Comparison of data integration methods for neural network solution of the inverse problem of exploration geophysics

The inverse problems of exploration geophysics consist in reconstructing the spatial distribution of the medium properties in the thickness of the earth from the geophysical fields that are measured on its surface. This study considers the inverse problems of magnetometry, gravimetry, and magnetotelluric sounding, and also their integration, which means simultaneous use of various geophysical fields to reconstruct the desired distribution. Here we compare the results of neural networks (NN) directly trained on the data of several geophysical methods at once with the results of stacking of NN trained on the data of each geophysical method separately.