The research, led by Arpita Biswas of the Indian Institute of Technology Kharagpur together with Sankar Kumar Nath of the University of Burdwan, drew on an extraordinary geotechnical and geophysical database of approximately 6,000 investigation sites across the basin. These included Standard Penetration Test boreholes, downhole seismic surveys, multichannel analysis of surface waves, ambient noise measurements and resistivity surveys, supplemented by laboratory tests on soil samples for grain size, moisture content, plasticity and density. The sheer scale of the dataset allowed the team to move beyond the city-level case studies — focused previously on Kolkata, Dhaka and Chittagong — that had dominated liquefaction research in the region.

The geological setting could hardly be more favourable for liquefaction. The basin contains the world’s deepest Neogene-to-recent sedimentary accumulation, exceeding 18 kilometres in thickness, and about 77 percent of the study area is blanketed in Quaternary alluvium — loose silt, sand, clay and gravel deposited by the Ganges, Brahmaputra and Meghna rivers. Statistical analysis of the borehole data revealed that the upper 15 metres of basin sediment is very loose to medium dense, with corrected penetration values averaging between 6 and 35 depending on depth, and a mean fines content of just 36.8 percent, indicating predominantly coarse-grained soils. Grain-size curves for sites in Dhaka, Rajshahi, Kalyani and Bhubaneswar showed that most soil layers fall within the internationally recognised boundaries of ‘most liquefiable’ and ‘potentially liquefiable’ soils.

Groundwater compounds the danger. Post-monsoon water tables across the basin range from just 0.27 to 4.71 metres below the surface, and shallow groundwater dramatically lowers the effective confining stress on sediments, creating ideal conditions for pore-pressure buildup during shaking. Historical precedent is sobering: the 1885 Bengal earthquake of magnitude 6.8, the 1897 Shillong earthquake of magnitude 8.1, the 1918 Srimangal earthquake of magnitude 7.6 and the 1934 Bihar-Nepal earthquake of magnitude 8.1 all triggered documented liquefaction, leaving ground fissures, sand boils, lateral spreading and soil slumping in their wake. Even moderate events such as the 2003 Kolabunia earthquake of magnitude 5.6 produced ground rupture along riverbanks in Rangamati.

Methodologically, the study is a tour de force of integrated hazard modelling. Because instrumental recordings are absent for the great historical earthquakes, the team used the EXSIM finite-fault stochastic simulation technique, which discretises a fault plane into numerous subsources and sums their contributions with realistic rupture propagation and travel-time delays. The resulting bedrock motions were then fed into two-dimensional nonlinear finite element site response analyses in PLAXIS 2D, capturing soil behaviour that simpler one-dimensional equivalent-linear methods miss. The simulated ground motions were validated against Modified Mercalli intensities derived from historical isoseismal maps, yielding a root-mean-square error of 0.91 — a respectable agreement for regional-scale work.

The probabilistic component is equally ambitious. Combining a comprehensive probabilistic seismic hazard model with fourteen ground-motion prediction equations, the researchers computed bedrock peak ground acceleration for a 475-year return period — the conventional design-basis earthquake — ranging from 0.08 to 0.58 g. Multiplying these values by spatially interpolated site amplification factors, which ranged from 1.30 to as much as 4.60 in the soft coastal deposits of the Sundarbans, produced surface-consistent hazard values reaching a remarkable 1.40 g near Sylhet in the northeast, close to the Dauki fault and the Indo-Burman subduction zone. Cities such as Mymensingh and Sylhet fell into the ‘severe’ category, while Dhaka, Kolkata and Khulna were rated ‘moderate’ and Bhubaneswar and Dhanbad ‘low’.

With these ground motions in hand, the team evaluated liquefaction using the stress-based framework of Idriss and Boulanger, computing factors of safety for every soil layer, then aggregating them into the Liquefaction Potential Index, the Liquefaction Risk Index, the probability of liquefaction and the probability of ground failure. The results are stark: under the probabilistic scenario, roughly 42 percent of the basin’s area falls into the ‘severe’ liquefaction potential class, with a further 35 percent rated ‘high’. Dhaka, Sylhet, Mymensingh, Chittagong, Barisal, Khulna and Rangpur all exceed the severe threshold of LPI greater than 15, while Rajshahi and Kolkata sit in the high class. The risk index map places nearly half the region — about 48 percent — in the ‘extremely high’ risk zone. At representative sites including Dhaka, Chittagong, Haldia and Kalna, probabilities of liquefaction exceeded 65 percent between depths of 5 and 20 metres, and cities from Dhaka to Barasat showed probabilities of ground failure above 0.9 — ‘extremely high to absolutely certain’.

Crucially, the modelled hazard aligns with reality. The predicted severity zones show an increasing trend in the number of documented liquefaction occurrences, providing qualitative validation of the approach. The spatial pattern also makes geological sense: the northeast around Sylhet and Mymensingh combines high seismic loading from nearby active faults with young floodplain deposits and shallow groundwater, while the southwest, underlain by older, stiffer sediments near the Indian Shield, fares considerably better. Notably, the study shows that strong shaking alone does not guarantee severe liquefaction — the Chittagong Fold Belt’s uplifted, folded sediments partially offset high seismic demand, whereas the unconsolidated Holocene deltaic plains of Khulna and Barisal remain highly susceptible despite more moderate loading.

Perhaps the most practically valuable contribution is the back-analysis of target penetration resistance. By reversing the standard liquefaction triggering equations, the researchers calculated the improved SPT-N values — denoted N-improved — needed at each site to guarantee no liquefaction under the probabilistic scenario, assuming a target factor of safety of 1.5. Mapped at depth intervals from 5 to 20 metres, these values range from as low as 1 in the stable west to 50 at 20-metre depth in the most demanding northeastern locations. Such maps give engineers, contractors and planners a first-order estimate of the densification or reinforcement required at specific sites, bridging the persistent gap between regional hazard assessment and ground-level engineering decisions.

The authors are candid about limitations: the empirical triggering relationships were calibrated on global case histories rather than Bengal Basin-specific data, uncertainties were not formally propagated through the entire framework, and the maps represent regional trends rather than site-specific design criteria. Ground improvement techniques — vibro-compaction, stone columns, gravel drains, grouting, dewatering and reinforcement — each have their place depending on geology, land use and economics, and the study suggests densification suits the young deltaic south while targeted foundation measures may be more realistic in congested megacities like Dhaka. Still, as a unified, basin-scale portrait of liquefaction threat across one of the world’s most populous regions, the work offers an essential foundation for risk-informed urban planning — and a sobering reminder that beneath the streets of Dhaka and the fields of Sylhet, the ground itself may be the region’s most insidious seismic hazard.

Subject of Research: Assessment of earthquake-induced soil liquefaction potential and mitigation targets across the Bengal Basin using probabilistic seismic hazard analysis and historical earthquake scenarios

Article Title: Probabilistic seismic hazard and historical scenario-based assessment of liquefaction potential in the Bengal Basin for liquefaction mitigation strategies

Article References: Biswas, A., & Nath, S. K. (2026). Probabilistic seismic hazard and historical scenario-based assessment of liquefaction potential in the Bengal Basin for liquefaction mitigation strategies. Discover Geoscience, 4(1), Article 293. https://doi.org/10.1007/s44288-026-00672-6

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00672-6

Keywords: Bengal Basin, soil liquefaction, seismic hazard, liquefaction potential index, probabilistic seismic hazard analysis, ground motion simulation, site amplification, Standard Penetration Test, ground improvement, earthquake engineering, Ganges-Brahmaputra delta, geotechnical investigation

The study, published in Natural Hazards and Earth System Sciences by Veronica Gironelli and colleagues at the Istituto Nazionale di Geofisica e Vulcanologia (INGV) in Milan, addresses a stubborn gap in seismic hazard assessment. Italy possesses one of the richest macroseismic archives in the world, with intensity data points — IDPs — stretching back nearly a millennium. Yet the association between large historical earthquakes and the specific geological structures that generated them has traditionally rested on qualitative interpretation. The new approach replaces that subjectivity with a standardised pipeline that can, in principle, be applied systematically to every strong Italian earthquake since 1117.

The workflow begins with a deceptively simple but crucial step: cleaning the data. Macroseismic intensity is an ordinal measure, graded on scales such as the Mercalli–Cancani–Sieberg (MCS) or the European Macroseismic Scale 1998 (EMS-98), each with twelve degrees of increasing shaking severity. Because intensities are assigned from documentary descriptions of effects on people, buildings and the environment, individual data points can be anomalously high or low relative to the expected pattern of attenuation with distance. Such outliers can poison downstream analyses. The team therefore compares each observed intensity with the value predicted by a regionally calibrated intensity prediction equation (IPE), and flags as outliers any observation deviating by more than three standard deviations — roughly 2.25 intensity units — from the model prediction. This conservative threshold respects the inherent uncertainty of about one intensity degree in the data while removing only the most egregious discrepancies.

The importance of this pre-processing is vividly illustrated by the 5 May 1990 Potenza earthquake in the Southern Apennines. The macroseismic epicentre computed from the raw intensity field sat tens of kilometres from any instrumental location, a mismatch serious enough that the solution was excluded from Italy’s parametric earthquake catalogue. The residual analysis revealed 116 outliers, including suspiciously high intensities clustered around Naples — far from the true source — that turned out to reflect cumulative damage from the catastrophic 1980 Irpinia earthquake a decade earlier. Once these contaminated points were removed, the recalculated macroseismic epicentre moved to within 16.7 kilometres of the instrumental location, down from a 37.3-kilometre misfit. A similar exercise on the well-documented 2016 Amatrice earthquake showed that removing a single inflated intensity value at Pescara del Tronto, where damage had been amplified by unstable landslide deposits, brought the macroseismic magnitude down from 6.46 to 6.26 — a difference with major consequences, since the larger value would imply a rupture area of roughly 270 square kilometres, far exceeding the approximately 180 square kilometres constrained by satellite radar and geological data.

With a cleaned macroseismic field in hand, the workflow moves into its constructive phase. Earthquake parameters — epicentral location and magnitude — are recomputed using the Boxer code, the same tool underpinning the Italian Parametric Earthquake Catalogue CPTI15. Boxer locates the epicentre as the barycentre of the sites recording the highest intensities and estimates magnitude from the areas enclosed by isoseismal contours. Fault dimensions are then derived from magnitude using the widely adopted empirical scaling relationships of Wells and Coppersmith, and a rectangular three-dimensional fault plane is constructed around the hypocentre, with its orientation, dip and slip direction drawn from authoritative repositories including the Italian Seismic Bulletin, the CLASS catalogue of absolute locations, and the Database of Individual Seismogenic Sources (DISS).

The critical innovation lies in how candidate sources are tested. For each proposed fault geometry, the team simulates ground shaking using the ITA18 ground motion model, calibrated for shallow crustal earthquakes in Italy, computing predicted peak ground acceleration and velocity across the affected region. These ground motion parameters are converted into expected intensities using two independent ground-motion-to-intensity conversion equations, allowing the simulated shaking scenario to be compared directly, point by point, with the observed macroseismic field. The best-fitting source is the one that minimises two statistical metrics: the between-event residual, which captures systematic bias, and the root-mean-square error, which captures scatter. Thresholds of half an intensity degree for the former and one intensity degree for the latter define when a source can be considered reliable.

The method was validated on three instrumentally recorded earthquakes of different character: the 1984 Monti della Meta event (moment magnitude 5.86), the 1989 Prealpi Vicentine earthquake (magnitude 4.85), and the 2008 Parmense earthquake (magnitude 5.36). For Monti della Meta, nine alternative fault configurations were tested, and the winning model — built on the CLASS hypocentral location and the global centroid moment tensor magnitude — not only minimised the intensity residuals but also aligned its surface projection with the Barrea–Castelnuovo fault system mapped in the ITHACA database of capable faults. Crucially, where accelerometric recordings existed, the residuals between predicted and recorded ground motions matched the intrinsic standard deviation of the ground motion model itself, providing an independent stamp of validation.

The true test, however, comes from the deep past. The team applied the workflow to the 3 June 1781 Cagli earthquake, one of the most damaging events to strike the interior of the Marche region, and one with no clear seismotectonic interpretation. Two alternative, equally credible macroseismic fields exist for this event, and no capable faults are mapped at the surface near the epicentral area. Four source hypotheses were simulated: two consistent with the Montiego thrust proposed in the DISS database, and two aligned with a system of shallow normal faults in the Monte Nerone area documented by field geologists. The residual analysis favoured the sources built from the alternative intensity dataset, and suggested that a slightly lower magnitude — 6.33 rather than 6.51 — fits the observations better. In effect, the workflow allowed the researchers to adjudicate between competing historical datasets and competing tectonic hypotheses using quantitative evidence rather than expert judgement alone.

The implications extend well beyond Italian borders. Seismic hazard models in many regions of the world depend on catalogues of historical earthquakes whose magnitudes and locations are derived from exactly the kind of documentary evidence used here — and whose assigned faults are frequently matters of conjecture. By providing standardised acceptance thresholds, a transparent chain of processing, and validation against multiple independent benchmarks including instrumental recordings and mapped geological structures, the workflow offers a template that other countries with long documentary records could adapt. The authors note that the approach can be systematically applied to reconstruct the seismogenic sources of all Italian historical earthquakes of moment magnitude 5.5 and above for which consistent intensity distributions are available.

There is also a forward-looking payoff. Once a historical earthquake’s source is credibly identified, realistic shaking scenarios can be generated for the event — showing modern planners precisely which towns and infrastructure would be shaken if the same fault ruptured again. In a country where centuries-old faults still lie beneath densely populated cities, turning the written testimony of past survivors into a map of future risk may prove one of the most consequential applications of historical seismology in decades.

Subject of Research: A methodological workflow for identifying earthquake seismogenic sources from macroseismic intensity data

Article Title: A methodological workflow for the identification of earthquake sources from macroseismic data

Article References: A methodological workflow for the identification of earthquake sources from macroseismic data. (n.d.). https://doi.org/10.5194/nhess-26-4569-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-4569-2026

Keywords: macroseismic data, seismogenic sources, earthquake intensity, historical seismology, seismic hazard assessment, intensity prediction equation, ground motion simulation, Boxer method, Italian earthquakes, fault geometry, outlier detection, shaking scenarios

The research, led by Guan Chen of the Earth Observatory of Singapore and the National University of Singapore, together with Siau Chen Chian, Michael Beer, and Shengji Wei, addresses a persistent gap in seismic hazard analysis. Engineers can already estimate how strong ground shaking will be at a given distance from an earthquake of a certain size using ground-motion prediction equations. What has been harder to quantify is the impulsive character of near-fault motion — how long the dominant pulse lasts, how large its velocity amplitude is, and how these quantities vary with the seismological setting. The new work links pulse parameters directly to magnitude, distance, and site stiffness, giving hazard analysts a physically grounded way to generate realistic near-fault ground motions for any scenario of interest.

The first technical challenge the team tackled was identification. Not every record close to a fault contains a clear pulse, and distinguishing pulse-like from non-pulse-like ground motions has traditionally relied on wavelet-based classification methods, most famously the approach introduced by Jack Baker in 2007. Many existing identification procedures, however, tend to extract a single type of pulse from a record, which can miss the richer structure of real near-fault motions. Records from large continental earthquakes frequently contain multiple pulses, arising from forward-directivity effects as the rupture front races toward the site, from fling-step associated with permanent tectonic deformation, or from constructive interference of seismic waves with basin geometry. A method that captures only one pulse type risks underrepresenting the destructive potential of the motion.

To overcome this limitation, the researchers developed a novel extraction procedure that integrates a parametric pulse model with the generalized continuous wavelet transform, or GCWT. The parametric model provides a mathematical description of the pulse shape, while the wavelet transform scans the velocity time series across scales and times to locate the energy concentration that corresponds to the pulse. By combining the two, the procedure can capture various types of pulses within a single record, preserving both the diversity and the accuracy of the extracted signals. The authors report that the pulses extracted by their method effectively match the pulse characteristics visible in the original records, a critical validation step before any statistical modeling can proceed.

Armed with this identification tool, the team mined the PEER NGA-West2 database, one of the largest curated collections of strong-motion recordings from shallow crustal earthquakes worldwide. Out of 21,222 records examined, 815 were identified as pulse-like ground motions. This carefully classified dataset forms the empirical backbone of the study. Each pulse-like record was decomposed into its pulse component and a residual component, and the parameters describing the pulse — including its period, amplitude, and the number of significant cycles — were tabulated alongside the corresponding earthquake magnitude, source-to-site distance, and Vs30 for each station.

The second major contribution is a set of empirical regression models that predict the pulse parameters from these three explanatory variables. The functional form of the models was developed with attention to the underlying physics: larger earthquakes produce longer rupture durations and hence longer pulse periods, pulses attenuate with distance as seismic energy spreads and is absorbed, and soft soil sites can amplify and lengthen the dominant velocity cycles relative to stiff rock. Model uncertainty was quantified through residual analysis, following the random-effects regression framework that has become standard in ground-motion modeling since the work of Abrahamson and Youngs in 1992. The proposed models were then validated against existing pulse models and against independent datasets not used in the calibration, providing confidence that the predictions generalize beyond the fitting sample.

A particularly valuable practical outcome of the study is a workflow for synthesizing near-fault broadband ground motions. The approach combines the empirical pulse model with a stochastic representation of the residual component — the high-frequency, incoherent part of the motion that remains after the coherent pulse is removed. Because the pulse and the residual are treated separately, the resulting artificial ground motions retain the distinctive impulsive features that govern structural response while simultaneously matching the spectral velocity and spectral acceleration of benchmark recordings across a broad frequency range. This two-component strategy echoes the spectral representation method in stochastic process simulation and offers a computationally efficient alternative to full physics-based rupture simulations, which remain expensive for routine hazard applications.

The implications for earthquake engineering are substantial. Previous research has demonstrated that near-fault pulses impose large deformation demands on flexible structures: moment-resisting frames subjected to pulse-like records accumulate damage differently than those subjected to ordinary shaking, and offshore wind turbines, tunnels, and seismically isolated buildings all show heightened vulnerability when the impulsive content of the motion is ignored. By providing pulse parameters conditioned on magnitude, distance, and site class, the new model allows analysts to construct scenario-consistent suites of near-fault motions without waiting for the next large earthquake to supply fresh recordings. The authors also note that the framework was informed by recent events, including the 2023 moment magnitude 7.8 Kahramanmaras earthquake in Turkey, whose near-fault recordings have become an important test case for pulse identification and simulation methods.

Accessibility is another notable feature of the work. The processed dataset, together with MATLAB code for identifying the optimal pulse in near-fault ground motions and for generating the empirical pulse model, has been released publicly through Zenodo, while the raw waveforms remain freely downloadable from the PEER NGA-West2 database. This open approach lowers the barrier for practicing engineers and researchers in seismically active regions to incorporate directivity and fling effects into their hazard and risk calculations. The study also acknowledges insightful discussions with Professor Jack W. Baker of Stanford University regarding pulse types and the functional form of the model, and it was supported by the Ministry of Education, Singapore, under the InVEST program for integrating volcano and earthquake science and technology.

For cities that sit directly above or beside active faults — from Los Angeles and Wellington to Istanbul and Kathmandu — the difference between ordinary shaking and a coherent velocity pulse can be the difference between repairable damage and collapse. By transforming decades of strong-motion recordings into a compact, predictive, and openly available model, this research moves the field closer to hazard assessments that reflect not just how hard the ground will shake, but how it will shake. As computational tools for seismic risk analysis grow more sophisticated, models of this kind provide the physically consistent input that engineers need to design structures capable of withstanding the sharpest blows an earthquake can deliver.

Subject of Research: Empirical modeling of near-fault velocity pulses in earthquake ground motion

Article Title: Empirical pulse model for near-fault ground motion considering magnitude, distance, and site characteristics

Article References: Chen, G., Chian, S. C., Beer, M., & Wei, S. (2026). Empirical pulse model for near-fault ground motion considering magnitude, distance, and site characteristics. Bulletin of Earthquake Engineering. https://doi.org/10.1007/s10518-026-02650-2

Image Credits: AI Generated

DOI: 10.1007/s10518-026-02650-2

Keywords: near-fault ground motion, velocity pulse, pulse-like records, NGA-West2, wavelet transform, seismic hazard, ground motion simulation, rupture directivity, fling-step, Vs30, earthquake engineering, Bulletin of Earthquake Engineering

The response spectrum is the central currency of earthquake engineering. It summarizes, for a range of vibration periods, the peak response a single-degree-of-freedom oscillator would experience when subjected to a particular accelerogram, and design codes around the world specify shaking demands in exactly these terms. A spectrum-compatible time history is therefore a record whose computed response spectrum matches a code spectrum or a scenario-specific prediction across the period range of interest. Classical approaches to this problem, including wavelet adjustment and frequency-domain optimization, iteratively tweak an existing record until its spectrum fits. Those methods work, but they can distort the phase characteristics, duration, and energy distribution of the original motion, sometimes producing signals that are mathematically compliant yet physically implausible, with unrealistic pulses or smeared energy that mislead nonlinear structural analyses.

The new framework attacks the problem in the time-frequency domain. The authors convert ground motion records into short-time Fourier transform, or STFT, representations, which describe how the amplitude content of the signal evolves across both time and frequency. This representation is crucial because earthquake shaking is inherently nonstationary: high-frequency P- and S-wave energy arrives first, lower-frequency surface waves dominate later, and the amplitude envelope rises and falls in ways that strongly affect structural response. By training a generative model on STFT amplitudes conditioned on a target response spectrum and a latent random variable, the model learns the conditional distribution of time-frequency amplitudes that real, spectrum-consistent accelerograms occupy. Once trained, it can sample that distribution, drawing an effectively unlimited family of distinct accelerograms for any given target spectrum, each with its own plausible waveform realization.

The architecture combines two of the most influential ideas in modern generative machine learning. A variational autoencoder provides a structured latent space from which diverse samples can be drawn, while a generative adversarial network, trained with the Wasserstein distance and a gradient penalty, sharpens the realism of the generated time-frequency amplitudes by forcing them to be statistically indistinguishable from those of genuine records. The Wasserstein formulation, introduced by Arjovsky and colleagues and refined by Gulrajani and colleagues, was chosen for its well-behaved gradients, and the authors report that training histories showed no numerical divergence and no abrupt collapse of sample diversity, a failure mode that plagues conventional adversarial training. The Griffin-Lim style signal reconstruction then maps the generated amplitude information back into the time domain to yield acceleration time histories.

The most distinctive component, however, is what the authors call the response spectrum consistency network, or RSCN. Rather than hoping that spectrum compatibility emerges implicitly from training on real records, the team embedded an explicit spectrum consistency loss directly into the training objective. The RSCN acts as a differentiable surrogate for the computation of the response spectrum, allowing gradients to flow from a period-weighted measure of spectral mismatch back into the generator. In ablation experiments on an event-independent test subset, adding the RSCN measurably reduced response spectrum residuals and improved period-specific calibration compared with an otherwise identical model trained without it. In other words, the network does not merely imitate the spectral statistics of its training data; it actively learns to hit the specific target spectrum handed to it at generation time.

Where do those target spectra come from? The study couples the generative model with a validation-derived, period-weighted ensemble ground motion model, or GMM, that supplies scenario-dependent targets. Ground motion models, the empirical equations that predict spectral accelerations as a function of magnitude, distance, and site conditions, have themselves been undergoing a machine-learning transformation, with neural networks, XGBoost, and hybrid recurrent architectures increasingly used to capture regional attenuation and site effects. By ensembling such predictions and weighting them across the period range, the framework produces a target spectrum tailored to a specific earthquake scenario, and the STFT-VAEGAN then fills in the missing physics: the full stochastic waveform structure that a spectrum alone cannot specify. This linkage between scenario-based spectrum prediction and stochastic time history generation is, according to the authors, the central contribution of the work.

The training and validation data were drawn from the NGA-West2 strong-motion database, the peer-maintained compilation of shallow crustal earthquake recordings curated at the Pacific Earthquake Engineering Research Center, which has become the de facto standard for empirical ground motion modeling worldwide. The authors evaluated the model on a test subset held out by event, a stricter protocol than holding out individual records, because it prevents the network from memorizing the characteristics of particular earthquakes. Test examples demonstrated that the generated accelerograms carry response spectra compatible with their targets while retaining nonstationary time-frequency structure. The team also applied the framework in a case study of the Chi-Chi earthquake, one of the most extensively recorded events in modern seismology, showing that the model can produce multiple distinct accelerograms appropriate to that scenario.

The study builds on a rapidly growing body of work applying deep generative models to seismology. Earlier efforts include conditional GANs that synthesize nonstationary ground motion in the time-frequency domain, data-driven broadband synthesis of earthquake records, and adversarial neural operators for broadband ground-motion synthesis. Hu and colleagues had previously developed a conditional variational autoencoder with adversarial training for spectrum-compatible artificial accelerograms, published in Engineering Structures, and the new Bulletin of Earthquake Engineering paper extends that line by integrating the explicit spectrum consistency loss and the ensemble ground motion model into a single end-to-end pipeline. The distinction matters for practice: a generator that can accept an arbitrary, scenario-derived target spectrum and return a family of conforming, realistic time histories is precisely the tool that performance-based earthquake engineering workflows need for large ensembles of nonlinear structural analyses.

The practical implications extend across seismic hazard analysis, code-based design verification, and the growing field of physics-informed machine learning for earthquake engineering. Fragility curves, which estimate the probability of structural damage as a function of shaking intensity, depend on suites of ground motions that are both numerous and spectrally appropriate; generative sampling offers a way to build such suites without the record-scaling artifacts that contaminate many existing studies. The authors note that the codes developed in the work will be made publicly available on GitHub, and that the underlying records come from the openly accessible NGA-West2 database, lowering the barrier for other research groups to adopt and stress-test the approach. Funded by the National Natural Science Foundation of China under grants 52192675 and 52378541, the work signals a broader shift in which the stochastic simulation of earthquake shaking, once the province of hand-tuned spectral matching algorithms, is increasingly delegated to neural networks that learn the statistics of real ground motion directly from data, and then generate exactly the shaking scenarios engineers ask for.

Subject of Research: Generative AI-based simulation of spectrum-compatible earthquake ground motion time histories

Article Title: Generative AI for spectrum-compatible ground motion time history simulation

Article References: Hu, X., Chen, S., Ding, Y., Liu, X., Fu, L., Zhao, Q., & Li, X. (2026). Generative AI for spectrum-compatible ground motion time history simulation. Bulletin of Earthquake Engineering. https://doi.org/10.1007/s10518-026-02699-z

Image Credits: AI Generated

DOI: 10.1007/s10518-026-02699-z

Keywords: generative artificial intelligence, STFT-VAEGAN, ground motion simulation, response spectrum matching, earthquake engineering, variational autoencoder, generative adversarial network, Wasserstein GAN, NGA-West2 database, ground motion model, nonstationary waveforms, seismic hazard analysis