Traditional geothermal development in the Paris basin typically involves drilling deviated doublets to intersect permeable oolite layers. To increase geothermal production, also because of the limited surface footprint, multi-lateral wells were drilled in this project to more than double the production. 3D Far-field (3D FF) sonic technology was applied in these wells with two objectives:

1) Determine lateral continuity of the (sub-seismic resolution) layers for optimal well placement.

2) Establish whether 3D FF sonic can be used to eliminate the pilot hole drilling.

3D Far-field sonic provides excellent data quality in this region, a result of clear reflections between fast dolomite layers and the slower oolite layers. This provides the opportunity to see reflectors up to 40 m from the borehole. The challenge of the geothermal laterals is to make sure the well is landed in these highly permeable oolite sections. Combining unique azimuthal sensor capabilities with innovative proprietary workflows to extract dip and azimuth, we managed to position horizontal layers effectively above or below the well. Using this technology, a dune structure was identified 10 meters below the well. Despite many decades of exploration in this basin, this structure was not previously known to exist in this play and now forms an exciting new potential aquifer target.

Minimizing the occurrence of noticeable seismicity is a key aspect to increase the public acceptance of geothermal energy production in Germany. This requires a fundamental understanding about the subsurface response to pressure changes at reservoir depths, as well as a precise estimation of the impact of associated induced seismicity at the Earth’s surface.

We use numerical simulations to predict maximum surface veolcity amplitudes (peak ground velocity values, PGV) in the target areas. Here, the accuracy of the PGV-estimates highly depends on the quality of the seismic velocity model, wich is setup using information from 3D seismic explorations and stratigraphic boundaries from the regional GeORG model. We further include H/V measurements and 1D velocity profiles, calculated from dispersion curves, to improve the accuracy of the near-surface model. The results of the H/V mesurements show that the thick sediments in the Upper Rhine Graben do not feature significant velocity contrasts at depths that are relevant to induced seismicity (no classical site effects). However, the modelling studies still reveal a significant impact of the shallow velocity distribution on the final PGV values.

In a final step we simulate different (realistic) source mechanism scenarios and combine the results to derive worst-case scenario PGV maps.

Earthquake localization using Distributed Acoustic Sensing (DAS) remains challenging due to the directional limitations of single-component DAS systems. Here, we present a wavefield-based localization approach that employs dense DAS recordings and known subsurface heterogeneity to constrain source positions. The method relies on full-waveform simulations to generate synthetic DAS wavefield images across a range of possible earthquake locations. A deep convolutional neural network (CNN), based on a U-Net architecture, is trained to associate these wavefield patterns with corresponding source coordinates, without requiring the identification and picking of P- and S-wave arrivals.

The approach is evaluated using a synthetic case study with DAS data from a single vertical borehole and tested across multiple velocity models of increasing geological complexity. The results show that the CNN effectively learns location-dependent waveform features influenced by structural heterogeneity. Localization accuracy improves further with the inclusion of a second DAS borehole, which also helps to resolve ambiguity due to noise and uncertainties in the velocity model. Although the study is based on idealized two-dimensional models, the results present a promising approach for improved microseismic monitoring in settings where detailed structural information, such as that from seismic surveys, is available.