利用多元地震属性预测测井特性

Using Multi-attribute to Predict Log Properties from Seismic Data

摘要:

通过寻找井旁地震数据与测井曲线的关系,将这一关系应用到远离井的区域(只有地质数据,但无测井)来预测测井的有关特性,其方法有单属性分析和多属性分析[1]。本文通过实例描述了多属性分析的特点及预测结果。从单属性回归到多属性预测、再到神经网络预测过渡时,预测能力持续提高。同时对地震属性的选择和有效性进行了讨论,将结果应用到整个二维地震剖面上,能更好地确定井以外区域的测井特性。

Abstract:

Through forming a relationship between log and seismic data, and appling the relationship into area far from the logging well, for where there are only geological data without logging, the log property in the area can be predicted by the data from the logged well. There are two methods for this target： single-attribute analysis and multi-attribute analysis. In this paper, by an example, the characteristics of multi-attribute analysis and result of the method are described. The predictive power improves more and more when we use single-attribute regression, linear multi-attribute prediction and neural network prediction in order. Meanwhile how to choose the seismic attributs and determine their validities are discussed also. Applying this result into 2-D seismic profile could determine the log properties in the area outside the logged well.