Search

Earthquake site response is an essential part of seismic hazard assessment, especially in densely populated areas. The shallow geology of the Netherlands consists of a very heterogeneous soft sediment cover, which has a strong effect on the amplitude of ground shaking. Even though the Netherlands is a low- to moderate-seismicity area, the seismic risk cannot be neglected, in particular, because shallow induced earthquakes occur. The aim of this study is to establish a nationwide site-response zonation by combining 3D lithostratigraphic models and earthquake and ambient vibration recordings. As a first step, we constrain the parameters (velocity contrast and shear-wave velocity) that are indicative of ground motion amplification in the Groningen area. For this, we compare ambient vibration and earthquake recordings using the horizontal-to-vertical spectral ratio (HVSR) method, borehole empirical transfer functions (ETFs), and amplification factors (AFs). This enables us to define an empirical relationship between the amplification measured from earthquakes by using the ETF and AF and the amplification estimated from ambient vibrations by using the HVSR. With this, we show that the HVSR can be used as a first proxy for site response. Subsequently, HVSR curves throughout the Netherlands are estimated. The HVSR amplitude characteristics largely coincide with the in situ lithostratigraphic sequences and the presence of a strong velocity contrast in the near surface. Next, sediment profiles representing the Dutch shallow subsurface are categorised into five classes, where each class represents a level of expected amplification. The mean amplification for each class, and its variability, is quantified using 66 sites with measured earthquake amplification (ETF and AF) and 115 sites with HVSR curves. The site-response (amplification) zonation map for the Netherlands is designed by transforming geological 3D grid cell models into the five classes, and an AF is assigned to most of the classes. This site-response assessment, presented on a nationwide scale, is important for a first identification of regions with increased seismic hazard potential, for example at locations with mining or geothermal energy activities.

Deliverable from WP3 of TURNkey project

Measurements of pore pressure variations contain important information for the production of hydrocarbon reservoirs. Here, we couple surface-wave velocity changes directly to pore pressure change using sensitivity kernels. Using these kernels for realistic pore pressure changes in the Harlingen chalk gas reservoir, we modelled yearly Rayleigh-wave velocity changes. The velocity variations appear sufficiently large to be detectable on a yearly basis using passive image interferometry. This implies that, for specific formations, field-wide pore pressure monitoring is feasible with continuous seismic noise recordings at the surface.

A comprehensive database that has been used to develop ground-motion models for induced earthquakes in the Groningen gas field is provided. The database includes more than 8,500 processed ground-motion recordings from 87 earthquakes of local magnitude ML between 1.8 and 3.6, obtained from a large network of surface accelerographs and borehole geophones at 50 m intervals to a depth of 200 m. The 5%-damped pseudo acceleration spectra and Fourier amplitude spectra of the records are also provided. Measured shear-wave velocity (VS) profiles, obtained primarily from seismic CPTs, are provided for 80 of the ~100 recording stations. A site response model is also applied for the entire gas field plus a 5 km onshore buffer zone, specifying lithology, VS and damping for all layers above the reference baserock horizon located at about 800 m depth. For both the recording station locations and the field-wide model, the database includes transfer functions and non-linear frequency-dependent amplification factors from the reference rock horizon to the surface. The database provides a valuable resource for further refinement of induced seismic hazard and risk modeling in Groningen as well as for generic research in site response of thick, soft soil deposits and the characteristics of ground motions from small-magnitude, shallow-focus induced earthquakes. The data herein are presented and discussed in the paper of Ntinalexis et al. (2022), which has been submitted to Earthquake Spectra.

This report details the procedures that have been applied, developed or improved in Task 3.4 for the rapid generation of ground shaking maps (i.e., shake-maps). The report revolves around several actions and technical results, which are organized as follows:  Section 3 summarizes the state-of-the-art of current shake-map algorithms and systems, based on a review by Guérin-Marthe et al. (2021). The USGS ShakeMap v4 approach is compared to a method based on Bayesian updating, which is put forward as one of the technical solutions to be implemented in the TURNkey platform.  Section 4 is based on previous work carried out in Task 3.3.2: in each TB, various GMMs (Ground-Motion Models) are evaluated and ranked, via different scoring metrics. The selected GMMs may then be used for the generation of shake-maps in each TB.  Section 5 details novel research efforts, where the GMM coefficients are updated in order to match the observations. This approach is useful to rapidly update a GMM in a given area for a given earthquake, so that the updated GMM may be reused for subsequent events (e.g., in a seismic sequence) in order to improve the accuracy of ground-shaking estimates. Two parallel and complementary methods are presented, namely a direct calibration of the coefficients (EUC) and a Bayesian updating of the uncertain coefficients in the shake-map (BRGM).  Section 6 discusses site amplification models that are available in each TB: besides local models, Vs30 maps generated at the European level (in the European project SERA) are extracted for each TB. Recommendations are given for each TB on which model should be applied, and whether the SERA model represents a satisfying approximation.  Section 7 explores additional sources of observations that may be used to characterise shake-maps: collection and aggregation of felt reports by EMSC; and extraction of Twitter data and integration as soft evidence in the shake-maps by BRGM. The Bayesian approach for the derivation of shake-maps has been implemented in a Python code, which is briefly described in the companion deliverable report D3.9.

Botswana is a country with relatively low seismic activity that experienced an unexpected Mw 6.5 earthquake on 3 April 2017. Using data from the first countrywide seismic network, we established a Botswana earthquake catalog for the period January 2014 to February 2018. Two areas of elevated seismic activity were detected. The first one is the Okavango Rift Zone in northern Botswana, an area that is known to be active. The other one is associated with the 2017 mainshock and its aftershocks in central Botswana; it follows the Paleoproterozoic suture between the Limpopo Belt and the Kaapvaal Craton. Double-difference relocation of these events revealed a reactivated fault system with a northwesterly strike with the aftershocks occurring at shallower depths than the mainshock at 29 km. The focal mechanisms of the mainshock and selected aftershocks are of normal faulting type with strikes similar to the orientation of the fault system. The unidirectional rupture of the mainshock in the lower crust combined with the upward migration of the aftershocks along the Moiyabana Fault and a thin low velocity anomaly the uppermost mantle are consistent with the events being produced by eclogitization of a dry metastable granulite facies rock by fluid intrusion from the mantle in an extensional stress regime. The Okavango Rift Zone is generally interpreted as the terminus of the southwestern branch of the East African Rift System. We suggest that the recent earthquakes in central Botswana may be considered as the southern continuation of this branch.

Similar to horizontal earthquake motions, vertical motions are amplified depedent on the local site conditions which can be critical for the safety of certain structures. Production of natural gas in Groningen, the Netherlands, results in reservoir compaction causing low magnitude, shallow earthquakes which are recorded with a borehole seismic network. These recordings form an excellent data set to understand how shallow unconsolidated subsurface geology influences the amplification behaviour of compressional waves (P-waves). First, we present borehole and single-station techniques (amplification factors, empirical transfer functions (ETF) and V/H spectral ratio implementations) to quantify vertical amplification. We show that vertical-wave incidence is a reasonable assumption. All techniques are capable of emphasising the sites with strong amplification of vertical ground motion during an earthquake. Subsequently, we compare ETF with single-station methods with the aim to develop proxies for vertical site-response using spectral ratios. In a second step, we link vertical site-response with shallow subsurface conditions, like the P-wave velocity and peat content. To better understand the amplification mechanisms, we analytically simulate P-wave propagation. In the simulations, we compute synthetic transfer functions using realistic subsurface conditions and make a comparison with the ETF. The simulations support the hypothesis that thin layers of shallow gas, originating from the Holocene peat, result in wave amplification. We observe strong vertical site-response in particular in the eastern part of Groningen, with industrial facilities and pipeline infrastructure in the region. Here, if high vertical amplifications are persistent at large earthquake magnitudes, appreciable levels of vertical loading may be expected. This study demonstrates that vertical motions should be assessed separately from horizontal motions, given that the amplification behaviour of P-waves is affected by distinctive mechanisms.

Previous studies examining the relationship between the groundwater table and seismic velocities have been guided by empirical relationships only. Here, we develop a physics-based model relating fluctuations in groundwater table and pore pressure with seismic velocity variations through changes in effective stress. This model justifies the use of seismic velocity variations for monitoring of the pore pressure. Using a subset of the Groningen seismic network, near-surface velocity changes are estimated over a four-year period, using passive image interferometry. The same velocity changes are predicted by applying the newly derived theory to pressure-head recordings. It is demonstrated that the theory provides a close match of the observed seismic velocity changes.

We use a novel technique named global-phase seismic interferometry (GloPSI) to image the lithospheric structure, and in particular the Moho, below two parallel north-south transects belonging to the GANSSER network (2013–2014). The profiles cross the Himalayan orogenic wedge in Bhutan, a tectonically important area within the largest continent-continent collision zone on Earth that is still undergoing crustal thickening and represents a challenging imaging target for the GloPSI approach. GloPSI makes use of direct waves from distant earthquakes and receiver-side reverberations with near vertical incidence. Reflections are isolated from earthquake recordings by solving a correlation integral and are turned into a reflectivity image of the lithosphere below the arrays. Our results compare favorably with first-order features observed from a previous receiver function (RF) study. We show that a combined interpretation of GloPSI and RF results allows for a more in-depth understanding of the lithospheric structure across the orogenic wedge in Bhutan., Frontiers in Earth Science, 9, ISSN:2296-6463

Since 1991, induced earthquakes have been observed and linked to gas production in the Groningen field. Recorded waveforms are complex, resulting partly from a Zechstein salt layer overlying the reservoir and partly from free-surface reverberations, internal multiples, interface conversions, guided waves, and waves diving below the reservoir. Therefore, picking of polarities or amplitudes for use in moment tensor inversion is problematic, whereas phase identification may be circumvented employing full waveform techniques. Although moment tensors have become a basic tool to analyze earthquake sources, their uncertainties are rarely reported. We introduce a method for probabilistic moment tensor estimation and demonstrate its use on the basis of a single event within the Groningen field, concentrating on detailed tests of input data and inversion parameters to derive rules of good practice for moment tensor estimation of events recorded in the Groningen field. In addition to the moment tensor, event locations are provided. Hypocenters estimated simultaneously with moment tensors are often less sensitive to uncertainties in crustal structure, which is pertinent for the application to the Groningen field, because the task of relating earthquakes to specific faults hitherto suffers from a limited resolution of earthquake locations. Because of the probabilistic approach, parameter trade-offs, uncertainties, and ambiguities are mapped. In addition, the implemented bootstrap method implicitly accounts for modeling errors affecting every station and phase differently. A local 1D velocity model extracted from a full 3D velocity model yields more consistent results than other models applied previously. For all velocity models and combinations of input data tested, a shift in location of 1 km to the south is observed for the test event compared to the public catalog. A full moment tensor computed employing the local 1D velocity model features negative isotropic components and may be interpreted as normal fault and collapse at reservoir level.

Recent developments in the densification of the seismic network covering the Groningen gas field allow a more detailed study of the connection between induced seismicity and reactivated faults around the gas reservoir at 3 km depth. With the reduction of the average station distance from 20 km to 4–5 km, a probabilistic full-waveform moment tensor inversion procedure could be applied, resulting in both improved hypocenter location accuracy and full moment tensor solutions for events of M≥2.0 recorded in the period 2016–2019. Hypocenter locations as output from the moment tensor inversion are compared to locations from the application of other methods and are found similar within 250 m distance. Moment tensor results show that the double-couple (DC) solutions are in accordance with the known structure, namely normal faulting along 50°–70° dipping faults. Comparison with reprocessed 3D seismic sections, extended to a depth of 6–7 km, demonstrate that (a) most events occur along faults with a small throw and (b) reactivated faults in the reservoir often continue downward in the Carboniferous underburden. From non-DC contributions, the isotropic (ISO) component is dominant and shows consistent negative values, which is expected in a compacting medium. There is some indication that events connected to faults with a large throw (>70 m) exhibit the largest ISO component (40%–50%).

Up to now, almost all of the ground motion modeling and hazard assessment for seismicity in the Netherlands focuses on horizontal motion. As a rule of thumb, the strength of vertical ground motions is taken as 2/3 of that of horizontal ground motions. In reality of course, amplifications and V/H ratios are site-dependent and thus vary regionally. Recent studies have indeed shown that vertical ground motion is not always simply 2/3 of the horizontal motion. However, these studies are performed in areas with high magnitude (Mw>5.0) earthquakes and the question is whether vertical motion is relevant to be included in seismic hazard assessment for low magnitude earthquakes (to date, max Mw=3.6 in Groningen).In the Netherlands, the top part of the soils is practically always unconsolidated, so the elastic waves generated by deeper (~3000m) seated earthquakes will be subject to transformation when arriving in these layers. Recordings over a range of depth levels in the Groningen borehole network show the largest amplification to occur in the upper 50 meters of the sedimentary cover. We not only observe a strong amplification from shear waves on the horizontal components, but also from longitudinal waves on the vertical component. A better understanding of vertical motion of low magnitude earthquakes aims to support the design of re-enforcement measures for buildings in areas affected by low magnitude seismicity. Furthermore, interference between the longitudinal -and shear waves might contribute to damage on structures.This study presents observations of longitudinal wave amplification in the frequency band 1-10 Hz, corresponding to resonance periods of Dutch buildings. From 19 seismic events, with a minimum of magnitude two, we retrieved transfer functions (TFs) from the vertical component, showing a strong site response at certain locations. In addition, we calculate event V/H ratios and VH factors from the surface seismometer. These results are compared with the TFs and show a similar pattern in terms of site response. Furthermore, the sites with highest vertical amplification correspond to very low (800-900 m/s) P-wave velocities. Our study shows that vertical amplification is very site dependent. However, the question whether the vertical motion is significant enough to form a real hazard can only be answered through cooperation between seismologist and structural engineer.

The characterization of the megaregolith on the Moon has been investigated in various ways including analysis of lunar meteorites, remote sensing of mineralogy and gravity, and deriving a seismic velocity profile. In this study, we propose a method for analyzing azimuthal anisotropy of the megaregolith. We call this method deep-moonquake seismic interferometry applied to S-wave coda (DMSI-S). DMSI-S can turn the records of deep moonquakes into recordings from virtual active sources. The retrieved virtual sources coincide with the station positions, and thus, we obtain virtual zero-offset (pulse-echo) measurements. DMSI-S is applied to seven clusters of deep moonquakes recorded at the Apollo 14 landing site, resulting in virtual zero-offset measurements at the Apollo station 14. We use the S-wave recordings retrieved from DMSI-S to analyze azimuthal anisotropy. We find weak anisotropy at the layer where the megaregolith is assumed to be present. We interpret our result to show that the megaregolith at this site is characterized by a layer (or layers) of impact material, following the Imbrium impact, with internal alignment of the crushed material.