Focal Mechanism Determination Research Articles

Determination of earthquake focal mechanism via multi-task learning

A multi-task learning-based focal mechanism network (MTFMN) is proposed for calculating parameters of the focal mechanism of earthquakes by regression with incorporating expert prior knowledge. The model automatically learns feature representations of seismic waveforms and transforms the inversion task of the focal mechanism into multi-task learning. Experimental results suggest that MTFMN outperforms traditional methods in the task of earthquake focal mechanism and improves the accuracy of parameter estimation of focal mechanism. In addition, comparative experiments demonstrate MTFMN’s enhanced robustness and generalization capabilities compared to other methods. Our proposed methodology presents a more precise regression approach for focal mechanism inversion, with the potential to provide a better understanding of seismic events.

Inversion techniques for focal mechanism determination of small-magnitude earthquakes: a comparative study using the Burewala earthquake ($$M_w$$ $$\sim $$ 4) in Pakistan

Inversion techniques for focal mechanism determination of small-magnitude earthquakes: a comparative study using the Burewala earthquake ($$M_w$$ $$\sim $$ 4) in Pakistan

The 4.5 mb Amargosa earthquake of 30 August 2020, Bahia, Brazil: Implications for the intraplate stress field across the continental margin

The 4.5 mb Amargosa earthquake was the largest event to occur in the state of Bahia, NE Brazil, in more than 100 years. It caused intensity VI MM in the epicentral area and was felt up to more than 200 km away. A previously published focal mechanism determination had suggested a normal faulting event at about 10 km depth. However, our detailed analyses of regional and teleseismic records show, in fact, a reverse faulting mechanism and a very shallow depth of about 1 km. The shallow hypocentral depth was consistently determined by modeling of teleseismic P waves, as well as waveform inversion of regional stations for moment tensor solution. Our focal mechanism solution is also consistent with the relative location of the aftershocks which shows the fault plane to be oriented ESE-WNW dipping to the SSW. The P axis is oriented about NNE-SSW, parallel to the coast. In the interior of the continent, the maximum horizontal stresses tend to be roughly E-W oriented. The margin parallel P axis of the Amargosa earthquake is similar to other SHmax estimates from breakout data in the marginal basins and is yet another example of stress rotation due to local effects associated with the continental/oceanic crustal transition.

Focal mechanism of a Late Triassic large magnitude earthquake along the Longmen Shan fault belt, eastern Tibetan Plateau

Stress states near active fault zones are key to understand their seismogenic setting, geodynamic evolution and future seismic activity. The Longmen Shan Fault Belt (LSFB) is a prominent fault zone that has had a long history of activity and major recent earthquakes in the eastern Tibetan Plateau. However, we have no records of major events prior to the 2008 Wenchuan earthquake, and this hinders progress in understanding both the past and future seismicity of the LSFB. A Late Triassic pseudotachylyte, recently discovered and dated, provides an opportunity to compare modern stress states with those acting ca. 230 Ma ago. Here we use a new paleoseismological approach, based on the magnetic fabric of the fault rocks, to determine the focal mechanism of Late Triassic seismic events, each recorded in pseudotachylyte veins at the Bajiaomiao outcrop. The anisotropy of magnetic susceptibility (AMS) of pseudotachylytes and cataclasites arises from the shape-preferred orientation of coseismic neoformed magnetite and monoclinic pyrrhotite, whereas that of protocataclasite arises mainly from the crystallographic preferred orientation of inherited clastic magnetite. The AMS of fault rocks shows well-defined magnetic foliation and lineation. The pseudotachylyte generation vein formed along the plane where seismic slip took place, while the AMS obliquity with respect to this plane indicates seismic slip with ∼248° trend and 38° dip. This result is consistent with the attitude of macroscopically visible subhorizontal striations on the pseudotachylyte plane itself. The consistent asymmetry of multiple injection veins with respect to the pseudotachylyte generation vein, along with the asymmetry of the AMS fabric, shows that seismic deformation resulted from left-lateral, strike-slip motion. This study provides the oldest example, to date, of focal mechanism determination using the AMS method. The kinematics of the Late Triassic event departs from that of the 2008 Wenchuan earthquake thrust motion. Yet, with a displacement estimated between ∼4.4 and 17.4 m, the magnitude of the Late Triassic event was likely Mw 7.5–7.9.

Focal Mechanism Determination of Event S1222a and Implications for Tectonics Near the Dichotomy Boundary in Southern Elysium Planitia, Mars

AbstractOn 4 May 2022 the InSight seismometer SEIS‐VBB recorded the largest marsquake ever observed, S1222a, with an initial magnitude estimate of 4.6. Understanding the depth and source properties of this event has important implications for the nature of tectonic activity on Mars. Located ∼37° to the southeast of InSight, S1222a is one of the few non‐impact marsquakes that exhibits prominent surface waves. We use waveform modeling of body waves (P and S) and surface waves (Rayleigh and Love) to constrain the focal mechanism, assuming a double‐couple source, and find that S1222a likely resulted from reverse faulting in the crust (source depth near 22 km). We estimate the scalar moment to be 2.5 × 1015–3.5 × 1015 Nm (magnitude MW 4.2–4.3). Our results suggest active compressional tectonics near the dichotomy boundary on Mars, likely due to thermal contraction from planetary cooling.

Determination of intraplate focal mechanisms with the Brazilian Seismic Network: A simplified Cut-and-Paste approach

The feasibility of a simplified Cut-and-Paste approach (CAP) for routine determination of intraplate earthquake mechanisms in Brazil is demonstrated with data acquired by the Brazilian Seismographic Network (RSBR). Our approach models three-component seismic recordings through: (i) computation of synthetics seismograms with a single, average velocity model, and (ii) independent alignment of observed and synthetic waveforms through waveform cross-correlation. No partitioning of regional waveforms into their corresponding body- and surface-wave trains is considered. The empirical alignment successfully accounts for lateral variations along specific source-station paths, along with location uncertainties and timing errors, and effectively augments the number of waveforms available for inversion. The performance of the proposed approach is demonstrated through modeling of two recent intraplate earthquakes: the Mw 4.4 2017 Maranhão earthquake and the Mw 4.1 2020 Amargosa earthquake. For the Maranhão event, we obtain a strike-slip focal mechanism (340° ± 0°, 70° ± 0°, 0° ± 0°) and a centroid depth of 9 km with magnitude of Mw 4.4, fully consistent with previous studies. For the Amargosa earthquake, we obtain a normal-fault mechanism (20° ± 4.6°, 60° ± 16.9°, -30° ± 14.4°) with a preferred centroid depth of 10 km, which is in general agreement with P-wave polarity data and the orientation of the nearby Salvador Shear Zone. We thus conclude that our simplified CAP approach is an effective tool for modeling regional waveforms in Brazil and, most likely, in sparse networks in general.

PolarCAP – A deep learning approach for first motion polarity classification of earthquake waveforms

The polarity of first P-wave arrivals plays a significant role in the effective determination of focal mechanisms specially for smaller earthquakes. Manual estimation of polarities is not only time-consuming but also prone to human errors. This warrants a need for an automated algorithm for first motion polarity determination. We present a deep learning model - PolarCAP that uses an autoencoder architecture to identify first-motion polarities of earth-quake waveforms. PolarCAP is trained in a supervised fashion using more than 130,000 labelled traces from the Italian seismic dataset (INSTANCE) and is cross-validated on 22,000 traces to choose the most optimal set of hyperparameters. We obtain an accuracy of 0.98 on a completely unseen test dataset of almost 33,000 traces. Furthermore, we check the model generalizability by testing it on the datasets provided by previous works and show that our model achieves a higher recall on both positive and negative polarities.

Using Machine Learning Techniques with Incomplete Polarity Datasets to Improve Earthquake Focal Mechanism Determination

Abstract Earthquake focal mechanisms are traditionally produced using P-wave first-motion polarities and commonly require well-recorded seismicity. A recent approach that is less dependent on high signal-to-noise exploits similar waveforms to produce relative polarity measurements between earthquake pairs. Utilizing these relative polarity measurements, it is possible to produce composite focal mechanisms for clusters within microseismic sequences using regional networks. However, missing or low-confidence polarity measurements still limit our ability to calculate high-quality composite focal mechanisms. Here, we replaced unreliable polarity measurements with estimates using iterative random forests, an unsupervised ensemble machine learning method. Using the imputed (“replaced”) polarity data, we then categorically clustered the events into families. As a case study, we applied this modified composite mechanism workflow to a multistation template matched catalog of an earthquake swarm that occurred during 2020 near the Maacama fault in northern California. We found that our modified methodology produced higher-quality earthquake families and improved composite focal mechanisms, with fault-plane uncertainties <35° for 94% of the families compared with 34% of families using the previous methodology.

Geologic aspects (faults, landslides and volcanic dikes) of the Nateekin River shallow earthquake swarm of 15 June 2020 near Makushin Volcano, Unalaska Island, Alaska, U.S.A.-siftdesk

Starting 15 June 2020, a shallow earthquake swarm occurred in the Nateekin River valley just east of Makushin Volcano of Unalaska Island, Alaska, U.S.A. Focal mechanism determinations were made on the larger earthquake events. These correlated with four known fault groups originally recognized during 1980-85 field work in this region. These are (a) numerous N 54° W ± striking normal faults, (b) prominent N 74° W ± striking normal faults, (c) N 50° E ± strike-slip faults, and (d) a N 36° E ± striking reverse faults. These fault groups correlate with known dikes for the region, which suggest magmatic intrusions are associated with this earthquake swarm. These faults reflect a predictable fault structure for a regional maximum horizontal tectonic stress direction of N 54° W ± with a localized spreading rift zone associated with Makushin Volcano that is striking N 74° W ±. This rift zone has a reactivated landslide and numerous nearby small landslides caused by this June/July earthquake swarm. This swarm sheds insight into the faulting, volcanic and geothermal processes of the region. Key words: Shallow earthquake swarm in a volcanic rift zone; four different seismic focal (fault) mechanisms recognized; associated active faults and landslides observed; related Holocene dikes and volcanic activity; regional maximum horizontal tectonic stress and magmatic processes; extensive geothermal resources.

Focal mechanism and source determination of the Padang Lawas earthquake april 30, 2020, using moment tensor inversion

The earthquake focal mechanism is one of the essential parameters in understanding the characteristics of an earthquake source. For example, the earthquake mainshock at the Padang Lawas Regency and South Tapanuli Regency on April 30, 2020, caused some damage to houses, schools, and mosques, especially in the South Tapanuli Regency. BMKG released the first information that the earthquake was 1.17N, 99.53E, 24 kilometers northwest of Padang Lawas, North Sumatra province, at a depth of 10 kilometers, Thursday (30/4/2020) at 15.20.25 Local Time with a magnitude of 5.6. The purpose of this study is to determine the focal mechanism of the Padang Lawas earthquake mainshock on April 30, 2021, and determine the fault nodal plane that best suits local tectonic conditions. We used the moment tensor inversion method by using waveform velocity data from the BMKG mini regional station seismic network installed in 2019 and close to the epicenter, namely TTSM, PLSM, LTSM, and RRSI. We also used the velocity model from AK135. In this study, we applied the filter technique using a Butterworth bandpass filter with a lower limit of 0.05 Hz, an upper limit of 0.10 Hz, and a cut frequency of 0.13 Hz. The result of the moment tensor inversion shows the variance of the observation and calculation data is 0.3824. The earthquake source parameters show that the Nodal I plane with strike/dip/slip values is 100.1/88.3/7.5 and the Nodal II plane with strike/dip/slip values is 9.9/82.5/178.3. The value of Seismic Moment (Mo) is 0.4653E+17 Nm or Mw 5. Based on the distribution of aftershocks and the mechanism pattern of the Tapanuli earthquake source on April 30, 2020, it is consistent with the nodal plane II. The mechanism is in the form of a strike-slip fault and corresponds to the movement mechanism of the Sumatran fault system in the Toru segment. Therefore, the tectonic conditions in the southern part of Padang Lawas Regency and South Tapanuli Regency are classified as active. Thus, the government and society need to be aware and make mitigation efforts against the impact of the next earthquake in the future.

Real-time determination of earthquake focal mechanism via deep learning

An immediate report of the source focal mechanism with full automation after a destructive earthquake is crucial for timely characterizing the faulting geometry, evaluating the stress perturbation, and assessing the aftershock patterns. Advanced technologies such as Artificial Intelligence (AI) has been introduced to solve various problems in real-time seismology, but the real-time source focal mechanism is still a challenge. Here we propose a novel deep learning method namely Focal Mechanism Network (FMNet) to address this problem. The FMNet trained with 787,320 synthetic samples successfully estimates the focal mechanisms of four 2019 Ridgecrest earthquakes with magnitude larger than Mw 5.4. The network learns the global waveform characteristics from theoretical data, thereby allowing the extensive applications of the proposed method to regions of potential seismic hazards with or without historical earthquake data. After receiving data, the network takes less than two hundred milliseconds for predicting the source focal mechanism reliably on a single CPU.

Insights into Mechanical Properties of the 1980 Irpinia Fault System from the Analysis of a Seismic Sequence

Seismic sequences are a powerful tool to locally infer geometrical and mechanical properties of faults and fault systems. In this study, we provided detailed location and characterization of events of the 3–7 July 2020 Irpinia sequence (southern Italy) that occurred at the northern tip of the main segment that ruptured during the 1980 Irpinia earthquake. Using an autocorrelation technique, we detected more than 340 events within the sequence, with local magnitude ranging between −0.5 and 3.0. We thus provided double difference locations, source parameter estimation, and focal mechanisms determination for the largest quality events. We found that the sequence ruptured an asperity with a size of about 800 m, along a fault structure having a strike compatible with the one of the main segments of the 1980 Irpinia earthquake, and a dip of 50–55° at depth of 10.5–12 km and 60–65° at shallower depths (7.5–9 km). Low stress drop release (average of 0.64 MPa) indicates a fluid-driven initiation mechanism of the sequence. We also evaluated the performance of the earthquake early warning systems running in real-time during the sequence, retrieving a minimum size for the blind zone in the area of about 15 km.

МЕХАНІЗМ ВОГНИЩА ЛИТОВСЬКОГО ЗЕМЛЕТРУСУ НА ОСНОВІ ІНВЕРСІЇ ХВИЛЬОВИХ ФОРМ

The aim: Determination of focal mechanism of Lithuanian earthquake of 12.06.15 (t0 = 08:18:26.4; 55.52° N, 21.40° E; hs = 0.9 км.; ML = 2.6) by waveform inversion using direct waves and a limited number of stations. Method: Matrix method is used for modelling of seismic wave propagation in the medium modelled as horizontally layered heterogeneous elastic structure. There were obtained the relations of displacement waves on the free surface that were used for seismic tensor determination using only direct P- and S- waves. Determination of seismic tensor and the focal mechanism on the base of developed method for a point source is described. Thus, based on forward modeling, numerical techniques are developed for the inversion of observed waveforms for the components of moment tensor. Results: In the paper, a method is presented for the focal mechanism determination of Lithuanian earthquake of 12.06.15 (ML = 2.6) by waveform inversion using limited number of stations. The focal mechanism is determined using the data from two stations: PABE, SLIТ and from three stations: PABE, MTSE, SLIТ. These seismic stations are the part of BAVSEN (BalticVirtualSeismicNetwork). Scientific novelty: 1. In the paper, a method is presented for moment tensor inversion for the focal mechanism determination of events with a low seismicity. The East Baltic region (EBR) is the region with low seismicity. 2. The focal mechanism is determined using the data from a limited number of stations. Practical significance: The results of focal mechanism determination can be used to study seismicity for regions with a low seismicity using a limited number of stations.

Focal mechanism of the great 2004 Sumatra earthquake by joint inversion of GPS deformation field and GRACE gravity change

In this study, the gravity changes observed by GRACE satellites along with the deformation field determined from GPS velocity vectors are used to estimate the fault parameters of 2004 Sumatra-Andaman Earthquake (SAE) by the inversion of PSGRN/PSCMP dislocation model. To do so, the co-seismic gravity changes are determined by fitting a time-series function to the GRACE monthly gravity fields and the effects of hydrology are suppressed with the aid of Global Land Data Assimilation (GLDAS) model. To increase the accuracy of inversion process, the GPS observed co-seismic and post-seismic displacements from 90 stations scattered across the deformed area are also used to constrain the inversion algorithm.The inversion is performed via the Particle Swarm Optimization (PSO) method that minimizes the error between the model-predicted and observed deformation and gravity changes. To this end, the PSGRN/PSMCP dislocation model predicts the deformation and gravity changes for each candidate fault in the PSO iterative process. It is shown that the joint use of the GPS and GRACE data in determination of focal mechanism, results in a fault model that is better consistent with the geometry of the existing subduction zone compared to the inverted fault using only GPS or GRACE data. Due to the complex geometry of the Sunda Trench, a discretized rectangular fault system is considered that allows the depth, slip, strike angle and dip angle vary between the sub-faults. In fact, the estimated fault model consists of 100 sub-faults with a total moment magnitude of Mw = 9.23 that matches the curved geometry of the Sunda Trench. The fault model reveals that the slip has a greater magnitude in segments near the northern Sumatra and Nicobar islands while for the segments of the fault, near the Andaman Islands, the slip is relatively smaller and has its greater magnitudes at depth. The depths and dip angles of the estimated fault model are compared to the SLAB2 3D geometry provided for the Sumatra and Java subduction zones and through this comparison, the RMSEs of 6.0 km for the depth and 3.1 ° for the dip angle are obtained.

Method for Determination of Focal Mechanisms of Magnitude 2.5–4.0 Earthquakes Recorded by a Sparse Regional Seismic Network

ABSTRACTFocal mechanisms of earthquakes with magnitudes Mw 4.0 and less recorded by a sparse seismic network are usually poorly constrained due to the lack of an appropriate method applicable to finding these parameters with a sparse set of observations. We present a new method that can accurately determine focal mechanisms of earthquakes with Mw (3.70–3.04) using data from a few regional seismic stations. We filter the observed seismograms as well as synthetic seismograms through a frequency band of 1.5–2.5 Hz, which has a good signal-to-noise ratio for small earthquakes of the magnitudes with which we are working. The waveforms are processed to their envelopes to make the waveforms relatively simple for modeling. To find the optimal focal mechanism for an event, a nonlinear moment tensor inversion in addition to a coarse grid search over the possible dip, rake, and strike angles at a fixed value of focal depth and a fixed value of scalar moment is performed. We tested the method on 18 aftershocks of Mw (3.70–2.60) of the 2011 Mw 5.7 Mineral, Virginia, earthquake and on five aftershocks of Mw (3.62–2.63) of the 2013 Mw 4.5 Ladysmith, Quebec, earthquake. Our method obtains accurate focal mechanisms for 16 out of the 21 events that have previously reported focal mechanisms. Tests of our method for different crustal models show that event focal mechanism determinations vary with an average Kagan angle of 30° with the different crustal models. This means that the event focal mechanism determinations are only somewhat sensitive to the uncertainties in the crustal models tested. This study confirms that our method of modeling envelopes of seismic waveforms can be used to extract accurate focal mechanisms of earthquakes with short-time functions (Mw<4.0) using at least three regional seismic network stations at epicentral distances of 60–350 km.

Resolving Complicated Faulting Process Using Multi‐Point‐Source Representation: Iterative Inversion Algorithm Improvement and Application to Recent Complex Earthquakes

AbstractComplicated faulting processes of large earthquakes may involve simultaneous or sequential rupture of multiple fault planes, often with different focal mechanisms. Determination of these subevent focal mechanisms is essential for multifault parameterization and complex rupture process analysis. Iterative inversion for Multi‐Point‐Source (MPS) faulting representations using broadband teleseismic body waves is an established strategy to estimate the focal mechanism variation but tends to be an unstable procedure. We enhance an iterative inversion algorithm developed by M. Kikuchi and H. Kanamori by specifying extra a priori constraints in the inversion. These constraints include the specific search time window, location, and source time function of each point source subevent to enable controls over the temporal and spatial properties of each point source. We apply the improved algorithm to seven recent earthquakes with complex faulting: the 2008 Wenchuan, 2009 Samoa, 2010 El‐Mayor Cucapah, 2010 Darfield, 2012 Indian Ocean, 2016 Kaikoura, and 2018 Gulf of Alaska earthquakes. The faulting complexity of these events has been independently constrained using geodetic, seismic, tsunami, and/or field observations, enabling tests of the MPS inversion algorithm performance. The improved iterative inversion resolves varying focal mechanisms of multiple point sources with specific timing consistent with detailed solutions, but spatial resolution tends to be relatively low (>20 km). Summation of subevent point‐source solutions recovers 30% to 95% of the long‐period point‐source seismic moments, varying with different levels of parameterization. The improved iterative MPS inversion algorithm is a promising approach for prompt (within a few hours) analysis of primary focal mechanism variation with reasonable overall space‐time distribution for complicated events.

An In‐Depth Seismological Analysis Revealing a Causal Link Between the 2017 MW 5.5 Pohang Earthquake and EGS Project

AbstractHydraulic injection by the Pohang enhanced geothermal systems (EGSs) has been suspected to trigger the 2017 moment magnitude (MW) 5.5 Pohang earthquake in South Korea. The last stimulation experiment in the EGS was conducted only 2 months before the disaster, which has led to this suspicion. In this study, we conducted a seismic analysis on the earthquakes that have occurred around the EGS site in the past 10 years. The study included the construction of a velocity model, earthquake detection, the determination of hypocenters, magnitudes, focal mechanisms, and stress inversion, and a clustering analysis. No seismic activity was detected near the study area until November 2015 when there was a loss of a large quantity of heavy drilling mud. For three stimulations of a geothermal well, earthquakes sequentially migrated to the southwest along a fault plane, leading to the location of the mainshock. The delineated fault plane crossed the injection well at approximately 3,800 m, which corresponds to the borehole interval of not only the mud loss but also the breakage of the well's casing due to the mainshock rupture. These findings can be treated as empirical evidence for the hypothesis that the 2017 MW 5.5 Pohang earthquake was initiated on a critically stressed fault zone by the anthropogenic activity of fluid injection, consequentially releasing accumulated strain energy via tectonic loading.

Earthquake relocation, focal mechanism and stress field determination in central Brazil

In this work, we perform a seismological investigation in the central Brazil region for focal mechanisms and stress. This large area (910.000 km2) is surrounded by several geological provinces: Eastern Amazon craton; southwestern Parnaíba basin; western São Francisco craton; northern Paraná basin and almost the whole Tocantins province. The area is crossed, in an SW-NE direction, by the continental-scale discontinuity - Transbrasiliano lineament (LTB). In the Brazilian seismic catalog (1720–2019), it is possible to distinguish a seismic belt following the LTB lineament and some dispersed events in the northern, western and eastern areas. The study area is characterized mainly by low magnitude events (M < 4); only 11 events reported in the catalog with M > 4 occurred, and among them just one with magnitude 5. Based on quality criteria a set of 118 events from 338 present in the catalog were selected for a thorough analysis. This work was divided in three parts: (i) Events relocation with the iLoc code and Regional Seismic Travel Time (RSTT) velocity model; (ii) Focal mechanisms determination of 10 events using waveform envelopes inversion and polarities, and; (iii) Inversion of focal mechanisms for stress field. The study shows that the seismicity of the area is mainly concentrated in two relatively narrow belts and that the major compressional stress axis of the whole zone is well resolved with azimuth about 133° which is compatible with previous studies of stress in central Brazil.

Implications of 3D Seismic Raytracing on Focal Mechanism Determination

AbstractThe purpose of this study is to investigate apparent first‐motion polarities mismatch at teleseismic distances in the determination of focal mechanism. We implement and compare four seismic raytracing algorithms to compute ray paths and travel times in 1D and 3D velocity models. We use the raytracing algorithms to calculate the takeoff angles from the hypocenter of the 24 August 2016 Mw 6.8 Chauk earthquake (depth 90 km) in central Myanmar to the stations BFO, GRFO, KONO, and ESK in Europe using a 3D velocity model of the upper mantle below Asia. The differences in the azimuthal angles calculated in the 1D and 3D velocity models are considerable and have a maximum value of 19.6°. Using the takeoff angles for the 3D velocity model, we are able to resolve an apparent polarity mismatch where these stations move from the dilatational to the compressional quadrant. The polarities of synthetic waveforms change accordingly when we take the takeoff angles corresponding to the 3D model into account. This method has the potential to improve the focal mechanism solutions, especially for historical earthquakes where limited waveform data are available.

Quick determination of earthquake source parameters from GPS measurements: a study of suitability for Taiwan

SUMMARY We test the feasibility of GPS-based rapid centroid moment tensor (GPS CMT) methods for Taiwan, one of the most earthquake prone areas in the world. In recent years, Taiwan has become a leading developer of seismometer-based earthquake early warning systems, which have successfully been applied to several large events. The rapid determination of earthquake magnitude and focal mechanism, important for a number of rapid response applications, including tsunami warning, is still challenging because of the limitations of near-field inertial recordings. This instrumental issue can be solved by an entirely different observation system: a GPS network. Taiwan is well posed to take advantage of GPS because in the last decade it has developed a very dense network. Thus, in this research, we explore the suitability of the GPS CMT inversion for Taiwan. We retrospectively investigate six moderate to large (Mw6.0 ∼ 7.0) earthquakes and propose a resolution test for our model, we find that the minimum resolvable earthquake magnitude of this system is ∼Mw5.5 (at 5 km depth). Our tests also suggest that the finite fault complexity, often challenging for the near-field methodology, can be ignored under such good station coverage and thus, can provide a fast and robust solution for large earthquake directly from the near field. Our findings help to understand and quantify how the proposed methodology could be implemented in real time and what its contributions could be to the overall earthquake monitoring system.