Chapter 1 Introduction

Map of the recomputed moment tensors in the North Sea region.

Figure 1.1: Map of the recomputed moment tensors in the North Sea region.

The success of CCS depends heavily on understanding the present-day stress field to anticipate reservoir and cap rock response to fluid injection. Despite its importance, many proposed carbon storage sites in the North Sea are located in areas with little to no borehole stress data available, presenting a significant challenge.

Within the ACT project SHARP Storage framework, we have addressed this gap by generating a comprehensive earthquake bulletin for the North Sea, revealing spatial clusters of seismic events with the majority of earthquakes with ML < 4. Focal mechanisms of earthquakes are excellent indicators of crustal dynamics, which are essential for assessing the present-day stress field. Therefore, to improve the understanding of the in-situ stress conditions, we created a comprehensive workflow to evaluate focal mechanisms based on data from the North Sea (Kettlety et al. 2024). This guide represents an explanation of the workflow, which was used to evaluate focal mechanisms of the earthquakes with a data of sufficient quality.

1.1 Probabilistic moment tensor inversion

Due to the limited availability of waveform data and station distribution in the North Sea region we tested the complex model space using of a probabilistic approach for the moment tensor estimation, which will allow for uncertainty estimations of the retrieved source parameters. For moment tensor computations, we used software Grond (Heimann et al. 2018).

Based on data analysis and methodological tests, we computed deviatoric moment tensor (MT) inversions for the selection of the events. By doing so, we hope to resolve a stable double-couple (DC) component of the moment tensor, with some erroneous parts in the compensated linear vector dipole (CLVD) component due to inversion artefacts. It should be noted that the deviatoric solution excludes mechanisms with a volumetric change in source.

The Grond algorithm uses forward modelling of numerous potential solutions instead of directly attempting to invert for the source parameters. Synthetic seismograms (forward simulation on Figure 1.2) are compared to observations of the target earthquake (observed displacement on Figure 1.2), and the procedure is iterated for different moment tensor configurations using a Bayesian bootstrap optimiser to retrieve an optimal moment tensor solution. The developed code efficiently navigates complex model spaces, analyses the trade-offs and uncertainties of source parameters and allows for probabilistic methodology and error estimations. The latter is especially crucial for the later integration of information in further modelling of the response of reservoir and caprock to large-scale fluid injection over prolonged periods. More detailed information on the code implementation can be found in Heimann et al. (2018).

The principle of the bootstrap-based probabilistic Grond moment tensor optimisation code.

Figure 1.2: The principle of the bootstrap-based probabilistic Grond moment tensor optimisation code.

1.2 Input files

The input files required for the probabilistic moment tensor inversion using Grond.

Figure 1.3: The input files required for the probabilistic moment tensor inversion using Grond.

The moment tensor inversion requires the following input data prepared in the right format: Green’s function store (see Chapter 3), waveform files in miniSEED format (see Chapter 4) or other file formats supported by Pyrocko (Heimann et al. 2017), station files (see Chapter 4) and event files (see Chapter 7).

The filtered and windowed waveforms with reasonable data quality as well as the corresponding station files containing instrument responses are used for data fittings during the optimisation process to compute moment tensor solutions.

1.3 Getting started

Before starting computing focal mechanisms for the selected earthquake, the first point of action is to prepare and install all the required Python packages (Chapter 2). After that, you need to collect the required data for the computations (Chapter 4). Then, you can start running probabilistic moment tensor inversions (Chapter 8) and evaluating the results (Chapter 9). To compute focal mechanisms using this guide’s chapters, use the arrows to move to the following Chapter or use the main menu to jump directly into the Chapter of interest.

References

Heimann, Sebastian, Marius Isken, Daniela Kühn, Henriette Sudhaus, Andreas Steinberg, Simon Daout, Simone Cesca, Hannes Vasyura-Bathke, and Torsten Dahm. 2018. Grond - A Probabilistic Earthquake Source Inversion Framework. GFZ Data Services. https://doi.org/10.5880/GFZ.2.1.2018.003.
Heimann, Sebastian, Marius Kriegerowski, Marius Isken, Simone Cesca, Simon Daout, Francesco Grigoli, Carina Juretzek, et al. 2017. “Pyrocko - An Open-Source Seismology Toolbox and Library.” GFZ Data Services. https://doi.org/10.5880/GFZ.2.1.2017.001.
Kettlety, Tom, Evgeniia Martuganova, Daniela Kühn, Johannes Schweitzer, Cornelis Weemstra, Brian Baptie, Trine Dahl-Jensen, et al. 2024. “A Unified Earthquake Catalogue for the North Sea to Derisk European CCS Operations.” First Break 42 (5): 31–36. https://doi.org/10.3997/1365-2397.fb2024036.