The potential of the electromagnetic induction method to monitor temperature changes in the near-surface

Geothermal heat production might pose the risk of degrading groundwater quality due to temperature changes, which may lead to ecological and economic impacts. Monitoring and quantifying spatial and temporal changes in the groundwater temperature are challenging but necessary for reliable environmental evaluations. Electromagnetic Induction (EMI) measurements have been extensively used in environmental monitoring, since Electrical Conductivity (EC) is very sensitive to changes in the groundwater properties, such as the presence of contaminants or changes in fluid temperature. Subsurface EC estimates can be obtained through the inversion of EMI measurements in a non-invasive and cost-effective way. However, these estimates are inherently ambiguous because of the ill-posed nature of the inverse problem.

In this project, we investigate how to provide an accurate EC estimation of the subsurface using EMI measurements, with the aim of understanding the near-surface groundwater temperature evolution. We test two different inversion methodologies to estimate horizontally layered EC models. The first, using a search in a pre-computed and stored database containing a discrete version of the full solution space finds the model that fits the data best in the least squares sense. The second, using a gradient descent optimization. We tested the method using different numerical scenarios. We show that for a horizontally 3-layered model using a gradient descent optimization might be insufficient to obtain an accurate estimate. Moreover, we demonstrate that the in-phase part of the measurement contains information about the electrical conductivity that might be useful to include in the estimation. Finally, our results give insight into the challenges and limitations of the estimation of EC horizontally layered models using frequency domain EMI data in the context of geothermal operations in the Netherlands.