Monitoring the seismicity induced by fluid injections in deep geothermal reservoirs is often limited by the significant anthropogenic ambient noise level inherent to most enhanced geothermal system (EGS) projects in urban contexts. We report on the performance of a monitoring network made of three small aperture (72 m) seismic arrays composed of 21 3D nodes. We test the setup for four months (from 12 December 2020 to 8 April 2021) of the Strasbourg sequence of induced earthquakes (2019–2022) related to the EGS Georhin project (Vendenheim, France). The deployment starts a few days after the MLv = 3.6 induced earthquake of 4 December 2020 and covers the early shut-in period of the wells. We use a beamforming technique to characterize the main noise sources, which consist of slow apparent velocities of surface waves emitted from mobile anthropogenic sources (motorway and railway traffic). We detect events with a phase-weighted stacking method, which is efficient when wavefronts illuminate the arrays with a high apparent velocity. Earthquakes associated with these detections are located using a matched field processing technique. The obtained catalog includes 216 seismic events, which represent four times more events than the reference catalog from Le Bureau Central Sismologique Français–Renass (the national academic agency in charge of seismicity monitoring in France), and a reduction of the completeness magnitude from 0.3 to −0.5. The clustering of the seismicity is analyzed using waveform correlation. The enriched catalog reveals intermittent seismic activity during the slow and continuous decrease in fluid pressure after shut-in.