reactivation. Our results show that small and local variations in stress before fault failure can be inferred using coda wave interferometry for time-lapse monitoring, as coda waves are more sensitive to small perturbations in a medium than direct waves. Hence, these signals can be used as precursors to laboratory fault slip and to give insight into reactivation mechanisms. Our results show that time-lapse monitoring of coda waves can be used to monitor local stress changes associated with fault reactivation in this laboratory setting of fluid-induced fault reactivation. This is a critical first step toward a method for continuous monitoring of natural fault zones, contributing to seismic risk mitigation of induced and natural earthquakes.

Magnesite (MgCO3) entering the lower mantle together with the subducted oceanic crust is an important carbon carrier. The reaction between magnesite and mantle minerals has been documented, but its influence on the density and velocity profiles of lower mantle remains unexplored. To decipher the deep carbon transportation and its associated effect, here we determined the thermal equations of state of magnesite up to 120 GPa and 2600 K using X-ray diffraction in laser-heated diamond anvil cells. The obtained thermal elastic parameters of magnesite facilitated a comprehensive understanding on the influence of magnesite-SiO2 reaction, variation of carbon and SiO2 content, and temperature on the origin of lower-mantle scatterers at 1,000–1,800 km depth. Our modeling revealed that the depth of the lower-mantle V S scatterers is mainly controlled by the Al2O3 content in SiO2, while its magnitude depends on the SiO2 content. Along normal geotherm, the magnesite-SiO2 reaction would occur before the post-stishovite transition, consuming substantial SiO2 in the subducted oceanic crust. Depending on the amount of residual SiO2, the post-stishovite transition can produce a 2.5–5.2 (2)% V S reduction, compatible with the observed seismic scatterers in Izu-Bonin and Mariana subduction zones. Along slab geotherm, this reaction occurs after the post-stishovite transition, generating a greater V S reduction of 4.4–6.4 (4)%. We thus propose that the reaction between sinking MgCO3 and SiO2 in the slab is one of the potential factors influencing the magnitude of the lower-Vs scatterers at 1,000–1,900 km depth. Our results provide new insights into the deep-mantle carbonate transportation influencing regional geophysics.

The thermal properties of the Earth's primordial magma are the key factors that constrained crystallization and other thermo-chemical processes in Earth's primitive magma ocean and therefore controlled the Earth's long-term evolution. Thermal conductivity of the primordial magma is conventionally assumed to be a constant of about 4 W m−1 K−1 under the high pressure-temperature conditions of the primitive magma ocean. Here we measured the lattice thermal conductivity of a variety of basaltic and silicate glasses at high pressures and a wide range of temperatures. Our results suggest that the primordial magma, if it is indeed represented by basaltic melts, had a thermal conductivity of ∼1.0–1.9 W m−1 K−1, much lower than previously thought. Such low thermal conduction reduced heat loss and thus prolonged the cooling time of the early magma ocean, promoting convection in the solidifying mantle and preventing a global overturn. Moreover, if the seismic ultralow velocity zones presently observed in the lowermost mantle are made of basaltic melts, originating either from remnants of the primitive magma ocean or pieces of subducted crust, the material in these zones must have an ultralow thermal conductivity, which would reduce cooling and thus influence the thermo-chemical evolution of the present day core-mantle boundary.

Mass transfer across the crust-mantle boundary is a fundamental process governing planetary differentiation, the evolution of geochemical reservoirs and ore formation, controlled by physicochemical conditions at the crust-mantle interface. In situ trace-element, clinopyroxene 87Sr/86Sr and garnet Fe3+/ΣFe of kimberlite-borne eclogite xenoliths from the deep (∼50 km) crust-mantle transition below the ca. 1.2–1.0 Ga Namaqua-Natal Fold Belt (southwestern Kaapvaal craton margin) were determined to elucidate their origin and evolution, and to constrain the oxygen fugacity of this pivotal but largely inaccessible environment. Based on a garnet source signature (NMORB-normalized Er/Lu > 1) in pristine “gabbroic” eclogites with pronounced positive Eu, Sr, and Pb anomalies, the suite is interpreted as originating as plagioclase-rich cumulates in oceanic crust from melts generated beneath mature oceanic lithosphere, subsequently subducted during the Namaqua-Natal orogeny. Enriched eclogites have higher measured 87Sr/86Sr in clinopyroxene (up to 0.7054) than gabbroic ones (up to 0.7036), and show increasing bulk-rock Li, Be and Pb abundances with increasing δ18O in clinopyroxene, and muted Eu-Sr-Pb anomalies. These systematics suggest interaction with a siliceous fluid sourced from seawater-altered oceanic sediment in a subduction mélange setting. Garnet Fe3+/ΣFe in deep crustal eclogites is extremely low (0.01–0.04, ±0.01 1σ), as inherited from the plagioclase-rich cumulate protolith, and owing to preferred partitioning into clinopyroxene at low temperatures (∼815–1000°C). Average maximum oxygen fugacities (∆logƒO2(FMQ) = −3.1 ± 1.0 to −0.5 ± 0.7 relative to the Fayalite-Magnetite-Quartz buffer) are higher than in deeper-seated on-craton eclogite xenoliths, but mostly below sulfate stability, limiting the role of S6+ species in oxidizing the mantle wedge.

The eastern Tibetan Plateau (TP) is crucial to the exploration of plateau outgrowth mechanisms. Teleseismic attenuation may provide constraints on anelasticity in the lithosphere-asthenosphere system and may therefore improve the understanding of plateau outgrowth and material extrusion. This study use dense array data across the eastern TP to measure relative attenuation from teleseismic P-wave phases. These observations are then used to invert a 2D relative attenuation map and a 3D Qp−1 model. Weak attenuation is observed beneath the Sichuan Basin and most of the Dianzhong Block at depths of both 0–100 km and 100–200 km (Qp−1 < 0.003). Strong attenuation is observed beneath the eastern TP and southwestern Dianzhong Block at depths of 0–100 km (Qp−1 ∼ 0.01–0.13), but underneath the eastern TP only at depths of 100–200 km (Qp−1 ∼ 0.013–0.015). These results suggest a cratonic lithosphere underneath the Dianzhong Block similar to that beneath the Sichuan Basin. We assume that this is the cooled lithosphere remnant of the Permian Emeishan plume. In contrast, strong attenuation beneath the eastern TP would indicate a thick lower crust and asthenosphere with intense heating. The boundary between strong and weak attenuation lies close to the Longmenshan-Xiaojinhe faults. We infer that the materials extruded eastward from the eastern TP cross this boundary into the southwestern Dianzhong Block in the lower crust, but stop at this boundary in the asthenosphere.

The relationship between deep crustal structure and the deformation in the overlying sedimentary wedge in Santos basin, Brazil is not well understood, and the origin and evolution of the salt-related “Albian Gap (AG)” remain a topic of debate. We investigate the deep structure using three-dimensional inversion of full tensor marine magnetotelluric (MT) data (of 10−1 to 104 s period bandwidth) from 92 stations along three NW–SE lines in 50–1,700 m water depth, one crossing the Cretaceous hinge line (CHL), AG, and Cabo Frio Fault (CFF). The geological validity of the resulting MT resistivity models was determined using resistivity logs from 11 wells and seismic data. The model shows two regionally persistent electrically conductive layers (C2 and C3) related to key Cenozoic and Cretaceous unconformities in the upper part of the sedimentary wedge. Beneath this wedge, the resistive continental crust is ∼35 km thick across the CHL until the 200 m isobath and thereafter thins rapidly seaward to ∼21 km over a lateral distance of ∼80 km defining a domain of highly extended and faulted crust. Our models show a mantle-associated basement high and evidence of significant uplift of the lower part of the sedimentary wedge at 100–150 km distance along our central profile which spatially coincides with the AG and a previously proposed Moho high. This implies a mantle-driven deformation of the crust and basin fill. We propose that mantle flow and magmatism may have played a significant role in the inferred displacement at the AG.

The Central Andes subduction has been the theater of numerous large earthquakes since the beginning of the 21st Century, notably the 2001 Mw = 8.4 Arequipa, 2007 Mw = 8.0 Pisco and 2014 Mw = 8.1 Iquique earthquakes. We present an analysis of 47 permanent and 26 survey global navigation satellite system (GNSS) measurements acquired in Central-South Peru between 2007 and 2022 to better understand the frictional properties of the megathrust interface. Using a trajectory model that mimics the different phases of the cycle, we extract a coherent interseismic GNSS field at the scale of the Central Andes from Lima to Arica (12–18.5°S). Interseismic models on a 3D slab geometry indicate that the locking level is relatively high and concentrated between 20 and 40-km depth. Locking distributions indicate a high spatial variability of the coupling along the trench, with the presence of many locked patches that spatially correlate with the seismotectonic segmentation. Our study confirms the presence of a creeping segment where the Nazca Ridge is subducting; we also observe a lighter apparent decrease of coupling related to the Nazca Fracture Zone (NFZ). However, since the Nazca Ridge appears to behave as a strong barrier, the NFZ is less efficient to arrest seismic rupture propagation. Considering various uncertainty factors, we discuss the implication of our coupling estimates with size and timing of large megathrust earthquakes considering both deterministic and probabilistic approaches. We estimate that the South Peru segment could have a Mw = 8.4–9.0 earthquake potential depending principally on the considered seismic catalog and the seismic/aseismic slip ratio.

Teleseismic back-projection imaging has emerged as a powerful tool for understanding the rupture propagation of large earthquakes. However, its application often suffers from artifacts related to the receiver array geometry. We developed a teleseismic back-projection technique that can accommodate data from multiple arrays. Combined processing of P and pP waveforms may further improve the resolution. The method is suitable for defining arrays ad-hoc to achieve a good azimuthal distribution for most earthquakes. We present a catalog of short-period rupture histories (0.5–2.0 Hz) for all earthquakes from 2010 to 2022 with M W  ≥ 7.5 and depth less than 200 km (56 events). The method provides automatic estimates of rupture length, directivity, speed, and aspect ratio, a proxy for rupture complexity. We obtained short-period rupture length scaling relations that are in good agreement with previously published relations based on estimates of total slip. Rupture speeds were consistently in the sub-Rayleigh regime for thrust and normal earthquakes, whereas a tenth of strike-slip events propagated at supershear speeds. Many rupture histories exhibited complex behaviors, for example, rupture on conjugate faults, bilateral propagation, and dynamic triggering by a P wave. For megathrust earthquakes, ruptures encircling asperities were frequently observed, with downdip, updip, and balanced patterns. Although there is a preference for short-period emissions to emanate from central and downdip parts of the megathrust, emissions updip of the main asperity are more frequent than suggested by earlier results.

No abstract is available for this article.

The continental mid-lithosphere discontinuity (MLD) is widely detected within cratons, with the dominant depth range of 70–100 km and a significant reduction of shear-wave velocity of 2%–12%. However, the formation mechanism and corresponding strength of the MLD are widely debated, which may strongly affect the roles of MLD in craton evolution. The comparisons among variable mechanisms indicate that the strength of the MLD varies from the relatively high viscosity of wet olivine to the rather low viscosity of antigorite. Thus, systematic numerical modeling has been conducted with the MLD of contrasting strengths, that is, the wet olivine-induced MLD or antigorite-induced MLD, to investigate the roles of MLD in the craton instability under variable tectonic regimes (stable, extension, compression, mantle flow traction, or mantle plume). The models show that the cratonic lithosphere with wet olivine-induced MLD maintains its stability under all the tectonic regimes. In contrast, the antigorite-induced MLD with lowest viscosity could significantly promote the decoupling of lithosphere, and facilitate the lithospheric deformation. However, lithospheric delamination only occurs with the rather weak MLD interacting with the sub-plate asthenosphere upwelling during craton extension or mantle plume activity. The sufficient amount of melts is essential for this process, which requires a large amount of extension or a mantle plume with rather high temperature anomaly and large size. Therefore, craton destruction is still difficult, and requires additional strict conditions. This may explain the general stability of most cratons with widespread MLDs.

To enhance our comprehension of the dynamic processes associated with complex plate subduction, volcanic magmatism, and lithosphere deformation beneath the Ryukyu–Taiwan–Philippines region, we utilized nonuniform inversion grids for Pn velocity and anisotropy tomography, and obtained the uppermost mantle structure of this area. The results demonstrate remarkable characteristics: cold oceanic subducting plates display high Pn velocities, whereas volcanic arcs, the extinct mid-ocean ridge in the South China Sea, and the Palawan region exhibit low Pn velocities. The extinct mid-ocean ridge also displays a low-velocity anomaly, indicating that residual heat persists even after approximately 15 million years since seafloor spreading. The occurrence of a discontinuous low-velocity beneath the Ryukyu arc supports the presence of a slab window at approximately 123°E. The volcanic arcs of all subduction zones within the study area displayed trench-parallel Pn anisotropy. The observed Pn fast directions beneath the Taiwan orogenic belt are consistent with crustal anisotropy, providing evidence for the crust–mantle coupled deformation. Moreover, our results shed light on the deep structural characteristics of the complex subduction zones beneath the Philippines region. Plate subduction causes partial melting due to dehydration; then, the melt ascends and accumulates at the uppermost mantle, showing low Pn velocities. However, the low-velocity anomaly is not widely distributed but corresponds to a narrow band. In particular, the east–west bidirectional subduction of the Philippines and Negros formed separate low-velocity anomalies. Low Pn velocities and trench-parallel anisotropy indicate the location of different subduction zones and finely characterize their impact on the uppermost mantle structure.

The Ries impact structure (Germany) contains well-preserved ejecta deposits consisting of melt-free lithic breccia (Bunte Breccia) overlain by suevite. To test their emplacement conditions, we investigated the magnetic properties and microstructures of 26 polymict breccia clasts and a stratigraphic profile from the clasts into the suevite at the Aumühle quarry. Remanent magnetization directions of the Bunte Breccia clasts fall into two groups: those whose directions mostly lie parallel to the reversed field during impact carried mostly by magnetite, and those whose directions vary widely among each clast carried by titanohematite. Basement clasts containing titanohematite acquired a chemical remanent magnetization (CRM) during the ejection process and then rotated during turbulent deposition. Clasts of sedimentary rocks grew magnetite after turbulent deposition, with CRM directions lying parallel to the paleofield. Suevite holds a thermal remanent magnetization carried by magnetite, except for ∼12 cm from the contact with the Bunte Breccia, where hematite concentrations increase due to hydrothermal alteration. These observations lead us to propose a three-stage model of (a) turbulent deposition of the melt-free breccia with clast rotation <580°C, (b) deposition of the overlying suevite, which acted as a semi-permeable barrier that confined hot (<300°C) oxidizing fluids to the permeable breccia zone, and (c) prolonged hydrothermal activity producing further alteration which ended before the next geomagnetic reversal. Basement outcrops have significantly different magnetic properties than the Bunte Breccia basement clasts with similar lithology. Two basement blocks situated near the inner ring may have been thermally overprinted up to 550°C.

Damaging aftershock sequences often exhibit considerable spatio-temporal complexity. The stress changes associated with coseismic slip and aseismic afterslip are commonly proposed to drive aftershock sequences, but few systematic studies exist and do not always support strong, universal driving relationships. To investigate the roles that these two sources of stress changes may play in driving aftershocks, we assess the spatio-temporal relationships between coseismic slip, afterslip, and on-fault (within 5 km) aftershock density following seven M w 6.0–7.6 continental-settings earthquakes, using available high-quality slip models and regional seismic data. From previous empirical work and frictional considerations, near the mainshock we expect coseismic slip and afterslip to be anti-correlated, and aftershocks to occur where coseismic slip is low/zero, near high slip gradients, and/or to migrate with afterslip. However, we find that spatial relationships between afterslip and coseismic slip, and between afterslip and aftershock density differ between earthquakes. Aftershock density correlates with coseismic slip following five of the earthquakes, and with total cumulative slip (coseismic slip + afterslip) following six: indicating that on-fault aftershock distributions may be approximated by total slip (at current resolutions). Additionally, we find that the gradients of coseismic slip and afterslip (proxies for new stress concentrations) do not clearly correlate with aftershock distributions and that the choice of spatial domain over which relationships are tested can affect results significantly. A possible explanation of these results is that fault zones contain considerable fine-scale structural and frictional heterogeneity. Nonetheless, the empirical evidence for frequently assumed relationships between coseismic slip, afterslip and on-fault aftershocks is mixed.

Distributed acoustic sensing (DAS) using existing optical fiber cables facilitates high-density seismic observation. However, few studies have examined the reliability of the seismic waveform amplitude recorded by DAS. In this study, a DAS network was connected to optical fiber cables installed over a distance of 75 km along a high-speed train (Shinkansen) railway in the Kumamoto prefecture, Japan. We successfully observed strong motions of the Mj6.6 earthquake (approximately 150 km from the fiber) on 22 January 2022, in Hyuga-nada, in addition to several small local earthquakes. The observed strong motions from the Mj6.6 earthquake, using DAS, exhibited cycle skipping (clipping) issues due to dynamic range limitations at numerous channels. To address this, we estimated the shaking map, representing maximum strain distributions for Mj6.6, by replacing the clipped data with information from nearby unclipped channels and scaling their RMS amplitudes based on S-coda (unclipped). Furthermore, we verified the reliability of the amplitude information obtained from DAS by estimating the distance attenuation of seismic waves while correcting for the differences in the structure type and coupling as much as possible. The distance attenuation property of local earthquakes was consistent with that of the peak ground velocities obtained from seismometers, indicating that DAS data acquired using fibers installed on infrastructure (various structures) can also be utilized to assess the spatial distribution of the relative amplitude values along the fiber. Obtaining high-density seismic motion distributions is important for earthquake early warning and accurate damage estimation of strong motions.

Processes of magma generation and transportation in global mid-ocean ridges are key to understanding lithospheric architecture at divergent plate boundaries. These magma dynamics are dependent on spreading rate and melt flux, where the SW Indian Ridge represents an end-member. The vertical extent of ridge magmatic systems and the depth of axial magma chambers (AMCs) are greatly debated, in particular at ultraslow-spreading ridges. Here we present detailed mineralogical studies of high-Mg and low-Mg basalts from a single dredge on Southwest Indian Ridge (SWIR) at 45°E. High-Mg basalts (MgO = ∼7.1 wt.%) contain high Mg# olivine (Ol, Fo = 85–89) and high-An plagioclase (Pl, An = 66–83) as phenocrysts, whereas low-Mg basalts contain low-Mg# Ol and low-An Pl (Fo = 75–78, An = 50–62) as phenocrysts or glomerocrysts. One low-Mg basalt also contains normally zoned Ol and Pl, the core and rim of which are compositionally similar to those in high-Mg and low-Mg basalts, respectively. Mineral barometers and MELTS simulation indicate that the high-Mg melts started to crystallize at ∼32 ± 7.8 km, close to the base of the lithosphere. The low-Mg melts may have evolved from the high-Mg melts in an AMC at a depth of ∼13 ± 7.8 km. Such great depths of magma crystallization and the AMC are likely the result of enhanced conductive cooling at ultraslow-spreading ridges. Combined with diffusion chronometers, the basaltic melts could have ascended from the AMC to seafloor within 2 weeks to 3 months at average rates of ∼0.002–0.01 m/s, which are the slowest reported to date among global ridge systems and may characterize mantle melt transport at the slow end of the ridge spreading spectrum.

Distributed Acoustic Sensing (DAS) is becoming a powerful tool for earthquake monitoring, providing continuous strain-rate records of seismic events along fiber optic cables. However, the use of standard seismological techniques for earthquake source characterization requires the conversion of data in ground motion quantities. In this study we provide a new formulation for far-field strain radiation emitted by a seismic rupture, which allows to directly analyze DAS data in their native physical quantity. This formulation naturally accounts for the complex directional sensitivity of the fiber to body waves and to the shallow layering beneath the cable. In this domain, we show that the spectral amplitude of the strain integral is related to the Fourier transform of the source time function, and its modeling allows to determine the source parameters. We demonstrate the validity of the technique on two case-studies, where source parameters are consistent with estimates from standard seismic instruments in magnitude range 2.0–4.3. When analyzing events from a 1-month DAS survey in Chile, moment-corner frequency distribution shows scale invariant stress drop estimates, with an average of Δσ = (0.8 ± 0.6) MPa. Analysis of DAS data acquired in the Southern Apennines shows a dominance of the local attenuation that masks the effective corner frequency of the events. After estimating the local attenuation coefficient, we were able to retrieve the corner frequencies for the largest magnitude events in the catalog. Overall, this approach shows the capability of DAS technology to depict the characteristic scales of seismic sources and the released moment.

Geoscientists use observed data to estimate properties of the Earth's interior. This often requires non-linear inverse problems to be solved and uncertainties to be estimated. Bayesian inference solves inverse problems under a probabilistic framework, in which uncertainty is represented by a so-called posterior probability distribution. Recently, variational inference has emerged as an efficient method to estimate Bayesian solutions. By seeking the closest approximation to the posterior distribution within any chosen family of distributions, variational inference yields a fully probabilistic solution. It is important to define expressive variational families so that the posterior distribution can be represented accurately. We introduce boosting variational inference (BVI) as a computationally efficient means to construct a flexible approximating family comprising all possible finite mixtures of simpler component distributions. We use Gaussian mixture components due to their fully parametric nature and the ease with which they can be optimized. We apply BVI to seismic travel time tomography and full waveform inversion, comparing its performance with other methods of solution. The results demonstrate that BVI achieves reasonable efficiency and accuracy while enabling the construction of a fully analytic expression for the posterior distribution. Samples that represent major components of uncertainty in the solution can be obtained analytically from each mixture component. We demonstrate that these samples can be used to solve an interrogation problem: to assess the size of a subsurface target structure. To the best of our knowledge, this is the first method in geophysics that provides both analytic and reasonably accurate probabilistic solutions to fully non-linear, high-dimensional Bayesian full waveform inversion problems.

Repeating earthquakes repeatedly rupture the same fault asperities, which are likely loaded to failure by surrounding aseismic slip. However, repeaters occur less often than would be expected if these earthquakes accommodate all of the long-term slip on the asperities. Here, we assess a possible explanation for this slip discrepancy: partial ruptures. On asperities that are much larger than the nucleation radius, a fraction of the slip could be accommodated by smaller ruptures on the same asperities. We search for partial ruptures of repeating earthquakes in Parkfield using the Northern California earthquakes catalog. We find 3991 individual repeaters which have 4468 partial ruptures. The presence of partial ruptures suggests that the asperities of repeating earthquakes are much larger than the nucleation radius. However, we find that partial ruptures could accommodate only around 25% of the slip on repeating earthquake patches. A 25% increase in the slip budget can explain only a small portion of the long recurrence intervals of repeating earthquakes.

We use sparse dictionary learning to develop transformations between seismic velocity models of different resolution and spatial extent. Starting with data in the common region of both models, the method can enhance a regional lower-resolution model to match the style and resolution of local higher-resolution results while preserving its regional coverage. The method is demonstrated by applying it to two-dimensional V S and three-dimensional V P and V S regional and local velocity models in southern California. The enhanced reconstructed regional results exhibit clear visual improvements, especially in the reconstructed V P /V S ratios, and better correlations with geological features. Moreover, the reconstructed regional V P , V S models outperform the original ones in comparison of simulated earthquake waveforms to observations. The improved fitting to observed waveforms extends beyond the domain of the overlapping region. The developed dictionary learning approach provides physically interpretable results and offers a powerful tool for additional applications of data enhancement in earth sciences.