Inversion of 1D frequency- and time-domain electromagnetic data with convolutional neural networks

Inversion of electromagnetic data finds applications in many areas geophysics. The inverse problem is commonly solved with either deterministic optimization methods (such as the nonlinear conjugate gradient or Gauss-Newton) which are prone to getting trapped a local minimum, probabilistic very computationally demanding. A recently emerging alternative employ deep neural networks for predicting subsurface model properties from measured data. This approach entirely data-driven, does not traditional misfit and provides guess instantaneously. In this study, we examine feasibility using convolutional inversion marine frequency-domain controlled-source (CSEM) well onshore time-domain (TEM) Our yields accurate results both on synthetic real them Using several combining their outputs various training epochs can also provide insights into uncertainty distribution, found be higher regions where resistivity anomalies present. proposed method opens up possibilities estimate distribution exploration scenarios time.