地震相干性是由地质构造、地层、岩性、孔隙度和烃类物质变化而引起地震响应横向变化的一种测量。它不像晕渲地形图根据层位拾取能对断层和河道进行3-D可视化,地震相干分析是利用地震资料本身,因而不受解释人员和自动拾取软件偏差的影响。本文提出了一种比较稳健的、多道、基于相似性的相干算法,允许我们去分析质量比原来的三道互相关算法还低的地震数据。这第二代基于相似性的相干算法,其垂直分辨率超过我们原来的零均值互相关算法,减少了上覆或下伏地层特征的混淆。一般而言,我们尽可能用一个狭长的时窗分析地层特征,一般是由地震数据中可利用的最高频率决定的时窗;在极限情况下,任何人都可以有信心将我们这个新的基于相似性的算法应用到一个样点大小的数据体中,该数据体是沿一常规拾取地层层位得到的。而相应于波峰或波谷的层位,其幅值比地震环境噪声大得多。比较起来,近于垂直的构造特征,如断层,当使用一个相当于输入地震数据中可得到的最低频率的较长分析时窗时,其特征得到较好地增强。对整个地震数据体反射层倾角/方位角的计算,使我们能把常规的复数遗属性(包括包络、相位、频率和带宽)的计算推广到复反射层属性的计算(相干分析时窗内,将输入数据沿反射层倾角倾斜叠加而产生)。利用常规地质统计、聚类和分割算法,把这些稳健的复反射层属性同相干性和倾角/方位角结合起来,从而提出一个完整的多属性分析方法。 Earthquake coherence is a measure of the lateral change in seismic response caused by changes in geologic formations, formations, lithology, porosity, and hydrocarbon species. Unlike 3-D visualization of faults and channels by flotation topographic maps, which are based on the horizon, seismic coherence analysis utilizes the seismic data itself and is therefore immune to deviations from interpreter and automated pick-up software. This paper presents a robust, multi-channel, similarity-based coherence algorithm that allows us to analyze seismic data of lower quality than the original three-channel cross-correlation algorithm. This second-generation coherent algorithm based on similarity, whose vertical resolution exceeds our original zero-mean cross-correlation algorithm, reduces the confusion of overlying or underlying stratigraphic features. In general, we analyze the stratigraphic features as often as possible using a long narrow time window, which is generally the time window determined by the highest frequency available in the seismic data. Under extreme conditions, anyone can be confident that our new, The similarity algorithm is applied to a sample-sized volume of data obtained along a conventional layer-by-layer basis. And corresponding to the peak or trough of the horizon, its amplitude is much larger than the seismic environment noise. In comparison, near-perpendicular tectonic features, such as faults, are characterized better by using a longer analysis window equivalent to the lowest frequency available in the input seismic data. The calculation of the dip / azimuth of the entire seismic data body reflector allows us to generalize the computation of the conventional complex properties (including envelope, phase, frequency and bandwidth) to the calculation of the properties of the complex reflector (coherent analysis time window , The input data obliquely stacked along the reflective layer resulting). Using the conventional geostatistical, clustering and segmentation algorithms, these robust properties of complex reflector are combined with coherency and dip / azimuth to provide a complete multi-attribute analysis approach.