Toward False Event Detection and Quarry Blast versus Earthquake Discrimination in an Operational Setting Using Semiautomated Machine Learning

Abstract

Small-magnitude earthquakes shed light on the spatial and magnitude distribution of natural seismicity, as well as its rate and occurrence, especially in stable continental regions where natural seismicity remains difficult to explain under slow strain-rate conditions. However, capturing them in catalogs is strongly hindered by signal-to-noise ratio issues, resulting in high rates of false and man-made events also being detected. Accurate and robust discrimination of these events is critical for optimally detecting small earthquakes. This requires uncovering recurrent salient features that can rapidly distinguish first false events from real events, then earthquakes from man-made events (mainly quarry blasts), despite high signal variability and noise content. In this study, we combined the complementary strengths of human and interpretable rule-based machine-learning algorithms for solving this classification problem. We used human expert knowledge to co-create two reliable machine-learning classifiers through human-assisted selection of classification features and review of events with uncertain classifier predictions. The two classifiers are integrated into the SeisComP3 operational monitoring system. The first one discards false events from the set of events obtained with a low short-term average/long-term average threshold; the second one labels the remaining events as either earthquakes or quarry blasts. When run in an operational setting, the first classifier correctly detected more than 99% of false events and just over 93% of earthquakes; the second classifier correctly labeled 95% of quarry blasts and 96% of earthquakes. After a manual review of the second classifier low-confidence outputs, the final catalog contained fewer than 2% of misclassified events. These results confirm that machine learning strengthens the quality of earthquake catalogs and that the performance of machine-learning classifiers can be improved through human expertise. Our study promotes a broader implication of hybrid intelligence monitoring within seismological observatories.

Advancing Local Distance Discrimination of Explosions and Earthquakes With Joint P/S and ML‐MC Classification

“…Joint discrimination could also benefit future development of new discrimination methods, including machine‐learning based techniques. The success of our application offers unique insight into the “black‐box” classification resulting from machine‐learning‐based discrimination algorithms by highlighting specific parts of the wavefield that are diagnostic of different source types (e.g., Kortström et al., 2016; Linville et al., 2019; Renouard et al., 2021). The high‐transportability of our method may also mitigate potential challenges with network‐based discriminations, where location pattern recognition or site‐specific source effects could unintentionally dominate source type determination.…”

“…

Accurate classification of explosive seismic sources and earthquakes is a key task for agencies monitoring compliance with nuclear test ban treaties (e.g., Bowers & Selby, 2009) and for hazard mitigation agencies monitoring earthquake activity (e.g., Renouard et al, 2021). Seismic discrimination of underground explosions has been successful for larger magnitude events that are typically observed at teleseismic distances (i.e., >3,000 km), however discrimination at local distances (<200 km) remains challenging.

…”

New CNN-based tool to discriminate anthropogenic from natural low magnitude seismic events

“…The use of features extracted from the seismic signal is a widely used technique in machine learning. Renouard et al (2021), for instance, use an hybrid approach including both human expertise and machine learning algorithms to achieve this objective. They trained a Random Forest algorithm by extracting selected features from the seismic signals and by implementing them in a decision tree.…”

“…To overcome these problems, such as finding relevant features to classify the data, machine learning tools can be implemented. In the last few years, these approaches were widely used for the detection of events (Yoon et al 2015;Ross et al 2018;Mousavi et al 2020), automatic picking of seismic phase on raw data (Pardo et al 2019;Woollam et al 2019;Zhu & Beroza 2019), but also for signal classification using features extracted directly from the data (Del Pezzo et al 2003;Meier et al 2019;Renouard et al 2021), or relying on deep neural networks (Li et al 2018;Linville et al 2019). In the latter, training algorithms with a sufficiently large database allow thoses programs to recognize, like humans, natural objects and to make expert-level decisions.…”

Unsupervised Clustering of Continuous Ambient Noise Data to Get Higher Signal Quality in Seismic Surveys

“…In this study we show that it is possible to identify noise windows in which body wave energy dominates with the aid of machine learning methods, speci cally unsupervised clustering. (Shen & Shen, 2021) and event discrimination (Renouard et al, 2021;Linville et al, 2019). However, most of the above-mentioned studies can be classi ed as "supervised learning" approaches because they require a training dataset and information about the data beforehand.…”