Imaging technologies, such as reverse-time migration (RTM), generate migration images for complex structures. However, the migrations can be distorted due to undersampled acquisition geometry, limited recording aperture and illumination effects. In order to partially correct the distortions, we present multi-iteration image-domain least-squares migration (LSM) by approximating the Hessian through point spread functions (PSFs). Since information in a migration image is insufficient to uniquely determine a reasonable reflectivity model and LSM itself can cause artifacts, we include an L p -norm regularization of the difference between the migration image and the reflec-tivity model and a total variation (TV) regularization of the reflectivity model. The L p -norm regularization reduces artifacts and makes the reflectivity model to be similar to the migration image, while the TV regularization helps to maintain structural continuity. We illustrate the LSM through RTM. Given a migration velocity model, synthetic data for a grid of scatterer points is generated through Born modeling operators and, then, is migrated to calculate PSFs. A reflectivity model is convolved with the PSFs to match a migration image. We use a nonlinear conjugate gradient method to seek an optimal reflectivity model. Preliminary results from numerical examples show that the scheme is helpful to improve resolution and reduce artifacts in image domain and broadens spectrum in wavenumber domain.
Summary Correctly injecting recorded multicomponent seismic data to generate receiver-side wavefield is a major step in elastic reverse time migration (RTM). In land reflection survey, the data are often recorded at the free surface where the traction free condition is naturally maintained. Based on the representation theorem, we propose a numerical approach to inject receiver-side signals from a free surface to reconstruct the receive-side wavefield. By injecting the traction-free time-reversed multicomponent seismic records from the surface, the method can generate correct down-going wavefields, which can further be used in the elastic RTM. Numerical examples are used to validate the injection and wavefield reconstruction procedure. Synthetic dataset is calculated in a model with a free surface boundary condition, and then re-injected to the model with a traction-free boundary condition. The reconstructed wavefields are carefully compared to the original forward propagated waves to evaluate the accuracy of the boundary treatment. By combining the regenerated receiver-side wavefield, the source-side wavefield and a vectorized image condition, we also tested the multicomponent elastic RTM. The PP, PS, SP and SS images are calculated.
In the last decade, important oil and gas deposits were discovered in complex mountainous areas in Tarim basin, western China, but many of them were accompanied by serious HSE (health, safety and environment) risks due to poor topographic conditions there. Extremely complex terrains limited ground access and made deployment of sources and receivers difficult, in particular for the dynamite-source seismic exploration, where transportation of drilling rigs was a tremendous challenge and raised serious safety concerns. It is long expected to reduce the exposure of field personnel to hazardous situations and environments by significantly reducing the fieldwork load and reasonably planning implementation route, both reducing the construction risk and improving the survey efficiency. With the application of high precision remote sensing information, especially the DEM (digital elevation model), many important results were achieved. In this paper, we present a terrain risk classification method based on the DEM, and introduce its applications, such as optimized deployment of source and receivers, planning survey routes and lift locations of helicopters, in mountainous areas of western China.
