Temperature Correction for Long-Time DC Monitoring Experiments

Time-lapse DC monitoring is widely performed to study electrical conductivity (EC) variations over long periods that might derive from subsoil changed conditions, potentially interesting for hydrogeological, geotechnical, or for general environmental purposes. The inversion of time-lapse models differs from the classic inversion techniques mainly due to the implementation of techniques to reduce systematic measurement errors and obtain compact changes in subsequent inversions. For time-lapse surveys which last across seasonality, the temperature variations will affect the subsoil EC conditions, representing therefore a source of noise that might hide the electrical changes of interest. To avoid such an effect, in this study the model vector contains the resistivity at a reference temperature and the temperature as parameters. The effective EC is then computed cell-by-cell before running the forward response. A synthetic experiment is performed simulating two different ground temperature models associated to two different resistivity models, to show the comparison of classic inversion, without temperature modelling, and time lapse inversion with temperature modelling. A real case scenario is also proposed to confirm first the temperature effect on the shallowest subsoil EC, then to present the results of the temperature modeling on the conductivity final model.