Contribution of SeismoCitizen Raspberry Shake dense network in monitoring induced seismicity in northern Alsace (France)

PrESENCE ANR project (2022-2025) focuses on seismic hazards induced by deep geothermal operations in northern Alsace, France, and their associated societal perception. Seismological observations are obtained using a large number of low cost internet-connected equipment (Raspberry Shake seismic station and associated open access data). The breakthrough strategy of the project relies on the deployment of the stations in residences or administrative buildings of non-seismologist volunteer citizens or authorities. The aim is to use those stations to densify the french permanent seismic network, and to improve the detection and location of seismic events, in particularly small ones. Our presentation will be focused on the Soultz-sous-Forêts and Rittershoffen areas (northern Alsace, France), which are sites of deep geothermal operations. 

 

The topology of the seismological network was determined by the location of permanent stations, from Epos-France permanent network (4) and public stations belonging to geothermal operators (2), the number of low-cost stations (35) to be deployed in the region, the location of deep geothermal power plants (Soultz and Rittershoffen) and the location of volunteer citizen hosts. Volunteer citizens were selected initially by word of mouth, then by a call for applications (through social networks, flyers, local newspapers). Twenty-one stations are currently (end of 2023) hosted in the area. About ten additional stations are planned to be deployed early 2024 in the area.

 

Based on our past experience in deploying similar networks in other contexts and regions (Mayotte, Vosges massif, Mulhouse, etc.), we have consolidated the installation of these stations to ensure reliable data acquisition and, in particular, to achieve better data completeness (acquisition directly at the station using the Seedlink protocol via a VPN, hardware watchdog). We use Ansible (an open source IT automation platform) to facilitate the deployment of Raspberry Shake stations configuration and management tasks, ensuring rapid and consistent production deployment.

 

The workflow for building the seismicity catalog benefits from our advances in the use of new artificial intelligence tools, such as PhaseNet, a deep learning automatic picking method, as well as in the development of a deep learning method for discrimination between earthquakes, quarry blasts and explosions. Our tests over the year 2023 show that even if the stations are installed in urban areas (and therefore in a noisy environment), the network is able to automatically detect and locate many small induced earthquakes, including around 250 with a high level of confidence, compared with the ten detected or so by the standard procedure of BCSF-Renass, the French National Observation Service.