Interpreting the cause of bound earthquakes at underground injection experiments

Abstract. Constraining the maximum possible magnitude (M_MAX) of an induced earthquake sequence is a challenging process with important implications for managing risks. CAP-tests are a suite of statistical tests that can infer, quantify, and select best-fitting M_MAX models via an earthquake catalogue’s magnitudes. We use CAP-tests to discern between bound/unbound earthquake sequences at underground laboratories, where high-resolution and near-field geophysical observations are abundant. There, we find clear evidence for bound sequences, where magnitude growth was restricted during stimulation. Furthermore, bound sequences tend to be associated with stimulations that occurred within intact rock. On the other hand, unbound sequences tended to be associated with stimulations where hydraulic fractures interacted with relatively large pre-existing faults/fractures. We further examine bound sequences by fitting magnitude growth to a generalized family of M_MAX functions. This process appears to be able to aggregate bound sequences into categories consistent with theoretical considerations (e.g., tectonic, tensile-crack, or shear-crack). These results provide a basis for validating and interpreting bound sequences in controlled experiments, which is important for extrapolating to larger-scale observations. Overall, CAP-tests appear to be a promising avenue for constraining M_MAX from earthquake catalogue data.