A supervised learning method to estimate basement depth from potential fields

To estimate the basement depth from potential fields, we develop a Deep Learning method based on a neural network of the Feed Forward (FFNN) type. The data used to train and test the network are related to the Bishop synthetic model. The training was organized using a moving window along the profiles in the N-S and E-O direction, associating the corresponding depth values of the basement to each window. The moving window step is the data spacing, this allowing a data overlapping. Subsequently, after obtaining the trained net in this way, a test was carried out, checking the basement estimate related to a synthetic model built in a similar way as the Bishop model, to verify the robustness of the trained net. The method is then applied to the isostatic anomaly of a sedimentary basin in Nevada, the Frenchman Flat. The results are consistent with previous interpretation of the area, that was based on 3D gravity inversion constrained by two gamma-gamma density logs.