Search for a command to run...
Abstract The first detected induced earthquake in the Netherlands was in 1986. Since then, the seismic network has largely been expanded and geared towards recording induced seismicity. Nearly 2000 events have been located and assigned a local magnitude. The impact of these events on the Earth’s surface can be assessed with help of ground motion prediction equations (GMPEs). In this study, we derive these GMPEs. Previously, GMPEs were derived especially for the Groningen gas field, using data recorded at the surface and until approximately 35 km distance. In this study, we include induced seismicity from three other provinces in the north of the Netherlands, and we use recordings until 110 km distance. We compile a database with approximately 45,000 ground motions, most of which are ’free field’ and all of which are on unconsolidated sediments. This database does not only contain ground motions recorded at ground level, but also motions recorded at 200 m depth. These deep recordings have been made in boreholes situated in quite uniform deposits and hence are less perturbed by varying site conditions. This allows derivation of GMPEs that describe well the distance-dependent attenuation resulting from the deep wave propagation. A correction term is implemented when the focal depth is below a thick salt layer, resulting in lower amplitudes near the epicenter. The GMPEs at the Earth’s surface are derived as scaled versions of the deep GMPEs, with the option to add various site terms. The GMPEs have an empirical base for low-magnitude seismicity (1.0 $$\le $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> M $$\le $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 3.6), epicentral distances between 0 and 110 km and nucleation depths between 2.3 and 4.0 km. Body waves are responsible for the peak ground motions that have been recorded. These body waves yield a complicated attenuation pattern that is incorporated into the GMPEs. Furthermore, the amplification in the top 200 m soil column is found to have a distance dependence, which is included in the surface-level GMPEs. Other tested site terms yield a modest to negligible improvement in explaining the measured ground motions. The models are validated using records from the 2025 Zeerijp M3.4 event, and compared with other recent GMPEs for a similar seismotectonic setting.