Abstract:Gravity data, which are critical in mineral, oil, and gas exploration, are obtained from the vertical component of the gravity field, while gravity gradient data are measured from changes in the gravity field in three directions. However, few studies have sought to improve exploration techniques by integrating gravity and gravity gradient data using inversion methods. In this study, we developed a new method to integrate gravity and gravity gradient data in a 3D density inversion using the non-linear conjugate gradient (NLCG) method and the minimum gradient support (MGS) functional to regularize the 3D inverse problem and to obtain a clear and accurate image of the anomalous body. The NLCG algorithm, which is suitable for solving large-scale nonlinear optimization problems and requires no memory storage, was compared to the Broyden–Fletcher–Goldfarb–Shanno (BFGS) quasi-Newton algorithm and the results indicated that the convergence rate of NLCG is slower, but that the storage requirement and computation time is lower. To counteract the decay in kernel function, we introduced a depth weighting function for anomalous bodies at the same depth, with information about anomalous body depth obtained from well log and seismic exploration data. For anomalous bodies at different depths, we introduced a spatial gradient weighting function to incorporate additional information obtained in the inversion. We concluded that the spatial gradient weighting function enhanced the spatial resolution of the recovered model. Furthermore, our results showed that including multiple components for inversion increased the resolution of the recovered model. We validated our model by applying our inversion method to survey data from Vinton salt dome, Louisiana, USA. The results showed good agreement with known geologic information; thus confirming the accuracy of this approach.

Inversion of Gravity Anomaly Based on AlexNet Deep Neural Network

3D Inversion of Gravity Anomaly – an Application of Technology of Apparent Density Imagery

3D gravity inversion based on deep learning

Magnetic Anomaly Inversion Based on Convolutional Neural Network

3-D Gravity Inversion Based on Cross Dimensional Deep Neural Networks

Fast Forward Approximation and Multi-Task Inversion of Gravity Anomaly Based on UNet3+

Deep Learning for 3-D Inversion of Gravity Data

A Deep Learning Gravity Inversion Method Based on a Self-Constrained Network and Its Application

The inversion of apparent density based on the separation and continuation of potential field

3-D Gravity Inversion Based on Deep Convolution Neural Networks

Three-Dimensional Gravity Inversion Based on Attention Feature Fusion

Joint Gravity and Magnetic Inversion Using CNNs' Deep Learning

Three-dimensional Basin Basement Boundary Inversion of Gravity Anomalies Based on EfficientNetV2 Network

Prediction Method of Underground Density Abnormal Body in Gravity Data Based on Convolutional Neural Network

3-D Gravity Data Inversion Based on Enhanced Dual U-Net Framework.

Two-dimensional magnetotelluric inversion method based on deep learning

3D gravity inversion using cycle-consistent generative adversarial network

A Multitask Deep Learning for Simultaneous Denoising and Inversion of 3-D Gravity Data

Self-Supervised Deep Learning for 3D Gravity Inversion

Improved Gravity Inversion Method Based on Deep Learning with Physical Constraint and Its Application to the Airborne Gravity Data in East Antarctica

Integrated Gravity and Gravity Gradient 3D Inversion Using the Non-Linear Conjugate Gradient