Double-Difference Elastic-Waveform Inversion with Weighted Gradients for Monitoring EGS Reservoirs

Abstract

Reliably monitoring for enhanced geothermal systems (EGS) is crucial for placement of production wells and optimizing production of geothermal resources. We develop an improved double-difference elastic-waveform inversion method with weighted gradients for monitoring EGS reservoir changes utilizing time-lapse seismic data. In the conventional waveform inversion of time-lapse seismic data, reservoir changes are obtained by subtracting the results of independently inversion of each dataset. Our double-difference elastic-waveform inversion method jointly inverts for reservoir changes using time-lapse seismic data. The method consists of two steps: Inversion of the baseline data for the initial compressional- (P-) and shear-wave (S-wave) velocity models, and reconstruction of geophysical property changes within an EGS reservoir using differences of time-lapse seismic data. We employ the gradients weighted by P- and S-wave energy in elastic-waveform inversion to improve reconstructed P- and S-velocity models using the baseline seismic data. Our new method well reconstructs the deep structures as well as the shallow structures while the conventional waveform inversion produces poor reconstruction for the deep structures. We apply the weighted gradients to our double-difference elastic-waveform inversion method, and make use of a priori knowledge of the locations of EGS reservoirs to further improve quantification of changes in EGS reservoirs. Our numerical example using a Brady's EGS model shows that our double-difference elastic-waveform inversion with weighted gradients can quantitatively reconstruct time-lapse changes within EGS reservoirs. In addition, our new method significantly reduce image noise outside the EGS reservoir compared to those obtained using the conventional waveform inversion of time-lapse seismic data.