摘要
特殊的成藏条件和低幅度构造圈闭致使柴达木盆地三湖地区第四系存在大量的低饱和度气藏.储层物性较差,储层厚度偏薄,受围岩和测井仪器分辨率的限制,难以准确划分储层;高泥质含量、高束缚水饱和度、高地层水矿化度和粘土矿物的影响,致使测井曲线在低饱和度气层表现出许多模糊性,使低饱和度气层的识别显得尤为困难.针对这一问题,文章采用小波分析对测井曲线进行重构,以提高测井曲线纵向分辨率,并与成像测井资料进行对比分析,利用提高分辨率之后的测井曲线对储层进行准确划分;同时使用决策树建立低饱和度气层的预测模型,根据决策树学习过程的透明性、学习结果可理解性,并综合储层实际特征,对预测模型进行修正,达到准确识别低饱和度气层的目的 .实际应用表明小波分析技术和决策树方法有效地解决了研究工区储层难以划分和低饱和度气层识别困难的问题.
The particular reservoir condition and low-amplitude structural trap generate the abundant low saturation natural gas in the Quaternary of the Sanhu area in the Qaidam basin. It is difficult to accurately delineate reservoirs because of the poor reservoir properties, thin reservoir thickness and limitations of surrounding rocks and logging instrument resolution. The effects of the high shale content, high irreducible water saturation, high formation water salinity, and clay minerals result in the log curves show much ambiguity at low-saturation natural gas, so that the identification of low-saturation natural gas is particularly difficult. To solve this problem, this work uses wavelet analysis to reconstruct log curves in order to improve the vertical resolution, makes a comparative analysis with the imaging logging data, and uses improved log curves to accurately delineate reservoirs. At the same time, we employ the decision tree to set up the predictive model of low-saturation natural gas based on the transparency of learning process and intelligibility of study results of the decision tree. This study amends the predictive model based on actual characteristics of reservoirs and achieves the purpose of an accurate identification of low-saturation natural gas.Practical application shows that the wavelet analysis and decision tree can effectively solve the reservoir delineation and identification of low-saturation natural gas problem in the research area.
出处
《地球物理学进展》
CSCD
北大核心
2011年第1期240-245,共6页
Progress in Geophysics
基金
国家高技术研究发展计划863项目的部分研究成果(2009AA062802)资助
关键词
小波分析
纵向分辨率
储层划分
决策树
低饱和度气层
预测模型
识别
wavelet analysis, vertical resolution, reservoir delineation, decision tree, low-saturation natural gas,predictive model, identify
