Special Report on the Westerwijtwerd Earthquake 22Nd May 2019

Special Report on the Westerwijtwerd Earthquake 22nd May 2019

Datum May 2019 Editors Jan van Elk and Dirk Doornhof

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Contents 1 Summary...... 5 2 Introduction ...... 6 3 Analysis of Measurements and Observations of the Westerwijtwerd Earthquake ...... 7 3.1 Processing of the Earthquake records ...... 7 3.2 Surface Ground Motions measured by KNMI Network ...... 8 3.2.1 Peak Ground Accelerations and Velocities ...... 11 3.2.2 Ground-Motion Durations ...... 16 3.2.3 Spectral Accelerations and Comparison with the GMM ...... 20 3.3 Concluding Remarks Surface Ground Motions ...... 24 3.4 Determination of Hypocentre Location ...... 25 3.4.1 Standardised Operational Method ...... 25 3.4.2 Full Waveform Inversion (FWI) method, ...... 25 3.5 Production and Pressures ...... 28 3.6 Reported Building Damage ...... 34 3.7 TNO Household Sensors ...... 35 3.8 Analysis ...... 38 4 MRP status 22nd of May 2019 ...... 41 4.1 Activity rate ...... 42 4.2 Earthquake density ...... 44 4.3 PGA and PGV ...... 48 4.4 Damage state ...... 48 4.5 Other patterns and considerations ...... 49 4.5.1 Loppersum trends ...... 49 4.5.2 Probability earthquakes with higher magnitude and b-factor ...... 50 5 Intervention measures and their estimated effect on seismicity ...... 52 6 References ...... 53 7 Appendix A – Letter Evaluatie: Westerwijtwerd aardbeving ...... 56

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

1 Summary This Special Report, which was announced in NAM’s letter of 24th May 2019 (NAM, 2019a), has been prepared in accordance with the Measurement and Control protocol (MRP) (NAM, 2017a). It addresses the earthquake on the 22nd May 2019 at 5:49 local time (CEST), near the village of Westerwijtwerd.

The Westerwijtwerd earthquake had a magnitude of ML 3.4 on the Richter Scale, as established by KNMI. The earthquake was widely felt and caused (DS1) damage to buildings in the region around the epicentre. No falling objects (e.g. chimneys) or physical injuries have been reported. In this report, the measurements and observations obtained during the Westerwijtwerd earthquake are analysed in chapter 3. The assessment of these measurements and observations of ground motions and building damage showed that in all aspects, the Westerwijtwerd earthquake was in line with the current understanding of the field. No special characteristics of the Westerwijtwerd earthquake have been identified, or characteristics that deviate from current modelling results. On this basis, no revision of the Hazard and Risk Assessment (HRA) is required at this time. In chapter 4, the report provides an overview of the status and trend of all parameters used in the MRP framework. This showed the “waakzaamheidsniveau” (vigilance level) in the MRP has been reached for PGV. The other parameters are in the monitoring range. The expected effects of mitigation measures on seismicity and the parameters in the MRP are discussed in chapter 5. The analyses described above, which have been more extensive than those prepared in the first 48 hours after the earthquake, do not lead to a requirement to revise and restate NAM’s assessment of the situation, as described in its letters to Staatstoezicht op de Mijnen (SodM) and Ministry of Economic Affairs and Climate Policy (MEAC) of 24nd May 2019.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

2 Introduction This Special Report has been prepared in accordance with the Measurement and Control protocol (MRP) (NAM, 2017a), following the earthquake on the 22nd May 2019 at 5:49 local time (CEST and 3:49:00 UTC), near the village of Westerwijtwerd. This was the third largest earthquake in Groningen, th following the Westeremden earthquake of 8 August 2006, with a magnitude of ML 3.5 on the Richter th Scale, and the Huizinge earthquake of 16 August 2012, with a magnitude of ML 3.6. This earthquake has the same magnitude as the Zeerijp earthquake of 8th January 2018. During this earthquake, the largest ground acceleration, 0.042 g was recorded by one of the KNMI stations, station G170, located at 1.7 km from the epicentre. The largest PGV was registered by station G210, located at 3.6 km from the epicentre. This largest PGV was 1,0 cm/s. As a result of this PGV, the “waakzaamheidsniveau” (vigilance level) threshold of the Measurement and Control Protocol (MRP) was exceeded. The MRP requires the Risk Coordination Team (RCT)1 to meet within a few days when the ‘vigilance’ level is triggered. In this case NAM elected to treat the earthquake as a very significant event, based on the magnitude and the societal impact, and convened the RCT in line with the highest level in the escalation structure of the MRP. The RCT of NAM met that same day to analyse the event.

The Westerwijtwerd earthquake had a magnitude of ML 3.4 on the Richter Scale, as established by KNMI. The earthquake was widely felt and caused (DS1) damage to buildings. Using the methodology developed by USGS, “Did you feel it?“ (USGS, 2011), the earthquake was felt at a large distance (~18km) from the epicentre, including parts of the city of Groningen. Some 115,000 houses are located in the area where the earthquake was felt. Especially in the area near Westerwijtwerd, where the largest vertical accelerations have been recorded, the earthquake was experienced as frightening. The number of buildings that have been exposed to a peak ground velocity larger than 2 mm/s (corresponding to the damage contour of the TU Delft) is about 44,500. Within the contour based on the SBR Guideline (masonry in normal state) some 6,500 buildings are located. On 28 May 2019, the Tijdelijke Commissie Mijnbouwschade Groningen (TCMG) reported to have received some 2,600 damage claims. After preliminary analysis of the available data, letters (NAM, 2019a) were prepared, detailing the initial assessment of the impact of the earthquake and providing the Minister of Economic Affairs and Climate and the regulator SodM. The current “Special Report on the Westerwijtwerd Earthquake – 22nd May 2019” presents the technical analysis carried out during the 2-week period following the Westerwijtwerd earthquake. In detail, some of the technical conclusions have been further refined since the 48-hour letter was prepared, although no fundamental new insights have come to light since then. It is possible that future studies carried out as part of the Studies and Data Acquisition Plan will generate new insights.

1 The MRP requires the RCT to meet within a few days when the ‘vigilance’ level is triggered. In this case NAM elected to treat the earthquake as a very significant event, based on the magnitude and the societal impact. NAM convened the RCT within 24 hours, in line with the highest level in the escalation structure of the MRP. 6 Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3 Analysis of Measurements and Observations of the Westerwijtwerd Earthquake 3.1 Processing of the Earthquake records In this section of the Special Report, the measurements and observations of the Westerwijtwerd Earthquake are reported and discussed. An initial analysis of these measurements and observations is provided, focussing on the assurance of the hazard and risk modelling. The records presented in this section of the report are the final product of a processing procedure designed to maximise the quality and usability of the data. Modern digital accelerographs do not record acceleration directly; instead, they record the change in voltage of a transducer inside the accelerometer, which occurs in order to provide a force balance to the acceleration experienced. The proportion of voltage to acceleration is called the instrument gain. The voltage change recorded is also affected by the frequency characteristics of the sensor; together with the instrument gain, this is called the instrument response. The pure ground-motion record is obtained by removing the instrument response, i.e. performing the instrument correction. The KNMI provides the records in the form of raw volt counts in mini-seed files, and the instrument response in the form of a zero-pole-gain transfer function in an .xml or dataless seed (dseed) inventory. The instrument correction is performed by deconvolving the response function from the Fourier Spectra of the records. The record obtained from that process contains acceleration in m/s2 and is unaffected by the frequency characteristics of the sensor. The next step necessary to acquire accurate values of PGV and long-period response spectral accelerations is to apply a low-frequency (high-pass) filter to the records. The records presented here were processed using the methodology prescribed by Ntinalexis et al. (2015), i.e., using an 8th-order acausal Butterworth low-cut filter. The filter corner-frequency was determined separately for each record by identifying the frequency at which the decay of the Fourier spectra towards lower frequencies deviates from the f2 trend. After each record is filtered, its displacement trace is computed and checked visually for noise. If noise can still be observed in the displacement trace, in terms of deviations, a lower frequency is chosen and filtering is repeated, until no more noise is apparent. The horizontal components of each record are always filtered with the same filter parameters, using the lower corner-frequency of the two, in order to preserve their common time basis. The application of a high-frequency (high-cut) filter to the records is not necessary, as high-frequency spectral acceleration values can be accurately obtained also without removing the high-frequency noise, as they are less sensitive to the filter cut-off than long-period ordinates. This can be understood as a result of the relation between Fourier amplitude and response spectra, and the fact that high- frequency response spectral ordinates are poorly correlated with high-frequency Fourier spectra amplitudes (Bora et al., 2016). Different studies (e.g., Akkar et al., 2011; Douglas & Boore, 2010) have in fact demonstrated that high-cut filters are required only in particular circumstances and are only desirable for records obtained from the older, analogue sensors, which have undergone digitisation (Boore & Bommer, 2005). The processing of the earthquake records is a time-consuming effort requiring focus and time. KNMI aims to make spectral accelerations and PGV available on their RRSM website very quickly after an earthquake occurs and therefore have set up a process of automatic processing of the records. Part of that processing involves applying two 4th order Butterworth filters to the records, one with corner

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019 frequencies of 0.05 – 40 Hz and another with corner frequencies of 0.3 – 80 Hz. The rather low high- cut corner frequency of the former filter has led to some differences between the processed recordings presented in this chapter and those reported by KNMI. The most significant difference is for the recording of station G170 located closest to the epicentre of the Westerwijtwerd earthquake. Due to its proximity to the epicentre, and because the ruptured fault is a steep normal fault, the recording displays a strong P-wave arrival, which produces the maximum amplitude in the recording. The energy of this arrival is concentrated at a frequency of 60 Hz in the Fourier spectra; hence, the application of the bandpass filter removes the energy of the P-wave from the record. After the application of the filter, the PGA computed corresponds to the S-wave. The vertical acceleration for station G170 is also the recording of the largest PGA for the Westerwijtwerd earthquake 0.07 g. The largest horizontal PGA evaluated for the G170 station by KNMI is 0.034 g. After processing of the data without application of the high-cut filter, NAM evaluated the largest horizontal PGA to be slightly higher at 0.042 g. For stations further from the epicentre, the amplitude of the P-wave in the recordings diminishes and the peak of the record corresponds to the arrival of the S-wave, whose energy is concentrated in lower Fourier spectral frequencies. Therefore, the influence of the application of the bandpass filter is significantly smaller, as is the difference in the values computed following the two procedures. The KNMI reported horizontal accelerations and velocities will be used in discussing the Monitoring and Control protocol (MRP). This difference does not impact the functioning of the MRP for the Westerwijtwerd earthquake. 3.2 Surface Ground Motions measured by KNMI Network On Wednesday 22 May 2019 at 03:49:00 UTC (11 minutes to 6 am local time), an earthquake occurred near the village of Westerwijtwerd in the municipality of Loppersum (Figure 3.1). In common with all induced earthquakes in the Groningen field, a focal depth of 3 km was assigned by KNMI, who reported a local magnitude of ML 3.4. Only two larger earthquakes have occurred in the Groningen field, the largest being the ML 3.6 Huizinge earthquake of August 2012 and the second largest the ML 3.5 Westeremden earthquake of August 2006. One earthquake of the same magnitude has occurred in Zeerijp in January 2018.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.1 Epicentre of Westerwijtwerd earthquake (green star) together with epicentres of previous earthquakes of ML ≥ 2.5 (red stars) and of ML 1.8-2.4 (blue stars) In keeping with the trend during more recent earthquakes (Figure 3.2), the latest earthquake has triggered a large number of accelerograms, as a direct result of the expansion of the strong-motion recording monitoring network in the Groningen field operated by KNMI (Dost et al., 2017).

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.2 Diagram illustrating the timing of earthquakes of ML ≥ 2.5 in the Groningen field and the number of records yielded by the permanent KNMI network (B-stations, blue) and by the expanded borehole geophone network (G-stations, red). The Westerwijtwerd earthquake has added another 83 records to the database of ground motions.

The KNMI portal (http://rdsa.knmi.nl/opencms/nl-rrsm) made accelerograms from the earthquake available within an hour of the event and 83 three-component recordings were downloaded and processed for this preliminary assessment of the motions. Figure 3.3 shows these recordings in the magnitude-distance occupied by the database used to derive the current ground-motion model used for seismic hazard and risk analyses in the Groningen field. This report presents an overview of the recorded motions in terms of their amplitudes and durations and discusses how the recorded amplitudes of motion compare with predictions from the ground-motion models.

Figure 3.3 Magnitude-distance distribution of the Groningen strong-motion database including the recordings of the 2019 Westerwijtwerd earthquake.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3.2.1 Peak Ground Accelerations and Velocities Figures 3.4 and 3.5 show the horizontal values of PGA and PGV of three component definitions from each recording obtained during the Westerwijtwerd earthquake plotted against the distance of the recording site from the epicentre. The largest amplitudes were obtained at the G170 station located 1.7 km from the epicentre: the PGAs recorded at this station are 40.70 cm/s2 (0.042 g) on the H1 component and 18.29 cm/s2 on the H2 component. The largest PGV values are at the station G210, 3.61 km from the epicentre, and are 0.99 cm/s (H2) and 0.50 cm/s (H1).

Figure 3.4 As-recorded horizontal components of PGA recorded during the Westerwijtwerd earthquake plotted against epicentral distance

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.5 As-recorded horizontal components of PGV recorded during the Westerwijtwerd earthquake plotted against epicentral distance

Figure 3.6 shows the geometric mean values of PGA and PGV of the historical earthquakes, with the peaks from this latest earthquake superimposed, plotted as a function of magnitude. This figure confirms the information already conveyed in Figures 3.4 and 3.5, namely that the recorded amplitudes from this latest earthquake are comfortably within the range of previously observed values of PGA and PGV. Indeed, the maximum PGA and PGV values recorded in the Westerwijtwerd earthquake have been exceeded several times in previous earthquakes, including the Zeerijp earthquake of 8th January 2018, which was of the same magnitude. However, it does need to be noted that there are fewer recordings from very short epicentral distances in the case of the Westerwijtwerd event.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.6 Geometric mean horizontal components of PGA (upper) and PGV (lower) recorded during the Westerwijtwerd earthquake (red) and in previous earthquakes (blue) plotted against local magnitude. Figure 3.7 shows the horizontal components of PGA and PGV obtained within 7 km of the epicentre, from which it can be appreciated that the very strong polarisation often observed in Groningen recordings (e.g., Bommer et al., 2017a) is also apparent in some of near-source recordings of this event, particularly those displaying the highest absolute amplitudes of motion.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.7 Horizontal components of PGA (upper) and PGV (lower) recorded during the Westerwijtwerd earthquake at epicentral distances of less than 7 km; units are cm/s2 and cm/s, respectively.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

An interesting feature of these recordings, which has also been observed in previous events in the field, is the occurrence of the maximum horizontal peak of acceleration at the closest station (G170) at the very beginning of the recording (Figure 3.8). As can be appreciated from the traces in Figure 3.8, the strongest component of this accelerogram is the vertical component and the maximum amplitudes in both the horizontal and vertical time-histories are almost exactly coincident at the start of the traces. Figure 3.8 also shows that there appears to be a second sub-event occurring about 2 seconds after the first wave arrivals, which is discussed below in the context of the duration of motion. The Huizinge earthquake of 2012 showed a similar feature. Despite the small number of records for this earthquake, these motions were extensively studied (Wentinck 2018a and b). Figure 3.8 indicates that the maximum horizontal peaks of velocity are related to this second wave train.

Figure 3.8 Horizontal and vertical components of acceleration and velocity from the G170 station

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

The Westerwijtwerd earthquake has contributed a large number of low-amplitude recordings, a feature also clearly visible for the ML 2.6 Slochteren, ML 3.1 Hellum and ML 3.4 Zeerijp earthquakes, reflecting the expansion of the recording networks in the Groningen field (Figure 3.6). As already shown in Figures 3.4 and 3.5, the amplitudes decay rapidly with distance although the effect of simultaneous arrivals of direct and critically refracted/reflected waves leads to an increase in amplitudes at some locations between 12 and 20 km from the epicentre. However, these effects do not lead to significant absolute amplitudes at those distances and it is clear from Figures 3.6 and 3.7 that outside of the epicentral area the motions are generally of very low amplitude: < 0.01g for PGA and < 0.3 cm/s for PGV. Using the most recent version of the empirical GMPE for the prediction of PGV (Bommer et al., 2019a), the residuals show a strongly negative event term for this earthquake (Figure 3.9), which indicates that on average the amplitudes of motion from this latest earthquake are low compared with the average trends in the current Groningen ground-motion database. However, as Figure 3.9 clearly shows, there have been previous events with similar or even larger negative event- terms but the event-terms for the Westerwijtwerd earthquake are the lowest among the four largest earthquakes that have so far occurred in the field. In this context, it may be relevant to note that KNMI reports a significant uncertainty of ± 0.25 associated with the estimation of the magnitude of this earthquake (Bernard Dost, personal communication, 2019).

Figure 3.9 Event-terms for three different horizontal component definitions, with the Westerwijtwerd earthquake indicated by the red triangles, relative to predictions from the empirical GMPEs of Bommer et al. (2019a). 3.2.2 Ground-Motion Durations The maximum amplitude of ground shaking, whether represented by PGA or PGV, provides a simple indication of the strength of the motion but the potential for adverse effects—such as damage to masonry buildings or triggering liquefaction—also depends on the duration or number of cycles of the motion. A feature that has been consistently observed in the Groningen ground motions is a very pronounced negative correlation between PGA and duration, with high amplitude motions being associated with shaking of very short duration and more distant recordings being associated with low amplitudes and generally longer durations (Bommer et al., 2016). The same pattern is observed in the recordings of the Westerwijtwerd earthquake, as shown in Figure 3.10. Figure 3.11 shows how this pattern is consistent with general trends in the Groningen ground-motion database.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.10 Individual components of PGA and duration from the Westerwijtwerd event, the symbols indicating the epicentral distance of the recording.

Figure 3.11 Pairs of PGA and significant duration for individual components of the Westerwijtwerd records, superimposed on a plot of the same data for the full V6 database.

The largest value of PGA, recorded on the EW component at the G170 station, is associated with a duration of less than two seconds (1.9 s), which is not exceptionally short and would seem to be a consequence of the ‘double event’ discussed earlier (Figure 3.8). The horizontal components of both acceleration and velocity from this station are shown in Figure 3.12, which also shows the build-up of

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Arias intensity (which is a measure of the energy in the motion) over time. The plot indicates that the relatively ‘long’ duration for this recording is due to the interval separating the two sub-events.

Figure 3.12 Horizontal components of acceleration and velocity from the G170 station; the upper frame shows the accumulation of Arias intensity (energy) over time.

The two apparent sub-events are only observed in this recording. Figure 3.12 shows a similar plot for the G210 recording — the fifth closest instrument to the epicentre but source of the second largest value of PGA — where the larger amplitude component is associated with a significant duration of just 0.56 s.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.13. Horizontal components of acceleration and velocity from the G210 station; the upper frame shows the accumulation of Arias intensity (energy) over time.

Figure 3.14 shows the event terms of significant duration—based on the interval of accumulation from 5% to 75% of the total Arias intensity—with respect to the V5 GMM (Bommer et al., 2017b). The figure shows that the Westerwijtwerd durations have a positive event-term indicating that they are slightly longer, on average, than the median values predicted by this model. However, it is also clear that the event term is in no way exceptional and it is very close to that of the ML 3.4 Zeerijp event of January 2018.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.14 Event terms of duration, with the red triangle indicating the average offset of the durations from the Westerwijtwerd earthquake

In conclusion, the durations of motion recorded in this most recent event are entirely consistent with the trends observed in previous earthquakes, despite the influence of an apparent double event. Taking into account the slightly longer-than-average durations with respect to the predictive model and the pronounced inverse correlation between duration and amplitude observed in the Groningen field, it would be expected that the recorded amplitudes will be slightly lower than average with regards to predictive models derived from the Groningen ground-motion database; this has so far been confirmed for PGV (see Figure 3.9). 3.2.3 Spectral Accelerations and Comparison with the GMM The database for the GMM extended with the earthquake data collected in 2018 from the Zeerijp and Garsthuizen earthquakes (Bommer et al., 2018), which, in common with all of the previous versions, predicts the geometric mean horizontal components of motions. Figure 3.15 shows the residuals of the PGA (which are equivalent to the response spectral acceleration at an oscillator period of 0.01 second) and PGV. The residuals are plotted against distance in this figure and the key observation that can be made is the residuals are generally negative: one can easily infer a general offset downwards, which is consistent with a negative event term, as discussed above. Residuals are calculated from the logarithmic ratio of observed to predicted amplitudes, so a negative value corresponds to a ratio of less than 1, indicating the observed amplitude is over-predicted by the model. The conclusion that begins to emerge then is that the motions from the Westerwijtwerd earthquake are in no way exceptional or unusual among Groningen earthquakes, but within the general trends reflected indicated by the predictive models derived from the recordings in the database, this latest earthquake displays slightly longer than average durations of motions coupled with slightly lower than average peak amplitudes of acceleration and velocity.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.15 Residuals of PGA (left) and PGV (right) with respect to the central branch of the GMM database extended with the earthquakes in 2018: Zeerijp and Westeremden, plotted against rupture distance.

The fragility functions used in the estimation of seismic risk in the Groningen field are defined in terms of response spectral accelerations at various oscillator periods (Crowley et al., 2017). The horizontal acceleration response spectra from the G170 and G270 recordings of the Westerwijtwerd earthquake are shown in Figure 3.16. The spectra show significantly different shapes and amplitudes, which reflects different distances and azimuths of these recording stations with respect to the earthquake source, as well as different near-surface geological profiles and corresponding amplification factors, even though these two stations have similar values of VS30. Figure 3.17 shows the residuals of geometric mean horizontal accelerations at 8 different oscillator periods, plotted against distance in the same way as the PGA and PGV residuals in Figure 3.15.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.16 Horizontal response spectra from the G170 (left) and G210 (right) station recordings; note the different y-axis scales in the two plots.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.17 Residuals of Sa(T) at 8 response periods, T, relative to the GMM database extended with the earthquakes in 2018: Zeerijp and Westeremden, against distance

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

In all cases, as with PGA and PGV, there is considerable scatter but not clear trends with respect to distance. For the periods of 0.05 and 0.1 s, the model seems to be a good fit, with the residuals more or less centre around zero. At longer oscillator periods, there is tendency towards a negative event- term, which increases with period with significant over-prediction at periods on the order of 1 second. In each case, the residual is the natural logarithm of the ratio of the observed (recorded) to the median predicted value, so a residual of 0.7 indicates that the recorded value was underestimated by a factor of 2 by the model and a residual of -0.7 would indicate over-prediction by a factor of 2. 3.3 Concluding Remarks Surface Ground Motions nd The ML 3.4 Westerwijtwerd earthquake of 22 May 2019 has contributed a large body of ground- motion recordings that will inform and enrich the ongoing work of developing hazard and risk estimation models. The largest component of PGA recorded in this earthquake is 0.04g. However, the largest value of PGV—which is generally considered a better indicator of the damage potential of the motion—recorded in this latest event is 0.99 cm/s, which is significantly smaller than the largest value of the database, a 3.46 cm/s recorded in the Huizinge earthquake. The average amplitudes of the motion are generally over-predicted by the GMM V5 extended by the data collected during the earthquakes in 2018 (Zeerijp and Garsthuizen) while the durations, as would be expected in such a case, are slightly longer than average. However, the motions are consistent with previous recordings from Groningen earthquakes and there is no reason to conclude that this is an exceptional earthquake in any sense.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3.4 Determination of Hypocentre Location 3.4.1 Standardised Operational Method Event locations as reported by KNMI are calculated using the first arrival times of the P waves at the geophone or accelerometer stations. In the calculations, a generalized 1D velocity model over Groningen is used. With this method, results can be delivered very quickly, in an automated fashion. The vertical (depth) resolution however is rather poor; the events are therefore set at a fixed depth of 3 km, the average depth of the Groningen gas reservoir. The lateral positioning uncertainty is about 500 m, but can be as large as 1,000 m, in the most unfavourable circumstances. These circumstances are determined by local geology (Zechstein salt geometry, presence of geologic structure such as grabens, etc), the magnitude of the event (signal/noise ratio) and the number and position of stations that have recorded the event. Specifically, the quality of the earthquake location can be highly dependent on the number and availability of the nearby stations, within a couple of kilometres form the epicentre. 3.4.2 Full Waveform Inversion (FWI) method, With the aim of obtaining a higher fidelity event location and computing a focal mechanism solution, we apply the Full Waveform Inversion (FWI) method. As the name suggests, the FWI method makes use the full earthquake record, including the S-wave arrival. With the aid of a detailed 3D local velocity model derived from the available 3D seismic data and sonic logs, a more accurate hypocentre location can be determined, including depth. The focal mechanism of an earthquake describes the deformation in the source region that generates the seismic waves. It numerically and visually describes the fault plane that slipped and the slip vector. Strike, dip and rake are the 3 angles that describe this mechanism. Focal mechanisms are derived from a solution of the moment tensor for the earthquake, which itself is estimated by the application of the FWI method to the observed seismic waveforms as provided by KNMI. Our automated Full Waveform event location and Moment Tensor Inversion workflow has produced results for the Westerwijtwerd ML 3.4 event (Figure 3.18). The earthquake has been extensively covered by measurements from the monitoring network, as described in section 3.1, and the phase arrivals show high signal to noise ratio (as would be expected for an event of this magnitude). In this report we present preliminary results from the Full Waveform Inversion method. At this time, we are continuing to analyse this event in more detail, and we intend to provide an update once we reach any significant improvements or new insights. The key finding thus far is that the location objective function space, from which we choose the optimal event location, is very noisy – much noisier than what would be expected from an M3.4 event. In general, there are several potential causes for this and we list them below: ▪ Low Signal to Noise ratio in the recordings – however not applicable for an event of this magnitude and good quality signal-to-noise records ▪ Poorly sampled wavefield or in other words lack of azimuthal coverage – generally not the case as this event was well recorded (as discussed in section 3.1); however, the records within the first couple of kilometres around the epicentre are sparse and that can have an impact on the stability of the inversion ▪ Deviations from the velocity model used in the Green’s Function modelling – due to the proximity of a geologic structure i.e. the “mini-graben”, it is possible that subsurface velocities are poorly determined especially in the Carboniferous; also, any increase in anisotropy in the

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

local region not captured in the model could have an effect on the final results. Other possibilities also exist and we are continuing to look into these.

Figure 3.18 Left: The preliminary FWI location (black dot) and the KNMI location (red dot); Right: the preliminary focal mechanism solution shown on top of the fault map indicating normal fault slip on a NW-SE striking fault.

The location of the epicentre for the Westerwijtwerd as determined by KNMI and NAM using the FWI are very close together. The epicentre is located near a mapped fault. The source mechanism indicates normal fault movement, in line with the stress orientation and mapped fault throw. Figure 3.19 shows 11 events that have occurred in the vicinity of the Westerwijtwerd event. Although there is some uncertainty in the location of each events, especially for the smaller, older events before the network was densified to the present standard, these events would have contributed to any local change of stresses in the local fault structure system. The seismicity in the immediate vicinity of the Westerwijtwerd earthquake has been relatively low, with only two events higher than M1 occurring in 2014.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.19 KNMI catalogue of 12 local and nearby events between 02-03-2013 and 28-05-2019 in the Westerwijtwerd area. Picture from KNMI seismic data portal website

Event nr Date Magnitude 1. 02-03-13 0.9 2. 27-06-13 1.3 3. 18-02-14 1.7 4. 13-05-14 1.4 5. 08-12-15 0.8 6. 01-02-16 0.5 7. 04-04-16 0.5 8. 06-07-16 0.3 9. 10-08-16 0.5 10. 15-11-17 0.8 11. 22-05-19 3.4 Table 3.1 Summary table of the events in the closest vicinity of the hypocentre of the Westerwijtwerd earthquake (last entry).

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3.5 Production and Pressures The Westerwijtwerd hypocentre is located in the northwest area of the field, commonly referred to as the “Loppersum area”. Figure 3.20 shows the approximate hypocentre with reference to the surrounding well locations on a semblance map of base Zechstein (top reservoir) which brings out fault locations. It can be seen that there are no well penetration to the West of the identified fault. To the East, there are numerous wells (BRH-1, SDM-1, ODP-1) which have all been well matched in the dynamic reservoir simulator, building confidence in our understanding of the reservoir pressure behavior. However, due to the absence of well penetration in the Eastern fault block, the reservoir pressure in that fault block cannot be measured and is uncertain. In turn, this western block is bound to the west by a fault that fully offsets the Rotliegend reservoir Figure 3.25), which is believed to cause a pressure disconnect with the adjacent Rodewolt aquifer. Offtake from the nearby production clusters2 was constrained by the Minister of Economic Affairs on 17/1/20143, and production has been fully shut-in since 2/2/2018, Figure 3.21. Due to this absence of production, the pressure behaviour in this area is very gradual. Further away from the hypocentre, there have been production fluctuations prior to the event: to accommodate the relatively warm period in April followed by the colder period in May (Figure 3.22), Groningen production was ramped- down respectively ramped-up. Figure 3.23 shows the associated ramp-down/ramp-up in the East- Central area4, and the ramp-up in the South-West area5. Note that this production fluctuation was somewhat dampened by reducing the injection rate of Groningen gas into the Norg UGS (Figure 3.24). It can be observed from Figure 3.23 (right-hand panel) that the associated production clusters are over 15 km away from the hypocentre. The physical behaviour of a highly compressible fluid (gas) in a porous medium has a strongly dampening effect on pressure fluctuations with distance away from fluctuating producing wells further reduced by destructive interference. As a result, the pressure decline around the hypocentre was not affected by these far-away production fluctuations. Figure 3.25 shows a three-dimensional view from the dynamic reservoir simulator (NAM, 2018b), highlighting the approximate location of the hypocentre. Figure 3.26 shows the modeled pressure decline in the period leading up to the seismic event from gridblocks on both sides of the identified fault. The gridblock locations are highlighted on the left hand side of the panel. On the right hand side, the pressure decline can be seen to be very gradual in the period leading up to the event on 22/5/2019, with a pressure decline rate around 2 bar/year. To illustrate the uncertainty in reservoir pressure away from well control, the results from an alternative reservoir simulation model (NAM, 2016b) is given in Figure 3.27. This model matches the measured reservoir pressures at the well penetrations equally well, but alternative choices were made for the model permeability and fault transmissibility. Consistently, in this model realisation the simulated pressure decline is also very gradual.

2 Loppersum clusters: Leermens, Overschildt, Ten Post, De Paauwen, ‘t Zandt 3 Annual production constraint of 3.0 Bcm per year 4 East-Central area consists of clusters AMR, TJM, SDB, OWG and SCB. 5 South-West area consists of clusters EKR, SZW, ZPD, FRB, KPD, SLO, SAP, TUS, ZVN and SPI. 28

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.20 Wells and production clusters in the vicinity of the hypocentre (yellow circle). Locations near the epicentre are observation well locations and Loppersum clusters.

Figure 3.21 Daily production rates of clusters in the vicinity of the hypocentre

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.22 Temperatures at De Bilt in April and May 2019. Source: https://weerstatistieken.nl/de-bilt/2019/april

Figure 3.23 Daily production rates per region (coloured by production cluster). Regions are highlighted by color in the right-hand panel, with the approximate hypocentre location as a yellow circle.

Figure 3.24 Norg production/injection

Figure 3.25 3D visualisation of reservoir pressure (22/5/2019) from the full field model (V5) in the top of the Slochteren formation. Colorscale clipped at 150 bar. Approximate epicentre location is indicated as a yellow circle

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.26 Pressure trends over time for two gridblocks on either side of the suspect fault (Mores model V5)

Figure 3.27 Pressure trends over time for two gridblocks on either side of the suspect fault (Mores model V2.5)

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3.6 Reported Building Damage An earthquake of magnitude in the range of the Westerwijtwerd earthquake is expected to cause aesthetic damage (DS1) to buildings located near the epicentre. The probability of damage occurring is dependent on the level of ground shaking at each specific building location; PGV is typically used as an effective indicator for the intensity of ground movement for assessing DS1 damage potential. The median PGV near the epicentre of ~20 mm/s reduces to 5 mm/s at epicentral distances of ~4.3 km and to 3 mm/s at ~5.9 km. The DS1 contour areas are given in figure 3.27 below. These estimates are based on an evaluation of the ground motions Empirical GMPE March 2019, “SBR Trillingsrichtlijn A: Schade aan bouwwerken:2017” and the “TU Delft Schademethodiek versie 8” (TU Delft, 2009). Radius of DS1 Populated buildings contour area within range SBR criterium 5.4 km 6300 Metselwerk pand; Bouwkundige staat: normaal P75 PGV 5 mm/s SBR criterium 8.5 km 15000 Metselwerk pand; Bouwkundige staat: gevoelig P75 PGV 2.94 mm/s TU Delft criterium (versie 8) 12 km 44500 P75 PGV = 2 mm/s Table 3.2 Table of the region showing the DS1 damage contour for 3 criteria (with 25% probability of exceedance).

Figure 3.27 Map of the region showing the DS1 damage contour for 3 criteria (with 25% probability of exceedance).

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

During the first week following the Westerwijtwerd earthquake, TCMG reported on their website that some 2,603 damage claims have been received. As a comparison, in the first week after the Zeerijp earthquake of 8th January 2018 2,605 damage claims were received (NAM, 2018a). The map in Figure 3.28 shows those claims received per municipality (status 28 May 2019, 16:00 hrs). TCMG has also received 77 notices of acute unsafe situations (AOS); 66 have been visited, of which 13 have been confirmed.

Figure 3.28 Screen caption from TCMG website with EQ epicentre added. 3.7 TNO Household Sensors For the Westerwijtwerd event, also data from TNO household sensors has been collected and analysed. Figure 3.29 shows PGV values of the three definitions used in the Empirical PGV GMPE, from the records obtained from the B-, G- and Household networks, plotted against distance. It also shows the median predictions of the March 2019 Empirical PGV GMPE, adjusted using the event-terms, as well as the 16th and 84th percentile predictions, which correspond to one intra-event standard deviation (phi) on each side of the median. The event-terms used were calculated using only B- & G- network records. It has been shown that ground-floor level Household network sensors produce reliable PGV values, hence in this comparison data from all sensors of the houses, irrespective of whether they are on the ground-floor wall or in the cellar/basement/crawl space have been used.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 3.29 Westerwijtwerd records obtained by TNO household sensors and KNMI stations (B and G network) compared.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

The records of the three networks are compatible and show consistent trends of amplitude with distance. Also, the event-adjusted Empirical GMPE predicts the amplitudes very well throughout the distance range, while the bulk of the data are within one standard deviation of the adjusted median. This confirms the network-to-network consistency in terms of PGV, the good fit of the model to the data and its validity for predicting PGV values recorded at houses.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

3.8 Analysis For the quality control of the hazard and risk modelling, it is important that we analyse the measurements and observations and identify any potential characteristics of the recent Westerwijtwerd earthquake that do not fall within the anticipated frame of the modelling, which could prompt a potential re-evaluation of this modelling. The MRP (NAM, 2017a) in section 10.4, sets out this quality assurance process (Figure 3.30).

Figure 3.30 Cause/effect diagram for failures in the hazard and risk model for induced quakes in Groningen taken from MRP (NAM, 2017a) (Chapter 10.5, pg 36)

In this section of the Special Report (chapter 3), an overview and analysis of the measurements and observations of the Westerwijtwerd earthquake is presented. This includes: ▪ intensity of the measured ground motions (both velocity and acceleration), ▪ duration of the ground motions, ▪ magnitude of the earthquake, ▪ hypocentre location of the earthquake relative the faults in the reservoir, ▪ building damage reported. The assessment of the measurements and observations of the Westerwijtwerd earthquake showed that in all aspects, the Westerwijtwerd earthquake was within the anticipated range indicated by the hazard and risk modelling. No special characteristics of the Westerwijtwerd earthquake, deviating from the modelling, have been identified. In table 5.3 (on page 88) of the “Hazard, Building Damage and Risk Assessment – November 2017”

(NAM, 2017e) the probability of an earthquake with a magnitude larger than ML=3.6 during 2019, was estimated at 17% (Figure 3.31). Some 6 years and 9 months have lapsed since the Huizinge earthquake of magnitude ML = 3.6. Based on the higher volumes produced during these years, the Westerwijtwerd

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019 earthquake (the second earthquake of this magnitude during this period) was in this respect therefore not an unanticipated or surprising event, and not outside the modelling frame.

The seismological model is used to forecast the seismicity in terms of, the number location and magnitude of future earthquakes. The probability of an earthquake with a magnitude exceeding a given magnitude can be assessed. In table 5.3 the annual probability of an earthquake occurring with a magnitude exceeding the specified magnitude is given. For instance, the probability of an earthquake occurring in 2019 with a magnitude exceeding ML=3.6 (the magnitude of the Huizinge earthquake) is equal to 17%. YEAR M≥3.6 M≥4.0 M≥4.5 M≥5.0 2018 16.0% 6.6% 1.6% 0.4% 2019 17.0% 7.0% 1.6% 0.4% 2020 17.8% 7.5% 1.8% 0.4% 2021 19.3% 8.0% 1.9% 0.5% 2022 20.2% 8.7% 2.2% 0.6% Table 5.3 Table with annual probabilities for occurrence of earthquakes exceeding a set magnitude.

The probabilities for the occurrence of an earthquake with magnitude ML≥3.6, have not changed since the assessment for Winningsplan 2013. For the larger magnitudes, there is a slight reduction in the probability of occurrence since the assessment for Winningsplan 2013. Over time, these probabilities slightly increase when the field is produced at a constant gas production rate. However, over these longer times a-seismic relaxation of stresses in the reservoir might reduce seismicity below this forecast, as this effect has not been included in the model.

Figure 3.31 Excerpt from the “Hazard, Building Damage and Risk Assessment – November 2017” (NAM, 2017e) showing table 5.3 on page 88 of this document. These probabilities are based on a production scenario of 24 Bcm/year.

In the HRA report of March 2019 (NAM, 2019a) the probabilities of occurrence of an earthquake exceeding a magnitude were updated based on the production profile GTS-raming 2019. This shows that despite the reduction in production implemented in the profile GTS-raming 2019, the probability of an earthquake exceeding 3.6 for 2019 is still 12.57 % or 12.25% for operating strategy 1 and 2 respectively. The Westerwijtwerd earthquake was recorded by 83 stations of the seismic monitoring network. The recordings were made available by KNMI within 1 hour of the earthquake. The accelerations, velocities and durations of the earthquake records fit comfortably within the ranges of earlier earthquakes and of the models based on these earthquakes. The ground motions recorded were not unremarkable for Groningen earthquakes. During the Westerwijtwerd earthquake the largest PGA observed was 40.70 cm/s2. This PGA is about half of the PGA measured during the Huizinge earthquake in 2012 with magnitude 3.6 and only some 40% of the largest PGA recorded during the Zeerijp earthquake in 2018 which also had a magnitude of 3.4. However, although the Zeerijp earthquake was observed by about the same number of accelerometer stations of the KNMI, it does need to be noted that there are fewer recordings from very short epicentral distances in the case of the Westerwijtwerd event. This possibly explains why, despite having similar magnitudes, the largest PGA observed during the Westerwijtwerd earthquake is lower than the largest PGA observed during the Zeerijp earthquake.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

The observed ground motions were within the predictions of the Ground Motion Model (GMM). The observations showed the same short durations and the same strong negative correlation between PGA and duration. The hypocentre of the Westerwijtwerd earthquake was located near an identified fault. The number of damage claims received by the TCMG is published at their web-site. On the 28 May, TCMG had received 2,600 damage claims. This is virtually the same number of damage claims as received after the Zeerijp earthquake of 8th January 2018. In this report a technical analysis of the Westerwijtwerd earthquake is presented focussing on the quality control as part of the Measurement and Control Protocol. In addition to the calculation of personal safety, there is also another relevant safety component, the perception and feeling of safety (Royal HaskoningDHV 2016, 2017a and 2017b, NAM 2018c). According to the OVV, this describes the feeling of safety, peace of mind and the imminent disruption of the enjoyment of life among the residents of Groningen. The instemmingsbesluit also explicitly refers to the social consequences and feelings of fear. We recognise this is also an important component of safety, that is not addressed in this technical report.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4 MRP status 22nd of May 2019 The status of the MRP dashboard right after the Westerwijtwerd earthquake is shown in table 4.1. This chapter, including the dashboard, is based on the data provided by KNMI as part of the operational seismic monitoring of the Groningen field. The difference in data processing approaches between KNMI and NAM may lead to small differences in the presented data, as is explained in section 3.1. These differences do not affect the MRP-level after the Westerwijtwerd earthquake in the dashboard. The status of the MRP is used to provide an overall assessment of the current (seismic) situation.

MRP status

th th May 23 May 8 2019 2019 Activity Rate 11 11 (# Earthquakes, M ≥ 1.5) EQ density -2 -1 0.12 0.10 (km jr , M ≥ 1.0) Horizontal PGA 0.03* - ** (in “g”, last event M ≥ 2.0) Horizontal PGV (most recent maximum, in 10.4* - ** mm/s) Data not Data not Damage State available available Other patterns - - Table 4.1 MRP status in May 2019. *(Westerwijtwerd, M=3.4, 22-5-2019) **(For the Garsthuizen, M=2.81, on 13-4-2018 the horizontal PGA was 0.04 g and the horizontal PGV 7.6 mm/s. As this earthquake was over year prior to the reference date this earthquake did not impact the MRP dashboard.)

The activity rate remained constant at 11, on MRP-level ‘Monitoring’. Earthquake density increased to a value of 0.12 km-2yr-1, well below the upper level for monitoring, which is 0.17 km-2yr-1. The maximum horizontal PGA associated with the Westerwijtwerd earthquake was 0.04 g, at level ‘Monitoring’. The horizontal PGV associated with the Westerwijtwerd earthquake was 10.4 mm/s and fell within the range of the MRP-“waakzaamheidsniveau” (vigilance level).

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4.1 Activity rate After the Westerwijtwerd earthquake the activity-rate remained at the ‘Monitoring’-level. The activity rate reached a value of 11, which is the same as the value at 1st April 2019 reported in the half-yearly review (NAM, 2019b). Figure 4-1 shows the observed 12-month numbers for two magnitude classes, as well as the production over this period. The activity rate for magnitude class ML ≥ 1.5 decreased from a plateau around 20 in 2018 to a level around 11 early 2019. The activity rate for magnitude class

ML ≥ 1.0 has decreased steadily from mid-2018, when it was near 50 to about 20 in April 2019. Since April is has increased from 21 to 24 earthquakes in the previous 12 months.

Figure 4-1 Temporal evolution of the activity rate for the last 12 months for two magnitude classes.

In Figure 4-2 and Figure 4-3 the 12-month numbers are displayed for two different magnitude classes for a longer time period. The 12-month number for magnitude class ML ≥ 1.5 has been declining and is at its lowest since early 2017. The 12-month number for magnitude class ML ≥ 1.0 has been declining and was at the lowest since 2015, before increasing by 3 this May.

12 MONTHS NR - M≥1.5 40 35 30 25 20 15 10 5

02 07 12 00 01 03 04 05 06 08 09 10 11 13 14 15 16 17 18

------

sep sep sep sep sep sep sep sep sep sep sep sep sep sep sep sep sep sep sep

Figure 4-2 Historical development of the 12-month number for magnitude class ML ≥ 1.5.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

12-MAANDS-GETAL - M ≥ 1 60

16 17 18 19

15 18 15 16 16 17 17 18

16 17 18 19

- - - -

------

- - - -

feb feb feb feb

aug aug aug aug

nov nov nov nov

mei mei mei mei Figure 4-3 Historical development of the 12-month number for magnitude class ML ≥ 1.0.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4.2 Earthquake density The earthquakes that contributed to the current earthquake density value of 0.12 km-2yr-1 are shown in Table 4.2. Earthquake density maps for different reference times are given in Figure 4-4. This choice of calculation method (and MRP threshold values) for earthquake-density leads to a deliberate early and conservative triggering of this MRP parameter with the intent of early picking up signals of changing subsurface conditions (the other intent of this parameter is to simply pick up an increase of concentration of earthquakes in a certain area without special underlying cause but potentially causing nuisance nevertheless).

Figure 4-4 Earthquakes Density in Groningen based on the year preceding 1st of April, 1st of May and 23rd of May 2019.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Location Magnitude Municipality Date Westerwijtwerd 3.4 Loppersum 22/05/2019 04:49 Wirdum 1.0 Loppersum 19/05/2019 23:42 Appingedam 1.3 Appingedam 11/05/2019 14:27 Uithuizermeeden 1.0 Het Hogeland 08/05/2019 09:50 Loppersum 1.0 Loppersum 03/05/2019 19:21 Ten Post 1.8 Groningen 25/04/2019 20:27 Noordwolde 1.6 Het Hogeland 15/04/2019 08:03 Zuidbroek 1.4 Midden-Groningen 10/04/2019 05:40 Spijk 1.4 Delfzijl 09/04/2019 20:26 Kantens 1.5 Het Hogeland 29/03/2019 13:37 Toornwerd 1.6 Loppersum 28/02/2019 23:16 Appingedam 1.6 Appingedam 19/02/2019 12:15 Zeerijp 1.4 Loppersum 16/02/2019 19:22 Luddeweer 1.0 Midden-Groningen 01/02/2019 20:41 Lageland 1.3 Midden-Groningen 23/12/2018 23:16 Garrelsweer 1.2 Loppersum 22/12/2018 04:54 Bedum 1.0 Bedum 19/12/2018 18:55 Eppenhuizen 1.6 Eemsmond 25/11/2018 19:37 Eppenhuizen 1.6 Eemsmond 09/11/2018 15:41 Siddeburen 1.1 Midden-Groningen 16/10/2018 11:31 Sappemeer 1.2 Midden-Groningen 17/09/2018 15:01 Appingedam 1.8 Appingedam 09/08/2018 09:01 Appingedam 1.9 Appingedam 08/08/2018 03:55 Sappemeer 1.1 Midden-Groningen 13/07/2018 09:05 Siddeburen 1.1 Midden-Groningen 11/07/2018 14:52 Noordbroek 1.0 Midden-Groningen 07/07/2018 04:55 Scharmer 1.6 Midden-Groningen 27/06/2018 15:32 't Zandt 1.6 Loppersum 22/05/2018 00:27 Garrelsweer 1.6 Loppersum 06/05/2018 17:39 Garsthuizen 2.8 Loppersum 13/04/2018 22:31 Table 4.2 Earthquakes in the Groningen field since April 2018, Magnitude ML ≥ 1.0

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 4-5 Earthquake numbers and magnitudes in the Groningen field, region South West, LOPPZ, East and Eemskanaal (top to bottom).

Figure 4-5 shows earthquake numbers and magnitudes for specific regions over a longer period. Figure 4-6 shows the temporal development of earthquake density different areas. It shows that the earthquake density in the LOPPZ region is low, even though the Westerwijtwerd ML = 3.4 is included in the dataset.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Figure 4-6 Evolution of the earthquake density map over time.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4.3 PGA and PGV Figure 4-7 shows the PGA value associated with the Westerwijtwerd earthquake. It shows that this earthquake caused a relatively low measured PGA (0.03 g). Figure 4. shows the PGV value for the Westerwijtwerd earthquake. The PGV value of 10.4 mm/s remained below the signalling level, for which the threshold value is 50 mm/s. Both values seem low for a Magnitude ML = 3.4 event, but not anomalously low.

Figure 4-7 Plot of PGA values observed in the Groningen field (complete for M ≥ 2.0).

Figure 4.8 Plot of PGV values observed in the Groningen field (complete for M ≥ 2.0). 4.4 Damage state At this moment, no observations have been made to suggest that DS2 (structural damage) has occurred as a result of the Westerwijtwerd earthquake.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4.5 Other patterns and considerations 4.5.1 Loppersum trends Figure 4.9 shows the number of earthquakes in the Loppersum and East areas for earthquakes with a magnitude of 1.5 and higher.

Figure 4.9 Yearly number of earthquakes for the Loppersum area and for area “East”. 2019 data is included up until 24th of May 2019.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

4.5.2 Probability earthquakes with higher magnitude and b-factor The probability of higher magnitude earthquakes was reported in reference (Van Elk, Mar-Or, Geurtsen, Uilenreef, & Doornhof, 2019) and is reproduced here for discussion purposes. It shows that the probability for having one or more earthquakes with a magnitude of 3.6 or higher is 12.57 % for 2019. That probability keeps decreasing for consecutive years. Year P(M>=3.6) P(M>=4.0) P(M>=4.5) P(M>=5.0) 2019 12.57% 4.69% 0.97% 0.18% 2020 10.62% 3.93% 0.86% 0.15% 2021 9.74% 3.70% 0.76% 0.14% 2022 7.84% 3.01% 0.64% 0.14% 2023 6.64% 2.53% 0.53% 0.11% 2024 6.01% 2.24% 0.48% 0.08% 2025 5.52% 2.15% 0.45% 0.08% 2026 4.94% 1.90% 0.40% 0.08% 2027 4.36% 1.57% 0.36% 0.07% 2028 4.21% 1.56% 0.32% 0.07% Table 2.4 Annual probabilities of seismic events per magnitude class for operational strategy 1 (Van Elk, Mar-Or, Geurtsen, Uilenreef, & Doornhof, 2019).

Year P(M>=3.6) P(M>=4.0) P(M>=4.5) P(M>=5.0) 2019 12.25% 4.63% 0.99% 0.20% 2020 9.99% 3.78% 0.77% 0.13% 2021 9.00% 3.28% 0.74% 0.14% 2022 7.62% 2.86% 0.65% 0.12% 2023 6.57% 2.51% 0.56% 0.10% 2024 6.11% 2.24% 0.48% 0.09% 2025 5.76% 2.15% 0.46% 0.09% 2026 5.09% 1.88% 0.42% 0.09% 2027 4.92% 1.88% 0.37% 0.08% 2028 4.52% 1.72% 0.38% 0.05% Table 4.3 Annual probabilities of seismic events per magnitude class for operational strategy 2 (Van Elk, Mar-Or, Geurtsen, Uilenreef, & Doornhof, 2019).

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

The calculated b-factor after the Westerwijtwerd earthquake lies within the uncertainty bounds of the b-value as calculated just before the Westerwijtwerd earthquake (see Figure 4-8). Thus the Westerwijtwerd earthquake does not suggest a change in the b-value.

Figure 4-8 Estimates of the “b-factor”, before and after the Westerwijtwerd 3.4 earthquake for the whole field, with a

magnitude of completeness of ML = 1.5. The top left panel displays the cumulative distribution function for the b-value just before the Westerwijtwerd Earthquake. The top right panel displays the observed number of earthquakes for different magnitudes just before the Westerwijtwerd Earthquake. Above the top right panel

is the P50 of the b-value displayed, as well as the P10 and P90 between brackets. The bottom left panel displays the cumulative distribution function for the b-value just after the Westerwijtwerd Earthquake. The bottom right panel displays the observed number of earthquakes for different magnitudes just after the

Westerwijtwerd Earthquake. Above the bottom right panel is the P50 of the b-value displayed, as well as the P10 and P90 between brackets.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

5 Intervention measures and their estimated effect on seismicity The PGV parameter of the MRP has exceeded the alertness level. Other parameters remained within the monitoring range. This Special Report was prepared in response to the exceedance of this level by the PGV monitoring parameter. As discussed in Chapter 3 and 4, the following observations have been made for the Westerwijtwerd earthquake: 1. The maximum PGV of the Westerwijtwerd earthquake reached the “waakzaamheidsniveau” (vigilance level). 1. All other parameters monitored in the MRP remain in the monitoring range. 2. The duration of this seismic event was comparable of that of earlier earthquakes (e.g. Huizinge earthquake of August 2012), and as expected. 3. The epicentre of the earthquake is at a location far from any currently produced locations. 4. In 2017, the probability of an earthquake of the magnitude of Westerwijtwerd occurring during 2019 was estimated at >17%. Based on the new production profile GTS-raming 2019, this probability was in March 2019 estimated at > 12 – 13% (depending on the Operational Strategy). 5. At the time of writing this report, no confirmed DS2 damage caused by the Westerwijtwerd earthquake has yet been reported. 6. No new area has become seismically active. 7. There are no special observations with respect to faults. 8. The Westerwijtwerd earthquake caused significant anxiety and social upheaval. In summary even though the event has been analysed by NAM as significant in the context of the MRP, it is unlikely to lead to a new quantitative safety assessment (which would have made the event more significant). The epicentre of the earthquake is located on an identified fault in the reservoir at a considerable distance from the currently active production locations. The most effective measure to reduce the number of earthquakes in the Groningen field remains the reduction of the production of gas. Due to gas production, the pressure in the gas-filled pores of the reservoir declines. Reduced production will, after some time has elapsed to allow pressure difference to equilibrate over the reservoir, also lead to reduced seismicity in the Loppersum area. The incidence of further larger earthquakes cannot be ruled out. From the above conclusions and the detailed analyses in this report, no new information has come to light that would lead us to revise the assessment made in our 48-hour letter. The Minister’s existing production policy and plans will lead to reduction of the number of earthquakes and will reduce the probability of large earthquakes proportionally. NAM sees no reason to revise any of the information it provided to the Minister in preparation for Instemmingsbesluit 2018 and/or Vaststellingsbesluit 2019 (i.e. Operational Strategy). NAM will continue to act in line with these Ministerial decisions and apply its best technical assessment of events like the Westerwijtwerd earthquake in line with the Measurement and Control Protocol.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

6 References 1. Akkar, S., Ö. Kale, E. Yenier, & J.J. Bommer (2011). The high‐frequency limit of usable response spectral ordinates from filtered analogue and digital strong‐motion accelerograms. Earthquake Engineering & Structural Dynamics, 40(12), 1387-1401. 2. Bommer, J.J., B. Dost, B. Edwards, P.J. Stafford, J. van Elk, D. Doornhof & M. Ntinalexis (2016). Developing an application-specific ground-motion model for induced seismicity. Bulletin of the Seismological Society of America 106(1), 158-173. 3. Bommer, J.J., P.J. Stafford, B. Edwards, B. Dost, E. van Dedem, A. Rodriguez-Marek, P. Kruiver, J. van Elk, D. Doornhof & M. Ntinalexis (2017a). Framework for a ground-motion model for induced seismic hazard and risk analysis in the Groningen gas field, The Netherlands. Earthquake Spectra 33(2), 481-498. 4. Bommer, J.J., B. Edwards, P.P. Kruiver, A. Rodriguez-Marek, P.J. Stafford, B. Dost, M. Ntinalexis, E. Ruigrok & J. Spetzler (2017b). V5 Ground-Motion Model for the Groningen Field. 30 October, 161 pp. 5. Bommer, J.J., P.J. Stafford and M. Ntinalexis, Empirical Ground-Motion Prediction Equations for Peak Ground Velocity from Small-Magnitude Earthquakes in the Groningen Field Using Multiple Definitions of the Horizontal Component of Motion - Updated Model for Application to Smaller Earthquakes, November 2017, 2017c 6. Bommer, J.J., B. Edwards, P.P. Kruiver, A. Rodriguez-Marek, P.J. Stafford, B. Dost, M. Ntinalexis, E. Ruigrok, J. Spetzler, V5 Ground-Motion Model for the Groningen Field, 2017d. 7. Bommer, J.J., B. Dost, B. Edwards, P.P. Kruiver, M. Ntinalexis, A. Rodriguez-Marek, P.J. Stafford & J. van Elk (2018). Developing a model for the prediction of ground motions due to earthquakes in the Groningen gas field. Netherlands Journal of Geoscience. 8. Bommer, J.J., P.J. Stafford & M. M. Ntinalexis (2019a). Updated empirical GMPEs for PGV from Groningen earthquakes. Revision 1, 4 March 2019, 15 pp. 9. Boore, D.M. & J.J. Bommer (2005). Processing strong-motion accelerograms: needs, options and consequences. Soil Dynamics & Earthquake Engineering 25(2), 93-115. 10. Bora, S.S., F. Scherbaum, N. Kuehn & P. Stafford (2016). On the relationship between Fourier and response spectra: implications for the adjustment of empirical ground-motion prediction equations (GMPEs). Bulletin of the Seismological Society of America 106(3), 1235-1253. 11. Crowley, H., B. Polidoro, R. Pinho & J. van Elk (2017). Framework for developing fragility and consequence models for local personal risk. Earthquake Spectra 33(4), 1325-1345. 12. Dost, B., E. Ruigrok & J. Spetzler (2017). Development of probabilistic seismic hazard assessment for the Groningen gas field. Netherlands Journal of Geoscience 96(5), s235-s245. 13. Douglas, J. & D. M. Boore (2010). High-frequency filtering of strong-motion records. Bulletin of Earthquake Engineering, 9(2), 395-409. 14. Kruiver, P. P., E. van Dedem, E. Romijn, G. de Lange, M. Korff, J. Stafleu, J.L. Gunnink., A. Rodriguez-Marek, J.J. Bommer, J. van Elk & D. Doornhof (2017). An integrated shear-wave velocity model for the Groningen gas field, The Netherlands. Bulletin of Earthquake Engineering 15(9), 3555-3580. 15. NAM (Jan van Elk and Dirk Doornhof, eds), Technical Addendum to the Winningsplan Groningen 2016 - Production, Subsidence, Induced Earthquakes and Seismic Hazard and Risk Assessment in the Groningen Field, PART V - Damage and Appendices, Nederlandse Aardolie Maatschappij BV 1st April 2016, 2016a.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

16. NAM, Ulan Burkitov, Henk van Oeveren and Per Valvatne, Groningen Field Review 2015 - Subsurface Dynamic Modelling Report, 2016b 17. NAM, Meet- en Regelprotocol Groningen veld, June 2017a. 18. NAM, Rapportage Seismiciteit Groningen - 1 november 2017, November 2017b. 19. NAM, Trillingsschade aan gebouwen, Informatiedocument – versie 1.0, NAM, June 2017, 2017c. 20. NAM, Van Oeveren, Henk, Valvatne, Per and Geurtsen, Leendert, Groningen Dynamic Model Updates 2017, Assen : NAM, 2017. EP201708205454, 2017d 21. NAM, J. van Elk and D. Doornhof, Hazard, Building Damage and Risk Assessment for the Groningen Area, November 2017, 2017e. 22. NAM, Rapportage Seismiciteit Groningen - 1 November 2017, 2017f 23. NAM, Special Report on the Loppersum Earthquakes, December 2017, 2017g 24. NAM, Geurtsen, Leendert and Valvatne, Optimisation of the Production Distribution over the Groningen field to reduce Seismicity, 2017h 25. NAM, Evaluatie en aanbevelingen voorbeheersmaatregelen: Zeerijp aardbeving, Letter 19-1- 2018, Brief Ref.:EP201801201646, 2018a 26. NAM, Quint de Zeeuw and Leendert Geurtsen, Groningen Dynamic Model Update 2018 – V5, June 2018, EP201806200206, 2018b 27. NAM, Zeerijp beving beheersmaatregelen, Letter 17-1-2018, Brief Ref. EP201801203705, 2018c 28. NAM, Maatschappelijke Effecten Inventarisatie van aardbevingen in Noordoost-Groningen 2017 - Derde inventarisatie over de periode 2017 en Q1 2018 Vervolg op MEI 2015 en MEI 2016, Geske Barendregt, June 2018c. 29. NAM, Letter to the Minister of Economic Affairs and Climate Policy: Evaluatie: Westerwijtwerd aardbeving, 24 May 2019 30. TNO-034-DTM-2009-04435, Kalibratiestudie schade door aardbevingen, 11 November 2009 31. Noorlandt, R.P., P.P. Kruiver, M.P.E. de Kleine, M. Karaoulis, G. de Lange, A. Di Matteo, J. von Ketelhodt, E. Ruigrok, B. Edwards, A. Rodriguez-Marek, J.J. Bommer, J. van Elk & D. Doornhof (2018). Characterisation of ground-motion recording stations in the Groningen gas field. Journal of Seismology DOI: 10.1007/s10950-017-9725-6. 32. Ntinalexis, M., B. Polidoro, J.J. Bommer & B. Edwards (2015a). Selection of processing procedures for the accelerograph recordings of induced seismicity in Groningen. Report prepared for NAM, 24 August 2015, 169 pp. 33. Royal HaskoningDHV, Maatschappelijke Effecten Inventarisatie van aardbevingen in Noordoost Groningen, Feb 2016. 34. Royal HaskoningDHV, Tweede Maatschappelijke Effecten Inventarisatie van aardbevingen in Noordoost Groningen, Mar 2017. 35. Royal HaskoningDHV, Tweede Maatschappelijke Effecten Inventarisatie van aardbevingen in Noordoost Groningen - Bijlagen, Mar 2017. 36. USGS, “Did you feel is?” Internet micro-seismic intensity maps, Wald, Quitoriano, Worden, Hopper and Dewey, Annals of Geophysics 54, 6, 2011. 37. Wentinck, H.M., Kinematic modelling of large tremors in the Groningen field using extended seismic sources - Huizinge Earthquake Part 1, 2017 38. Wentinck, H.M., Kinematic modelling of large tremors in the Groningen field using extended seismic sources - Huizinge Earthquake Part 2, 2018 The NAM reports can be downloaded from:

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019 https://www.nam.nl/feiten-en-cijfers/onderzoeksrapporten.html#

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

7 Appendix A – Letter Evaluatie: Westerwijtwerd aardbeving

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Staatstoezicht op de Mijnen De heer ir. T. Kockelkoren, MBA Henri Faasdreef 312 2492 JP Den Haag Brief ref.: EP201905203882

Datum: 23-05-2019

Betreft: Evaluatie: Westerwijtwerd aardbeving

Geachte heer Kockelkoren,

Westerwijtwerd beving Deze rapportage is geschreven naar aanleiding van de beving op de ochtend van 22 mei 2019 met een magnitude van 3.4 op de Richter Schaal, waarbij het epicentrum is vastgesteld nabij Westerwijtwerd. Met deze beving wordt het PGV Waakzaamheidsniveau bereikt van het Meet- en Regelprotocol (verder: MRP). Formeel zou de NAM, op grond van de gemeten waarden, binnen 1 week6 een rapportage moeten sturen aan u en de Minister van Economische Zaken en Klimaat. Echter, gezien de maatschappelijke onrust en de impact die deze aardbeving heeft gehad op de veiligheidsbeleving van de mensen in de Provincie Groningen, acht de NAM het correct om dezelfde uitgangspunten te hanteren als bij de Zeerijp beving, met een rapportage binnen 48 uur. Deze rapportage wijkt af van de rapportage naar aanleiding van de Zeerijp beving, aangezien de NAM geen enkele rol meer speelt in het “regel” deel van het MRP. De minister bepaalt inmiddels expliciet het winningsniveau en de wijze van winning uit het Groningen gasveld. De NAM heeft daarbij een winplicht. In het “Instemmingsbesluit Groningen gasveld 2018-2019” (verder: Instemmingsbesluit) staat hierover opgenomen: Dit besluit is gebaseerd op de huidige wetgeving en op basis daarvan kan het gebruik van het huidige MRP worden voorgezet. Eventuele overschrijdingen van de signaalparameters in het gasjaar 2018-2019 zullen gedocumenteerd worden door de NAM zoals afgesproken in het huidige MRP. Voor wat betreft mogelijk ingrijpen op basis van die overschrijdingen constateer ik, gelet op de reeds genomen maatregelen en een winningsniveau op hoogte van leveringszekerheid, dat er sneller dan voorheen sprake zal moeten zijn van een politiek-bestuurlijke afweging door mij die dan zal leiden tot wijziging van dit besluit. Daarom heb ik in artikel 5 bepaald dat de NAM niet alleen aan SodM, maar ook aan mij rapporteert. Het MRP blijft een middel om de ontwikkeling van de seismiciteit te volgen, waarbij ook ik word geïnformeerd. Indien

6 Zie hiervoor pagina 28 van het MRP. 57

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

de ontwikkeling van de seismiciteit mij aanleiding geeft tot het nemen van aanvullende maatregelen, zal ik de regio Groningen daarbij betrekken. In artikel 5 van het Instemmingsbesluit is vervolgens opgenomen: Artikel 5 (meet- en regelprotocol) NAM rapporteert op basis van het Meet- en regelprotocol aan SodM en aan de Minister van Economische Zaken en Klimaat over de in dit protocol benoemde signaalparameters. De NAM zal in deze rapportage dan ook enkel rapporteren over de Westerwijtwerd aardbeving en geen voorstellen doen voor eventuele beheersmaatregelen voor het Groningen gasveld. De NAM zal wel toetsen of er uit de analyse van de bevingen technische inzichten ontstaan die de basis onder het Instemmingsbesluit, inclusief de operationele strategie, in twijfel zouden kunnen trekken en waar de minister dus mogelijk elementen van het Instemmingsbesluit zou kunnen heroverwegen. Effect van de aardbeving De aardbeving op 22 mei bij Westerwijtwerd had een magnitude van 3,4 op de Richter Schaal met een gemeten maximale grondversnelling van 0,03 g op 1,4 kilometer van het epicentrum van de beving. De grootst gemeten maximale grondsnelheid is 1,04 cm/s op 4,12 kilometer van het epicentrum. De beving is over een groot gebied gevoeld. Rond het epicentrum van de beving was de duur van de beving 1 of 2 seconden. Zoals ook vastgesteld bij eerdere bevingen neemt de sterkte van de beving af maar de duur van de beving toe op grotere afstand van het epicentrum. Alle bovenstaande meetgegevens zijn verzameld door het KNMI-aardbevingsmonitoringsnetwerk. De vastgestelde metingen van het TNO-gebouwensensorennetwerk stemmen hier goed mee overeen. Na de Westerwijtwerd beving hebben de signaalparameters van het MRP7 de volgende waarden:

Signaleringsparameter Waarde op 22 mei Niveau van signalering PGA 0,03 g Monitoring Activity Rate (aantal aardbevingen 12 Monitoring in laatste 12 maanden) Aardbevingsdichtheid 0,116( km2 . jaar)-1 Monitoring PGV 1,04 cm/s Waakzaamheids-niveau De PGV daalt tot beneden de 5 mm/s (SBR criterium voor een metselwerk pand; bouwkundige staat: normaal8) op een afstand van 4,3 kilometer rondom het epicentrum. In dit gebied bevinden zich ongeveer 5200 bewoonde panden. De PGV daalt tot beneden de 3 mm/s (SBR criterium voor een metselwerk pand; bouwkundige staat: gevoelig) op een afstand van 5,9 kilometer. In dit gebied bevinden zich ongeveer 7100 bewoonde panden. In de volgende paragrafen wordt nader ingegaan op de vraag of sprake was van een opvallende of onverwachte gebeurtenis als bedoeld in het MRP en vervolgens wordt nader ingegaan op schade en veiligheid. Daarbij wordt ook aandacht besteed aan de veiligheidsbeleving van mensen.

7 De signaalparameter “Damage State’ is niet opgenomen omdat niet de NAM maar de TCMG de schademeldingen behandelt. 8 Zie ook pagina 55 van de “SBR Trillingsrichtlijn A: Schade aan bouwwerken:2017”. 58

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Duiding analyses Alle analyses van de beschikbare data zijn afgerond binnen 48 uur na de beving. Hoewel de NAM daarbij alle mogelijke zorgvuldigheid heeft betracht, is het uiteraard mogelijk dat de analyse met voortschrijdend inzicht de komende weken nog iets bijgesteld kan worden. Onverwachte of opvallende gebeurtenis als bedoeld in het MRP? Algemeen In het MRP wordt voorgeschreven dat niet alleen moet worden gekeken naar de vijf parameters waar signaleringniveaus voor zijn vastgesteld, maar ook naar opvallende of onverwachte gebeurtenissen. Een onverwachte gebeurtenis wordt in het MRP gedefinieerd als een gebeurtenis die niet past binnen de door de NAM gebruikte modellen dan wel een gebeurtenis die een significante afwijking van een trend vormt. Om te voorkomen dat de werking van het MRP te afhankelijk is van een model, wordt daarnaast gekeken naar opvallende gebeurtenissen. Dat zijn factoren die niet zijn meegenomen in het model maar wel relevant kunnen zijn. Hierbij kan bijvoorbeeld worden gedacht aan de locatie en duur van de beving. In deze paragraaf wordt gekeken in hoeverre hier sprake van is geweest bij de Westerwijtwerd beving. Onverwacht? De beving vond plaats na een relatief rustige periode van het Groningen gasveld voor wat betreft seismiciteit (zie ook de halfjaarlijkse rapportage van 1 mei 2019). In de twaalf maanden van 1 mei 2018 tot 1 mei 2019 had de zwaarst gemeten aardbeving een magnitude van 1.9 op de Richter Schaal (8 augustus nabij Appingedam). Echter, de berekende kans op een beving met een magnitude van 3,6 of hoger in 2019 is, zelfs na de productiereducties van de afgelopen jaren, nog steeds circa 13 procent (zie tabel 4.1 in “Seismic Hazard and Risk Assessment Groningen Field update for Production Profile GTS - raming 2019”). Met deze eerder berekende kans is wat ons betreft dan ook geen sprake van een onverwachte gebeurtenis als bedoeld in het MRP. Opvallend? Het epicentrum van de Westerwijtwerd aardbeving ligt in het westelijk gedeelte van het meest aardbevingsgevoelige gebied van het Groningen gasveld rond Loppersum. De productie uit het Groningen gasveld in deze regio is al sinds 2014 sterk gereduceerd en sinds 2018 volledig gestopt. Echter door de voortdurende productie uit vooral het zuidoostelijke deel van het Groningen gasveld is in dat gebied de gasdruk verder gedaald. Hierdoor is naar verwachting gas vanuit noordwesten van het veld naar het zuidoosten blijven stromen, met als gevolg een daling van de gasdruk in het noordwesten en verdere bodemdaling en kans op aardbevingen rond Loppersum. De gemeten grondversnellingen voor deze beving waren circa 50 procent lager dan de gemeten grondversnellingen na de Zeerijp beving die een vergelijkbare magnitude had. Echter, de gemeten grondversnelling waarden vallen binnen de ervaringsband van eerdere aardbevingen en de onzekerheidsband van de modellen. De duur van de beving was niet afwijkend van wat verwacht mag worden van een beving van deze magnitude. Als laatste wordt nog opgemerkt dat het onwaarschijnlijk is dat een grote drukval over de breuk heeft geleid tot deze aardbeving. De Westerwijtwerd beving heeft plaatsgevonden op een breuk met een offset kleiner dan de reservoirdikte. De ervaring is dat over dit soort breuken de drukval gewoonlijk erg klein is.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

De conclusie is dan ook dat er met deze beving geen sprake is van een opvallende seismische gebeurtenis als bedoeld in het MRP. Schade Schade ten gevolge van de gaswinning uit het Groningen gasveld wordt sinds maart 2018 afgehandeld door de Tijdelijke Commissie Mijnbouwschade Groningen (verder: TCMG). Op haar website heeft het TCMG op donderdag 23 mei 2019 het volgende bericht gezet: Zo'n 36 uur na de beving in Westerwijtwerd staat de stand van het aantal schademeldingen bij het schadeloket op een totaal van 1096. Er kwamen 47 meldingen van een acuut onveilige situatie (AOS) binnen.

Op een normale dag ontvangen we circa 30 nieuwe aanvragen tot schadevergoeding. Aanvragen komen zowel telefonisch als via de website bij ons binnen. Het AOS-team bezocht tot nu toe 28 plaatsen waar een AOS was gemeld. Er werden 6 meldingen gegrond verklaard. Op deze adressen worden beveiligingsmaatregelen getroffen. In de speciale rapportage die de NAM binnenkort nog zal toesturen (zie de paragraaf “Beheersmaatregelen”) zullen wij een nadere analyse geven op grond van de beschikbare gegevens en daarbij aangeven of de te verwachten schade overeenkomt met de werkelijke schademeldingen. Veiligheid Persoonlijke Veiligheid De recente aardbeving bij Westerwijtwerd valt binnen de dreigings- en risicoanalyse gepresenteerd in de rapportage “Seismic Hazard and Risk Assessment Groningen Field update for Production Profile GTS - raming 2019 ” van maart 2019. Veiligheidsbeleving Naast de calculatieve inschatting van persoonlijke veiligheid is er ook een andere relevante component binnen veiligheid, de veiligheidsbeleving. Deze beschrijft volgens de OVV het gevoel van veiligheid, de gemoedsrust en de dreigende verstoring van het woongenot onder de bewoners van Groningen. Ook in het Instemmingsbesluit wordt nadrukkelijk gesproken over de sociale gevolgen en angstgevoelens. De NAM is zich er volledig van bewust dat een aardbeving met deze magnitude het gevoel van onveiligheid bij de bewoners die boven het Groningen gasveld wonen verder versterkt. Zo lang er sprake is van gaswinning uit het Groningen gasveld zullen er aardbevingen zijn en dit is voor ons een niet makkelijk te accepteren realiteit. Wij kunnen ons ook inleven in het onbegrip en de weerstand tegen de gaswinning in de regio. De minister heeft in maart 2018 het besluit genomen om het Groningen gasveld zo snel mogelijk in te sluiten en wij steunen dat besluit volledig. Tot die tijd heeft de minister bepaald dat er niet meer gas zal worden geproduceerd dan noodzakelijk is voor de leveringszekerheid. Op die wijze worden de seismische risico’s zo veel als mogelijk beperkt. NAM zal in deze periode blijven voldoen aan haar winplicht. De NAM spreekt de hoop uit dat, nu de keuzes voor de gaswinning en de wijze waarop wordt omgegaan met de gevolgen daarvan (met name schade en bouwkundig versterken) volledig bij de overheid liggen, er meer vertrouwen ontstaat bij de Groningers ten aanzien van de afbouw van de

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019 gaswinning en een betere de afhandeling van schade en bouwkundig versterken. De NAM is en blijft daarbij financieel aansprakelijk voor de schade die mensen leiden door de aardbevingen. Beheersmaatregelen Zoals aangegeven in de inleiding van deze rapportage doet de NAM geen voorstellen voor eventuele beheersmaatregelen voor het Groningenveld. Wel voert de NAM die beheersmaatregel uit die past bij onze rol, namelijk het maken van een speciale rapportage. Deze uitgebreide speciale rapportage zal (in lijn met de Zeerijp beving) twee weken na de beving (5 juni) door de NAM worden gedeeld. Hierin wordt onder andere een meer gedetaileerde analyse gegeven van de grondbeweging (inclusief de duur van de beving), de vastgestelde schade, de relatie tot breuken in het reservoir van het Groningen gasveld en het aardbevingsbronmechnisme. Tenslotte willen wij graag meedelen of naar onze mening uit de analyse van de beving nieuwe technische inzichten zijn ontstaan die de basis van het Instemmingsbesluit, inclusief de gehanteerde operationele strategie, in twijfel zouden kunnen trekken waardoor de minister dus mogelijk elementen van het Instemmingsbesluit zou kunnen heroverwegen. Het stoppen van de productie is uiteindelijk de enige maatregel die ertoe zal leiden dat de aardbevingen in het Groningen gasveld stoppen. Uitgangspunt van de huidige productie van het Groningen gasveld, zoals ook vastgelegd in het Instemmingsbesluit, is dat niet meer gas wordt geproduceerd dan nodig is voor de leveringszekerheid. Gelet op dit bestaande beleid van de minister wordt reeds maximaal ingezet op een zo beperkt mogelijke productie uit het Groningen gasveld. Daarnaast is verdeling van de productie over het Groningen gasveld al geoptimaliseerd waardoor de seismische risico’s zo veel als mogelijk worden beperkt. De nieuwste inzichten op dit gebied zijn weergegeven in de operationele strategie die de NAM op 15 maart 2019 heeft ingediend bij de minister. Het epicentrum van deze aardbeving ligt op grote afstand van de producerende productieclusters en meer in het bijzonder op grote afstand van de clusters die in de momenteel gehanteerde productiewijze preferentieel geproduceerd worden. De afstand tot het Eemskanaal cluster is 9,5 kilometer en de afstand tot de Kooipolder, Siddeburen, Slochteren en Bierum clusters is respectievelijk 14,7, 15,4, 16,1 en 16,9 kilometer. Door deze grote afstand en de hoge compressibiliteit van het gas op deze lagere drukken in het reservoir, zal de invloed van eventuele productiefluctuaties in deze clusters op de aardbevingslocatie nauwelijks waargenomen kunnen worden. Wij zien dan ook geen aanwijzingen in de analyse van deze beving dat een alternatieve verdeling van de productie tot een verminderd seismisch risico zou kunnen leiden. Afsluitend Als de NAM zijn we opgelucht dat, zover bekend, de aardbeving van Westerwijtwerd van woensdag 22 mei 2019 geen persoonlijk lichamelijk letsel tot gevolg heeft gehad. We realiseren ons dat deze beving de gevoelens van onveiligheid en boosheid onder Groningers verder heeft versterkt. Zo lang er geproduceerd wordt zullen er bevingen plaatsvinden en daarom blijven we meewerken aan het beleid van de minister om de productie uit het Groningen gasveld zo snel al mogelijk naar nul terug te brengen.

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019

Deze rapportage is conform de vereisten van het Instemmingbesluit tevens toegezonden aan de minister van Economische Zaken en Klimaat. Mocht u nog nadere vragen hebben, dan staan wij vanzelfsprekend tot uw beschikking.

Hoogachtend,

J. Atema NAM directeur

Special Report on the Westerwijtwerd Earthquake of 22nd May 2019