Abstract:ABSTRACT The clustering phenomenon in seismology draws significant attention from geophysicists as well as the public at large. The reason for this is rooted in the fundamental interest of reliable earthquake prediction; or otherwise, if earthquake prediction is possible. The investigations of earthquake spatial-magnitude clustering, especially if earthquake magnitudes are clustering within a controlling geological structure(s), are however, still of active debate. In this paper, we first present several discoveries on the magnitude clustering phenomenon via the interrogation of the laboratory acoustic emission (AE); after that, we shortly display the accompanying instigations on field-scale catalogs intrigued by the laboratory interrogation. In laboratory, the clustering of AE event magnitude can only unconditionally exist under shear stress; the tensile stress will remove the magnitude clustering phenomenon. This confirmation from laboratory interrogation intrigues the field investigation on several natural and anthropogenic catalogs crossing the North American continent. As the consequence that pure tensile stress condition can hardly exist in field-scale, magnitude clustering phenomenon for field-scale catalogs is found to be prominent crossing scales from the scale of hundreds of meters to the scale of hundreds of kilometers. INTRODUCTION The clustering phenomenon in space and time is a well-recognized feature of earthquakes. There are prominent examples of earthquakes as being spatial clustering, i.e. the aftershocks around a mainshock, and temporal clustering, i.e. the Omori-Utsu decay in the temporal productivity after a mainshock (Felzer & Brodsky, 2006; Utsu, Ogata, & Matsu’ura, 1995). However, the existence of clustering in earthquake magnitudes is still a matter of active debate. The earthquake magnitudes were thought to be independent until a set of studies reported magnitude correlations between sequential catalogued earthquakes (A. Corral, 2004; A. Corral, 2006; Lippiello, Godano, & de Arcangelis, 2007). Questions on these results had also been proposed by other researchers as the magnitude clustering observation could be influenced by catalog incompleteness (Davidsen & Green, 2011; Davidsen, Kwiatek, & Dresen, 2012). The reason for the extensive investigations on earthquake clustering phenomenon is rooted in the fundamental interest of reliable earthquake prediction; or otherwise, to address the question if earthquake prediction is really possible (Davidsen & Green, 2011). If magnitude clustering does exist, it can have practical meanings for the short-term earthquake forecasting. For instance, if similar magnitude events are clustered in aftershock sequences (Field et al., 2017; Nandan, Ouillon, & Sornette, 2019; Nichols & Schoenberg, 2014; Spassiani & Sebastiani, 2016). Specifically, contemporary forecasting approaches typically utilize the epidemic-type aftershock sequence (ETAS) methodology for simulating seismicity. This method has incorporated with the spatial and temporal clustering but has not included any form of magnitude clustering (Field et al., 2017; Hardebeck, 2013; Ogata, 1988).

Unsupervised clustering of catalog-driven features for characterizing temporal evolution of labquake stress

Attention network forecasts time-to-failure in laboratory shear experiments

Seismo-Magnitude Clustering is Imposed by Structural Constraint and Shear Stress

Characterizing Acoustic Signals and Searching for Precursors during the Laboratory Seismic Cycle Using Unsupervised Machine Learning

Spatial Continuity and Stress Transfer: Primary and Complementary Factors Shaping Pure-Swarm Laboratory Catalog into Mixed Burst-Like and Swarm-Like

Magnitude Clustering During Stick‐Slip Dynamics on Laboratory Faults

Spatial migration of temporal earthquake clusters driven by the transfer of differential stress between neighbouring fault/shear-zone structures

Machine Learning Predicts the Timing and Shear Stress Evolution of Lab Earthquakes Using Active Seismic Monitoring of Fault Zone Processes

Deep learning for laboratory earthquake prediction and autoregressive forecasting of fault zone stress

Aseismic strain localization prior to failure and associated seismicity in crystalline rock

Extracting acoustic emission features that precede hard rock instability with unsupervised learning

Spatial Correlation-Based Characterization of Acoustic Emission Signal-Cloud in a Granite Sample by a Cube Clustering Approach

Estimating the Physical State of a Laboratory Slow Slipping Fault from Seismic Signals

Intermittent Criticality Multi‐Scale Processes Leading to Large Slip Events on Rough Laboratory Faults

An Unsupervised Machine Learning Approach for Ground-Motion Spectra Clustering and Selection

High-resolution source imaging and moment tensor estimation of acoustic emissions during brittle creep of basalt undergoing carbonation

Machine learning opportunities to conduct high-fidelity earthquake simulations in multi-scale heterogeneous geology

Seismic clusters and fluids diffusion: a lesson from the 2018 Molise (Southern Italy) earthquake sequence

Seismic magnitude clustering is prevalent in field and laboratory catalogs

Stress Evolution Before the 2014 Ludian M6.5 Earthquake: Insights from Groundwater and Geodetic Measurement

Impact of Environmental Stressors on Jointed Rock Cliffs by Acoustic Emission Sensing: Preliminary Findings from the Acuto Field Laboratory (Central Italy)