OR/17/033 Research
| Baptie, B. 2017. Earthquake seismology 2016/2017 - BGS seismic monitoring and information service. British Geological Survey Internal Report, OR/17/033. |
Unconventional oil and gas development: understanding and monitoring induced seismic activity
A study to improve understanding of the levels of induced seismic activity that could be associated with unconventional oil and gas activities in Scotland was commissioned by the Scottish Government (Baptie et al., 2016[1]). This also examined regulatory and non-regulatory actions that can be taken to mitigate any noticeable effects on communities.
Scotland is characterised by low levels of earthquake activity. Historical observations of earthquake activity date back to the 16th century, and show that despite many accounts of earthquakes felt by people, damaging earthquakes are relatively rare. The largest recorded earthquake in Scotland had a magnitude of 5.2 ML and only two other earthquakes with a magnitude of 5.0 ML or greater have been observed in the last 400 years. As a result, the risk of damaging earthquakes is low.
Most earthquake activity in Scotland is north of the Highland Boundary Fault, on the west side of mainland Scotland, and there are fewer earthquakes in northern and eastern Scotland. It is rarely possible to associate these earthquakes with specific faults because of uncertainties both in the earthquake location estimates, which are typically several kilometres, and our limited knowledge of faulting below the surface.
Earthquake activity in the Midland Valley of Scotland is lower than that north of the Highland Boundary Fault, and many of the recorded earthquakes in this area in the 1970s, 1980s and 1990s were induced by coal-mining. Most of these mining induced earthquakes are small (the largest in Scotland had a magnitude of 2.6 ML) and since the decline of the coal-mining industry in the 1990’s, very few mining-induced earthquakes have been recorded.
These mining induced events represent a temporary perturbation and they need to be removed from the earthquake catalogue so that an accurate measure of natural earthquake activity rates can be established. We did this by defining a simple spatial filter based on the Mining Reporting Areas, as issued by the Coal Mining Authority. All events from within these areas are removed from the catalogue.
The revised earthquake activity rate for Scotland determined from 1970 to present suggests that, on average, there are eight earthquakes with a magnitude of 2.0 or above (which is roughly the minimum magnitude felt by people) somewhere in Scotland every year. Activity rates calculated for the Midland Valley are lower, although the small number of observed earthquakes for this area means the values have large uncertainties. This suggests that earthquake hazard in the Midland Valley is lower than elsewhere in Scotland.
Existing catalogues of earthquake activity in Scotland are incomplete at magnitudes below 2 ML, from 1970 to present, and for higher magnitudes prior to this. This is due to the detection capability of the networks of seismometers that have operated in the study area over the last few decades. This, together with the low background activity rates, limits our ability to identify any areas that might present an elevated seismic hazard for any Unconventional Oil and Gas (UOG) operations based on seismic data alone. Similarly, limited information about the state of stress in the Earth’s Crust means that it is not possible to identify any particular parts of the study area where faults are more likely to be reactivated and that may present an elevated seismic hazard for any UOG operations.
Noise and detection capability
Ambient Earth noise is present in all recordings. It affects data quality and can limit the ability to detect and reliably locate small transient signals from earthquakes or other disturbances. We have analysed ambient noise levels at all sites across the UK and used the results to improve models of the detection capability of the network.
Seismograms always contain noise from ambient Earth vibrations as well as transient recordings from earthquakes. Seismic noise from human activity is often referred to as ‘cultural noise’ and originates primarily from the coupling of traffic and machinery energy into the Earth. This cultural noise propagates mainly as high-frequency surface waves (1–100 Hz) that attenuate within a few kilometres of the noise source and often shows very strong diurnal variations. The frequency content is similar to that for small and moderate local earthquakes. As a result, high noise levels can limit the ability to detect and reliably locate small transient signals from earthquakes or other disturbances.
We used power spectral density (PSD), calculated from one hour segments of continuous data, to characterize noise levels in a range of frequencies or periods at all stations in the UK network. A statistical analysis of the PSDs yields probability density functions (PDFs) of the noise power for each of the frequency bands at each station and component. We use the median, 5th and 95th percentiles of the PDF as the basis of median, low and high noise models for each station.
We find that noise can vary significantly even for stations that are close together. For example, the variations in RMS displacement amplitudes at frequencies above 1 Hz between stations in the Vale of Pickering, Yorkshire, can exceed two orders of magnitude. The quietest stations show RMS amplitudes of less than 1 nm, while noisier stations can show RMS amplitudes of almost 100 nm. This is primarily a result of proximity to cultural noise sources.
Similarly, RMS noise amplitudes for our low, median and high noise models show systematic variation across the UK that generally reflects proximity to noise sources and site geology. Sites on soft rock in the south east of England show high noise levels, whereas sites on hard rock in remote rural locations show low noise levels.
Previous models of the detection capability used constant noise levels at all stations, with 2 nm, 4 nm and 20 nm for the low, median and high noise models. We use our results to determine detection capability for a network where noise varies realistically. The results suggest a rather better detection capability in the UK than previously expected.
The Amatrice earthquake sequence
Following the devastating Amatrice earthquake in the Central Apennines of Italy, BGS secured funding from NERC to deploy 24 earthquake sensors in the affected area to supplement permanent and temporary stations deployed by the Istituto Nationale Geofisica e Vulcanologia (INGV). This provides an unparalleled dataset to analyse how each earthquake within the sequence contributes to the next, and how this behaviour evolves through space and time.
In August 2016, a destructive earthquake sequence, including at least five events with magnitude larger than 5.4 Mw, began to unfold in Central Italy. The events spanned a 50 km fault zone that has been active in both historical and modern times. The loss of life and the damage to buildings underline the pressing need to understand the complexity of the underlying physics of earthquake sequences, and to use this knowledge to anticipate the evolution of such sequences in future.
After the first earthquake in the sequence, on 24 August 2016, BGS, with funding from NERC, deployed 24 sensors to supplement the 28 permanent and 23 temporary stations deployed by Istituto Nationale Geofisica e Vulcanologia (INGV) and the 19 accelerometer stations operated by the Italian Department of Civil Protection. This network has an average station spacing of ~5 km and will provide an unparalleled dataset to analyse how each earthquake within the sequence contributes to the next, and how this behaviour evolves through space and time.
A NERC funded project is focussing on the development of testable forecast models for informed decision-making and the creation of a scientific protocol for stress-based modelling applicable to global seismicity.
This research has been carried out in collaboration with the University of Edinburgh and INGV-Rome. Recent scientific results were presented in the British Seismological Meeting and in the Annual Meeting of the Seismological Society of America.
The development of stress-based models for aftershock forecasting of evolving sequences presents a clear advantage over easier statistical approaches that rely on empirical knowledge but do not improve our understanding of earthquake physics. Instead physics-based approaches, as shown in the figure below, allow us to improve our knowledge on earthquake nucleation and the conditions under which large earthquakes nucleate.
The unprecedented, for Europe, dataset has been the basis for an international collaboration with UK (BGS, Universities of Edinburgh and Bristol), USA (University of Stanford, US Geological Survey, Lamont-Doherty Observatory of Columbia University) and European (INGV-Rome, EPOS) institutes. The project aims to explore the processes driving this destructive earthquake sequence and quantify how each earthquake in a series contributes to the next, and how this behaviour evolves through space and time.
Supporting self-recovery after disasters
‘Self-recovery’ refers to what most households affected by disasters do to ‘repair, build or rebuild their shelter themselves or through local builders’ (Schofield and Miranda Morel, 2017[2]). BGS are part of a consortium with CARE International, University College London and the Overseas Development Institute undertaking research to better understand how self-recovery can be better supported by the humanitarian sector, geoscientists and engineers.
Self-recovery (SR) tends to be the predominant route to recovery after disasters and often happens with little or no external assistance (Parrack et al., 2014[3]). It is crucial that this process is well-supported by scientific knowledge of geohazards and the environment to help communities build back safer and better and not ‘rebuild risk’.
BGS scientists have undertaken community-based fieldwork with humanitarian practitioners, engineers and social scientists to investigate self-recovery from a range of perspectives. So far, we have explored cases of self-recovery in rural communities following rapid-onset disasters in the Philippines (typhoons in 2013 and 2015) and Nepal (the 2015 Gorkha earthquake).
A strong awareness of the environment is common to the communities we visited in both the Philippines and Nepal. In the Philippines, people’s understanding of geohazards appears to come primarily from first-hand experience (typhoons occur regularly) and through transfer of ancestral knowledge, with more varying and limited direct input from scientific organisations. There is evidence that individuals’ awareness of geohazards and perceptions of event frequency have influenced some rebuilding.
In Nepal, the focus was on rural communities in Dhading District that had been severely affected by the 25 April 2015 Gorkha earthquake. Besides the direct impact of the earthquake and its aftershocks on shelter, many of these communities and the roads leading to them were, and continue to be, affected by landslides. There were also many reports of water supplies being disrupted by the earthquake. This, and ongoing damage to roads, is impeding the recovery process.
Rebuilding efforts focus on seismic resistance but it is clear that these communities are now exposed to a multiple geohazards at places that were previously considered safe. There is very limited scientific input into the self-recovery decision-making process although some information regarding safe siting of houses is given by the government.
The two cases show the impact of the natural environment on SR and the limited extent to which scientific knowledge supports this process. Finding ways for geoscience to better support SR is therefore crucial.
Correcting local magnitude estimates discrepancies at near-event distances
A Local Magnitude scale is used throughout the BGS earthquake catalogue. The scale is similar to the original Richter Scale. Recent research has shown that amplitude measurements from epicentral distances of less than 15–20 km considerably overestimate event magnitudes compared to more distant observations. We have revised the existing magnitude scale to correct for this effect.
Recent research has shown that amplitude measurements from epicentral distances of less than 15–20 km considerably overestimate event magnitudes compared to more distant observations (Butcher et al., 2017[4]). Similarly, magnitudes calculated for earthquakes induced by hydraulic fracturing at Preese Hall, Lancashire (Clarke et al., 2014[5]) using ground motions recorded on seismometers at distances of a few kilometres away were unrealistically high.
A detailed examination of the BGS earthquake catalogue shows that individual station magnitudes for stations within 5 km of an earthquake are up to an order of magnitude higher than station magnitudes at other stations (Luckett et al., 2017[6]). In many cases this would cause a considerable increase in the event magnitude, compared to the magnitude expected from macroseismic information. As a result, such amplitudes have not been included when calculating the magnitude.
The A0 term in Richter’s (1935) local magnitude relationship can be expressed as:
− log10 𝐴0 = 𝑎 log10 𝑟 + 𝑏 𝑟 + 𝑐
where r is the hypocentral distance and a, b and c are constants. The a and b terms represent the effect of geometrical spreading and attenuation respectively. Hutton and Boore (1987)[7] find the following is equivalent to the original Richter tables for California.
− log10 𝐴0 = 1.11 log10 𝑟 + 0.00189 𝑟 − 2.09
These values of the constants are currently used for determination of earthquake magnitude in the UK. Ottemöller and Sargeant (2013)[8] used data recorded on the BGS seismic network to develop an ML scale for the United Kingdom, finding a similar relationship to Hutton and Boore (1987).[7]
− log10 𝐴0 = 1.06 log10 𝑟 + 0.00121 𝑟 − 1.98
Butcher et al. (2017)[4] suggest that the magnitude discrepancy is a result of higher attenuation in near-surface geology, and requires a change in the attenuation term of the ML scale. They use data collected at distances of less than 10 km from a sequence of mining events near New Ollerton, Nottinghamshire, to determine new constants for the ML scale, finding the following values
− log10 𝐴0 = 1.17 log10 𝑟 + 0.0514 𝑟 − 3.0
Butcher et al. (2017)[4] suggest that the increase in the attenuation term 0.00189 to 0.0514 is representative of a raypath within a slower, more attenuating sedimentary layer compared to the continental crust and that this magnitude scale should be used when local monitoring networks are within 5 km of the event epicentres. Strictly, this scale is only valid for data from the New Ollerton sequence, however, Butcher et al. (2017)[4] show that it gives reasonable results when applied to the earthquakes induced by hydraulic fracturing at Preese Hall. Additionally, the scale cannot be used above the suggested cut-off distance of 5 km as it will result in incorrect estimates of magnitude. This cut-off distance is not well constrained.
Luckett et al. (2017)[6] suggest that the higher than predicated amplitudes at distances of less than 10–20 km are a result of high amplitude surface waves. These are an important part of the waveform for shallow sources at distances of less than 20 km, but attenuate quickly with distance. They suggest adding an extra exponential term to account for this effect, and determine the following expression for a UK data set.
− log10 𝐴0 = 1.11 log10 𝑟 + 0.00185 𝑟
− 1.16𝑒−0.2𝑟 − 2.09
This expression results in a significant reduction in residuals when compared to the Hutton and Boore (1987)[7] relationship. Luckett et al. (2017)[6] also apply this relationship to data from Preese Hall and New Ollerton and find that the results are in good agreement with those of Clark et al. (2014)[9]and Butcher et al. (2017).[4]
References
- ↑ Baptie, B, Segou, M, Ellen, R, and Monaghan, A. 2016. Unconventional Oil and Gas Development: Understanding and Monitoring Induced Seismic Activity. British Geological Survey Open Report, OR/16/042.
- ↑ Schofield, H, and Miranda Morel, L. 2017. Whose recovery? Power, roles and ownership in humanitarian shelter assistance, Humanitarian Exchange, no. 69, 29–30
- ↑ Parrack, C, Flinn, B, and Passey, M. 2014. Getting the Message Across for Safer Self- Recovery in Post-Disaster Shelter. Open House International, vol.39 (3)
- ↑ 4.0 4.1 4.2 4.3 4.4 4.5 Butcher, A, Luckett, R, Verdon, J P, Kendall, J‐M, Baptie, B, and Wookey, J. 2017. Local magnitude discrepancies for near‐event receivers: implications for the U.K. Traffic‐ Light Scheme. Bulletin of the Seismological Society of America, 107, 2, 532–541.
- ↑ Clarke, H., Eisner, L., Styles, P. and Turner, P., 2014. Felt seismicity associated with shale gas hydraulic fracturing: The first documented example in Europe. Geophysical Research Letters, 41, 23, 8308–8314.
- ↑ 6.0 6.1 6.2 6.3 Luckett, R, Baptie, B, Butcher, A, and Ottemöller, L. 2017. Correcting Local Magnitudes at for Near-Event Distances. Submitted to Bulletin of the Seismological Society of America.
- ↑ 7.0 7.1 7.2 7.3 Hutton, L K, and Boore, D M. 1987. The ML scale in southern California, Bulletin of the Seismological Society of America, 77, 2074–2094.
- ↑ 8.0 8.1 Ottemöller, L, and Sargeant, S. 2013. A local magnitude scale ML for the United Kingdom, Bulletin of the Seismological Society of America, 103, 2884–2893.
- ↑ Clarke, H, Eisner, L, Styles, P, and Turner, P. 2014. Felt seismicity associated with shale gas hydraulic fracturing: The first documented example in Europe. Geophysical Research Letters, 41, 23, 8308–8314.