We use the Hamiltonian Monte Carlo (HMC) algorithm to estimate the posterior probability distribution of a number of earthquake source parameters. This distribution describes the probability of these parameters attaining a specific set of values. The efficiency of the HMC algorithm, however, can be improved through the formulation of a geologically constrained prior probability distribution. The primary objective of the presented study is, therefore, to assess the role of the prior probability in the application of the HMC algorithm to recordings of induced seismic events in the Groningen gas field.@en

We present an efficient probabilistic workflow for the estimation of source parameters of induced seismic events in three-dimensional heterogeneous media. Our workflow exploits a linearized variant of the Hamiltonian Monte Carlo (HMC) algorithm. Compared to traditional Markov chain Monte Carlo (MCMC) algorithms, HMC is highly efficient in sampling high-dimensional model spaces. Through a linearization of the forward problem around the prior mean (i.e., the “best” initial model), this efficiency can be further improved. We show, however, that this linearization leads to a performance in which the output of an HMC chain strongly depends on the quality of the prior, in particular because not all (induced) earthquake model parameters have a linear relationship with the recordings observed at the surface. To mitigate the importance of an accurate prior, we integrate the linearized HMC scheme into a workflow that (i) allows for a weak prior through linearization around various (initial) centroid locations, (ii) is able to converge to the mode containing the model with the (global) minimum misfit by means of an iterative HMC approach, and (iii) uses variance reduction as a criterion to include the output of individual Markov chains in the estimation of the posterior probability. Using a three-dimensional heterogeneous subsurface model of the Groningen gas field, we simulate an induced earthquake to test our workflow. We then demonstrate the virtue of our workflow by estimating the event's centroid (three parameters), moment tensor (six parameters), and the earthquake's origin time. Using the synthetic case, we find that our proposed workflow is able to recover the posterior probability of these source parameters rather well, even when the prior model information is inaccurate, imprecise, or both inaccurate and imprecise.@en

Estimating earthquake parameters, including their uncertainty, requires probabilistic sampling or inversion using Bayesian algorithms. One such Bayesian algorithm known to be highly efficient is the Hamiltonian Monte Carlo (HMC) algorithm, and modifying the algorithm with an additional linearization step can further increase this efficiency. However, the modified HMC relies heavily on accurate prior information to effectively sample non-linear earthquake parameters (e.g., hypocenter and origin time). Furthermore, the ability of the modified HMC to estimate non-linear parameters diminishes with respect to the high degree of non-linearity that is inherent to some types of events, such as induced earthquakes. To address this, we adjust the modified HMC to be run in multiple stages, combined with pre-determined initial prior sets. We test this adjustment using synthetic and real data from an induced earthquake event in the Groningen gas field in the Netherlands. We start by obtaining an initial estimate of the prior information and use it to draw multiple initial prior sets. We then run the HMC for each initial prior set in multiple stages where the results from the current stage serve as the prior for the next stage. As the final step, we form the final posterior distributions by selecting results that give the best fit between the observed and modeled data. Within this approach, we estimate ten earthquake parameters those are the six components of a full moment tensor solution, the centroid (three coordinate components), and the earthquake's origin time (including the static time corrections for each recording station). After obtaining the final results, we compare our findings with those of an existing earthquake catalog and several other research results. Given the available fault map of Groningen's subsurface, we found that our results have a higher degree of correlation with respect to the major subsurface faults.@en