基于模型正演的地震属性优选与加权融合技术应用——以四川盆地中部深层碳酸盐岩储层预测为例  

Forward modeling-based seismic-attribute optimization and weighted attribute fusion:Application to predicting deep carbonate reservoirs,central Sichuan Basin

作　　者：邹定永 邵佳 代瑞雪 杨晋蓉 许翔 马乾 吕文正 梅花浪雪 ZOU Dingyong;SHAO Jia;DAI Ruixue;YANG Jinrong;XU Xiang;MA Qian;LYU Wenzheng;MEI Hua-angxue(Exploration and Development Research Institute,PetroChina Southwest Oil&Gasfield Company,Chengdu,Sichuan 610041,China;Southwest Geophysical Research Institute,CNPC BGP Inc.,Chengdu,Sichuan 610213,China)

机构地区：[1]中国石油西南油气田公司勘探开发研究院,四川成都610041 [2]中国石油集团东方地球物理公司西南物探研究院,四川成都610213

出　　处：《天然气勘探与开发》2024年第4期46-54,共9页Natural Gas Exploration and Development

基　　金：中国石油集团公司攻关性应用性科技专项(编号:2023ZZ16)。

摘　　要：四川盆地中部PL地区D1井区震旦系灯影组四段埋深从5 600 m到7 000 m不等,储层发育,属超深储层。如何在地震分辨率相对较低的超深储层,寻找到最有利的优质储层是当前勘探开发工作的重点。通过以实钻井储层厚度及分布位置等数据为基础的正演模型,探寻不同地质条件下的储层响应特征,并基于正演模型对吸收系数、波形长度、反射强度、瞬时频率等地震单属性进行优选;在优选属性的基础上进行多属性融合分析,即通过多个单属性与储层厚度的相关性分析,确定融合属性数量及权重系数,结合加权求和方式,获取融合属性。研究结果表明:(1)基于实际井资料的模型正演,可以有效地指导地震属性的提取与优选,避免了盲目的大量属性提取;(2)融合属性相比于单一属性与储层厚度之间关系具有更高的相关性,并提高了储层预测符合率。该技术流程可大大缩短盲目提取、筛选常规属性的时间,并有效提高了属性的定性预测精度。With the burial depth ranging from 5,600 m to 7,000 m and well-developed reservoirs,the fourth member of Sinian Dengy-ing Formation in D1 well block,PL area,central Sichuan Basin,belongs to ultra-deep reservoirs.Seeking for the most favorable high-quality reservoirs in such ultra-deep strata even with a bit low seismic resolution has received great concerns in current explo-ration and development.Thus,based on such drilling data as reservoir thickness and extensional position,the forward modeling was conducted to identify reservoir response under geological conditions,and accordingly several seismic attributes containing absorption coefficient,waveform length,reflection intensity and instantaneous frequency were individually optimized to analyze the multi-attri-bute fusion,namely that the correlation of individual attribute to reservoir thickness was explored to fix on both number and weight coefficient in these attributes,and finally to clarify the fusion attributes in combination with the weighted sum.Results show that(i)the forward modeling may effectively guide attribute extraction and optimization,and avoid excessive extraction blindly;and(ii)com-pared with a single attribute,the fusion attributes have higher correlation with reservoir thickness,leading to an increasing accuracy in reservoir prediction.This process is time-saving in both extraction and screening of conventional attributes,and raises the accuracy to predict the attributes qualitatively.

关 键 词：模型正演 地震属性 属性优选 融合属性 储层预测 

分 类 号：P618.13[天文地球—矿床学] P631.4[天文地球—地质学]