Earthquake Early Warning Research Articles

Quasi-Real-Time Inversion of Large Earthquake Rupture Process Based on Earthquake Early Warning System in China: Applications to 2017 MW 6.5 Jiuzhaigou and 2022 MW 6.7 Menyuan Earthquakes

The study of seismic ruptures is crucial for understanding major earthquake events. Historically, research on large earthquake rupture processes has required a long time to be published. However, recent advancements in earthquake early warning (EEW) systems have allowed for more rapid analyses of earthquake rupture inversions. The widespread and dense distribution of EEW stations provides in China an opportunity to study and invert rupture processes in near-real time. Two notable earthquakes were studied using this technology: the 2017 Jiuzhaigou MW 6.5 earthquake in Sichuan Province, China, and the 2022 Menyuan MW 6.7 earthquake in Qinghai Province, China. In both instances, numerous strong-motion sensors captured the seismic events and transmitted waveform data to data analysis centers in real time. Following automated site selection and preparation procedures, the rupture processes of these earthquakes were analyzed and released within 30[Formula: see text]min of the event origin. This rapid response time demonstrates that the seismic rupture process determined through inversion using existing EEW systems can serve as a guide for emergency rescue work after earthquakes.

SCDetect: A SeisComP Module for Real-Time Waveform Cross-Correlation-Based Earthquake Detection

Abstract Enhanced earthquake catalogs based on waveform cross correlation (template matching) have become routine when studying regional or sequence specific seismicity. Currently, there is no existing open-source cross-correlation software that is designed to be fully integrated in real-time operations of seismic networks. To fill this gap, we introduce SCDetect, a software that implements real-time earthquake detection based on waveform cross correlation in the time domain. SCDetect with the extension module scdetect-cc is an open-source SeisComP package written in C++. scdetect-cc can be used to process both archived waveform data (playback mode), and real-time data. In the real-time application, waveforms are fetched through one of the SeisComP RecordStream interfaces, and its output (picks, origin times, amplitudes, and magnitudes) are sent to the SeisComP messaging system. The new origins are associated either with existing events detected by other pick-based SeisComP modules, or create new events. Optionally, the hypocenter location can be refined by downstream application of existing SeisComP modules. scdetect-cc offers two magnitude estimation methods that are based on the amplitudes of the template earthquakes and the new detections. In the real-time application, scdetect-cc can be scaled to handle thousands of templates without overloading the application or becoming too latent, unable to keep up with the data flow. In the playback mode, we applied scdetect-cc to three recent earthquake sequences occurring in Switzerland and surrounding regions between 2019 and 2020. Two scenarios are tested to simulate its performance in real time. The first scenario cross correlates the signals of the nearest station, whereas the second requires four stations. In both cases, we successfully detected most of the cataloged events and added hundreds of new detections. Overall, scdetect-cc is a computationally efficient and highly customizable tool to detect earthquakes for regional networks that implement the SeisComP system for earthquake monitoring.

Earthquake Early Warning Systems by Using Cloud and IoT

Earthquake early warning systems play a crucial role in minimizing the impact of seismic events on human life and infrastructure. This research paper explores the integration of Cloud Computing and Internet of Things (IoT) technologies to enhance the effectiveness of earthquake early warning systems. The proposed system aims to provide real-time data collection, analysis, and dissemination of alerts through a distributed network, leveraging the scalability and flexibility of cloud-based platforms. This paper discusses the architecture, components, and benefits of the proposed system, along with a case study to illustrate its potential in earthquake-prone regions.

Monitoring the Workflow of Earthquake by Using Machine Learning

This study explores the innovative application of machine learning techniques in monitoring and analyzing the workflow associated with earthquakes. The primary objective is to enhance the accuracy and efficiency of earthquake detection, prediction, and response mechanisms the development of machine learning technology enables more robust real-time earthquake monitoring through automated implementations. However, the application of machine learning to earthquake location problems faces challenges in regions with limited available training data. This paper proposes a machine learning-based approach to monitor seismic workflows efficiently. We apply various Machine Learning algorithms, including deep learning neural networks and decision trees, to process and interpret vast amounts of seismic data. These algorithms were trained on historical earthquake records to identify patterns and anomalies that precede seismic events. Our approach not only focused on the detection of seismic activities but also on the prediction of potential aftershock locations and magnitudes. Earthquake monitoring and response systems play a crucial role in mitigating potential damages and ensuring rapid response to seismic events. Traditional methods rely on sensor networks and manual analysis, often leading to delays in detection and response. The system leverages real-time data from seismic sensors, incorporating advanced machine learning algorithms for data analysis and pattern recognition. By employing deep learning models, such as convolution neural networks (CNNs) and recurrent neural networks (RNNs), the system can detect and classify seismic activities accurately. Additionally, anomaly detection techniques enable the identification of unusual seismic patterns that might indicate impending earthquakes or aftershocks. The system integrates geographical information systems (GIS) to map the seismic activities spatially, aiding in the visualization and understanding of seismic patterns across regions. The machine learning models are trained on historical seismic data to enhance their predictive capabilities and adaptability to varying seismic behaviors. This proposed system offers a real-time monitoring solution that can assist in early earthquake detection, accurate event classification, and timely response coordination. Its ability to continuously learn from incoming data ensures adaptability to changing seismic behaviors, thereby improving overall efficiency and reliability in earthquake monitoring workflows.

Earthquake monitoring using deep learning with a case study of the Kahramanmaras Turkey earthquake aftershock sequence

Abstract. Seismic phase picking and magnitude estimation are fundamental aspects of earthquake monitoring and seismic event analysis. Accurate phase picking allows for precise characterization of seismic wave arrivals, contributing to a better understanding of earthquake events. Likewise, accurate magnitude estimation provides crucial information about an earthquake's size and potential impact. Together, these components enhance our ability to monitor seismic activity effectively. In this study, we explore the application of deep-learning techniques for earthquake detection and magnitude estimation using continuous seismic recordings. Our approach introduces DynaPicker, which leverages dynamic convolutional neural networks to detect seismic body-wave phases in continuous seismic data. We demonstrate the effectiveness of DynaPicker using various open-source seismic datasets, including both window-format and continuous recordings. We evaluate its performance in seismic phase identification and arrival-time picking, as well as its robustness in classifying seismic phases using low-magnitude seismic data in the presence of noise. Furthermore, we integrate the phase arrival-time information into a previously published deep-learning model for magnitude estimation. We apply this workflow to continuous recordings of aftershock sequences following the Turkey earthquake. The results of this case study showcase the reliability of our approach in earthquake detection, phase picking, and magnitude estimation, contributing valuable insights to seismic event analysis.

Machine Learning-Based Rapid Epicentral Distance Estimation from a Single Station

Abstract Rapid epicentral distance estimation is of great significance for earthquake early warning (EEW). To rapidly and reliably predict epicentral distance, we developed machine learning models with multiple feature inputs for epicentral distance estimation using a single station and explored the feasibility of three machine learning methods, namely, Random Forest, eXtreme Gradient Boosting, and Support Vector Machine, for epicentral distance estimation. We used strong-motion data recorded by the Japanese Kyoshin network within a range of 1° (∼112 km) from the epicenter to train machine learning models. We used 30 features extracted from the P-wave signal as inputs to the machine learning models and the epicentral distance as the prediction target of the models. For the same test data set, within 0.1–5 s after the P-wave arrival, the epicentral distance estimation results of these three machine learning models were similar. Furthermore, these three machine learning methods can obtain smaller mean absolute errors and root mean square errors, as well as larger coefficients of determination (R2), for epicentral distance estimation than traditional EEW epicentral distance estimation methods, indicating that these three machine learning models can effectively improve the accuracy of epicentral distance estimation to a certain extent. In addition, we analyzed the importance of different features as inputs to machine learning models using SHapley additive exPlanations. We found that using the top 15 important features as inputs, these three machine learning models can also achieve good results for epicentral distance estimation. Based on our results, we inferred that the machine learning models for estimating epicentral distance proposed in this study are meaningful in EEW.

Incorporating Intensity Distance Attenuation Into PLUM Ground‐Motion‐Based Earthquake Early Warning in the United States: The APPLES Configuration

AbstractWe develop Attenuated ProPagation of Local Earthquake Shaking (APPLES), a new configuration for the United States West Coast version of the Propagation of Local Undamped Motion (PLUM) earthquake early warning (EEW) algorithm that incorporates attenuation into its ground‐motion prediction procedures. Under APPLES, instead of using a fixed radius to forward‐predict observed peak ground shaking to the area surrounding a seismic station, the forward‐predicted intensity at a location depends on the distance from the station using an intensity prediction relationship. We conduct conceptual tests of maximum intensity distribution predictions in APPLES and PLUM using a catalog of ShakeMaps to confirm that the attenuation relationship in APPLES is appropriately modeling shaking distributions for West Coast earthquakes. Then, we run APPLES and PLUM in simulated real‐time tests to determine warning time performance. Finally, we compare real‐time alert behavior during the 2022 M6.4 Ferndale, California, earthquake and other recent events. We find that APPLES presents two potential improvements to PLUM by reducing over‐alerting during smaller magnitude earthquakes and by increasing warning times in some locations during larger earthquakes. APPLES can produce missed and late alerts in locations that experience shaking intensities close to the level used to issue alerts, so preferred alerting strategies with APPLES would use alert thresholds that are lower than the intensities targeted for EEW alerts. We find alerts using APPLES are also similar to those for the source‐based approaches currently used in the ShakeAlert EEW system, which will make APPLES easier to integrate into the system.

Event Detection Techniques and Optimization for Earthquake Early Warning Application

An efficient, robust, reliable Earthquake Early Warning (EEW) can play a vital role in hazard mitigation by reducing false and missed alarms. This study represents a comparative analysis of state-of-the-art real-time event detection techniques for the EEW. In real-time event detection, a trade-off between detection reliability and delay in detection exists dominantly. This trade-off is the motivation behind the implementation and performance optimization of event detection techniques, namely Short-time-average over Long-time-average (STA/LTA), Multi-window Algorithm (MWA), and Modified Energy Ratio (MER) for EEW applicability. Earthquake data from the Kyoshin Network (K-NET), Japan along with noise events recorded by the Seismic Sensing Nodes (SSN) at the NCR, India is used for analysis and validation of detection techniques for the EEW system. The critical parameters such as signal window duration and threshold for event declaration are optimized for the detection techniques. An introspection of their performance measures in terms of delay in detection and detection accuracy for optimized parameters is presented. The median delay in detection is best for STA/LTA (0.53[Formula: see text]s) followed by MER (0.55[Formula: see text]s) and MWA (3.23[Formula: see text]s). The comparative analysis shows that MER provides the best results among the candidate techniques with a normalized balanced accuracy of 0.98 followed by STA/LTA (0.97) and MWA (0.82).

Monitoring the workflow of Earthquake by using Machine Learning

The main goal is to improve the accuracy and efficiency of earthquake detection, prediction and response through automation and, over time, to improve machine learning techniques for earthquake monitoring. However, the application of machine learning to earthquake location problems faces challenges in areas with limited data. This article presents an engine-based approach to monitor earthquake performance. We use a variety of Machine Learning algorithms, including deep learning neural networks and decision trees, for processing and interpret large amounts of seismic data. These algorithms are trained on historical earthquake records to identify patterns and anomalies prior to seismic events. Our approach focuses not only on detecting seismic events, but also on predicting the location and magnitude of earthquakes. Earthquake monitoring and response systems play an important role in reducing potential losses and providing rapid response to seismic events. Conventional methods rely on sensor networks and manual analysis, often with delays in detection and response. The system receives real-time data from seismic sensors, incorporating advanced machine learning algorithms for data analysis and pattern recognition. The system can detect and describe seismic events using deep learning models such as convolutional neural networks (CNN) and random neural networks (RNN).In addition, anomaly detection techniques allow the identification of unusual seismic patterns that may indicate impending earthquakes or earthquakes. The system integrates geographic information systems (GIS) to comprehensively map seismic activity, helping to visualize and understand seismic patterns in the region. Machine learning models are trained on historical seismic data to improve their predictive capabilities and adapt to different seismic behaviors. This proposed system provides a real-time monitoring solution that can assist in early earthquake detection, accurate event classification, and timely response coordination. The ability to continuously learn from incoming data enables adaptation to change in seismic activity, thereby increasing the efficiency and reliability of earthquake monitoring operations.

TXED: The Texas Earthquake Dataset for AI

Abstract Machine-learning (ML) seismology relies on large datasets with high-fidelity labels from humans to train generalized models. Among the seismological applications of ML, earthquake detection, and P- and S-wave arrival picking are the most widely studied, with capabilities that can exceed humans. Here, we present a regional artificial intelligence (AI) earthquake dataset (TXED) compiled for the state of Texas. The TXED dataset is composed of earthquake signals with manually picked P- and S-wave arrival times and manually picked noise waveforms corresponding to more than 20,000 earthquake events spanning from the beginning of the Texas seismological network (TexNet) (1 January 2017) to date. These data are a supplement to the existing worldwide open-access seismological AI datasets and represent the signal and noise characteristics of Texas. Direct applications of the TXED datasets include improving the performance of a global picking model in Texas by transfer learning using the new dataset. This dataset will also serve as a benchmark dataset for fundamental AI research like designing seismology-oriented deep-learning architectures. We plan to continue to expand the TXED dataset as more observations are made by TexNet analysts.

Rapid report of the December 18, 2023 MS 6.2 Jishishan earthquake, Gansu, China

On December 18, 2023, the Jishishan area in Gansu Province was jolted by a MS 6.2 earthquake, which is the most powerful seismic event that occurred throughout the year in China. The earthquake occurred along the NW-trending Lajishan fault (LJSF), a large tectonic transformation zone. After this event, China Earthquake Networks Center (CENC) has timely published several reports about seismic sources for emergency responses. The earthquake early warning system issued the first alert 4.9 ​s after the earthquake occurrence, providing prompt notification that effectively mitigated panics, injuries, and deaths of residents. The near real-time focal mechanism solution indicates that this earthquake is associated with a thrust fault. The distribution of aftershocks, the rupture process, and the recorded amplitudes from seismic monitoring and GNSS stations, all suggest that the mainshock rupture predominately propagates to the northwest direction. The duration of the rupture process is ∼12 ​s, and the largest slip is located at approximately 6.3 ​km to the NNW from the epicenter, with a peak slip of 0.12 ​m at ∼8 ​km depth. Seismic station N0028 recorded the highest instrumental intensity, which is 9.4 on the Mercalli scale. The estimated intensity map shows a seismic intensity reaching up to IX near the rupture area, consistent with field survey results. The aftershocks (up to December 22, 2023) are mostly distributed in the northwest direction within ∼20 ​km of the epicenter. This earthquake caused serious casualties and house collapses, which requires further investigations into the impact of this earthquake.

Unsupervised anomaly detection for earthquake detection on Korea high-speed trains using autoencoder-based deep learning models

We propose a method for detecting earthquakes for high-speed trains based on unsupervised anomaly-detection techniques. In particular, we utilized autoencoder-based deep learning models for unsupervised learning using only normal training vibration data. Datasets were generated from South Korean high-speed train data, and seismic data were measured using seismometers nationwide. The proposed method is compared with the conventional Short Time Average over Long Time Average (STA/LTA) model, considering earthquake detection capabilities, focusing on a Peak Ground Acceleration (PGA) threshold of 0.07, a criterion for track derailment. The results show that the proposed model exhibit improved earthquake detection capabilities than STA/LTA for PGA of 0.07 or higher. Furthermore, the proposed model reduced false earthquake detections under normal operating conditions and accurately identified normal states. In contrast, the STA/LTA method demonstrated a high rate of false earthquake detection under normal operating conditions, underscoring its propensity for inaccurate detection in many instances. The proposed approach shows promising performance even in situations with limited seismic data and offers a viable solution for earthquake detection in regions with relatively few seismic events.

MACHINE LEARNING FOR FAST AND RELIABLE SOURCE-LOCATION ESTIMATION IN EARTHQUAKE EARLY WARNING

We develop a random forest (RF) model for rapid earthquake location with an aim to assist earthquake early warning (EEW) systems in fast decision making. This system exploits Pwave arrival times at the first five stations recording an earthquake and computes their respective arrival time differences relative to a reference station (i.e., the first recording station). These differential P-wave arrival times and station locations are classified in the RF model to estimate the epicentral location. We train and test the proposed algorithm with an earthquake catalog from Japan. The RF model predicts the earthquake locations with a high accuracy, achieving a Mean Absolute Error (MAE) of 2.88 km. As importantly, the proposed RF model can learn from a limited amount of data (i.e., 10% of the dataset) and much fewer (i.e., three) recording stations and still achieve satisfactory results (MAE<5 km). The algorithm is accurate, generalizable, and rapidly responding, thereby offering a powerful new tool for fast and reliable source-location prediction in EEW.

SeisParaNet: A Novel Multitask Network for Seismic Source Characterization in Earthquake Early Warning

SeisParaNet: A Novel Multitask Network for Seismic Source Characterization in Earthquake Early Warning

Application of Machine Learning to Determine Earthquake Hypocenter Location in Earthquake Early Warning

Application of Machine Learning to Determine Earthquake Hypocenter Location in Earthquake Early Warning

Academics came together in Üsküdar for the Earthquake Early Warning System

Academics came together in Üsküdar for the Earthquake Early Warning System

Real-Time Seismic Intensity Prediction Using Self-Supervised Contrastive GNN for Earthquake Early Warning

Real-Time Seismic Intensity Prediction Using Self-Supervised Contrastive GNN for Earthquake Early Warning

Potential of Earthquake Strong Motion Observation Utilizing a Linear Estimation Method for Phase Cycle Skipping in Distributed Acoustic Sensing

AbstractDistributed 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.

Analysis of Strong Ground Motion Characteristics for the Jishishan MS6.2 Earthquake December 18, 2023

The MS6.2 earthquake occurred in Jishishan, Gansu Province, China, on December 18, 2023. Based on some strong seismic acceleration time histories captured by some stations of China National Strong Earthquake Observation Network and National Earthquake Early Warning Network, the spatial distribution characteristics, rupture directivity effect, hanging wall and footwall effect, response spectrum characteristics, and attenuation of the peak ground acceleration (PGA) were analyzed in this paper. It can be used as reference for the site determination of post‐earthquake reconstruction and seismic design of engineering structures. The analysis shows that the spatial distribution of strong ground motion of the Jishishan MS6.2 earthquake is mainly controlled by the distribution of causative fault. Near the fault, the effect of rupture directivity is obvious. The effect of hanging wall and footwall significantly affects the distribution of PGAs for this earthquake, causing a pronounced asymmetric pattern about PGAs relative to the fault, with slower attenuation on the hanging wall compared to the footwall. The type of site obviously affects the characteristics of response spectrum. The ground motion intensity observed at the epicenter of this earthquake significantly exceeded the levels of rare and extremely rare seismic events in the current seismic design codes. The attenuation relationship of PGA obtained by regression analysis shows that ones in the 5th generation Chinese ground motion parameter zoning map for the Qinghai‐Tibet region underestimate the horizontal PGA.

UNSUPERVISED MACHINE LEARNING FOR SEISMIC ANOMALY DETECTION: ISOLATION FOREST ALGORITHM APPLICATION TO INDONESIAN EARTHQUAKE DATA

Indonesia's position on the seismically active Pacific "Ring of Fire" necessitates advanced methods for earthquake detection and preparedness. This paper introduces an application of the Isolation Forest algorithm—an unsupervised machine learning technique—for detecting seismic anomalies in Indonesia's complex geotectonic landscape. Unlike traditional methods that rely on predefined thresholds or patterns, the Isolation Forest algorithm isolates anomalies based on their rarity and distinctness without the need for labeled data. We applied this algorithm to a comprehensive dataset from the Indonesian Meteorology, Climatology, and Geophysical (BMKG) Agency, featuring a decade's worth of seismic events, to identify outliers that may signify potential seismic hazards. Our findings reveal that the Isolation Forest algorithm effectively identifies seismic anomalies, with 874 out of tens of thousands of events flagged as statistically significant outliers. A comparative analysis with traditional seismic anomaly detection models highlights the robustness of the Isolation Forest algorithm in handling the high-dimensional and noisy nature of seismic data, emphasizing its superiority in detecting subtle anomalies