https://hal-brgm.archives-ouvertes.fr/hal-01593405Jacob, ThomasThomasJacobBRGM - Bureau de Recherches Géologiques et Minières (BRGM)Rohmer, JeremyJeremyRohmerBRGM - Bureau de Recherches Géologiques et Minières (BRGM)Manceau, Jean-CharlesJean-CharlesManceauBRGM - Bureau de Recherches Géologiques et Minières (BRGM)Using surface and borehole time-lapse gravity to monitor CO 2 in saline aquifers: a numerical feasibility studyHAL CCSD2016[SDU.STU] Sciences of the Universe [physics]/Earth SciencesPouget, Anne-Marie2017-09-26 11:17:312022-07-25 03:49:332017-09-26 11:18:33enJournal articles10.1002/ghg.15321This numerical study focuses on the inversion of the CO2 plume shape and stored mass amount from both surface and borehole time-lapse gravity simulated measurements in saline aquifers. Solving for the density distribution at depth and hence the shape of the plume is an ill-posed inverse problem which requires regularization. We opt for regularized least-square inversion, using both L-Curve and general cross validation methods to obtain the regularization parameter. Numerical inversion experiments are set up by modeling the spatial distribution of injected gaseous CO2 using a 1D radial transport model, which allows testing for both injected mass amount and injection depth. We find that including surface data alone, the error on the position of the inverted plume is of the order of ∼30% for the shallowest aquifer depths of 800 to 1200 m and for injected CO2 mass larger than 10 Mt. The positioning error increases as depth increases and injection mass decreases. The addition of measurements from a single borehole to surface measurements, though stabilizing the inversion, only reduces the error on the position by a mere 5%. However, while the positioning error is high, the real CO2 mass within the inverted CO2 plume area is close to the total injected CO2 mass, with more than 90% of the injected mass being captured using surface and borehole data with the L-Curve method. This relatively high mass ratio shows that this method can be used to detect the main bulk of the injected CO2 plume.