论文详情
鉴于时不变子波的假设条件不能反映实际地震数据的非稳态特性,且传统分段提取时变子波的方法不能满足勘探所需的精度要求,提出了一种基于二阶自适应同步挤压S变换(ASST2)的时变子波提取方法,可以从非稳态地震数据中直接提取时变子波。首先对非稳态地震数据进行ASST2时频分析得到高质量的时频谱图,再通过建立局部谱的统计参数与加权指数型子波的解析关系实现逐点提取子波,最后基于时空域的时变子波矩阵建立非稳态地震记录褶积模型。与常规提取子波的方法相比,该方法不需要地震数据稳态或分段稳态的假设条件,根据数据自身特点自适应估计子波参数,克服了对子波形态的限制条件,同时避免了误差累计。模型数据测试和实际地震数据应用结果表明,该方法能够有效地提取时变子波,即使在相邻地层也能保证子波估计的准确性,从而提高地震资料处理的精度。
Conventional wavelet estimation methods involve the assumption that seismic data are stationary when a constant wavelet is considered;thus,they ignore the temporal variability of seismic wavelets.In reality,seismic data are nonstationary,and their frequency spectra change from shallow to deep formations.In this study,a time-varying wavelet extraction method was developed on the basis of the second-order adaptive synchro-squeezing S transform (ASST2).The ASST2 is a highly energy-concentrated time-frequency representation technique that allows for greater accuracy and stability of the proposed method.First,the ASST2 transform was carried out to obtain a high-quality time-spectrum image of the nonstationary seismic data;next,the wavelet was extracted point-by-point by establishing the analytic relationship between the statistical parameters of the local spectrum and the weighted exponential wavelet.Finally,the convolution model of the nonstationary seismic records was established based on the time-varying wavelet matrix in the time-space domain.The proposed method is suitable for handling the nonstationary nature of actual seismic data because it is a fully data-driven method;that is,the parameters of the time-varying wavelet are generated according to the nonstationary frequency spectrum.Moreover,synthetic tests demonstrate that the proposed method is reliable and robust,even under noise contamination.Finally,the application to field data confirms the superiority of the proposed method to conventional methods.
国家重点研发计划专项(2017YFB0202900)、国家自然科学基金(41674128)和国家重点研发计划(SQ2017YFGX030021)共同资助。