ArticlePublisher preview available

Forecasting of Induced Seismicity Rates from Hydraulic Fracturing Activities Using Physics-Based Models for Probabilistic Seismic Hazard Analysis: A Case Study

Authors:
Mauricio Reyes Canales
Mauricio Reyes Canales
Mauricio Reyes Canales
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Mirko van der Baan
Mirko van der Baan
Mirko van der Baan
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

One of the major challenges in seismic hazard analysis for induced seismicity is the forecasting of future seismicity rates, which are described by the Gutenberg–Richter parameters (a and b-values from the earthquake magnitude frequency distributions). In this study, we implement two methodologies in order to determine the Gutenberg–Richter parameters related to future induced seismicity: the Seismogenic Index and the Hydromechanical Nucleation model. We apply both methods in one recent case of induced seismicity: the Horn River Basin, Northeast B.C., Canada. We perform two tests to compare the predictions of both models with the observed seismicity. First, we compare the predicted number of earthquakes exceeding a certain magnitude per month with the observed number of earthquakes. In this test, both methods predict earthquake rates similar to the observed induced seismicity in the Horn River Basin. Second, we evaluate how appropriate are the predictions for specific magnitude ranges (given by forecasted Gutenberg–Richter parameters). In this case, both models make inaccurate predictions for specific magnitude ranges (annual magnitude frequency distributions), resulting in an under- or overestimation of the hazard but often with contradicting forecasts, despite using shared observations. The predictions under- and overestimate the hazard at different time points, due to the complexity in the evolution of the seismicity, and the assumption of constant b-values. As a result, these incorrect forecasts for future magnitude-frequency distributions lead to biased seismic hazard and ground motion predictions. More research effort is required in order to forecast more accurate Gutenberg–Richter parameters, particularly changes in the b-value, as observed in the Horn River Basin induced seismicity case.
Sketch of the methodology for the generation of synthetic ground motion catalogs, seismic hazard curves, and seismic hazard maps. a Calculate all distances between a given site and the location of the synthetic events (source-to-site distance). b Given the distances and magnitudes of the synthetic events, we use the GMPEs to estimate the corresponding ground motions at one site. c Seismic hazard curves (rate of exceedance as a function of ground motions) from ground motion catalogs. d Seismic hazard curves calculated at different sites. e Define a common rate of exceedance level and obtain the corresponding ground motion for each seismic hazard curve. f Knowing the ground motion at each site, it is possible to build a seismic hazard map
Sketch of the methodology for the generation of synthetic ground motion catalogs, seismic hazard curves, and seismic hazard maps. a Calculate all distances between a given site and the location of the synthetic events (source-to-site distance). b Given the distances and magnitudes of the synthetic events, we use the GMPEs to estimate the corresponding ground motions at one site. c Seismic hazard curves (rate of exceedance as a function of ground motions) from ground motion catalogs. d Seismic hazard curves calculated at different sites. e Define a common rate of exceedance level and obtain the corresponding ground motion for each seismic hazard curve. f Knowing the ground motion at each site, it is possible to build a seismic hazard map
… 
a Location of the seismic events (red dots) in the Horn River Basin, Northeast BC, using the catalog from Farahbod et al. (2015b). The blue square indicates the seismic source area used in this study. BC British Columbia, HRB Horn River Basin. b Monthly injection rates in the Horn River Basin. Notice the substantial increase in the injection volumes after Dec. 2009. From Farahbod et al. (2015a)
a Location of the seismic events (red dots) in the Horn River Basin, Northeast BC, using the catalog from Farahbod et al. (2015b). The blue square indicates the seismic source area used in this study. BC British Columbia, HRB Horn River Basin. b Monthly injection rates in the Horn River Basin. Notice the substantial increase in the injection volumes after Dec. 2009. From Farahbod et al. (2015a)
… 
Evolution of the a Seismogenic Index, and bb-value at the Horn River Basin, using the earthquake catalog from Farahbod et al. (2015b) and injection rates from Farahbod et al. (2015a). The light red (a) and green (b) curves show the mean Seismogenic Index ± error and the mean b-values ± error, respectively. The dashed purple lines show the 95 % confidence level for the Seismogenic Index and b-values given by bootstrapping techniques (see Langenbruch et al. (2020) and Gulia and Wiemer (2019) for details). c Monthly number of earthquakes with magnitude M≥2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M\ge 2.5$$\end{document} given by the observed seismicity (green bars) and the Seismogenic Index predictions (red line)
Evolution of the a Seismogenic Index, and bb-value at the Horn River Basin, using the earthquake catalog from Farahbod et al. (2015b) and injection rates from Farahbod et al. (2015a). The light red (a) and green (b) curves show the mean Seismogenic Index ± error and the mean b-values ± error, respectively. The dashed purple lines show the 95 % confidence level for the Seismogenic Index and b-values given by bootstrapping techniques (see Langenbruch et al. (2020) and Gulia and Wiemer (2019) for details). c Monthly number of earthquakes with magnitude M≥2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M\ge 2.5$$\end{document} given by the observed seismicity (green bars) and the Seismogenic Index predictions (red line)
… 
The column on the left shows a comparison between the GR parameters predicted by the Seismogenic Index (red line) and the GR parameters of the recorded seismicity (blue line, MLM fitting) for the years 2008–2011. The upper and lower curves shows the uncertainty of the GR parameter estimation. The column on the right shows the probable number of earthquakes in the range m=[2.5,3.6]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m=[2.5,3.6]$$\end{document}, for the years 2008–2011, given by the GR parameters of the Seismogenic Index (red line, prediction) and the observed earthquake catalog (blue line, MLM fitting). The light red and blue curves are the probability distributions given by the upper and lower GR curves. MLM=Maximum Likelihood method
The column on the left shows a comparison between the GR parameters predicted by the Seismogenic Index (red line) and the GR parameters of the recorded seismicity (blue line, MLM fitting) for the years 2008–2011. The upper and lower curves shows the uncertainty of the GR parameter estimation. The column on the right shows the probable number of earthquakes in the range m=[2.5,3.6]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m=[2.5,3.6]$$\end{document}, for the years 2008–2011, given by the GR parameters of the Seismogenic Index (red line, prediction) and the observed earthquake catalog (blue line, MLM fitting). The light red and blue curves are the probability distributions given by the upper and lower GR curves. MLM=Maximum Likelihood method
… 
Evolution of the aR(t) parameter, and ba-value (monthly) at the Horn River Basin, given by the Hydromechanical Nucleation approach. c Comparison between the monthly number of earthquakes larger than m≥2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m\ge 2.5$$\end{document}, given by the observed seismicity and the Hydromechanical Nucleation model
+4
Evolution of the aR(t) parameter, and ba-value (monthly) at the Horn River Basin, given by the Hydromechanical Nucleation approach. c Comparison between the monthly number of earthquakes larger than m≥2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m\ge 2.5$$\end{document}, given by the observed seismicity and the Hydromechanical Nucleation model
… 
Figures - available from: Pure and Applied Geophysics
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Springer Nature.
Learn more
Content available from Pure and Applied Geophysics
This content is subject to copyright. Terms and conditions apply.
Forecasting of Induced Seismicity Rates from Hydraulic Fracturing Activities Using Physics-
Based Models for Probabilistic Seismic Hazard Analysis: A Case Study
MAURICIO REYES CANALES
1
and MIRKO VAN DER BAAN
1
Abstract—One of the major challenges in seismic hazard
analysis for induced seismicity is the forecasting of future seis-
micity rates, which are described by the Gutenberg–Richter
parameters (aand b-values from the earthquake magnitude fre-
quency distributions). In this study, we implement two
methodologies in order to determine the Gutenberg–Richter
parameters related to future induced seismicity: the Seismogenic
Index and the Hydromechanical Nucleation model. We apply both
methods in one recent case of induced seismicity: the Horn River
Basin, Northeast B.C., Canada. We perform two tests to compare
the predictions of both models with the observed seismicity. First,
we compare the predicted number of earthquakes exceeding a
certain magnitude per month with the observed number of earth-
quakes. In this test, both methods predict earthquake rates similar
to the observed induced seismicity in the Horn River Basin. Sec-
ond, we evaluate how appropriate are the predictions for specific
magnitude ranges (given by forecasted Gutenberg–Richter param-
eters). In this case, both models make inaccurate predictions for
specific magnitude ranges (annual magnitude frequency distribu-
tions), resulting in an under- or overestimation of the hazard but
often with contradicting forecasts, despite using shared observa-
tions. The predictions under- and overestimate the hazard at
different time points, due to the complexity in the evolution of the
seismicity, and the assumption of constant b-values. As a result,
these incorrect forecasts for future magnitude-frequency distribu-
tions lead to biased seismic hazard and ground motion predictions.
More research effort is required in order to forecast more accurate
Gutenberg–Richter parameters, particularly changes in the b-value,
as observed in the Horn River Basin induced seismicity case.
Keywords: Forecasting seismicity rates, induced seismicity,
physics based models, time-dependent Gutenberg–Richter param-
eters, Monte-Carlo simulations, Horn River Basin, Canada.
1. Introduction
Probabilistic seismic hazard analysis (PSHA) has
been largely used for assessing hazards related to
natural seismicity. PSHA quantifies the possible
ground motion at one location, in a period of time,
caused by earthquake shaking (Cornell 1968; Baker
2008). PSHA outputs (e.g., seismic hazard curves and
seismic hazard maps) are used by governments and
industry in applications for life and property safety,
such as developing building code requirements,
deciding the security criteria for critical facilities like
dams, hydroelectric plants, nuclear plants, and
determining earthquake insurance rates (Baker 2008;
Mulargia et al. 2017).
Recent studies (Ellsworth 2013; Atkinson et al.
2015,2016; Langenbruch and Zoback 2016; van der
Baan and Calixto 2017) have shown increased seis-
micity in geologically stable basins in North
America, thought to be associated with hydraulic
fracturing treatments and/or waste water disposal
wells. As a result, it has been necessary to quantify
the seismic hazard related to induced activities
associated with shale oil and gas production. How-
ever, there are some challenges in the implementation
of PSHA for induced seismicity, including the esti-
mation of future induced seismicity rates and its non-
stationary behavior. Reyes Canales and Van der Baan
(2019) developed a methodology to include non-sta-
tionary seismicity rates in the seismic hazard
analysis, where the changing rates result from time-
dependent Gutenberg–Richter (GR) parameters.
However, estimating and forecasting GR parameters
for induced seismicity is still a major challenge due to
the lack of recorded events and the time dependency
of rates. In contrast, this is a lesser problem for nat-
ural seismicity, since seismic hazard analysis
assumes stationary GR parameters based on long-
term historical catalogs.
1
Department of Physics, University of Alberta, Edmonton,
Canada. E-mail: reyescan@ualberta.ca
Pure Appl. Geophys. 178 (2021), 359–378
Ó2021 The Author(s), under exclusive licence to Springer Nature Switzerland AG part
of Springer Nature
https://doi.org/10.1007/s00024-021-02661-x Pure and Applied Geophysics
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
... As an extension of these statistical analyses, physics-based models like the seismogenic index (Langenbruch & Zoback, 2016;Shapiro et al., 2010), as well as the hydromechanical nucleation approach (Norbeck & Rubinstein, 2018), can be used to forecast the number of earthquakes from upcoming operations. However, care should be taken that these approaches assume constant -values, thereby only forecasting event rates above a certain magnitude, thus providing only limited information on the seismic hazard (Reyes Canales & van der Baan, 2021). Nonetheless, such forecasting approaches can provide insight into likely future trends given operational scenarios, once calibrated against historical data. ...
... For instance, in the susceptibility analysis conducted by Wozniakowska and Eaton (2020) for the Montney Fm., the relevance of the pressure gradient as a proxy for induced seismicity was reduced in favor of other factors, particularly the distance to the cordillera belt, injection depth and distance to the basement. It is possible that some of these factors correlate to changes in the tectonic stresses, a causative factor that is explicitly ignored when solely considering pore pressure variations (Hager et al., 2021;van der Baan, 2021). ...
The Rise, Peak and Decline of the Seismic Hazard Related to Hydraulic Fracturing Activities in the Duvernay Play, Fox Creek Area, Alberta
Article
Full-text available
  • Jan 2022
We analyze the temporal evolution of the induced seismicity related to hydraulic fracturing activities in the Duvernay Formation, near Fox Creek, Alberta, Canada. For this analysis, we estimate annual Gutenberg‐Richter parameters, a(t) $a(t)$‐ and b(t) $b(t)$‐ values, and then calculate the annual likelihood of earthquakes greater than magnitude M>4 $M > 4$ from 2014 to 2020. The seismic hazard near Fox Creek has consistently decreased since 2015, from a 95% probability of an earthquake greater than magnitude M>4 $M > 4$ in 2015 to 4% in 2019 and less than 1% probability in 2020. The induced seismicity in Fox Creek is characterized by two actively seismic regions with distinctive features: (a) an Eastern region (∼220 events M>2 $M > 2$) with lower b‐values and higher hazard; (b) a Western region (∼210 events M>2 $M > 2$) with higher b‐values and lower seismic hazard. In contrast, extensive regions where hydraulic fracturing is performed, particularly East of the seismic cluster, remain non‐seismogenic. The overall decreasing seismic hazard, which contrasts with increasing operator activity, can be associated with (a) the intensification of hydraulic fracturing operations toward areas less susceptible to induced seismicity and (b) the reduction of seismic activity in the Eastern region, which exhibits the highest seismic hazard. We also find evidence of a minimum annual injection volume required to trigger induced seismicity in both the Western and Eastern regions. The minimum injection threshold increases over the years, implying increasingly successful mitigation strategies, likely due to regulatory implementations in the area, which has led the operators to exercise precaution in regions with significant seismic hazard and adapt treatment strategies to avoid triggering moderate magnitude size events during hydraulic fracturing stimulations.
View
Cost analysis of wastewater production from conventional and unconventional oil and gas wells
Article
Full-text available
  • Sep 2022
  • FUEL
Production of high volumes of flowback and produced water (FPW) from hydraulic fracturing (HF) is among the environmental concerns associated with hydrocarbon recovery from unconventional low-permeability formations. FPW management costs are a key factor for oil and gas companies in deciding the fate of FPW (treatment, disposal, recycling, and reuse). In this study, a comprehensive library of more than 20,000 oil and gas wells in the Montney Formation, Canada, is created to compare the FPW management costs from conventional and hydraulically fractured wells. The results indicate that for oil wells, both conventional and HF wells have similar volumetric oil–water ratios during the first two years of production. Since HF wells in the study region produce higher volumes of oil, on average, than conventional wells, the gross revenue from a typical HF oil well is higher than a typical conventional oil well. However, over the course of two years of production, conventional gas wells have higher volumetric gas–water ratios than HF gas wells. Estimation of the gross revenue for gas wells is sensitive to the average natural gas price and FPW management costs as HF wells typically generate more natural gas and FPW than conventional wells. Due to the fluctuating nature of oil and gas prices, we have created master curves for revenue (i.e., produced hydrocarbon value) and FPW management costs. The master curves cover oil and gas price ranges of $10–100/bbl and $0.5–5/MMBtu, respectively. Furthermore, since the FPW management costs may vary depending on the implemented technology or local labor costs, the master curves encompass a FPW management cost range of $5–100/m³. The provided master curves introduce a basis for the rapid analysis of FPW management costs of oil and gas wells, and can be incorporated with other well operational costs (such as drilling and completions, water withdraw permit, and well maintenance) for detailed evaluation of the profitability of hydrocarbon recovery processes.
View
Evolution of Short‐Term Seismic Hazard in Alberta, Canada, From Induced and Natural Earthquakes: 2011–2020
Article
Full-text available
  • Feb 2022
We generate short‐term seismic hazard maps for the province of Alberta, Canada, from 2011 through 2020. First, we describe the required adaptations to probabilistic seismic hazard analysis to generate short‐term seismic hazard maps, following the Monte Carlo simulation approach. Second, we identify the natural and induced seismic source areas in Alberta and estimate their earthquake recurrence parameters revealing considerable spatio‐temporal variations in the a $a$‐ and b $b$‐values in the different seismic clusters in the province. Areas with the highest short‐term seismic hazard during the last decade in Alberta are related to cases of induced seismicity, including hydraulic fracturing activities in the Duvernay Fm., near Fox Creek, and waste‐water disposal activities near the Musreau Lake. These maps provide a valuable tool to quantify the short‐term evolution in the seismic hazard, which is particularly important considering past and emerging cases of induced seismicity related to the energy sector in Alberta. Furthermore, using parameters from the previous year, we make a seismic hazard forecast for the year 2021. Our analysis provides a baseline of expected short‐term seismic hazard, with inheren uncertainty due to the assumption of unchanged recurrence parameters from the previous year; yet our study reveals pertinent seismicity patterns in line with changing human activities.
View
Research status of earthquake forecasting in hydraulic-fracturing induced earthquakes
Article
Full-text available
  • Jun 2021
In the new types of industrial activities including unconventional energy extraction associated with shale gas and hot dry rock, gas reservoir operations, CO2 geological storage, undergoing research on induced earthquake forecasting has become one of the forward positions of current seismology. As for the intense actual demand, the immature research on induced earthquake forecasting has already been applied in pre-assessment of site safety and seismic hazard and risk management. This work will review systematically recent advances in earthquake forecasting induced by hydraulic fracturing during industrial production from four aspects: earthquake occurrence probability, maximum expected magnitude forecasting, seismic risk analysis for engineering and social applications and key scientific problems. In terms of earthquake occurrence probability, we introduce statistical forecasting models such as an improved ETAS and non-stationary ETAS and physical forecasting models such as Seismogenic Index (SI) and hydro-mechanism nucleation. Research on maximum expected magnitude forecasting has experienced four stages of linear relationship with net injection volume of fluid, power exponential relationship and physical forecasting regarding fault parameters. For seismic risk analysis, we focus on probabilistic seismic hazard assessment and quantitative geological susceptibility model. Furthermore, this review is extended to key scientific problems that contain obtaining accurate fault scale and environmental stress state of reservoir, critical physical process of runaway rupture, complex mechanism of fault activation as well as physical mechanism and modeling of trailing effect. This work in understanding induced earthquake forecasting may contribute to unconventional energy development and production, seismic hazard mitigation, emergency management and scientific research as a reference.
View
Are Aftershock Sequences Pertinent to Long‐Term Seismic Hazard Assessments? Insights From the Temporal ETAS Model
Article
Full-text available
  • Jul 2020
Aftershocks are commonly removed from observed earthquake catalogs in probabilistic seismic hazard analyses, using declustering techniques. We use stationary and temporal Epistemic‐Type Aftershock Sequence (ETAS) models to generate aftershocks from background seismicity and preceding aftershocks. We assume that the mainshocks equal the background seismicity, to divide the synthetic earthquake catalogs into mainshock (declustered) and complete (nondeclustered) versions. Only mainshocks follow a Poissonian distribution. We then evaluate how accurately we can forecast the recurrence of the largest events based on the simulated catalogs. A single b value is derived from the simulated catalogs and used in the magnitude‐frequency forecast. When the b value of the mainshocks is considerably smaller than the b value of the aftershocks, the information derived from the mainshock catalog leads to accurate predictions of the occurrence of the largest events. Conversely, when the mainshock and aftershocks have comparable b values, only the complete catalog produces representative estimates for the occurrence statistics of the largest events. We also show that using Poisson statistics leads to representative assessment of long‐term recurrences, even if aftershocks have a non‐Poissonian distribution. Finally, we analyze a recent case of induced seismicity, Oklahoma, USA, where the complete catalog displays a kink in the magnitude‐frequency distribution. Declustering removes this kink, leading to better b value estimations for the largest magnitude events. We conclude that temporal declustering for seismic hazard assessment is only recommended in catalogs with a large number of earthquakes and in catalogs where the b values of the mainshocks are significantly different from the b values of the complete catalog.
View
Value at Induced Risk: Injection‐Induced Seismic Risk From Low‐Probability, High‐Impact Events
Article
Full-text available
The increasing number of damaging earthquakes, induced by underground injection of fluids, demonstrates the need of a structured methodology for critical thinking about and clear communication of induced seismic risk. Using the Pohang, South Korea, geothermal project and the damaging (US$75M) local magnitude ML 5.35 induced earthquake in 2017, we present a straightforward methodology that summarizes induced seismic risk in a single time-dependent number, the Value at Induced Risk (VaIR). VaIR quantifies the amount of economic losses that will not be exceeded at a given confidence level. Our method reveals that the Pohang project was exposed to a 2% chance of causing economic losses of US$140M or more (1% of US$3.2B or more). The high risk could have been identified during the sequence of reservoir stimulation, lasting almost 2 years. VaIR opens new ways to assess and communicate risk to all stakeholders for informed risk management decision making.
View
Insights From Micromechanical Modeling of Intact Rock Failure: Event Characteristics, Stress Drops, and Force Networks
Article
Full-text available
  • Dec 2019
We use a bonded‐particle method to investigate event characteristics, b values, internal force distributions, and stress drops in triaxial deformation tests. We simulate brittle through ductile deformation regimes. We find the following: (1) Event rates are proportional to anelastic axial strain: (i) significant and accelerating events rates only occur near peak stress (brittle deformation); (ii) event rates gradually increase until the anelastic strain plateaus (ductile deformation). (2) b value patterns show a systematic decrease when approaching peak stress, after which (i) they increase again (brittle case) or (ii) reach a constant minimum (ductile case). A decrease in b values is indicative of progressive internal damage, with an increasing event rate. (3) Weak‐ and strong‐force networks exist within the sample. Macrofailure of the sample occurs due to collapse of the strong‐force network. Acoustic emissions predominantly occur (i) within the weak‐force network allowing for tensile crack opening and closing despite the large compressive external stresses, and (ii) in areas with the largest spatial force gradients, close to the strong‐force networks. All internal compressive and extensional normal forces display exponential distributions despite uniform boundary stresses. (4) Stress drops of the largest events are inversely proportional to peak stress; the largest stress drops occur for brittle failure, progressively approaching a zero magnitude for ductile deformation. The systemic correlations between b values, the number of acoustic emissions, and stress drops with the maximum principal stress may offer opportunities to invert for changes in the stress state from these remote observables during earthquake cycles.
View
Real-time discrimination of earthquake foreshocks and aftershocks
Article
Full-text available
  • Oct 2019
  • NATURE
Immediately after a large earthquake, the main question asked by the public and decision-makers is whether it was the mainshock or a foreshock to an even stronger event yet to come. So far, scientists can only offer empirical evidence from statistical compilations of past sequences, arguing that normally the aftershock sequence will decay gradually whereas the occurrence of a forthcoming larger event has a probability of a few per cent. Here we analyse the average size distribution of aftershocks of the recent Amatrice–Norcia and Kumamoto earthquake sequences, and we suggest that in many cases it may be possible to discriminate whether an ongoing sequence represents a decaying aftershock sequence or foreshocks to an upcoming large event. We propose a simple traffic light classification to assess in real time the level of concern about a subsequent larger event and test it against 58 sequences, achieving a classification accuracy of 95 per cent.
View
Including Non-Stationary Magnitude–Frequency Distributions in Probabilistic Seismic Hazard Analysis
Article
Full-text available
  • Jun 2019
  • PURE APPL GEOPHYS
We describe a first principles methodology to evaluate statistically the hazard related to non-stationary seismic sources like induced seismicity. We use time-dependent Gutenberg–Richter parameters, leading to a time-varying rate of earthquakes. We derive analytic expressions for occurrence rates which are verified using Monte Carlo simulations. We show two examples: (1) a synthetic case with two seismic sources (background and induced seismicity); and (2) a recent case of induced seismicity, the Horn River Basin, Northeast British Columbia, Canada. In both cases, the statistics from the Monte Carlo simulations agree with the analytical quantities. The results show that induced seismicity affects seismic hazard rates but that the exact change greatly depends on both the duration and intensity of the non-stationary sequence as well as the chosen evaluation period. The developed methodology is easily extended to handle spatial source distributions as well as ground motion analysis in order to generate a complete methodology for non-stationary probabilistic seismic hazard analysis.
View
Induced Seismicity in Western Canada Linked to Tectonic Strain Rate: Implications for Regional Seismic Hazard
Article
Full-text available
  • Oct 2018
A rapid increase of injection-induced earthquakes (IIE) is often linked to a higher level of seismic hazard. In this study, we compare the geodetically defined moment rate to seismicity distribution in western Canada where significant IIE are observed. The regional seismic pattern is dominated by IIE, both in number and moment, along a 150-km wide NW-SE band of moderate strain rate in the easternmost Cordillera and foothills. The observed rate of moment release from local earthquakes is much closer to the tectonic moment rate in the IIE-dominated areas. We conclude that, on a regional scale, tectonic strain rate is an important control on IIE. Injection in areas with moderate tectonic strain may temporarily increase the local seismic hazard, but widespread IIE over an extended period of time may deplete the available tectonic moment and could, under the right conditions, have a limited long-term effect of reducing regional seismic hazard.
View
Physics-based forecasting of man-made earthquake hazards in Oklahoma and Kansas
Article
Full-text available
  • Sep 2018
Reinjection of saltwater, co-produced with oil, triggered thousands of widely felt and several damaging earthquakes in Oklahoma and Kansas. The future seismic hazard remains uncertain. Here, we present a new methodology to forecast the probability of damaging induced earthquakes in space and time. In our hybrid physical-statistical model, seismicity is driven by the rate of injection-induced pressure increases at any given location and spatial variations in the number and stress state of preexisting basement faults affected by the pressure increase. If current injection practices continue, earthquake hazards are expected to decrease slowly. Approximately 190, 130 and 100 widely felt M ≥ 3 earthquakes are anticipated in 2018, 2019 and 2020, respectively, with corresponding probabilities of potentially damaging M ≥ 5 earthquakes of 32, 24 and 19%. We identify areas where produced-water injection is more likely to cause seismicity. Our methodology can be used to evaluate future injection scenarios intended to mitigate seismic hazards.
View
Hydromechanical Earthquake Nucleation Model Forecasts Onset, Peak, and Falling Rates of Induced Seismicity in Oklahoma and Kansas
Article
Full-text available
  • Mar 2018
The earthquake activity in Oklahoma and Kansas that began in 2008 reflects the most widespread instance of induced seismicity observed to date. We develop a reservoir model to calculate the hydrologic conditions associated with the activity of 902 saltwater disposal wells injecting into the Arbuckle aquifer. Estimates of basement fault stressing conditions inform a rate-and-state friction earthquake nucleation model to forecast the seismic response to injection. Our model replicates many salient features of the induced earthquake sequence, including the onset of seismicity, the timing of the peak seismicity rate, and the reduction in seismicity following decreased disposal activity. We present evidence for variable time lags between changes in injection and seismicity rates, consistent with the prediction from rate-and-state theory that seismicity rate transients occur over timescales inversely proportional to stressing rate. Given the efficacy of the hydromechanical model, as confirmed through a likelihood statistical test, the results of this study support broader integration of earthquake physics within seismic hazard analysis.
View
2017 One‐Year Seismic‐Hazard Forecast for the Central and Eastern United States from Induced and Natural Earthquakes
Article
Full-text available
  • May 2017
We produce a one-year 2017 seismic-hazard forecast for the central and eastern United States from induced and natural earthquakes that updates the 2016 one-year forecast; this map is intended to provide information to the public and to facilitate the development of induced seismicity forecasting models, methods, and data. The 2017 hazard model applies the same methodology and input logic tree as the 2016 forecast, but with an updated earthquake catalog. We also evaluate the 2016 seismic-hazard forecast to improve future assessments. The 2016 forecast indicated high seismic hazard (greater than 1% probability of potentially damaging ground shaking in one year) in five focus areas: Oklahoma-Kansas, the Raton basin (Colorado/New Mexico border), north Texas, north Arkansas, and the New Madrid Seismic Zone. During 2016, several damaging induced earthquakes occurred in Oklahoma within the highest hazard region of the 2016 forecast; all of the 21 moment magnitude (M) ≥4 and 3 M ≥5 earthquakes occurred within the highest hazard area in the 2016 forecast. Outside the Oklahoma-Kansas focus area, two earthquakes with M ≥4 occurred near Trinidad, Colorado (in the Raton basin focus area), but no earthquakes with M ≥2.7 were observed in the northTexas or north Arkansas focus areas. Several observations of damaging ground-shaking levels were also recorded in the highest hazard region of Oklahoma. The 2017 forecasted seismic rates are lower in regions of induced activity due to lower rates of earthquakes in 2016 compared with 2015, which may be related to decreased wastewater injection caused by regulatory actions or by a decrease in unconventional oil and gas production. Nevertheless, the 2017 forecasted hazard is still significantly elevated in Oklahoma compared to the hazard calculated from seismicity before 2009.
View
Human-induced seismicity and large-scale hydrocarbon production in the USA and Canada
Article
We compare current and historic seismicity rates in six States in the USA and three Provinces in Canada to past and present hydrocarbon production. All States/Provinces are major hydrocarbon producers. Our analyses span three to five decades depending on data availability. Total hydrocarbon production has significantly increased in the past few years in these regions. Increased production in most areas is due to large-scale hydraulic fracturing and thus underground fluid injection. Furthermore, increased hydrocarbon production generally leads to increased water production, which must be treated, recycled, or disposed of underground. Increased fluid injection enhances the likelihood of fault reactivation, which may affect current seismicity rates. We find that increased seismicity in Oklahoma, likely due to salt-water disposal, has an 85% correlation with oil production. Yet, the other areas do not display State/Province-wide correlations between increased seismicity and production, despite 8–16-fold increases in production in some States. However, in various cases, seismicity has locally increased. Multiple factors play an important role in determining the likelihood of anthropogenic activities influencing earthquake rates, including (i) the near-surface tectonic background rate, (ii) the existence of critically stressed and favorably oriented faults, which must be hydraulically connected to injection wells, (iii) the orientation and magnitudes of the in situ stress field, combined with (iv) the injection volumes and implemented depletion strategies. A comparison with the seismic hazard maps for the USA and Canada shows that induced seismicity is less likely in areas with a lower hazard. The opposite, however, is not necessarily true.
View