【摘要】 本文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。研究表明,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。更多还原

【Abstract】 The paper systematically introduced and discussed five methods of S/N ratio evaluation:superposition of energy,frequency-spectrum evaluation method,power-spectrum evaluation method,correlation method and eigenvalue method. Model tests were used to correlate the influence of noise,time window,trace number,frequency band and structural attitude on the results of 5 methods. The results showed that the spectrum methods (both frequency spectrum and power spectrum) have higher computational precision for the low-quality data; the size of time window has less influences on S/N ratio calculation and influence of size of time window on power spectrum method is minimum; there are also the issues of lateral resolution and precision for the S/N ratio and it greatly reduces the computational precision of superposition of energy and eigenvalue method for the incline and curved formation; the frequency band of data is not main factor affecting the computational results of S/N ratio; the eigenvalue method is decomposition and combination of horizontal signal intrinsically,which is appropriate for flatter formulation. The study pointed out that it should choose correlation method when S/N ratio is selected for singularity detection of lateral prediction attribute of reservoir (fracture,fault,buried-hill structure etc).更多还原