Tectonically active fault networks are often inter-connected, but in the case of injection-induced seismicity, prior knowledge of fault architecture tends to be severely limited. In most cases, reactivated faults due to fluid injection are inferred, after-the-fact, by the spatial distribution of induced-seismicity hypocenters; such reliance on post-injection seismicity impedes any pre-operational risk analysis, as well as development of a more holistic understanding of fault-system models. By combining high-resolution, depth-migrated 3-D seismic data with a new focal-depth estimation method that reduces spatial uncertainty of hypocenters, this study pinpoints microearthquake fault activation within a buried thrust belt in the Montney Formation in western Canada (British Columbia). During hydraulic-fracturing operations, rupture nucleation occurred on seismically imaged thrust ramps that cut through the Debolt Formation, a massive carbonate layer that underlies the stimulated zone. High-resolution seismic images reveal transverse structures, interpreted as basement-controlled fold hinges or tear faults that transferred displacement between thrust faults during Late Cretaceous - Paleogene compressional shortening. The spatio-temporal pattern of induced seismicity suggests that these transverse structures provide permeable pathways for aseismic pore-pressure diffusion, thus connecting distinct thrust faults and enabling earthquake triggering on a timescale of days and at distances of up to 2 km from the injection wells. Inferred relationships highlight how the fault system is connected, including apparent stress concentrations at the intersections of transverse structures and orogen-parallel thrust ramps.