Obtaining accurate subsurface Q (quality factor) models using full-waveform inversion (FWI) methods remains a challenging task. The forward modeling problem of viscoelastic wave propagation can be solved by superimposing N rheological bodies of Maxwell or Zener type with generalized standard linear solid rheology. However, different approaches were proposed to calculate the attenuation sensitivity kernels in viscoelastic FWI. This study reviews and compares previous theories for constructing the viscoelastic sensitivity kernels. Furthermore, we derive the viscoelastic sensitivity kernels directly following the adjoint-state (or Lagrangian multiplier) method. Compared to previous approaches, we reveal that the Q sensitivity kernels can be calculated with adjoint memory strain variables. In the numerical experiments, different methods are used to calculate the viscoelastic sensitivity kernels for comparison. We have found that when simultaneously inverting for velocity and Q models, these methods can provide inversion results of comparable quality. However, in the event of inaccurate velocity structures, the Q sensitivity kernels calculated with memory strain variables can resolve the Q anomalies more clearly, while suffering from fewer parameter trade-offs.
Coupling factors of sources and receivers vary dramatically due to the strong heterogeneity of the near surface, and they are as important as the model parameters for the inversion success. We have adopted a full-waveform inversion (FWI) scheme that corrects for variable coupling factors while updating the model parameters. A linear inversion is embedded into the scheme to estimate the source and receiver factors and compute the amplitude weights according to the acquisition geometry. After the weights are introduced in the objective function, the inversion falls into the category of separable nonlinear least-squares problems. Hence, we could use the variable projection technique widely used in source estimation problems to invert the model parameter without knowledge of the source and receiver factors. The efficacy of the inversion scheme is demonstrated with two synthetic examples and one real data test.
8 August 2024 in IEEE Transactions on Geoscience and Remote Sensing
Seismic inversion is a critical research area in seismic data interpretation. Given the powerful feature extraction and representation capabilities of deep neural network (DNN), it has been widely adopted in the seismic reservoir parameters inversion. However, the majority of DNN-based inversion methods use 1-D models due to the scarcity of well-logging labels, which are only one-dimensional (1-D) time series. And the performance of higher-dimensional DNN-based inversion methods depends on the quality of the initial inversion results, leading to an interdependence between the model and data in the time and space dimensions. Here, we propose a model and data co-evolution method for seismic reservoir parameters inversion. It employs a 1-D DNN model based closed-loop model to generate initial reservoir inversion results. Then, the evolutionary two-dimensional (2-D) model learns spatial structural features constrained by the initial reservoir inversion results to improve the spatial continuity. We tested the proposed method on synthetic seismic data with multiple fault structures, achieving the lowest inversion error and highest inversion accuracy. It also exhibits the highest accuracy on blind well in real seismic data with the structural features of underground rivers being more pronounced.
The Ordos Basin is located in the western part of the North-China Platform, where China’s largest shale oil field was discovered. Most of the production originates from relatively thin and tight sand bodies within the Chang 7 member of the Triassic Yanchang Formation. This consequently requires accurate and high-resolution inversion data to characterize the reservoirs and define the horizontal well paths. In this paper, we adopt full-waveform inversion (FWI) to obtain the velocity model for time-to-depth inversion. Compared with conventional migration velocity analysis, FWI can produce velocity models with more high-wavenumber variations that reflect thin and localized geologic features. Geologic statistical inversion is also applied to resolve the thin beds beyond seismic resolution. A comprehensive index is then computed with the inverted parameters to evaluate the reservoir and build a sweet-spot map. We define the coordinates, direction, and length of the horizontal well on the maps. The well path is subsequently carefully designed on the sections of the time-to-depth converted inversion data. The effectiveness of the workflow is verified using the application case of the Qingcheng oil field at the Ordos Basin.