马尔可夫随机场反演在火山岩储层预测中的应用

常规反演方法识别火山岩储层通常会因为地质模型与地下地质情况差异较大而导致反演结果准确率不高。针对火山岩储层预测的难点,将波阻抗与测井曲线进行贝叶斯分级划分的岩相作为随机变量,分角度叠加地震数据作为连续变量,通过马尔可夫随机场模型建立两者的关系,并将其引入到火山岩地震叠前反演算法当中。该方法克服了叠前同时反演方法必须建立低频模型的缺点。利用楔形模型进行反演测试,反演结果与模型吻合且边界收敛。将该方法应用于长岭龙凤山火山岩内幕储层预测,通过合理的划分岩相,绕开了对火山岩的岩性预测而直接进行物性预测。预测结果表明,反映火山岩物性的密度体与实钻井结果吻合度高。

Application of Markov random field inversion in the prediction of volcanic reservoirs

The exploration and development of volcanic rock reservoirs has been a focus in the oil and gas geophysics community.The conventional inversion of volcanic rock reservoirs are usually inaccurate due to the large difference between the geological model and the underground geological conditions.In this study,the lithofacies divided by Bayes classification of impedance and log curves were regarded as the random variable,whereas angle stacking seismic data was regarded as the continuous variable.The two variables were related by the Markov random field model and then introduced into the algorithm of seismic pre-stack inversion of volcanic rocks.This process helped successfully overcome the dependence of the simultaneous inversion method on the low-frequency model,which was demonstrated by the test on a wedge model.The inversion results were more consistent with the model,while the boundary was more convergent.This method was applied to the prediction of the inner reservoirs of volcanic rock in Longfengshan,Changling.The performance of physical property prediction bypassed that of the lithology prediction,through a reasonable division of lithofacies.The prediction results showed that the density body that reflects the physical properties of the volcanic rock was highly consistent with the actual drilling results.