Corinna Roy*1, Andy Nowacki1, Xin Zhang2, Andrew Curtis2, Brian Baptie3
1School of Earth and Environment, University of Leeds, UK (firstname.lastname@example.org), 2School of Geosciences, University of Edinburgh, UK, 3British Geological Survey, UK
Industrial activities like mining, hydrofracturing and the subsurface disposal of waste fluid can cause felt earthquakes. Sufficiently large earthquakes may cause damage to buildings and infrastructure, therefore, regulations often require operations to cease if an earthquake above a certain magnitude occurs. The location and magnitude of an earthquake determine its ability to cause damage, but these parameters are inherently uncertain and biased, because of imperfect knowledge of the seismic velocity structure of the subsurface, which most crucially controls the inferred location and magnitude of micro-earthquakes.
To overcome this problem, we have developed a fully non-linearised tomographic method using Monte Carlo sampling to invert jointly for event locations and 3D velocity structure using body wave travel times and surface wave group velocities (Zhang et al. 2018, in review). Our Bayesian approach allows for the calculation of realistic probabilities for the velocity structure of the Earth beneath a set of seismic stations, and for earthquake locations that occur in that space.
We applied the algorithm to a synthetic model with two velocity perturbations as a testbed to explore its main characteristics and abilities. Furthermore, we applied this new inversion technique to the mining-induced events at New Ollerton in Leicestershire, UK from 2014. We find a fast velocity anomaly south-east of the coal seams, where the majority of the seismic events occur. Additionally, we observe a tradeoff between fast shear wave velocity and source locations in this fast velocity region, which emphasizes the interlinkage between subsurface velocity model and earthquake locations.