The Use of Elastic Wave Field Simulations to Aid in the Development of Ground Motion Prediction Equations, an Induced Seismicity Case Study at Groningen Gas Field

Here we present findings on the use of elastic finite difference waveform modeling to help construct ground motion prediction equations (GMPEs) at the Groningen gas field, The Netherlands. Over the past decade, seismicity induced by gas production at the Groningen gas field has increased and become a public concern. To better understand the risk and to quantify the hazard of the induced seismicity a Probabilistic Seismic Hazard Analysis (PSHA) workflow has been undertaken. One of the inputs into this workflow is a Ground Motion Prediction Equation (GMPE), which is a function that predicts ground motions for a given earthquake. The GMPEs are often empirically derived from a subset of local observations. In most cases the sample size of observations is such that in order to complete the GMPEs assumptions need to be made. Here we investigate the use of elastic waveform modeling as a tool that can be used to complement the empirical data in the development of GMPEs. To make our case we will present a comparison between simulated earthquakes and observed seismic field data. To this end a select number of earthquakes that occurred over the past year between 1.3Ml and 3.1Ml have been simulated. The seismic data comes from earthquakes recorded on a newly installed (Sept 2015) dense seismic array. This seismic array is part of the Royal Nederland Meteorological Institute (KNMI) seismic network and is composed of 67 site locations, with each location having a surface accelerometer and 4 downhole geophones at 50m, 100m, 150m and 200m respectively. Using this publically available seismic data it's our intent to show we can start to calibrate and predict to 1st order peak ground accelerations by modeling the elastic wave-field. Having demonstrated that we can reasonably start to predict to 1st order ground motions from a limited dataset the usefulness of using the simulated data to better understand magnitude scaling, path effects and spectral accelerations is explored.