A Synchronous Magnitude Estimation with P-Wave Phases' Detection Used in Earthquake Early Warning System

Dingwen Zhang, Jihua Fu, Zhitao Li, Linyue Wang, Jiale Li, Jianjun Wang

TL;DR

This paper introduces a novel method for estimating earthquake magnitudes synchronously with P-wave detection, aiming to improve the speed and accuracy of earthquake early warning systems (EEWS). By leveraging the signal-to-noise ratio (SNR) derived from the Short-Term Power to Long-Term Power (STP/LTP) ratio, the proposed PSNR method achieves rapid magnitude estimation within seconds of P-wave arrival, outperforming traditional methods in timeliness.

What Problem Does This Paper Address?

Earthquake Early Warning Systems (EEWS) are crucial for mitigating earthquake damage, but their effectiveness hinges on rapid and accurate magnitude estimation. A fundamental challenge is the trade-off between timeliness and accuracy: faster estimations often sacrifice precision, while more accurate methods require more data and time. Existing methods for magnitude estimation typically occur several seconds after the initial P-wave detection, creating a delay. This paper addresses this gap by proposing a method that estimates magnitude simultaneously with P-wave detection, aiming to bridge the timeliness-accuracy gap and enhance EEWS performance.

Methods

The core of this paper is the development of the PSNR (Peak Signal-to-Noise Ratio) method for magnitude estimation. This method builds upon the STP/LTP (Short-Term Power / Long-Term Power) technique, which is used for detecting P-wave arrivals.

Here's a breakdown of the approach:

  • P-wave Detection (STP/LTP): The paper utilizes the STP/LTP method to detect P-wave arrivals. This method calculates the ratio of short-term power (representing potential seismic signal) to long-term power (representing background noise) in seismic records. When this ratio exceeds a predefined threshold, a P-wave arrival is identified.
  • Magnitude Estimation (PSNR): Instead of using traditional parameters like peak displacement (Pd), the PSNR method uses the maximum STP/LTP value observed during the P-wave arrival as a proxy for seismic energy. This maximum STP/LTP value is termed PSNR.
  • Magnitude Model: A regression model is established relating the earthquake magnitude (M), the derived PSNR value, and the hypocentral distance (R). This model takes the form: $M = A \times \log(\text{PSNR}) + B \times \log(R) + C$.
  • Synchronous Estimation: The key innovation is that the PSNR value is determined immediately after the P-wave arrival is detected by the STP/LTP method. This allows for magnitude estimation to occur concurrently with P-wave identification, significantly reducing the overall time lag.

The paper also discusses the underlying principles of STP/LTP, which involves calculating characteristic functions based on the variance of seismic signals within short and long time windows.

Datasets

The study utilizes seismic data from Sichuan Province, China, a region known for frequent seismic activity.

  • Model Fitting and Performance Analysis: 75 earthquake records were used, including events from the Wenchuan M8.0 earthquake (2008) and other significant events like Jiuzhaigou M7.0 (2017) and Luxian M6.0 (2021). The dataset includes a range of magnitudes, from M3.0 to M8.0.
  • Model Testing: Eight additional earthquake events were selected for testing the developed model.
  • Small Earthquake Data: Data for small earthquakes (3.0 < M < 4.7) were provided by the National Key Research and Development Program of China, recorded by low-cost MEMS seismic sensors (MNSMS) at a 50 Hz sampling rate.
  • Large Earthquake Data: Data for large earthquakes (M = 6, 7, 8) were provided by the Institute of Engineering Mechanics, China Earthquake Administration, with a sampling rate of 200 Hz.

The paper also notes that for data selection, stations were chosen based on hypocentral distance: less than 50 km for M < 5.0 earthquakes and less than 100 km for M > 5.0 earthquakes.

Key Insights and Ideas

  • Synchronicity is Key: The most significant insight is that magnitude estimation can be effectively performed at the same time as P-wave detection. This fundamentally changes the paradigm for EEWS, moving from sequential processing to parallel processing.
  • PSNR as a Robust Energy Proxy: The PSNR, derived from the STP/LTP ratio, proves to be a robust indicator of seismic energy at the initial P-wave arrival, even in noisy environments. This is a departure from traditional methods relying on peak displacement, which can be more sensitive to noise and signal processing choices.
  • Timeliness Advantage: The PSNR method consistently yields magnitude estimates within 0-0.3 seconds of P-wave arrival, and in some cases, almost simultaneously. This is a dramatic improvement over methods that require several seconds post-P-wave arrival.
  • Dimensionless Parameter: PSNR is a dimensionless parameter, making it adaptable to different sensor types and specifications without requiring complex unit conversions, simplifying data processing and reducing costs.
  • Saturation Phenomenon: The paper acknowledges that PSNR, like other magnitude estimation methods, may exhibit a saturation phenomenon for very large earthquakes, though further study is needed to precisely define this limit.

Key Results

  • Improved Timeliness: The PSNR method achieves magnitude estimation within 2 seconds of P-wave arrival, with most estimates occurring within 0-0.3 seconds. This is significantly faster than traditional methods.
  • Accuracy: The PSNR method demonstrates high accuracy, with a fitting correlation coefficient of 0.8012 for the magnitude estimation model, which is slightly higher than the Pd method (0.7552). The variance of error for PSNR (0.3308) is also lower than for Pd (0.4074).
  • Robustness: The PSNR method inherits the robustness of the STP/LTP method, making it less susceptible to noise and instrument background variations.
  • Model Performance: Table 2 shows that the PSNR method can estimate magnitudes for events with as few as 2-3 stations, with estimated magnitudes (Me) closely matching actual magnitudes (Ma). For the Xingwen M5.7 event, the estimated magnitude converges to the actual magnitude as the number of contributing stations increases.

Limitations and Future Work

  • Saturation: The paper notes that PSNR may exhibit a saturation phenomenon for very large earthquakes, similar to other methods. Further research is needed to determine the specific magnitude at which this saturation occurs.
  • P-wave Detection Accuracy: The accuracy of the PSNR method is highly dependent on the accurate detection of P-wave arrivals. Errors in seismic phase identification could lead to false alarms.
  • Geological Specificity: The authors suggest that local historical records should be used for model training to improve accuracy and reduce false alarm rates, as geological structures and crustal movements vary by region.
  • Sensor Performance: The accuracy and stability of the PSNR value and the fitting model are directly influenced by the performance and stability of the seismic sensors.
  • Future Work: The paper implies that further refinement of the PSNR model and its application across diverse geological settings would be beneficial. The development of more robust P-wave detection algorithms would also enhance the PSNR method's reliability.

Summary updated

Abstract

How to estimate an earthquake’s magnitude rapidly and accurately is a challenge for any earthquake early warning system. In order to reach a balance between accuracy and timeliness, a synchronous magnitude estimation method with P-wave phases’ detection is proposed. In this method, the P-wave phases are detected by the changes of the signal-to-noise ratio (SNR) of the seismic records, where the SNRs are calculated by the short-term power and long-term power ratio (STP/LTP). Meanwhile, the variations of the SNR are applied to estimate the magnitude of the earthquake. By the statistics of some earthquake cases, a synchronous magnitude estimation model of the variation of the P-wave phases’ SNR, the earthquake magnitude, and the hypocentral distance was built. Compared with some other magnitude estimation methods, the suggested method inherits the robustness of the STP/LTP method and is more accurate and rapid than the peak displacement (Pd) method.
BibTeX
@inproceedings{zhang2022synchronous,
  author = {Dingwen Zhang and Jihua Fu and Zhitao Li and Linyue Wang and Jiale Li and Jianjun Wang},
  title = {{A Synchronous Magnitude Estimation with P-Wave Phases' Detection Used in Earthquake Early Warning System}},
  booktitle = {Sensors},
  year = {2022},
  url = {https://openreview.net/forum?id=yvNLWiQYmur},
}
Figure