This interactive webpage contains supplementary information for the publication by Kühn et al. 2020: "Probabilistic moment tensor inversion for hydrocarbon-induced seismicity in the Groningen gas field, the Netherlands, part 1: testing". It allows for an easy comparison between the various tests of inversion parameters and velocity models described for the analysis of the 11th of March 2017 Zeerijp ML 2.1 earthquake on the event induced in the Groningen gas field (Netherlands). Inversion runs collected here comprise the parameters employed for inversion (Problem Config), the inversion results and error estimates (Parameter Results) as well as a multitude of figures.
The analysis has been performed using the Grond software package (Heimann et al., 2018). The open source software for seismic source parameter optimization Grond implements a bootstrap-based method to retrieve solution sub-spaces, parameter trade-offs and uncertainties of earthquake source parameters. Green's functions (GFs) for three different velocity models were calculated with the orthonormal propagator method (QSEIS, Wang, 1999; see https://github.com/pyrocko/fomosto-qseis/). All GFs are stored in Pyrocko GF stores (Pyrocko toolbox, Heimann et al., 2017, Heimann et al. 2019). Green's functions were computed employing a tapered Heaviside wavelet, a sample rate of 25 Hz and a grid spacing of 50 m allowing for interpolation of Green's functions between nodes. The databases comprise source depths from 1 to 4 km and receiver depths from 0 to 200 m. We used a nearest neighbor interpolation inbetween grid points of the pre-computed GFs.
Synthetic and observed P- and S-phase waveforms from up to 10 km were restituted to displacement and filtered between 2 - 4 Hz (P-waves) and 1 - 3 Hz (S-waves), respectively, as well as windowed between [0 s; +0.5 s] or [-0.2 s; +0.5 s], respectively, from the expected phase arrival, given the tested candidate source model at each forward modeling step in the optimization. From the waveforms, different types of body wave attributes were calculated in order to be compared during the inversion, as amplitude spectra, envelopes, absolute amplitudes and cross-correlation traces. S-wave contributions were weighted half compared to P-wave contributions.