基于首次湖泛面的叠前波形指示反演在深层湖底扇储层预测中的应用

苏北盆地盐城凹陷ZJD地区深层扇三角洲砂岩沉积属于事件性沉积, 采用叠后反演方法无法有效识别深层扇三角洲相薄砂岩, 而采用“沉积模型+波形分类+地震驱动”的方式进行叠前波形指示反演, 根据叠前波形指示反演得到的杨氏模量和v_P/v_S对深层砂岩进行定量描述以及含气检测则是有效的方法。首先, 利用“两定+一验”方法识别首次湖泛面, 即根据测井、地震以及岩心等资料在纵向上确定沉积旋回边界、确定框架模型, 并根据不同模型下反演结果验证首次湖泛面的界线是否合理, 建立更为合理的沉积模型并根据沉积模型初步划分波形样式; 再根据与砂岩厚度相关性高的波形结构参数对波形做进一步分类, 确定不同厚度和不同期次砂岩波形旋回结构样式; 最后, 采取“人工干预+机器学习”的方式确定最终波形样式, 并将前期确定的时窗界限、波形分类数、样本数等关键参数作为约束条件加入到叠前波形指示反演过程中, 使预测结果在纵向上和平面上符合实钻井揭示的扇三角洲相砂岩沉积规律, 进一步降低叠前波形指示反演的随机性。该方法能够预测研究区厚度大于3m的砂岩, 平均误差为11.77%, 依靠预测结果部署的井均钻遇气层, 提高了研究区深层含气砂岩预测精度。

The thin sand body deposits of the deep fan delta front in the ZJD area of Yancheng Sag in the Subei Basin are event deposits, and post-stack inversion cannot effectively identify the thin sand bodies in the deep fan delta front.The "sequence frame constraint+waveform classification+seismic drive" method was adopted to perform pre-stack waveform indication inversion, and is an effective method to quantitatively describe deep sandstone and gas detection based on Young's modulus and v_P/v_s formed by pre-stack waveform indication inversion.First the "two determinations+one experiment"method was used to identify the first flood surface, which means the sedimentary cycle boundary and model were determined longitudinally by using logging, seismic and core data.Then the first flood surface was testified according to the inversion results of different models.Thus more reasonable sedimentary models were built, and the waveforms were preliminarily divided based on the sedimentary model.Second, the waveform was further classified according to the waveform structural parameters highly correlated with sandstone thickness, and the waveform cycle structure patterns of sand bodies with different thicknesses and periods were determined.Finally, using the method of "manual intervention and machine learning" to determine the waveform style, key parameters such as the time window limit, number of waveform categories, and number of samples determined in the early stage were added to the pre-stack waveform indication inversion process as constraints.The predicted results were in accordance with the sedimentary law of the fan-delta front sand body revealed by actual vertical and horizontal drilling, and the randomness of the pre-stack waveform inversion was further reduced.It can predict sandstone reservoirs greater than 3m in thickness with a mean error of 11.77%.The practical data application results showed that all the wells deployed based on these predictions hit the gas zones, indicating that the prediction accuracy of deep gas-bearing sandstone reservoirs in the study area was improved.