基于网格节点密度的重力正反演方法及其应用研究

发布时间：2018-06-23 15:35

本文选题：节点密度 + 重力正演　；参考：《中国地质大学(北京)》2016年博士论文

【摘要】：正演是反演的基础,重力正演计算效率和质量直接决定了反演方法的实用效果。基于直立六面体模型剖分的重力异常正演中存在严重的冗余计算,降低了模型正演效率,为此本研究引入网格节点密度的概念,对原有重力正演公式进行简化,得到基于网格节点密度的重力正演公式。不同计算环境下的计算速度对比实验结果表明,基于网格节点密度的重力正演方法有效消除了冗余计算,提高了正演计算效率,加速比可以达到4到8倍。通过对边界节点的特殊处理,基于网格节点密度的重力正演方法能够正确计算无限延伸,或具有真实密度模型的重力异常。此外,网格节点密度定义为块体密度的差分形式,具有良好的稀疏性,这在一定程度上有利于提高计算效率,节省物理内存消耗。利用上述正演计算方案,本文正演计算了Marmousi2密度模型、BP2004密度模型和SEG/EAGE三维盐丘密度模型的重力及重力梯度异常。在密度模型结构简单情况下,网格节点密度模型较块体密度模型具有更好的稀疏性。为此,本文在基于网格节点密度模型正演研究基础上,进一步开展了基于节点密度模型的稀疏反演方法研究。在反演过程中,采用柯西范数模型约束来得到稀疏节点密度模型解,通过引入深度加权约束改善模型深度分辨率。模型试验表明,该反演方法能够真实有效恢复密度异常体,异常体边界清晰,位置准确,有效改善了最小模型解、光滑模型解、聚焦约束模型解等反演结果分辨率低、异常体边界模糊的问题。此外,为改善重力物性反演深度分辨率,有效利用地震资料提供的地层深度信息,本文探索研究了地震层位约束的网格节点密度反演方法。通过在网格节点密度模型中引入地震层位加权,有效提高了反演结果的深度分辨率。在反演密度模型解释方面,提出通过构建真实密度模型,真实反映研究区地下密度结构信息,避免因异常分离导致的剩余密度异常解释上的误差。利用基于网格节点密度的重力正反演方法,结合曾母研究区区域地质、岩石物性等资料,开展研究区三维真实密度模型构建工作。利用已有地震层位信息构建背景密度模型,利用基于网格节点密度模型的重力正演剥离方法,获取研究区剩余重力异常,之后通过基于节点密度的稀疏反演方法获得研究区剩余密度异常,从而成功构建研究区真密度结构。
[Abstract]:Forward modeling is the basis of inversion. The efficiency and quality of gravity forward modeling directly determine the practical effect of inversion method. The gravity anomaly forward modeling based on vertical hexahedron model has serious redundancy calculation, which reduces the forward modeling efficiency. In this paper, the concept of grid node density is introduced to simplify the original gravity forward modeling formula. The forward gravity formula based on grid node density is obtained. The experimental results show that the gravity forward modeling method based on grid node density can effectively eliminate redundant computation and improve the efficiency of forward calculation. The speedup ratio can reach 4 to 8 times. Through special processing of boundary nodes, gravity forward modeling based on grid node density can correctly calculate infinite extension or gravity anomaly with real density model. In addition, the grid node density is defined as the difference form of block density, which has good sparsity, which is helpful to improve the computational efficiency and save physical memory consumption to some extent. In this paper, the gravity and gravity gradient anomalies of Marmousi2 density model, BP2004 density model and SEG / EAGE three-dimensional salt dome density model are calculated by using the forward modeling scheme mentioned above. When the density model is simple, the mesh node density model is more sparse than the block density model. Therefore, based on the forward modeling of grid node density model, the sparse inversion method based on node density model is further studied in this paper. In the inversion process, Cauchy norm model constraints are used to obtain the sparse node density model solution, and the depth resolution of the model is improved by introducing depth-weighted constraints. The model tests show that the inversion method can restore the density anomaly body effectively, the boundary of the anomaly body is clear, the location is accurate, and the resolution of the inversion results such as minimum model solution, smooth model solution and focus constrained model solution are improved effectively, and the resolution of inversion results is low. The problem of fuzzy boundary of abnormal body. In addition, in order to improve the depth resolution of the inversion of gravity physical properties and make effective use of the depth information provided by seismic data, the inversion method of grid node density with seismic horizon constraints is studied in this paper. The depth resolution of inversion results is improved effectively by introducing seismic horizon weighting into the grid node density model. In the interpretation of inversion density model, it is proposed to construct a real density model to truly reflect the underground density structure information in the study area, so as to avoid the error in the interpretation of residual density anomaly caused by abnormal separation. Using the method of gravity forward and inverse modeling based on grid node density and combining the data of regional geology and petrophysical properties of Zengmu study area, the 3D real density model of the study area is constructed. The background density model is constructed by using the existing seismic horizon information and the gravity forward stripping method based on the grid node density model is used to obtain the residual gravity anomalies in the study area. Then the residual density anomaly is obtained by sparse inversion method based on node density, and the true density structure of the studied area is constructed successfully.
【学位授予单位】：中国地质大学(北京)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P631.1

【参考文献】