The 9.1 km deep KTB (Kontinentale Tiefbohrung, Germany) drilling hole is one of the best investigated deep-drilling sites in the world. Among other parameters, in situ measurements revealed continuous profiles of principal stresses, pore fluid pressure and fracture geometry in the vicinity of the borehole. The present study combines these parameters with hydraulic and seismicity data obtained during fluid-injection experiments conducted at the KTB to derive a conceptual model for fluid-injection-induced seismicity at the KTB. This model rests on the well constrained assumptions that (1) the crust is highly fractured with a permeable fracture network between 9 km depth and the Earth's surface and (2) the crust is in near-failure equilibrium, whereby a large number of fracture planes are under near-critical condition.
During the injection experiment, the elevated pore fluid pressure remained well below the least principal stress and thus was too small to cause hydraulic opening of existing fractures. Consequently, the geometry of the fracture network was assumed to have not changed during fluid injection with induced seismicity occurring solely as a result of lowering of the effective normal stress, consistent with observed source mechanisms. The key parameter in the present model is the fracture permeability, which exhibits large spatial and directional variations. These variations are proposed to primarily control fluid migration paths and associated propagation of elevated fluid pressure during fluid injection. In contrast with common models based on isotropic fluid diffusion or spatially averaged permeability, highly permeable branches of the fracture network strongly affect the propagation of fluid pressure and prohibit the concept of a smooth ‘pressure front’. We find evidence that major fluid flow exists at comparatively low fluid pressure (below the critical pressure required to cause seismic failure) without being detected seismically. This might also explain the difference between 1011 J of hydraulic energy inserted into the system during fluid injection and ∼108 J of seismic energy: a major part of the hydraulic energy might be converted to potential energy of the ground water level caused by upward migrating fluid. From the fluid level response to changes of injection rate observed in a second borehole we estimate fluid signal velocities to be as large as 300 m d−1. Importantly, the suggested model also accounts for the occurrence of repeating earthquakes (multiplets), a large number of which were observed during the injection experiment. The present model also suggests that coseismic changes of the stress field caused by tectonic shear stress release are very local and of small magnitude. This is consistent with the observation that none of the larger induced events is followed by aftershock series that would be expected if coseismic processes had noticeably perturbed the local stress field.