单点数字检波器地震资料中弱信号特征分析及识别方法
|
摘要
弱信号的检测和识别是当今地球物理学界非常关注的一个技术问题.对于高密度单点资料究竟多弱的信号才是弱信号,如何检测和识别,以往学术界很少有这样的文献报道.本文以理论研究为主,结合胜利油田某高密度实际资料,对此做了分析和讨论,得出以下初步结论:①就视觉分辨率而言,当弱信号的信噪比S/N>2时,较易识别;S/N=1时,有可能识别错;S/N<0.5时,通过肉眼识别和解释已基本不可能.②对于薄储集层来说,S/N=2为计算薄层厚度的信噪比分界点.③单点资料中背景噪声会较大程度上影响深层弱信号,高密度资料弱信号的死亡值就是环境噪声的幅度.④单个弱信号,它所占的频谱成分很少,随机噪声主要影响频谱的高频端和低频端,即使S/N达到5,噪声对信号的频谱改造仍十分严重.⑤研究区高密度资料频带很宽,为5—210Hz;目标层高频衰减比较快,高频的死亡线在170Hz;深层20Hz以上信息基本与噪声变化规律非常相像,弱信号已很难检测.⑥对于混杂在噪声中的水平同相轴弱信号(S/N<1),经奇异值分解(SVD)法处理后,仍能有效地检测出.经研究确定,在动校正(NMO)后的共中心点道集(CMP)资料中S/N=0.5是能否用SVD方法进行处理的临界点;即使N/S达到3,仍能用曲波变换恢复出弱信号.这给我们一个启示:对于高密度单点资料,只要处理方法得当,仍有很大的潜力识别出更多的弱信号.
The detection and identification of weak signal is a well-known technical issue in today's geophysical industry.For high-density single sensor data,there is little information on how weak the signal will be called weak signal and how to detect and identify it in existing academic literatures.Based on theoretical study and combined with analyzing LJ high density data from Shengli Oilfield these questions were touched with and discussed in this paper.We draw the following conclusions:① In terms of visual resolution,the weak signal is more easily identified when signal to noise ratio S/N>2,it may be wrongly identified when S/N=1,and it is basically impossible by visual recognition and interpretation when S/N<0.5.② For thin reservoir,S/N=2 is the lower limit for estimating its thickness.③ Background noise will significantly affect the weak signals in deep part and the death value of high-density data weak signal is just the amplitude of environmental noise.④ A single weak signal shares less in the frequency spectrum.Random noise mainly affects high frequency and low frequency part of the spectrum,and the spectrum response is remarkably altered even if S/N comes up to 5.⑤ High-density data has a wide frequency band of 5—210 Hz.Target layer has faster high-frequency attenuation and the death value of high frequency is at 170 Hz.The signal above 20 Hz in deep layer shows similar variation with the noise and the weak signal is difficult to be detected.⑥ Horizontal co-phase weak signal mixed with noise(S/N>1) can still be effectively detected after processed with singular value decomposition(SVD),and the S/N=0.5 is the cut-off point determining whether SVD can be used to process the common midpoint(CMP) data after normal moveout(NMO) or not.Even if N/S reaches to 3,it can still be restored by curvelet transform.This gives us an enlightenment that,for high-density single-point data,there is still large potential of identifying more weak signals as long as we use a proper processing technique.
The detection and identification of weak signal is a well-known technical issue in today's geophysical industry.For high-density single sensor data,there is little information on how weak the signal will be called weak signal and how to detect and identify it in existing academic literatures.Based on theoretical study and combined with analyzing LJ high density data from Shengli Oilfield these questions were touched with and discussed in this paper.We draw the following conclusions:① In terms of visual resolution,the weak signal is more easily identified when signal to noise ratio S/N>2,it may be wrongly identified when S/N=1,and it is basically impossible by visual recognition and interpretation when S/N<0.5.② For thin reservoir,S/N=2 is the lower limit for estimating its thickness.③ Background noise will significantly affect the weak signals in deep part and the death value of high-density data weak signal is just the amplitude of environmental noise.④ A single weak signal shares less in the frequency spectrum.Random noise mainly affects high frequency and low frequency part of the spectrum,and the spectrum response is remarkably altered even if S/N comes up to 5.⑤ High-density data has a wide frequency band of 5—210 Hz.Target layer has faster high-frequency attenuation and the death value of high frequency is at 170 Hz.The signal above 20 Hz in deep layer shows similar variation with the noise and the weak signal is difficult to be detected.⑥ Horizontal co-phase weak signal mixed with noise(S/N>1) can still be effectively detected after processed with singular value decomposition(SVD),and the S/N=0.5 is the cut-off point determining whether SVD can be used to process the common midpoint(CMP) data after normal moveout(NMO) or not.Even if N/S reaches to 3,it can still be restored by curvelet transform.This gives us an enlightenment that,for high-density single-point data,there is still large potential of identifying more weak signals as long as we use a proper processing technique.
引文
陈遵德,段天友,朱广生.1994.SVD滤波方法的改进及应用[J].石油地球物理勘探,29(6):783-792.
郭小龙.2007.基于混沌理论的弱信号检测方法研究[D].南京:江苏科技大学:33-35.
陆文凯,李衍达.1998.SVD分解法提高地震资料的信噪比和分辨率[J].石油地球物理勘探,33(特刊1):145-149.
吕文彪,尹成,张白林,田继东,李大卫.2007.利用独立分量分析法去除地震噪声[J].石油地球物理勘探,42(2):132-139.
马宁,陈莉,王晓军,杨文伟.2009.小波变换在弱信号检测中的应用[J].哈尔滨工业大学学报,41(3):257-258.
彭才,朱仕军,孙建库,陈俊华,夏凌,黄东山.2007.基于独立成分分析的地震数据去噪[J].勘探地球物理进展,30(1):30-35.
石战结,田钢,薛建,王者江,曾绍发,赵维俊,颜延杰.2003.单点地震技术在浅覆盖区区域地质调查中的应用研究[J].世界地质,23(1):86-89.
王红玲,詹毅,张朝霞.2007.地震资料信噪比估算方法改进[J].物探化探计算技术,29(3):189-191.
王喜双,谢文导,邓志文.2007.高密度空间采样地震勘探技术发展与展望[J].中国石油勘探,12(1):49-53.
吴安楚.2009.无线单点检波器高密度地震采集[J].勘探地球物理进展,32(3):101-106.
夏颖,祝彩霞,孙灵群.2008.地震勘探仪器在高密度采集中的应用[J].物探装备,18(1):7-10.
徐晓刚,徐冠雷,王孝通,秦绪佳.2009.经验模式分解(EMD)及其应用[J].电子学报.37(3):581-585.
曾刚,侯祥博.2009.应用小波变换和双谱方法实现弱信号检测[J].数字信号处理,33(6):60-62.
詹毅,周熙襄.2004.小波包分析与奇异值分解(SVD)叠前去噪方法[J].石油地球物理勘探,39(4):394-397.
张军华,张瑞芳,王静,单联瑜,徐辉,付金荣,于海铖,步长城.2009a.高密度资料面元细分与速度分析关系研究[J].地球物理学进展,24(6):2079-2086.
张军华,梁晓腾,傅金荣,郭见乐,郑旭刚.2009b.Curvelet变换及其在地震资料去噪中的应用[C]∥CPG/SEG北京国际地球物理年会.
Candes E J,Demanet L,Donoho D L,Ying L.2005.Fast discrete curvelet transforms[DB/OL].1-44[2008-05-20].http:∥www.curvelet.org/
Neelamani R,Baumstein A I,Gillard D G.2008.Coherent and random noise attenuation using the curvelet transform[J].The Leading Edge,27(2):240-248.
郭小龙.2007.基于混沌理论的弱信号检测方法研究[D].南京:江苏科技大学:33-35.
陆文凯,李衍达.1998.SVD分解法提高地震资料的信噪比和分辨率[J].石油地球物理勘探,33(特刊1):145-149.
吕文彪,尹成,张白林,田继东,李大卫.2007.利用独立分量分析法去除地震噪声[J].石油地球物理勘探,42(2):132-139.
马宁,陈莉,王晓军,杨文伟.2009.小波变换在弱信号检测中的应用[J].哈尔滨工业大学学报,41(3):257-258.
彭才,朱仕军,孙建库,陈俊华,夏凌,黄东山.2007.基于独立成分分析的地震数据去噪[J].勘探地球物理进展,30(1):30-35.
石战结,田钢,薛建,王者江,曾绍发,赵维俊,颜延杰.2003.单点地震技术在浅覆盖区区域地质调查中的应用研究[J].世界地质,23(1):86-89.
王红玲,詹毅,张朝霞.2007.地震资料信噪比估算方法改进[J].物探化探计算技术,29(3):189-191.
王喜双,谢文导,邓志文.2007.高密度空间采样地震勘探技术发展与展望[J].中国石油勘探,12(1):49-53.
吴安楚.2009.无线单点检波器高密度地震采集[J].勘探地球物理进展,32(3):101-106.
夏颖,祝彩霞,孙灵群.2008.地震勘探仪器在高密度采集中的应用[J].物探装备,18(1):7-10.
徐晓刚,徐冠雷,王孝通,秦绪佳.2009.经验模式分解(EMD)及其应用[J].电子学报.37(3):581-585.
曾刚,侯祥博.2009.应用小波变换和双谱方法实现弱信号检测[J].数字信号处理,33(6):60-62.
詹毅,周熙襄.2004.小波包分析与奇异值分解(SVD)叠前去噪方法[J].石油地球物理勘探,39(4):394-397.
张军华,张瑞芳,王静,单联瑜,徐辉,付金荣,于海铖,步长城.2009a.高密度资料面元细分与速度分析关系研究[J].地球物理学进展,24(6):2079-2086.
张军华,梁晓腾,傅金荣,郭见乐,郑旭刚.2009b.Curvelet变换及其在地震资料去噪中的应用[C]∥CPG/SEG北京国际地球物理年会.
Candes E J,Demanet L,Donoho D L,Ying L.2005.Fast discrete curvelet transforms[DB/OL].1-44[2008-05-20].http:∥www.curvelet.org/
Neelamani R,Baumstein A I,Gillard D G.2008.Coherent and random noise attenuation using the curvelet transform[J].The Leading Edge,27(2):240-248.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心