Generalized nearest-neighbor distance and Hawkes point process modeling applied to mining induced seismicity DOI: 00.0000/000000000

Mohammadamin Sedghizadeh, Robert Shcherbakov, Matthew van den Berghe · 2025

Read Generalized nearest-neighbor distance and Hawkes point process modeling applied to mining induced seismicity DOI: 00.0000/000000000 by Mohammadamin Sedghizadeh, Robert Shcherbakov, Matthew van den Berghe (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-00768.txt.

Introduction

Mohammadamin Sedghizadeh, Robert Shcherbakov, and Matthew van den Berghe present a 2025 arXiv paper on modeling tremors set off by mining. Mine operators need clustering and rate models to judge hazard. The team generalizes the nearest-neighbor distance method, which works in a rescaled space of position, time, and size. Their version lets the size statistics follow a flexible two-part tapered Pareto law instead of the standard scaling.

A Hawkes point process model of the rate at a Saskatchewan potash mine found the shaking driven mostly by outside forces, with little quake-to-quake triggering. The authors discuss what the clustering pattern says about induced seismicity underground.

Catalog record

CREATORMohammadamin Sedghizadeh, Robert Shcherbakov, Matthew van den Berghe
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-00768

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Generalized nearest-neighbor distance and Hawkes point process modeling applied to mining induced seismicity DOI: 00.0000/000000000 Mohammadamin Sedghizadeh, Robert Shcherbakov, Matthew van den Berghe Abstract Modeling seismic activity rates and clustering plays an important role in studies of induced seismicity associated with mining and other resource extraction operations. This is critical for understanding the physical and statistical characteristics of seismicity and assessing the associated hazard. In this work, we introduce the generalization of the Nearest-Neighbor Distance (NND) method by incorporating an arbitrary distribution function for the frequency-magnitude statistics of seismic events. Operating within a rescaled hyperspace that includes spatial, temporal, and magnitude domains, the NND method provides an effective framework for examining seismic clustering. By integrating a mixture of the two tapered Pareto distributions, the generalized NND approach accommodates deviations from standard frequency-magnitude scaling when studying the clustering properties of seismicity. In addition, the application of the temporal Hawkes process to model the mining seismicity rate reveals that the seismicity is primarily driven by external factors and lacks pronounced interevent triggering. A case study from a potash mine in Saskatchewan is presented to illustrate the application of the generalized NND method and the Hawkes process to estimate the clustering properties and occurrence rates of induced microseismicity. The implications of observed temporal variations and clustering behavior are discussed, providing insights into the nature of induced seismicity within mining environments. \citethisauthor Sedghizadeh, M., R. Shcherbakov, and M. van den Berghe \vol 0 \iss 0 \recdate 00 Month 0000 1]Department of Earth Sciences, University of Western Ontario, London, ON, Canada, N6A 5B7. \auorc [https://orcid.org/]0000-0003-3713-3782(MS) 2]Department of Physics and Astronomy, University of Western Ontario, London, ON, Canada, N6A 3K7. \auorc [https://orcid.org/]0000-0002-3057-0851(RS) 3]Department of GeoServices, Nutrien Ltd., Saskatoon, SK, Canada, S7K 5R6. \corau *Corresponding author: [email protected] 1 Introduction Induced seismicity, the occurrence of seismic events that typically accompany energy/mineral resource extraction operations such as mining, hydraulic fracturing and waste water injection, reservoir impoundment, and others, has become a critical area of research due to its implications for operational safety, infrastructure integrity, and seismic hazard assessment ( Ellsworth 2013 ; Grigoli et al. 2017 ; Atkinson et al. 2020 ; Schultz et al. 2020 ) . These seismic events, while typically of lower average magnitudes than natural earthquakes, often exhibit a variety of spatial, temporal, and magnitude patterns that are influenced by anthropogenic factors rather than tectonic plate interactions ( Atkinson et al. 2016 ; Keranen and Weingarten 2018 ) . For example, mining-induced seismicity is frequently caused by the redistribution of stresses within rock masses as a result of excavation and ore extraction ( Gendzwill and Stead 1992 ; Gibowicz and Lasocki 2001 ; Lasocki and Orlecka-Sikora 2008 ; Gibowicz 2009 ) . Understanding the underlying mechanisms and characteristics of induced seismicity is essential for mitigating associated risks, improving forecasting models, and developing effective hazard assessment frameworks ( van der Elst et al. 2016 ; Shcherbakov et al. 2010 ; Shcherbakov et al. 2015 ; Langenbruch et al. 2018 ; Schultz et al. 2020 ; Zhou et al. 2024 ) . To accomplish these tasks the appropriate statistical models need to be formulated and used. In statistical seismology, the analysis of natural seismicity is typically based on the application of several well established statistical models and approaches. This includes the application of the Gutenberg-Richter (GR) law to model the frequency-magnitude statistics of earthquakes ( Gutenberg and Richter 1944 ) . In many instances, the analysis of the seismicity rates is performed using the temporal or spatio-temporal Epidemic Type Aftershock Sequence (ETAS) model ( Ogata 1988 ; Ogata 1998 ; Ogata 2024 ) . The clustering aspects of seismicity can be studied by using the Nearest-Neighbor Distance (NND) analysis ( Baiesi and Paczuski 2004 ; Zaliapin et al. 2008 ; Zaliapin and Ben-Zion 2013 ; Zaliapin and Ben-Zion 2022 ) . These models are typically used in a Probabilistic Seismic Hazard Analysis (PSHA) to forecast future seismic activity ( Llenos and Michael 2013 ; Llenos and Michael 2020 ; Mizrahi et al. 2024 ; Llenos et al. 2024 ) . While effective for natural seismicity, the above mentioned statistical models are not always applicable to some instances of induced seismicity. Deviations from GR scaling are commonly observed, necessitating the formulation of alternative statistical models ( Eaton et al. 2014 ; Urban et al. 2016 ; Igonin et al. 2018 ; Leptokaropoulos 2020 ; Herrmann et al. 2022 ; Sedghizadeh et al. 2023 ; Krushelnitskii et al. 2024 ) . While the ETAS model has proven effective for analyzing natural aftershock sequences, its applicability to induced seismicity is limited due to the absence or minimal presence of pronounced aftershock patterns commonly observed in tectonic environments. However, in some instances of mining seismicity, it was shown that events can generate aftershock sequences that follow the modified Omori law ( Kgarume et al. 2010 ; Vallejos and McKinnon 2010 ; Estay et al. 2020 ) . Meanwhile, the ETAS model has been applied to some cases of induced mining seismicity to forecast its evolution ( Martinsson and Törnman 2020 ; Gospodinov et al. 2022 ; Sedghizadeh et al. 2024 ) . While both the modified Omori law and ETAS models are effective for sequences driven by magnitude scaling, they are less suited for cases where external factors, such as operational activities, primarily govern seismicity rates. Beyond frequency-magnitude statistics, the study of seismic clustering – where events occur closely in space and time – is equally critical for understanding seismic dynamics ( Schoenball et al. 2015 ; Cochran et al. 2018 ; Li et al. 2021 ) . Clustering provides insights into the underlying stress redistribution processes and helps differentiate between background seismicity and triggered events. For example, induced seismicity often exhibits unique spatiotemporal clustering patterns, reflecting the influence of human activities rather than natural tectonic processes ( Schoenball and Ellsworth 2017 ; Cochran et al. 2020 ; Karimi and Davidsen 2023 ; Li 2025 ) . Among the tools adapted to study clustering aspects of natural and induced seismicity, the NND method can play a critical role ( Zaliapin and Ben-Zion 2022 ) . However, the standard NND method relies on the GR law to describe the frequency-magnitude distribution of earthquakes, which often fails to capture the statistical properties of mining-induced seismicity. The temporal Hawkes point process, of which the ETAS model is a special case, provides a mechanism for capturing the stochastic nature of the occurrence of random events. Introduced by Hawkes 1971 , this model quantifies the likelihood of an event triggering subsequent events, with its conditional intensity function reflecting the influence of past events. This approach can be well-suited for examining induced seismicity and has applications in fields ranging from financial transactions and social network dynamics to solar physics ( Zipkin et al. 2016 ; Ross 2020 ; Schoenberg 2023 ; Bernabeu et al. 2024 ) . In a recent study, a nonparametric Hawkes model has been developed to incorporate spatial inhomogeneity, anisotropy, and space-time non-separability, better reflecting the heterogeneous geological conditions observed in seismic data ( Kwon et al. 2023 ) . In the Hawkes point process, the event magnitudes do not influence the occurrence of subsequent events and this mechanism can explain some cases of

Read the complete text