摘要
系统地介绍和讨论了信噪比估算的5种方法,即能量叠加法、频谱估算法、功率谱估算法、相关法和特征值法。利用模型试算对比了噪声、时窗、道数、频带、构造形态对5种方法估算结果的影响。结果表明:对于低品质的资料,谱值法(频谱或功率谱)具有较高的计算精度;时窗大小对信噪比计算影响较小,功率谱法受时窗大小影响最小;信噪比也有横向分辨能力及精度问题,对于倾斜或弯曲地层,能量叠加法和特征值法的计算精度会大大降低;资料的频带并不是影响信噪比计算结果的主要因素;特征值法本质上是水平信号的分解和组合,适宜较平的地层。研究指出,当以信噪比作为储层横向预测属性进行奇异性检测(裂缝、断层、潜山构造等)时,应选用相关法。
Five methods for calculation of S/N ratio are studied and discussed systematically. These methods include energy stack, frequency spectrum estimation, power spectrum estimation, correlation and eigenvalue . Comparing the influence on the calculation of S/N of five different methods, it is shown that, for low quality seismic data, the frequency spectrum and power spectrum estimation method have higher calculating accuracy. The length of time window has smaller influence to the calculation of S/N ratio,which has minimum impact on power spectrum method. The S/N ratio also has horizontal resolution and precision, for dipping and bending layers, the calculation accuracy of energy stack and eigenvalue method is low. The frequency band of seismic data is not the main factor for calculating of S/N ratio. The nature of eigenvalue method is the good for horizontal strata withdecomposition of level signal. If the S/N ratio is used for lateral reservoir prediction or singularity detection, correlation method should be used.
出处
《油气地球物理》
2008年第4期9-14,共6页
Petroleum Geophysics
关键词
地震资料
信噪比估算
频谱
功率谱
相关
特征值
seismic data, signal-to-noise ratio, spectrum, power spectrum, correlation and eigenvalue.
