Discriminating Characteristics of Tectonic and Human‐Induced Seismicity

Abstract

We analyze statistical features of background and clustered subpopulations of earthquakes in different regions in an effort to distinguish between human-induced and natural seismicity. Analysis of end-member areas known to be dominated by humaninduced earthquakes (The Geyser geothermal field in northern California and TauTona gold mine in South Africa) and regular tectonic activity (the San Jacinto fault zone in southern California and the Coso region, excluding the Coso geothermal field in eastern central California) reveals several distinguishing characteristics. Induced seismicity is shown to have (1) higher rate of background events (both absolute and relative to the total rate), (2) faster temporal offspring decay, (3) higher rate of repeating events, (4) larger proportion of small clusters, and (5) larger spatial separation between parent and offspring, compared to regular tectonic activity. These differences also successfully discriminate seismicity within the Coso and Salton Sea geothermal fields in California before and after the expansion of geothermal production during the 1980s.

Source Spectral Properties of Small to Moderate Earthquakes in Southern Kansas

“…To examine the variability in Δ σ within and between prominent earthquake sequences, we use the method described by Zaliapin and Ben‐Zion () that partitions events into individual sequences based upon nearest‐neighbor space‐time distances. The nearest‐neighbors method has been shown to be effective in characterizing the space‐time clustering statistics of both tectonic and induced earthquake sequences (Schoenball et al, ; Zaliapin & Ben‐Zion, ). It defines the distance η i j between an event pair (parent i , daughter j ) to be the product of a rescaled time and rescaled distance , where d t i j is the difference in time in years, d r i j is the spatial distance in kilometers, M i is the magnitude of the parent event, and d = 1.6 is the assumed fractal dimension.…”

Spatiotemporal distribution of Oklahoma earthquakes: Exploring relationships using a nearest‐neighbor approach

“…In particular, analysis of the nearest‐neighbor distance distribution provides an insight on what type of earthquake clusters dominate the catalog of the studied region. Nearest‐neighbor method is shown to be applicable to areas with different types of dominant seismicity (natural, induced, or mixed), as shown in the present work and Hicks [], Zaliapin and Ben‐Zion [], Schoenball et al [], and Zaliapin and Ben‐Zion []. An advantage of the method lies in the fact that the separation of events into subpopulations is driven by the bimodal distribution of the nearest‐neighbor distance which does not assume any particular form of earthquake clusters.…”

“…The observed differences in the distribution of the nearest‐neighbor distance and properties of the clustered modes for both states are significant, in spite of the fact that the Southern California is known both for natural and induced seismicity. These differences can be attributed to the difference in the triggering mechanisms leading to the events occurrence and initial geologic settings (stress condition, hydrologic conductivity, and level of heat flow) of the two regions, as suggested in Goebel [] and Zaliapin and Ben‐Zion [] and deserve further investigation.…”

“…There are many regions in the continental U.S. where the induced seismicity is also of concern, e.g., Arkansas, California, Colorado, Kansas, Ohio, and Texas [ Majer et al , ; Majer and Peterson , ; Frohlich , ; Brodsky and Lajoie , ; Kim , ; Llenos and Michael , ; Buchanan et al , ; Darold et al , ; Rubinstein et al , ; Goebel , ; Rubinstein and Mahani , ]. Zaliapin and Ben‐Zion [] investigated the relationship between regional geologic attributes, presence of induced seismicity in the area, and observed clusters distribution. While both Oklahoma and California experience induced seismicity, the underlying differences in the geology, initial regional stress, and injection history make it difficult to predict the location, time, and rates of occurrence for future events.…”

Current challenges in monitoring, discrimination, and management of induced seismicity related to underground industrial activities: A European perspective

“…The union of these two features leads to the epidemic type aftershock sequence (ETAS) model which is a point process model that describes the seismicity rate observed in a region as a summation of a background rate of independent events and the aftershocks triggered by each event [ Ogata , ]. In recent years statistical methods for induced seismicity discrimination in the space‐time‐magnitude domain have been proposed [e.g., Zaliapin and Ben‐Zion , ]. Schoenball et al [] applied one of these approaches to the induced seismicity recorded at the Coso Geothermal field (California, USA).…”