Winschoten events, 19-11-2017

A succession of at least 4 events was measured on November 19, 2017, close to the city of Winschoten, at the south-eastern edge of the Groningen gas field. The largest amplitude recorded was at sensor G570: a peak particle acceleration of 2.45 mm/s. The waveforms had a signature different from the events initiated in the vicinity of the gas reservoir. Moreover, the Groningen gas field had not been seismically active at the south-eastern edge of the field. An alternative seismic source would be deformation in a salt cavern. The preliminary epicentre of the main event was situated on top of the Heiligerlee salt dome, in which 12 salt caverns have been mined. With a better constrained location and a spectral analysis, we assess whether the salt-cavern hypothesis is likely.

Using 5 improved P-wave picks and a detailed 3D velocity model for the area, the main event was mapped to a source location in the (lower) North Sea Group or upper Chalk Group, above the western flank of the Heiligerlee salt dome. At this location, brittle rock overlying the salt dome could have moved due to salt creep. The uncertainty ellipse includes the part of the salt dome where the caverns are present. However, a hypocentre at the depth range of the caverns is unlikely. With the assumption that the source is just within the Chalk Group, a moment magnitude of 1.28 ± 0.08 is found.

All nearby KNMI seismic stations recorded a strong 3.4 Hz resonance for the event. Due to large differences in receiver-side structure, the 3.4 Hz resonance likely originated from the source area. The assumption was tested whether one of the salt caverns would act as a resonating closed pipe. From the 12 caverns, 2 can possibly generate the recorded resonance: brine filled cavern HL-M over its height and nitrogen-filled cavern HL-K over its width. Another possible resonator, which is more in-line with the updated source location, would be the lower North Sea Group in the source area.

Waveforms two weeks prior to two weeks after November 19 were scanned to find possible similar, but smaller, events. The two largest subevents (W2 and W3) were used as a template. The scanning resulted in a few records with similarities just above the threshold of a 0.5 correlation coefficient. Subevent W4 turned out to be a close match to W3, with a correlation coefficient of 0.65.

The implemented scanning approach is not sufficient as a monitoring tool. Events are likely to occur for which a template does not yet exist. For detailed seismic monitoring of the Heiligerlee salt dome, currently there is no dedicated network, to allow both detection and location of very small seismic events. As a (temporary) alternative, we would suggest a monitoring tool solely based on sliding windows of root-mean-square amplitudes