| 基于数据空间和稀疏约束的三维重力和重力梯度数据联合反演 |
| |
| 引用本文： | 张镕哲,李桐林,刘财,李福元,邓馨卉,石会彦.基于数据空间和稀疏约束的三维重力和重力梯度数据联合反演[J].地球物理学报,2021,64(3):1074-1089. |
| |
| 作者姓名： | 张镕哲 李桐林 刘财 李福元 邓馨卉 石会彦 |
| |
| 作者单位： | 吉林大学地球探测科学与技术学院,长春 130026;中国地质调查局广州海洋地质调查局,广州 510760;长春工程学院,长春 130021 |
| |
| 基金项目： | 中国博士后科学基金资助项目;中国地质调查局项目;中国博士后科学基金特别资助(站前)项目;国家重点研发计划 |
| |
| 摘 要： | 随着重力和重力梯度测量技术的日趋成熟,基于重力和重力梯度数据的反演技术得到了广泛关注.针对反演多解性严重、计算效率低和内存消耗大等难点问题,本文开展了三维重力和重力梯度数据的联合反演研究,该方法结合重力和重力梯度两种数据,将L0范数正则化项加入到目标函数中,并在数据空间下采用改进的共轭梯度算法求解反演最优化问题.同时,本文摒弃了依赖先验信息的深度加权函数,引入了自适应模型积分灵敏度矩阵,用来克服因重力和重力梯度数据核函数随深度增加而衰减引起的趋肤效应问题.为了提高反演计算效率,本文又推导出基于规则网格化的重力和重力梯度快速正演计算方法.模拟试算表明,改进的共轭梯度法可以降低反演的迭代次数,提高反演的收敛速度;自适应模型积分灵敏度矩阵,可以有效解决趋肤效应,提高反演纵向分辨能力;数据空间和改进的共轭梯度算法结合,可以更好地降低反演求解方程的维度,避免存储灵敏度矩阵,有效地降低反演计算时间和内存消耗量.野外实例表明,该算法可以在普通计算机下快速地获得地下密度分布模型,表现出较强的稳定性和适用性.
|
| 关 键 词： | 重力数据 重力梯度数据 共轭梯度法 稀疏约束 数据空间 联合反演 |
Three-dimensional joint inversion of gravity and gravity gradient data based on data space and sparse constraints |
| |
| ZHANG RongZhe,LI TongLin,LIU Cai,LI FuYuan,DENG XinHui,SHI HuiYan.Three-dimensional joint inversion of gravity and gravity gradient data based on data space and sparse constraints[J].Chinese Journal of Geophysics,2021,64(3):1074-1089. |
| |
| Authors: | ZHANG RongZhe LI TongLin LIU Cai LI FuYuan DENG XinHui SHI HuiYan |
| |
| Institution: | (College of Geo-Exploration Sciences and Technology,Jilin University,Changchun 130026,China;Guangzhou Marine Geological Survey,China Geological Survey,Guangzhou 510760,China;Changchun Institute of Technology,Changchun 130021,China) |
| |
| Abstract: | With the increasing maturity of gravity and gravity gradient measurement technology,inversion technology based on gravity and gravity gradient data has been widely concerned.Aiming at the difficult problem of multiple inversions,low calculation efficiency,and large memory consumption,this paper has carried out a joint inversion study of 3D gravity and gravity gradient data.This method combines gravity and gravity gradient data,adds the L0 norm regulation term to the objection function,and uses an improved conjugate gradient algorithm to solve the inversion optimization problem in the data space.Meanwhile,this paper abandons the depth weighting function that relies on prior information and introduces the adaptive model integral sensitivity matrix to overcome the skin effect problem caused by the attenuation of the kernel function of gravity and gravity gradient data with depth.To improve the efficiency of inversion calculation,a fast forward calculation method of gravity and gravity gradient based on regular meshing is derived.Simulation experiments show that the improved conjugate gradient method can reduce the number of iterations of the inversion and improve the convergences speed of the inversion.The adaptive model integral sensitivity matrix,which can effectively solve the skin effect and can improve the longitudinal resolution of the inversion.The combination of data space and improved conjugate gradient method can better reduce the dimension of the inversion solution equation,avoid storing the sensitivity matrix,and effectively reduce the inversion calculation time and memory consumption.Field examples show that the proposed method can quickly obtain the underground density distribution model under ordinary computer,which shows strong stability and applicability. |
| |
| Keywords: | Gravity data Gravity gradient data Conjugate gradient method Sparse constraints Data space Joint inversion |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
