During the last five years, four hydraulic stimulations were carried out at the
geothermal site of Soultz-sous-Forêts (France). In order to monitor the seismic
activity induced by these massive injections of fluid, a surface seismological
network composed at least by 13 seismological short period and one broad-band
stations have been installed during each stimulation in complement of the 4 down-hole
accelerometers or velocimeters. Every hydraulic experiment generated several
thousands of seismic events among which more than 5,000 (September 2004 and February
2005 taken together) were located automatically by the surface network.
The location of the hypocenters is obtained through the classical Hypoinvers routine
adapted to the specific case of the network and with a variable Vp/Vs ratio or via a
simultaneous inversion together with the velocity structures. In this last case, the
uncertainty on the locations is 50 meters or less in every direction while it is the
double in the first. The magnitudes are deduced from the coda duration T following
the relation Ml = aLog T + b in which a and b were established from a set of events
whose magnitudes of moment have been obtained from the spectral analysis of the
broad-band records. The focal mechanisms are built from the polarities of P wave
first arrivals. It was always possible to built a double couple solution contrarily
at what is currently observed in regions where natural fluid circulation is present.
That doesn't mean that a non-double couple component is totally absent, but it is
always moderate.
The main features of the seismic activity will be presented and then we will focus on
the largest microseismic events, in the magnitude range 1.4 to 2.9, 1.4 is the
magnitude of the smallest events felt by the population and 2.9 is the magnitude of
the largest event ever produced during the hydraulic experiments in Soultz. For these
events, the arrival times of the P and S-waves recorded at the surface have been
manually picked and merged to the data from the downhole stations.
The main conclusions are:
1 - the non-overlapping of the seismic clouds induced by the successive stimulations;
2 - most of the mechanisms show normal faulting to strike-slip faulting in a stress
field compatible with the regional one deduced from in-situ measurements and focal
mechanisms;
3 - the “large” earthquakes underline the role of the large structures on the
circulation of the fluid within the reservoir;
4 - the relative amount of large events increases after the shut-in.

BRGM
3, avenue Claude-Guillemin
45060 Orléans cedex 02
France