摘要
地震几何属性被广泛用于提取地震数据中的几何结构特征,从而辅助解释相关的地质沉积和构造过程.本文提出基于多种递归滤波和构造导向滤波的地震几何属性快速算法,能显著提高地震相干和体曲率的计算效率和分辨率.递归滤波的计算效率远高于传统的加权求和计算,且其计算成本与平均窗口的大小无关;同时,利用高维滤波的可分离特性可以将其分解为多个一维滤波来进一步提升计算效率并有利于多线程并行运算.此外,使用构造导向滤波实现相干计算,可以有效地消除倾斜构造对结果的影响而无需传统方法中耗时的插值运算.三维实际地震数据的应用结果表明,本文提出的快速算法能将传统的相干和曲率属性计算速度提高10~30倍且对断层的刻画更加完整和连续.
Seismic geometric attributes are vital seismic interpretation tools for highlighting geometric structure features and thus help interpreters understand related depositional processes and tectonic movements.Although numerous seismic attribute techniques have been developed and achieved good results,there is still room awaiting further improvement on their efficiency and accuracy.The computational cost of recursive filters is independent on the sizes of averaging windows.Meanwhile,multi-dimensional filters can be separated into multiple 1D filters to further improve the computational efficiency and the separated 2D filtering also facilitates the multi-thread parallel computation.In addition,the structure-guided filters can effectively eliminate the effect of dipping structures in calculating the semblance without the need of computationally expensive interpolation.We first show that we can significantly improve the efficiency of calculating seismic semblance by implementing the conventional window-based algorithms and improved algorithms with recursive smoothing filters and structure-oriented filters.Then,we present a novel and efficient way to compute volumetric curvatures by simply applying recursive derivative filters to the precomputed seismic slopes.3D field examples show that our implementation of volumetric curvature is superior to conventional implementations in terms of computational cost and result resolution.Based on quantitative benchmark experiments,our implementation can speed up the semblance and curvature computation by 10~30 times,compared to the conventional implementations.Our study suggests that the recursive implementations of filters and structure-oriented filters can be used for efficient seismic geometric attributes analysis of a large volume of seismic data and therefore speed up the automated interpretation process.
作者
张文
伍新明
漆杰
ZHANG Wen;WU XinMing;QI Jie(School of Earth and Space Sciences,University of Science and Technology of China,Hefei 230026,China;Geophysical Insights,Houston 77024,TX,USA)
出处
《地球物理学报》
SCIE
EI
CAS
CSCD
北大核心
2023年第8期3374-3390,共17页
Chinese Journal of Geophysics
基金
国家自然科学基金"地震资料中断层的自动化智能解释"(41974121)资助。
关键词
地震属性
相干属性
曲率
递归滤波器
Seismic attributes
Coherence attribute
Curvatures
Recursive filters