The limits of earthquake early warning: Timeliness of ground motion estimates

Smartphones enabled up to 58 s strong-shaking warning in the M7.8 Türkiye earthquake

Public earthquake early warning systems (PEEWSs) have the potential to save lives by warning people of incoming seismic waves up to tens of seconds in advance. Given the scale and geographical extent of their impact, this potential is greatest for destructive earthquakes, such as the M7.8 Pazarcik (Türkiye) event of 6 February 2023, which killed almost 60,000 people. However, warning people of imminent strong shaking is particularly difficult for large-magnitude earthquakes because the warning must be given before the earthquake has reached its final size. Here, we show that the Earthquake Network (EQN), the first operational smartphone-based PEEWS and apparently the only one operating during this earthquake, issued a cross-border alert within 12 s of the beginning of the rupture. A comparison with accelerometer and macroseismic data reveals that, owing to the EQN alerting strategy, Turkish and Syrian EQN users exposed to intensity IX and above benefitted from a warning time of up to 58 s before the onset of strong ground shaking. If the alert had been extended to the entire population, approximately 2.7 million Turkish and Syrian people exposed to a life-threatening earthquake would have received a warning ranging from 30 to 66 s in advance.

Real-Time Seismic Intensity Measurements Prediction for Earthquake Early Warning: A Systematic Literature Review

With the gradual development of and improvement in earthquake early warning systems (EEWS), more accurate real-time seismic intensity measurements (IMs) methods are needed to assess the impact range of earthquake intensities. Although traditional point source warning systems have made some progress in terms of predicting earthquake source parameters, they are still inadequate at assessing the accuracy of IMs predictions. In this paper, we aim to explore the current state of the field by reviewing real-time seismic IMs methods. First, we analyze different views on the ultimate earthquake magnitude and rupture initiation behavior. Then, we summarize the progress of IMs predictions as they relate to regional and field warnings. The applications of finite faults and simulated seismic wave fields in IMs predictions are analyzed. Finally, the methods used to evaluate IMs are discussed in terms of the accuracy of the IMs measured by different algorithms and the cost of alerts. The trend of IMs prediction methods in real time is diversified, and the integration of various types of warning algorithms and of various configurations of seismic station equipment in an integrated earthquake warning network is an important development trend for future EEWS construction.

Instantaneous tracking of earthquake growth with elastogravity signals

Rapid and reliable estimation of large earthquake magnitude (above 8) is key to mitigating the risks associated with strong shaking and tsunamis¹. Standard early warning systems based on seismic waves fail to rapidly estimate the size of such large earthquakes^2–5. Geodesy-based approaches provide better estimations, but are also subject to large uncertainties and latency associated with the slowness of seismic waves. Recently discovered speed-of-light prompt elastogravity signals (PEGS) have raised hopes that these limitations may be overcome^6,7, but have not been tested for operational early warning. Here we show that PEGS can be used in real time to track earthquake growth instantaneously after the event reaches a certain magnitude. We develop a deep learning model that leverages the information carried by PEGS recorded by regional broadband seismometers in Japan before the arrival of seismic waves. After training on a database of synthetic waveforms augmented with empirical noise, we show that the algorithm can instantaneously track an earthquake source time function on real data. Our model unlocks ‘true real-time’ access to the rupture evolution of large earthquakes using a portion of seismograms that is routinely treated as noise, and can be immediately transformative for tsunami early warning.

A deep learning model trained on prompt elastogravity signal (PEGS) recorded by seismometers in Japan predicts in real time the final magnitude of large earthquakes faster than methods based on elastic waves.

Investigating the potential effectiveness of earthquake early warning across Europe

Here we assess the potential implementation of earthquake early warning (EEW) across Europe, where there is a clear need for measures that mitigate seismic risk. EEW systems consist of seismic networks and mathematical models/algorithms capable of real-time data telemetry that alert stakeholders (e.g., civil-protection authorities, the public) to an earthquake’s nucleation seconds before shaking occurs at target sites. During this time, actions can be taken that might decrease detrimental impacts. We investigate distributions of EEW lead times available across various parts of the Euro-Mediterranean region, based on seismicity models and seismic network density. We then determine the potential usefulness of these times for EEW purposes by defining their spatial relationship with population exposure, seismic hazard, and an alert accuracy proxy, using well-established earthquake-engineering tools for measuring the impacts of earthquakes. Our mapped feasibility results show that, under certain conditions, EEW could be effective for some parts of Europe.

The viability of earthquake early warning (EEW) in Europe is highly dependent on the magnitude of the ongoing earthquake and the ground-shaking threshold for alert issuance. The potential effectiveness of EEW is highest for Turkey, Italy, and Greece.

A framework for the prediction of earthquake using federated learning

Earthquakes are a natural phenomenon which may cause significant loss of life and infrastructure. Researchers have applied multiple artificial intelligence based techniques to predict earthquakes, but high accuracies could not be achieved due to the huge size of multidimensional data, communication delays, transmission latency, limited processing capacity and data privacy issues. Federated learning (FL) is a machine learning (ML) technique that provides an opportunity to collect and process data onsite without compromising on data privacy and preventing data transmission to the central server. The federated concept of obtaining a global data model by aggregation of local data models inherently ensures data security, data privacy, and data heterogeneity. In this article, a novel earthquake prediction framework using FL has been proposed. The proposed FL framework has given better performance over already developed ML based earthquake predicting models in terms of efficiency, reliability, and precision. We have analyzed three different local datasets to generate multiple ML based local data models. These local data models have been aggregated to generate global data model on the central FL server using FedQuake algorithm. Meta classifier has been trained at the FL server on global data model to generate more accurate earthquake predictions. We have tested the proposed framework by analyzing multidimensional seismic data within 100 km radial area from 34.708° N, 72.5478° E in Western Himalayas. The results of the proposed framework have been validated against instrumentally recorded regional seismic data of last thirty-five years, and 88.87% prediction accuracy has been recorded. These results obtained by the proposed framework can serve as a useful component in the development of earthquake early warning systems.

Reports on progress in physics the complex dynamics of earthquake fault systems: new approaches to forecasting and nowcasting of earthquakes

Charles Richter's observation that 'only fools and charlatans predict earthquakes,' reflects the fact that despite more than 100 years of effort, seismologists remain unable to do so with reliable and accurate results. Meaningful prediction involves specifying the location, time, and size of an earthquake before it occurs to greater precision than expected purely by chance from the known statistics of earthquakes in an area. In this context, 'forecasting' implies a prediction with a specification of a probability of the time, location, and magnitude. Two general approaches have been used. In one, the rate of motion accumulating across faults and the amount of slip in past earthquakes is used to infer where and when future earthquakes will occur and the shaking that would be expected. Because the intervals between earthquakes are highly variable, these long-term forecasts are accurate to no better than a hundred years. They are thus valuable for earthquake hazard mitigation, given the long lives of structures, but have clear limitations. The second approach is to identify potentially observable changes in the Earth that precede earthquakes. Various precursors have been suggested, and may have been real in certain cases, but none have yet proved to be a general feature preceding all earthquakes or to stand out convincingly from the normal variability of the Earth's behavior. However, new types of data, models, and computational power may provide avenues for progress using machine learning that were not previously available. At present, it is unclear whether deterministic earthquake prediction is possible. The frustrations of this search have led to the observation that (echoing Yogi Berra) 'it is difficult to predict earthquakes, especially before they happen.' However, because success would be of enormous societal benefit, the search for methods of earthquake prediction and forecasting will likely continue. In this review, we note that the focus is on anticipating the earthquake rupture before it occurs, rather than characterizing it rapidly just after it occurs. The latter is the domain of earthquake early warning, which we do not treat in detail here, although we include a short discussion in the machine learning section at the end.

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.

On-Site Earthquake Early Warning Using Smartphones

In this study, the measured accelerations of a single smartphone were used to provide an earthquake early warning system. In the presented system, after the smartphone is triggered, the triggering event is then classified as an earthquake event or not. Once an earthquake event is detected, the peak ground acceleration is then predicted every second until 10 s after the trigger. These predictions are made by the neural network classifier and predictor embedded in the smartphone, and an alert can be issued if a large peak ground acceleration is predicted. The proposed system is unique among approaches that use crowdsourcing ideas for earthquake early warning because the proposed system provides on-site earthquake early warning. In general, the accuracy rates of the earthquake classifications and peak ground acceleration predictions of the system were quite high according to the results of large amounts of earthquake and non-earthquake data. More specifically, according to said earthquake data, 96.9% of the issued alerts would be correct and 61.9% of the earthquakes that exceeded the threshold would have resulted in an alert being issued before the arrival of the peak ground acceleration. Among the false negative cases, approximate 97.8% would occur because of negative lead time. Using the shake table tests of worldwide and Meinong earthquake datasets, the proposed approach is confirmed to be quite promising.

Earthquake Detection in a Static and Dynamic Environment Using Supervised Machine Learning and a Novel Feature Extraction Method

Detecting earthquakes using smartphones or IoT devices in real-time is an arduous and challenging task, not only because it is constrained with the hard real-time issue but also due to the similarity of earthquake signals and the non-earthquake signals (i.e., noise or other activities). Moreover, the variety of human activities also makes it more difficult when a smartphone is used as an earthquake detecting sensor. To that end, in this article, we leverage a machine learning technique with earthquake features rather than traditional seismic methods. First, we split the detection task into two categories including static environment and dynamic environment. Then, we experimentally evaluate different features and propose the most appropriate machine learning model and features for the static environment to tackle the issue of noisy components and detect earthquakes in real-time with less false alarm rates. The experimental result of the proposed model shows promising results not only on the given dataset but also on the unseen data pointing to the generalization characteristics of the model. Finally, we demonstrate that the proposed model can be also used in the dynamic environment if it is trained with different dataset.

Seismogeodetic P-wave Amplitude: No Evidence for Strong Determinism

Whether the final properties of large earthquakes can be inferred from initial observations of rupture (deterministic rupture) is valuable for understanding earthquake source processes and is critical for operational earthquake and tsunami early warning. Initial (P-wave) characteristics of small to moderate earthquakes scale with magnitude, yet observations of large to great earthquakes saturate, resulting in magnitude underestimation. Whether saturation is inherent to earthquake dynamics or rather is due to unreliable observation of long-period signals with inertial seismic instrumentation is unclear. Seismogeodetic methods are better suited for broadband observation of large events in the near-field. In this study, we investigate the deterministic potential of seismogeodetically derived P-wave amplitude using a dataset of 14 medium-to-great earthquakes around Japan. Our results indicate that seismogeodetic P-wave amplitude is not a reliable predictor of magnitude, opposing the notion of strong determinism in the first few seconds of rupture.

Residents' Reactions to Earthquake Early Warnings in Japan

This article empirically examines the effectiveness of earthquake early warning (EEW) in Japan based on experiences of residents who received warnings before earthquake shaking occurred. In Study 1, a survey (N = 299) was conducted to investigate residents' experiences of, and reactions to, an EEW issued in Gunma and neighboring regions on June 17, 2018. The main results were as follows. (1) People's primary reactions to the EEW were mental, not physical, and thus motionless. Most residents stayed still, not for safety reasons, but because they were focusing on mentally bracing themselves. (2) Residents perceived the EEW to be effective because it enabled them to mentally prepare, rather than take physical protective actions, before strong shaking arrived. (3) In future, residents anticipate that on receipt of an EEW they would undertake mental preparation as opposed to physical protective actions. In Study 2, a survey (N = 450) was conducted on another EEW issued for an earthquake offshore of Chiba Prefecture on July 7, 2018. Results were in line with those of Study 1, suggesting that the findings described above are robust. Finally, given people's lack of impetus to undertake protective action on receipt of an EEW, this article discusses ways to enhance such actions.

Characterizing large earthquakes before rupture is complete

Whether earthquakes of different sizes are distinguishable early in their rupture process is a subject of debate. Studies have shown that the frequency content of radiated seismic energy in the first seconds of earthquakes scales with magnitude, implying determinism. Other studies have shown that recordings of ground displacement from small to moderate-sized earthquakes are indistinguishable, implying a universal early rupture process. Regardless of how earthquakes start, events of different sizes must be distinguishable at some point. If that difference occurs before the rupture duration of the smaller event, this implies some level of determinism. We show through analysis of a database of source time functions and near-source displacement records that, after an initiation phase, ruptures of M7 to M9 earthquakes organize into a slip pulse, the kinematic properties of which scale with magnitude. Hence, early in the rupture process-after about 10 s-large and very large earthquakes can be distinguished.

Crowdsourcing triggers rapid, reliable earthquake locations

In many cases, it takes several minutes after an earthquake to publish online a seismic location with confidence. Via monitoring for specific types of increased website, app, or Twitter usage, crowdsourced detection of seismic activity can be used to "seed" the search in the seismic data for an earthquake and reduce the risk of false detections, thereby accelerating the publication of locations for felt earthquakes. We demonstrate that this low-cost approach can work at the global scale to produce reliable and rapid results. The system was retroactively tested on a set of real crowdsourced detections of earthquakes made during 2016 and 2017, with 50% of successful locations found within 103 s, 76 s faster than GEOFON and 271 s faster than the European-Mediterranean Seismological Centre's publication times, and 90% of successful locations found within 54 km of the final accepted epicenter.

The Limits of Earthquake Early Warning Accuracy and Best Alerting Strategy

We explore how accurate earthquake early warning (EEW) can be, given our limited ability to forecast expected shaking even if the earthquake source is known. Because of the strong variability of ground motion metrics, such as peak ground acceleration (PGA) and peak ground velocity (PGV), we find that correct alerts (i.e., alerts that accurately estimate the ground motion will be above a predetermined damage threshold) are not expected to be the most common EEW outcome even when the earthquake magnitude and location are accurately determined. Infrequently, ground motion variability results in a user receiving a false alert because the ground motion turned out to be significantly smaller than the system expected. More commonly, users will experience missed alerts when the system does not issue an alert but the user experiences potentially damaging shaking. Despite these inherit limitations, EEW can significantly mitigate earthquake losses for false-alert-tolerant users who choose to receive alerts for expected ground motions much smaller than the level that could cause damage. Although this results in many false alerts (unnecessary alerts for earthquakes that do not produce damaging ground shaking), it minimizes the number of missed alerts and produces overall optimal performance.