Data processing of induced seismicity: estimation of errors and of their impact on geothermal reservoir models

Induced seismicity location and focal mechanisms are commonly used to image the sub-surface design in reservoirs among other tasks. In this Ph.D. the inaccuracies and uncertainties on earthquake location and focal mechanisms are quantified using a three-step method. The technique is applied to the geothermal sites of Soultz and Rittershoffen to investigate the effect of several criteria on the earthquake location. A good azimuthal seismic coverage and the use of seismic down-hole sensors seriously decrease the location uncertainty. On the contrary, velocity model uncertainties, represented by a 5% Gaussian distribution of the velocity model around the reference model, will multiply location uncertainties by a factor of 2 to 3. An incorrect knowledge of the sub-surface or the simplifications performed before the earthquake location can lead to biases of 10% of the vertical distance separating the source and the stations with a non-isotropic spatial distribution. Hence the sub-surface design may be distorted in the interpretations. To prevent from that fact, the calibration shot method was proved to be efficient. The study on focal mechanism errors seems to lead to different conclusions. Obviously, the angular bias may be increased by neglecting the fault in the velocity. But, it may also be the same as or even smaller than the bias calculated for the case simulating a perfect knowledge of the medium of propagation. Furthermore a better seismic coverage always leads to smaller angular biases. Hence, it is worth advising to use more than only earthquake location in order to image a reservoir. Other geothermal sites and reservoirs may benefit from the method developed here. (author)