Integrated Quantitative Distributary Channel Reservoir Characterization Based on Seismic Variational Mode Decomposition

Abstract

River-dominated delta front deposition in the shallow lacustrine basin is usually featured by small-scale and multiple-coeval distributary channels, spatial overlap formed complex interbedded sandstone and mudstone succession. As potential hydrocarbon reservoir, the architecture of sandbodies is difficult to characterize because of the strong heterogeneity. In order to establish a reliable correlation between seismic expression and geological features, seismic spectral decomposition is widely applied to such reservoir characterization recently. Routine seismic decomposition methods are constrained by interpretation parameters, so artificially generate “target” features based on human defined expressions, leading to artifacts and misinterpretation.

For more reasonable seismic component isolation, a data-driven decomposition is more appropriate and highly needed than an interpreter-defined decomposition method using predefined parameters. Variational mode decomposition (VMD) method is a data-driven decomposition method, which decomposes a non-stationary signal into an ensemble of band-limited intrinsic mode functions (IMF). VMD solves an optimization problem in the frequency domain to best isolate intrinsic seismic components, which concentrate to their own central frequencies but in the meanwhile well separated from other modes. In order to apply the VMD method to field applications, we modified it in two main aspects, lateral consistency, and 3D implementation. Then we adopt the following steps for the reservoir characterization: (1) construct the smallest seismic sequence stratigraphic units in 3-D volume, by correlating vertical facies-stacking in wireline logs and cycle patterns with seismic stratigraphic surfaces; (2) decompose seismic signal by VMD in stratigraphic framework; and (3) recognize the lateral boundaries of distributary channel sandbodies by integrating RGB color blending of IMFs with well logs analysis.

The results indicate though it doesn’t enhance high frequencies, the VMD enhances the signal to noise ratio (SNR) in the specific components which improve the ability to recognize lithological boundaries on different components by suppressing interference from other less representative components. The lateral boundaries of distributary channel sandstone, width is 200-300m and the thickness less than 6m, have been well recognized in this study, which contributes to characterizing the dimension, connection, and distribution of reservoir.

AAPG Datapages/Search and Discovery Article #90323 ©2018 AAPG Annual Convention and Exhibition, Salt Lake City, Utah, May 20-23, 2018