Seismology Research
Seismic hazard in the North of The Netherlands
January 2005
Torild van Eck, Femke Goutbeek, Hein Haak and Bernard Dost
Seismicity in The Netherlands occurs mainly in the southeastern part of the country and is related to tectonic
movements along the Roer Valley Graben (Dost and Haak, 2005). The Dutch seismological monitoring network of the
KNMI, however, also observes earthquakes in the North of The Netherlands since 1986. These earthquakes have been
classified as induced seismicity due to the exploration of oil- and gasfields. Here we find also one of the world's
largest gas reservoirs, the Groningen reservoir, which contains a reserve lasting at least several decades.
Occasionally earthquakes up to ML = 3.5 have caused minor damage (such as cracks in buildings), more often the felt
events are of general annoyance to the local population. Therefore, since January 1, 2003, the new Dutch mining
legislation requires for each concession a risk analysis and a monitoring plan. Up to now only general hazard
estimates were available, i.e. maximum possible earthquake and maximum possible Intensity. Within the context of
this new mining law we have been estimating site-specific engineering hazard parameters, i.e. ground motion that
can be associated to specific risks.
The challenge of such a hazard analysis is that it addresses the effects of small, shallow earthquakes. Small
earthquakes (ML ≤ 3.6) are generally irrelevant in risk analysis. However, these induced events occur at
shallow depths (< 4 km), as compared to 'natural' earthquakes (usually depths > 10 km) and may therefore cause
ground motions that reach significant amplitudes. In order to quantify possible risks (= hazard * vulnerability)
due to such earthquakes we first need to specify the hazard. In Van Eck et al. (2005) we estimated the seismic
hazard due to induced seismicity.
Observations
The Dutch seismological monitoring network of the KNMI consists currently of 21 permanent and mobile seismograph
stations and 20 accelerometers. Out of these, eleven borehole seismometer strings and 14 accelerometers are especially deployed to improve monitoring of the induced seismicity (Figure 1) to detect and locate all
earthquakes with ML ≥ 1.5. Since 1986 and up to 2004 the KNMI observed more than 340 induced events within a
magnitude range -0.8 ≤ ML ≤ 3.5. Approximately 60 of those events have been felt. Only nine of those had
magnitudes ML ≥ 3.0 and no intensities (EMS scale) larger than VI have been observed. All events in the
northern part of The Netherlands have been located in or in the direct vicinity of gas reservoirs.
Although it is generally accepted that induced earthquakes are caused by stress changes and fluid injection in
and around the reservoir (Zoback and Zinke, 2002) we have so far not succeeded in correlating changes in gas
production (rate) with the occurrence of seismicity (Van Eck et al., 2005; Van Eijs et al., 2005). We only see a
relation in general terms; an overall constant (stationary) rate of production causing a stationary rate of
seismic energy release (Figure 2) combined with an exponential distribution of numbers of events as a function
of size (Figure 3). This is comparable with statistics of natural events.
Figure 2. Cumulative of the square root of the earthquake energy in Gjoule (blue curve)
of all induced events with M > 1.4 in the northern part of The Netherlands as a function of time. The black
straight lines indicate the upper and lower energy boundaries of an assumed stationary seismic energy release.
The inlay figure compares the cumulative seismic energy release (red curve) with cumulative gas production on
land (green curve). Gas production numbers are kindly supplied by the NITG.
Figure 3. Cumulative annual frequency-magnitude model for all induced seismicity in
the North of The Netherlands for the period 1986 - 2003 (solid blue curve) and observations (blue crosses). Also
shown is the frequency-magnitude relation for all induced events in the Groningen field excluding (purple broken
curve) and including (red curve) three events 2.7 < M < 3.0 that occurred October-November 2003. This suggests
that the slope (b-value) of the frequency-magnitude model for a subset of 179 events has a tendency to approach
the b-value of the frequency-magnitude model for all earthquakes (340 events).
In order to obtain more ground motion measurements close to the epicentre the KNMI is deploying 14 accelerometers
in the North of The Netherlands. Among the nearly 30 acceleration records obtained within six years we observed
in some cases horizontal accelerations exceeding 0.3 g (gravity g = 9.8 m/sec2). This is considered significant
in earthquake engineering terms. However, its duration, determined by the distance and depth of the event, is in
our case usually very short, about one cycle. These short strong ground motion pulses have little damaging effect.
Hazard estimation approach
We used a standard probabilistic seismic hazard approach to obtain ground motion estimates. In this approach
statistical models of the seismicity distribution and the frequency-magnitude distribution are used in combination
with a ground motion prediction equation to obtain the probability of exceeding a ground motion at a specific site.
This analysis is repeated for a large number of grid points at the surface above and in the direct vicinity of
the hydrocarbon exploitation fields.
Seismicity distribution model
The induced seismicity occurs generally in and around the hydrocarbon reservoirs4) usually around 1.5 to 3.0 km
depth. Although we have strong indications that most seismicity is associated with existing faults, we are
currently unable to identify precisely the active faults. Consequently, the best seismicity model is currently a
homogeneous distribution of the seismicity at 2.5 km depth in the direct vicinity of a hydrocarbon reservoir that
has been identified as being seismically active.
Frequency-magnitude model
The frequency of occurrence versus size of all induced seismic events resembles nicely an exponential
distribution. Local variations do exist however. We observed, for example, in the period 1994-2004 only four
events in the Bergermeer field near Alkmaar, North-Holland, all had ML > 2.9. As we lack a specific physical
model explaining such behaviour we adhere to the general statistical frequency-magnitude model as shown in
Figure 3.
Ground motion prediction equation
Peak ground velocity and peak ground acceleration are two pragmatic parameters used to characterize seismic
hazard. Its amplitude can be predicted for a given magnitude and distance using a basic equation, which describes
the geometrical spreading and attenuation. The variables of this equation for our region has been estimated using
accelerometer and seismometer observations from small and shallow events in The Netherlands (Dost et al., 2004).
In our first approach we used a more general relation by Campbell (1997).
Results
We found that seismic hazard estimates due to induced events in the northern part of the Netherlands are best
given in terms of the Peak Ground Velocity (PGV) or, alternatively, the maximum in the 50% damped response
spectra at 10 Hz. The results are, among others, presented in maps that give the PGV values with 10% and 1%
annual probability of being exceeded once (Figure 4 and 5).
Figure 4. Estimated hazard at the surface above and around the Roswinkel field for a
return period of T=100 years, i.e. the peak ground velocity that can be exceeded with an annual probability of 1%.
The x- and y-axis indicate distance in kilometres. In the following figure we show the hazard only along a
section as indicated.
Figure 5. Seismic hazard sections. The estimated hazard at the surface as a function of
distance from the surface projection of the exploitation field for three reservoirs; The Groningen, the Bergermeer
and the Roswinkel field. The hazard is shown in terms of PGV for a return period of T=100 years (left) and T=10
years (right), i.e. annual probability of exceedance of 1% and 10% respectively.
Another result is that above the largest Dutch gas field, the Groningen field, we expect PGV of 20 and 30 mm/sec
that may be exceeded with a 10% probability in one and 10 years, respectively. Above some small (about 3-4 km2)
gas fields, Roswinkel and Bergermeer, we expect values around 35 and 60 mm/sec, respectively. These values would,
if they occur, exceed the Dutch building research (SBR) vibration guidelines. Another result is that our systematic
approach to estimate the hazard, which includes a sensitivity analysis, provides the decision maker with an insight
as to which relevant uncertainties may be decreased and which not. This can be and is consequently used to set
research and monitoring priorities.
Discussion
Our presented hazard approach provides first order estimates of (a) the strong ground motion at the surface due
to induced seismicity and (b) its probability of occurrence. These are increasingly important for risk analysis
in a society with high awareness of disturbances and increased vulnerability. Eventually we would like to replace
our analysis with more precise estimates, including local surface geology, exploitation information and detailed
local geological and tectonic information. To accomplish this we have and continue cooperation in this aspect with
the NITG. Wassing et al. (2005) added the local surface geology effect. Van Eijs et al. (2005) used the extensive
confidential concession and exploitation information, which companies have to provide to the NITG since 2003
within the new mining legislation, to take the first step to include exploitation and geological information.
Recently, the KNMI monitoring network has been extended with three strategically placed extra borehole stations
to improve detection, location and quantification. Six additional well-placed accelerometers will help us to
predict better the strong ground motion at the surface due to these small and shallow induced earthquakes in and
around hydrocarbon reservoirs.
References
- Campbell, K.W., 1997, Empirical near-source attenuation relationships for horizontal and vertical components of Peak Ground Acceleration, Peak Ground Velocity, and Pseudo-absolute acceleration response spectra, Seismol. Res. Lett., 68, 154-179.
- Dost, B., T. van Eck and H. Haak, 2004, Scaling peak ground acceleration and peak ground velocity recorded in The Netherlands, Bolletino di Geofisica, 45, 153 - 168.
- Dost, B. and H. W. Haak, 2005, Seismicity, In: Wong, Th.E., D.A.J. Batjes and J. de Jager (eds.) Geology of The Netherlands. in press.
- Van Eck, T., F.H. Goutbeek, H. Haak and B. Dost, 2005, Seismic hazard due to small shallow earthquakes in The Netherlands, Engineering Geology, in press.
- Van Eijs, R.M.H.E., F.M.M. Mulders, M. Nepveu, C.J. Kenter and B.C. Scheffers, 2005, Correlation between hydrocarbon reservoir properties and induced seismicity in The Netherlands, submitted to Engineering Geology.
- Wassing, B.B.T., T. van Eck and R.M.H.E. van Eijs, Seismisch hazard van geïnduceerde aardbevingen - Integratie van deelstudies, KNMI-publ. 208.
- Zoback, M.D. and J.C. Zinke, 2002, Production-induced Normal faulting in the Valhall and Ekofisk Oil fields, Pure and Applied Geophysics.
)
