当前位置: X-MOL 学术Comput. Geosci. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Iterative geostatistical seismic inversion incorporating local anisotropies
Computational Geosciences ( IF 2.1 ) Pub Date : 2020-05-20 , DOI: 10.1007/s10596-020-09966-1
Pedro Pereira , Inês Calçôa , Leonardo Azevedo , Rúben Nunes , Amílcar Soares

Geostatistical seismic inversion methods use stochastic sequential simulation as the model generation and perturbation technique. These stochastic simulation methods use a global variogram model to express the expected spatial continuity pattern of the subsurface elastic properties of interest. The conditioning to a single variogram model is not suitable for complex and non-stationary geological environments, resulting in poor inverted models unable to reproduce non-stationary features such as channels, folds, and faults. The proposed method uses a stochastic sequential simulation and co-simulation method able to cope with spatially varying information using local and independent variogram models. The information about the dip, azimuth, and ranges of the local variogram model is inferred directly from the observed data. First, local dip and azimuth structural volumes are computed from seismic attribute analysis. Then, local variogram models are fitted along the directions estimated from the previous step. This information is used as steering data during the inversion, acting as proxy of the true subsurface geological complexities. Application examples in synthetic and real datasets with complex geometries show the impact of using local anisotropy models in both the reproduction of the original seismic data and the reliability of the inverted models. The resulting inverted models show enhanced consistency, where small-to-large scale discontinuities and complex geometries are better reproduced, allowing reducing the spatial uncertainty associated with the subsurface properties. This work represents a step forward in integrating geological consistency into geostatistical seismic inversion, surpassing the limitation of using a single variogram model to reproduce complex geological patterns.

中文翻译:

包含局部各向异性的迭代地统计地震反演

地统计地震反演方法使用随机顺序模拟作为模型生成和微扰技术。这些随机模拟方法使用全局变异函数模型来表达感兴趣的地下弹性属性的预期空间连续性模式。对单个变异函数模型的条件处理不适用于复杂且非平稳的地质环境,从而导致不良的反演模型无法重现非平稳的特征,例如通道,褶皱和断层。所提出的方法使用能够利用局部和独立变异函数模型来应对空间变化信息的随机顺序仿真和协同仿真方法。有关局部变异函数模型的倾角,方位角和范围的信息是直接从观察到的数据中推断出来的。第一,根据地震属性分析计算局部倾角和方位角的结构体积。然后,沿上一步估计的方向拟合局部变异函数模型。该信息在反演期间用作导向数据,充当真实地下地质复杂性的代理。在具有复杂几何形状的合成数据集和真实数据集中的应用示例显示了在原始地震数据的再现和反演模型的可靠性中使用局部各向异性模型的影响。生成的反演模型显示出更高的一致性,可以更好地再现从小到大的不连续性和复杂的几何形状,从而可以减少与地下属性相关的空间不确定性。
更新日期:2020-05-20
down
wechat
bug