New Publication - Inverting airborne electromagnetic data with machine learning

CGI’s R&D team is constantly seeking new solutions to solve complex geoscientific problems combining professional expertise and cutting-edge technology!

Our latest study investigates how Neural Networks can provide an alternative to computationally expensive 3D airborne electromagnetic data inversions. We developed a Machine Learning model and a synthetic dataset that are tailored to inverting field data with varying acquisition parameters, requiring 3D forward modeling for training set generation only. We further demonstrate that our approach provides geoelectric models comparable to those from a 2D inversion of airborne electromagnetic data acquired over the Kaweah Subbasin in the Central Valley of California, U.S.A, instantaneously, while addressing the limitations on 1D inversion.

Read more on our website: https://bit.ly/3UJeU04