Abstract:When we use manifold seismic inforrnations to forecast the presence of hydrocarbon for 35 exploratory wells which are being drilled in South Sichuan, the success rate is above 70%;but when we use a single information, it is only 40-55%. Using manifold seismic informations, we do not qualitatively infer hydrocarbon, but analyse the presence of hydrocarbon in terms of confidence level in mathematical statistics, that is to say, evaluate the presence probability of statistical informations. According to the hydrocarbon control ability of each information factor, the weighted distributive formula of information factors is given on the basis of Bayes' Theorem. Six predicting methods are described in this paper. After a comment on them on the basis of confidence level and information ratio, the Bayes-L method is considered as the best one. As far as any one method is concerned, different numbers of information factors can lead to different prediction results. Experience shows that the decisive factors for predicting hydrocarbon are the apparent polarity, amplitude ratio, and instantancous phase of the reflection from the top of Yangxin formation, as well as the curvature of the upper surface of this formation.
引用本文:
高如曾. 预测油气富集的数理统计法[J]. 石油地球物理勘探, 1984, 19(4): 343-357,367.
Gao Ruzeng. A statistical method used for predicting rich hydrocarbon zones. OGP, 1984, 19(4): 343-357,367.