基于凸集投影方法的重磁数据规则缺失重建  

Reconstruction of gravity / magnetic data with the projection-onto-convexsets methods

作　　者：闫浩飞[1] 刘国峰[2] 薛典军[1] 王林飞[1] 

机构地区：[1]中国地质调查局国土资源航空物探遥感中心,北京100083 [2]中国地质大学(北京),北京100083

出　　处：《地球物理学进展》2016年第5期2192-2197,共6页Progress in Geophysics

基　　金：国家高技术研究发展(863)计划(2013AA063905-04)资助

摘　　要：在重磁等离散数据处理中,通常需要首先按预设网格对离散数据进行网格化,继而进行后续处理.当野外测量途径沼泽、水域、凹陷等不能直接获取测量数据时,网格化后的数据在该区域通常表现为数据空缺,常规处理如此尚可,但若资料进行场源反演等特殊计算时,需要将空缺补全,本文则针对此问题进行研究.本文介绍了一种基于2D傅立叶变换的凸集投影迭代算法,能够直接对网格数据进行计算,无需坐标信息,直接补全缺失数据.该方法具有计算直接、简单、精算精度高的优点.实际数据对比分析表明,在足够的计算迭代后,计算结果数值精度与最小曲率方法相当.During the processing of discrete data such as the dataset of gravity and magnetic methods, we always make the grids of the data regular so that we can do the next processing. But when filed measurement should walk through those places unable to reach such as marshes, waters and so on, the gridding data usually exists blank areas, which means that when we should do special processing such as inversion of the data, we need to complete those blanks. This article introduces a Projection-onto-Convex-Sets （POCS） method based on 2D Fourier, and this method is able to compute the gridding data directly so that we can fill those missing values. This method has the advantages of simple and direct calculation, and during the comparison and analysis of actual data, we can find that after the number of iterations satisfying the requirements, the precision of the method is similar to that of Minimum Curvature Method, which means that it has high accuracy.

关 键 词：数据重建 傅里叶变换 凸集投影 迭代计算 

分 类 号：P631[天文地球—地质矿产勘探]