Investigating a Deep Geothermal Reservoir Using Ambient Noise Correlation

Abstract

The ambient noise cross correlation technique is widely used in seismology and many examples of its can be found in the literature. The method determines the Green’s function between a pair of receivers by correlating sufficiently long seismic noise records. However, few applications of the method have been performed at a local scale and at periods lower than 5s, where the seismic noise is mostly dominated by anthropogenic sources. This represents the context of our study in the area of Rittershoffen (North-East of France) where a deep geothermal plant (2500m) is about to be installed (ECOGI project). The aim of this study is to build an image of the geothermal reservoir using ambient noise cross-correlation at high frequencies (0.2 to 5 Hz) and to follow the evolution of the reservoir during the production period. We have applied the standard surface-wave ambient noise cross-correlation technique to about 3 years of continuous data recorded by short-period permanent stations in this region. At periods below 0.5 s, the dispersion curves are poorly constrained and the correlation functions are less stable over time. It appears that the non uniform distribution of the noise sources plays a major role in the poor quality of the correlation functions in this period range. An on-going approach based on beamforming from local seismic arrays is shown to provide significant improvements.