Although infrequent, large (Mw7.5+) earthquakes can be extremely damaging and occur on subduction and intraplate faults worldwide. Earthquake early warning (EEW) systems aim to provide advanced warning before strong shaking and tsunami onsets. These systems estimate earthquake magnitude using the early metrics of waveforms, relying on empirical scaling relationships of abundant past events. However, both the rarity and complexity of great events make it challenging to characterize them, and EEW algorithms often underpredict magnitude and the resulting hazards. Here, we propose a model, M‐LARGE, that leverages deep learning to characterize crustal deformation patterns of large earthquakes for a specific region in real‐time. We demonstrate the algorithm in the Chilean Subduction Zone by training it with more than six million different simulated rupture scenarios recorded on the Chilean GNSS network. M‐LARGE performs reliable magnitude estimation on the testing data set with an accuracy of 99%. Furthermore, the model successfully predicts the magnitude of five real Chilean earthquakes that occurred in the last 11 years. These events were damaging, large enough to be recorded by the modern high rate global navigation satellite system instrument, and provide valuable ground truth. M‐LARGE tracks the evolution of the source process and can make faster and more accurate magnitude estimation, significantly outperforming other similar EEW algorithms. This is the first demonstration of our approach. Future work toward generalization is outstanding and will include the addition of more training and testing data, interfacing with existing EEW methods, and applying the method to different tectonic settings to explore performance in these regions.
This content will become publicly available on December 8, 2024
Earthquake early warning (EEW) systems aim to forecast the shaking intensity rapidly after an earthquake occurs and send warnings to affected areas before the onset of strong shaking. The system relies on rapid and accurate estimation of earthquake source parameters. However, it is known that source estimation for large ruptures in real‐time is challenging, and it often leads to magnitude underestimation. In a previous study, we showed that machine learning, HR‐GNSS, and realistic rupture synthetics can be used to reliably predict earthquake magnitude. This model, called Machine‐Learning Assessed Rapid Geodetic Earthquake model (M‐LARGE), can rapidly forecast large earthquake magnitudes with an accuracy of 99%. Here, we expand M‐LARGE to predict centroid location and fault size, enabling the construction of the fault rupture extent for forecasting shaking intensity using existing ground motion models. We test our model in the Chilean Subduction Zone with thousands of simulated and five real large earthquakes. The result achieves an average warning time of 40.5 s for shaking intensity MMI4+, surpassing the 34 s obtained by a similar GNSS EEW model. Our approach addresses a critical gap in existing EEW systems for large earthquakes by demonstrating real‐time fault tracking feasibility without saturation issues. This capability leads to timely and accurate ground motion forecasts and can support other methods, enhancing the overall effectiveness of EEW systems. Additionally, the ability to predict source parameters for real Chilean earthquakes implies that synthetic data, governed by our understanding of earthquake scaling, is consistent with the actual rupture processes.
more » « less- Award ID(s):
- 1835661
- NSF-PAR ID:
- 10479560
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Journal of Geophysical Research: Solid Earth
- Volume:
- 128
- Issue:
- 12
- ISSN:
- 2169-9313
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
Abstract Rapid earthquake magnitude estimation from real-time space-based geodetic observation streams provides an opportunity to mitigate the impact of large and potentially damaging earthquakes by issuing low-latency warnings prior to any significant and destructive shaking. Geodetic contributions to earthquake characterization and rapid magnitude estimation have evolved in the last 20 yr, from post-processed seismic waveforms to, more recently, improved capacity of regional geodetic networks enabled real-time Global Navigation Satellite System seismology using precise point positioning (PPP) displacement estimates. In addition, empirical scaling laws relating earthquake magnitude to peak ground displacement (PGD) at a given hypocentral distance have proven effective in rapid earthquake magnitude estimation, with an emphasis on performance in earthquakes larger than ∼Mw 6.5 in which near-field seismometers generally saturate. Although the primary geodetic contributions to date in earthquake early warning have focused on the use of 3D position estimates and displacements, concurrent efforts in time-differenced carrier phase (TDCP)-derived velocity estimates also have demonstrated that this methodology has utility, including similarly derived empirical scaling relationships. This study builds upon previous efforts in quantifying the ambient noise of three-component ground-displacement and ground-velocity estimates. We relate these noise thresholds to expected signals based on published scaling laws. Finally, we compare the performance of PPP-derived PGD to TDCP-derived peak ground velocity (PGV), given several rich event datasets. Our results indicate that TDCP-PGV is more likely than PPP-PGD to detect intermediate magnitude (∼Mw 5.0–6.0) earthquakes, albeit with greater magnitude estimate uncertainty and across smaller epicentral distances. We conclude that the computationally lightweight TDCP-derived PGV magnitude estimation is complementary to PPP-derived PGD magnitude estimates, which could be produced at the network edge at high rates and with increased sensitivity to ground motion than current PPP estimates.more » « less
-
Our research aims to improve the accuracy of Earthquake Early Warning (EEW) systems by means of machine learning. EEW systems are designed to detect and characterize medium and large earthquakes before their damaging effects reach a certain location. Traditional EEW methods based on seismometers fail to accurately identify large earthquakes due to their sensitivity to the ground motion velocity. The recently introduced high-precision GPS stations, on the other hand, are ineffective to identify medium earthquakes due to its propensity to produce noisy data. In addition, GPS stations and seismometers may be deployed in large numbers across different locations and may produce a significant volume of data consequently, affecting the response time and the robustness of EEW systems.In practice, EEW can be seen as a typical classification problem in the machine learning field: multi-sensor data are given in input, and earthquake severity is the classification result. In this paper, we introduce the Distributed Multi-Sensor Earthquake Early Warning (DMSEEW) system, a novel machine learning-based approach that combines data from both types of sensors (GPS stations and seismometers) to detect medium and large earthquakes. DMSEEW is based on a new stacking ensemble method which has been evaluated on a real-world dataset validated with geoscientists. The system builds on a geographically distributed infrastructure, ensuring an efficient computation in terms of response time and robustness to partial infrastructure failures. Our experiments show that DMSEEW is more accurate than the traditional seismometer-only approach and the combined-sensors (GPS and seismometers) approach that adopts the rule of relative strength.more » « less
-
ABSTRACT We explore the response of ground motions to topography during large crustal fault earthquakes by simulating several magnitude 6.5–7.0 rupture scenarios on the Seattle fault, Washington State. Kinematic simulations are run using a 3D spectral element code and a detailed seismic velocity model for the Puget Sound region. This model includes realistic surface topography and a near-surface low-velocity layer; a mesh spacing of ∼30 m at the surface allows modeling of ground motions up to 3 Hz. We simulate 20 earthquake scenarios using different slip distributions and hypocenter locations on a planar fault surface. Results indicate that average ground motions in simulations with and without topography are similar. However, shaking amplification is common at topographic highs, and more than a quarter of all sites experience short-period (≤2 s) ground-motion amplification greater than 25%–35%, compared with models without topography. Comparisons of peak ground velocity at the top and bottom of topographic features demonstrate that amplification is sensitive to period, with the greatest amplifications typically manifesting near a topographic feature’s estimated resonance frequency and along azimuths perpendicular to its primary axis of elongation. However, interevent variability in topographic response can be significant, particularly at shorter periods (<1 s). We do not observe a clear relationship between source centroid-to-site azimuths and the strength of topographic amplification. Overall, our results suggest that although topographic resonance does influence the average ground motions, other processes (e.g., localized focusing and scattering) also play a significant role in determining topographic response. However, the amount of consistent, significant amplification due to topography suggests that topographic effects should likely be considered in some capacity during seismic hazard studies.more » « less
-
more » « less
Observations of strong ground motion during large earthquakes are generally made with strong-motion accelerometers. These observations have a critical role in early warning systems, seismic engineering, source physics studies, basin and site amplification, and macroseismic intensity estimation. In this manuscript, we present a new observation of strong ground motion made with very high rate (>= 5 Hz) Global Navigation Satellite System (GNSS) derived velocities. We demonstrate that velocity observations recorded on GNSS instruments are consistent with existing ground motion models and macroseismic intensity observations. We find that the ground motion predictions using existing NGA-West2 models match our observed peak ground velocities with a median log total residual of 0.03-0.33 and standard deviation of 0.72-0.79, and are statistically significant following normality testing. We finish by deriving a Ground Motion Model for peak ground velocity from GNSS and find a total residual standard deviation 0.58, which can be improved by ~2% when considering a simple correction for Vs30.
