Three-dimensional time-lapse inversion of TEM data with application in an Icelandic geothermal site

The transient electromagnetic (TEM) method is widely used in resistivity mapping but rarely applied in monitoring experiments. In this study, we present an algorithm to invert time-lapse TEM data, with inversion of both synthetic data and field data. Three main novelties distinguish this new inversion algorithm: i) a multiple-mesh approach is used for the definition of model parameters and forward modelling, ii) the forward and jacobian computations are carried out in 3D and iii) two datasets, each composed of several TEM soundings, are inverted simultaneously with a generalized minimum support norm for time-lapse changes. In the synthetic example, dense and coarse acquisition layouts are modelled, to study the effect of data coverage on model retrieval. Coarse data coverage allows to retrieve the time-lapse anomaly, thanks to the 3D sensitivity of TEM data. However, dense data coverage over the anomalies gives better resolution, as expected. In the field example, we present the time-lapse inversion results of data collected at an Icelandic geothermal powerplant in 2019 and 2020. The TEM data were acquired for defining the baseline in a monitoring an experiment of Hydrogen sulphur sequestration planned for 2021. As expected, no variations were imaged by the time-lapse inversion.