The interactive web page contains supplementary information for a publication by Hensch et al. 2019: "Deep low-frequency earthquakes reveal ongoing magmatic recharge beneath Laacher See Volcano (Eifel, Germany)". Details on the analysis of three tectonic and nine deep low-frequency earthquakes are given, including parameter results, error estimates, and 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, Heimann et al., 2018) implements a bootstrap-based method to retrieve solution sub-spaces, parameter trade-offs and uncertainties of earthquake source parameters. Synthetic and observed P and S phase waveforms are restituted to displacement and filtered between 0.5 and 5 Hz in variable frequency ranges, depending on the signal-to-noise ratio (SNR) and the character of the signals. Station amplification factors and transfer functions have been evaluated before the restitution using an empirical calibration method (see Dahm et al., 2018). From waveforms, different types of body wave attributes were calculated, as amplitude spectra, envelopes, and amplitude spectral ratios.The Green's functions (GF) were calculated with the orthonormal propagator method (QSEIS, Wang, 1999; see https://github.com/pyrocko/fomosto-qseis/), for a 1 km grid spacing in a volume of 150 km source-receiver distances and 1 - 50 km source depths. The sampling rate was 40 Hz and the GF include near field terms. All GF are stored in a Pyrocko GF store (Pyrocko toolbox, see Heimann et al., 2017). We use a nearest neighbor interpolation in between the grid points of the pre-computed GF.Restituted observed and synthetic ground displacement time series are filtered and windowed between [-2 s; +3 s] from the expected phase arrival, given the tested candidate source model at each forward modeling step in the optimization. Additional to full waveforms, amplitude spectra, envelopes and spectral ratios between P-SV and SH-SV waves are compared. For spectral ratios, a water level approach was implemented to avoid bias from high noise. All components of the mixed inversion received a proper linear weighting with factors between 0.5 and 3, which was selected after running tests with some master events. Weighting and frequency range were defined different for earthquakes with magnitudes above or below ML 2. P and S phase arrivals have been picked to ensure correct selection of time windows during the centroid inversion, and station blacklists were considered event-wise, depending on the SNR.The plots show for every event the data fits and different types of solution plots. The naming of pages is self-explanatory, but more information can be found in the Grond documentation (https://pyrocko.org/grond/). In order to evaluate the ensembles of solutions for interpretation, we extended the standard statistical analysis of Grond to consider a cluster analysis of source mechanism distributions before statistical analysis. This is introduced because the best ensemble solutions of many of the DLF events show higher variability and groups of different mechanisms. A simple mean or median does not always represent the families of best performing solutions. We therefore declustered the ensemble of best solutions using the method of Cesca et al. (2013), applying the Kagan angle norm, and performed the statistical analysis for each individual cluster.
SAM ("Simplified Analytical Model") is a MatLab-based software that allows for fast and flexible simulations of three-dimensional dyke pathways in an elastic medium. The model was first introduced in "Mechanical modeling of pre-eruptive magma propagation scenarios at calderas" (Mantiloni, L. et al. 2023). In SAM, dykes are modelled as penny-shaped cracks of fixed radius, opening against the local direction of the least-compressive principal stress. The direction of propagation is determined by the gradient of the external stress normal to the crack's plane and the buoyancy force of the magma filling the dyke, calculated at a set of observation points along the crack's tipline. The model can also include a uniform internal pressure within the dyke and compute the stress intensity factor along the crack's tipline, comparing it to the fracture toughness of the host rock to determine if the dyke will advance. SAM needs a model for the stress field of the host rock as input, as well as magma and rock densities, rock elastic properties, the dyke's radius and the number of observation points. The model may be applied to simulate dyke pathways in realistic volcanic settings with different stress sources, and can perform large numbers of simulations in little time. The model does not, however, account for any viscous flow of magma within the dyke, nor the velocity of dyke propagation. Dykes cannot change shape or area during the propagation, and are always bound to be oriented normally to the local least-compressive principal stress axis. This repository also includes data and parameters of the synthetic scenarios discussed in "Mechanical modeling of pre-eruptive magma propagation scenarios at calderas".
This dataset presents the raw data from one experimental series (named CCEX, i.e., Caldera Collapse under regional Extension) of analogue models performed to investigate the process of caldera collapse followed by regional extension. Our experimental series tested the case of perfectly circular collapsed calderas afterward stretched under regional extensional conditions, that resulted in elongated calderas. The models are primarily intended to quantify the role of regional extension on the elongation of collapsed calderas observed in extensional settings, such as the East African Rift System. An overview of the performed analogue models is provided in Table 1. Analogue models have been analysed quantitatively by means of photogrammetric reconstruction of Digital Elevation Model (DEM) used for 3D quantification of the deformation, and top-view photo analysis for qualitative descriptions. The analogue materials used in the setup of these models are described in Montanari et al. (2017), Del Ventisette et al. (2019), Bonini et al., 2021 and Maestrelli et al. (2021a,b).