Integrated gravity and gravity gradient data focusing inversion

Gravity data contains a wealth of low frequency contents and gravity gradient data contains many high frequency contents. Thus, in theory, we can get a more reliable result through integrated gravity and gravity gradient inversion. Here, we propose a method to integrate gravity and gravity gradient data in inversion. The computation time and requirement for memory of the inversion are increased with multiple components included. Thus, we present a method to calculate sensitivity matrixes of different components to reduce the computational time. We adopt limited-memory BFGS quasi-Newton algorithm to solve the inverse problem. It uses curvature information from only the most recent iterations to construct the Hessian approximation. The requirement for storage is reduced in this way. A weighting scheme for resolution enhancement at depth is introduced through the re-weighted method. We estimate the depth of the anomalous body by single component inversion. Then the depth information is incorporated into the depth weighting functional. At last, we adopt re-weighted method to combine the depth weighting functional with the objective functional. We use a synthetic example to demonstrate the process. Although the estimate of depth is not accurate, we can get a more accurate inversion result by combine the depth information into the depth weighting function and apply it in inversion. We adopt a synthetic example to explore the advantages of integrated gravity and gravity gradient inversion. The results show that integrated gravity data and gravity gradient data in inversion, the addition noise which is inconsistent with the corresponding components can be identified. The recovered model is more reasonable. However, for different component combinations, the inversion results are similar, which indicates that the improvement of the recovered model is small. At last, we apply the method to real data of the Vinton salt dome, Louisiana, USA. The results indicate that the recovered model is improved through integrated gravity and gravity gradient inversion.