基于多t-x域联合的多震源地震数据混叠噪声分离方法

Blending interference noise separation method of simultaneous source seismic data based on multi-t-x domain combination

随着多震源同时激发采集方法在地震油气勘探领域的应用逐年增加,许多学者对其关键技术多震源地震数据混叠噪声分离进行了研究,就目前为止,仍需开展保真度更高的方法研究。基于以下设想:对多震源地震数据做正常时差校正(NMO),可以加强反射信号的线性分布程度(尤其在共中心点CMP道集内),并扩大反射信号和混叠噪声的分布差异;单域分离混叠噪声,仅针对混叠噪声和反射信号的某一分布特性差异进行分离,多t-x域联合分离可综合利用各t-x域内混叠噪声和反射信号在各域内的分布差异,其分离效果更好。首先,通过对多震源数据CMP道集做NMO以强化反射信号的线性程度、扩大随机分布混叠噪声的离散程度,利用中值滤波分离大部分混叠噪声,再在其他t-x域内(如共炮点CSG道集、共检波点CRG道集和共偏移距COG道集等)根据残余混叠噪声分布特征利用随机噪声衰减等方法进一步分离,并在整个分离处理中使用保幅处理方法。提出的多t-x域联合分离混叠噪声方法,经理论数据验证,相较于单域分离方法可以有效且保真分离多震源数据中的混叠噪声,并对其他噪声和横波有一定的压制作用;实际数据应用效果表明,该方法比单域方法分离效果较好,叠后成像效果品质较高。多t-x域联合分离混叠噪声方法,能够稳定、可靠和保真地分离多震源地震数据中的混叠噪声。

With the increasing application of simultaneous source seismic data acquisition method in seismic oil and gas exploration, many scholars have studied the key technology of simultaneous source seismic data blending interference noise separation. So far, research on methods with higher fidelity is still needed. This article is based on the following assumptions: Normal Moveout Correction(NMO) in simultaneous source seismic data can enhance the linear distribution degree of reflection(especially in Common Middle Point(CMP) gather), and expand the distribution difference of reflection and blending interference noise;single domain separation of blending interference noise only aims at a certain distribution characteristic difference of blending interference noise and reflection;multi-t-x domain combined separation can comprehensively utilize the distribution difference of blending interference noise and reflection in each t-x domain, and its separation effect is better. We first use NMO for simultaneous source seismic data CMP gather to enhance the linearity of the reflection, and expand the dispersion of randomly distributed blending interference noise. Secondly, we use median filtering to separate most of the blending interference noise. Finally, based on the residual blending interference noise distribution in other t-x domains(such as Common Shot Gather(CSG), Common Receiver Gather(CRG) and Common Offset Gather(COG) etc.), the random noise attenuation is used for further separation, and the amplitude-preserving method is used in the whole separation process. The multi-t-x domain combination separation method is verified by theoretical data, and compared with the single domain separation method, it can effectively separate blending interference noise from simultaneous source seismic data with high fidelity, and suppress other noises and shear waves to a certain extent. The actual data results show that the method has better suppression effect and better seismic profile results.