贝叶斯判别法在GD地区叠前地震属性交会中的应用

Application of Bayesian Discretion Method of Intersection of Pre-stack Seismic Attributes in GD Area

地震叠前反演技术可以提取大量的地震叠前属性,不同的地震属性及属性组合对于岩性或流体的敏感性不尽相同,必须对地震属性进行优选。传统的地震属性交会技术是一种有效的优选手段,但存在不确定性。为解决这一问题,引入贝叶斯判别法来量化这种不确定性。首先根据实际的储层岩性,利用地震属性交会技术将岩性分成不同的组,提取每一组的地震属性,对这些地震属性做分类统计;然后基于贝叶斯判别法计算出单个地震属性及地震属性组合的后验概率;最后依据贝叶斯法的决策准则求出各属性及属性组合的误判平均概率。利用这种方法对GD油田某目的层10种地震叠前属性及属性组合做流体识别能力分析,优选出对流体敏感的单属性及属性组合。

A large number of seismic attributes could be extracted by pre-stack seismic inversion.Different seismic attributes and their combinations should be optimized for different lithology and fluid sensitivities.Traditional intersection technology of seismic attribute was an effective one with uncertainty.To solve the problem,Bayesian discriminating method was used to quantify the uncertainty.First,based on the actual reservoir lithology,seismic attribute technology was used to divide the lithology into different groups to extract seismic attributes of each group to calculate these attributes based on Bayesian discriminating method;after that,the posterior probability of each seismic attribute and the attribute combinations were calculated based on Bayesian discriminating theory;finally,according to Bayesian decision rule,the erroneous of average probability of attributes and attribute combinations were derived.In this way,capability of fluid identification for 10kinds of pre-stack seismic attributes and attribute combinations are analyzed in the target zones of GD Oilfield,and the attributes and attribute combinations are selected.