The Node Density Based Gravity Forward And Inversion Method And Its Application

Forward is the basis of inversion, and the computation efficiency and quality of gravity forward method directly determines the practical application of the inversion method. Severe redundant computation exists in the block density based gravity forward method, which reduces the efficiency of the forward modeling approach. In this study, we introduce the mesh node density, and then simplify the classical gravity forward formula, and obtained the mesh node density based gravity forward formula. Computing speed tests in different computing environments show that the proposed gravity forward method effectively eliminates the redundant computations and improves the efficiency of the forward calculation, and the speedup ratio can be reach 4 to 8 times. A bonus of the mesh node density based forward approach is that this method can readily be used in computing the gravity anomalies of a horizontally or vertically infinite-extended model with real density by setting the corresponding boundary conditions. In addition, as defined as the difference of block density, mesh node density has a good sparsity. Removing all these zero node density values, leaving only the non-zero mesh node density points for the forward calculation, will help improve computational efficiency and reduce physical memory consumption. The proposed method is applied to the forward calculation of Marmousi2 density model, BP2004 density model and SEG / EAGE 3D salt density model. and a brief analysis of the correspondence between gravity, gravity gradient anomalies and subsurface density anomalies is provided.The mesh node density model has a better sparsity than the block density model when the density model has a simply model structure. To this end, this thesis studies the gravity inversion method based on the gravity forward method of node density model. In the inversion process, the Cauchy norm is used to get a sparse node density model solution, and the depth weighting constraint to improve the depth resolution. During the inversion approach, we use the L-curve method to determine the regularization parameter. Inversion model tests show that the inversion method can effectively restore the true density anomaly at the right location, and the boundary is clear. To improve the physical properties of gravity inversion depth resolution, and use of seismic data depth information, we introduced the seismic horizons constrained mesh node density inversion method. The 2D fault model test shows that by introducing weighting in the node density model, this method greatly improves the depth resolution of the final inversion results. In the interpretation of the inversion density model aspect, we construct the true density model, which effectively reflect the true density of the underground structure information, and avoid the error interpretation of the residual density anomaly caused by anomaly separation.Based on the use of mesh node density gravity forward and inversion method, and combining regional geology, petro-physical and other information of the Zengmu basin, we build the true density model of the study area. We construct the background density model with the seismic reflection interfaces and the physical properties of rocks. The residual gravity anomaly is obtained by removing the forward gravity anomaly of the background density model. The residual density model is inverted with the sparse constrained inversion approach, and the true density model is constructed with the residual density and background density model.