A. Górszczyk

Enhancing 3D post‐stack seismic data acquired in hardrock environment using 2D curvelet transform

Seismic data acquired in hardrock environment pose a special challenge for processing. Frequent lack of clear coherent events hinders imaging and interpretation. Additional difficulty arises from the presence of significant amount of cultural noise associated with production and processing of ore, which often remains in the processed, stacked data. Motivated by those challenges, we developed an efficient workflow of denoising 3D post-stack seismic data by using 2D discrete curvelet transform aimed at improving signal-to-noise ratio of the data. Our approach is based on the adjustment of the thresholds according to scales and angles in the curvelet domain, making parameterization flexible. We demonstrate effectiveness of our method using 3D post-stack volumes from the three different mining camps in Canada, which were characterized by variable data quality. Remarkable signal enhancement, confirmed by the improvements in the mean signal-to-noise ratio of the dataset, is obtained not only due to random energy attenuation but also by removal of certain features corrupting the data (e.g., acquisition footprint). Comparison with the F-X/F-XY deconvolution results shows the superiority of our algorithm in respect to signal enhancement, signal preservation, and amount of the removed noise. Imaged structures, even if initially dominated by random energy, are easier to follow after curvelet denoising and enhanced for interpretation. Therefore, our approach can significantly reduce interpretation uncertainties when dealing with the seismic data acquired in the hardrock environment.

Applications of Curvelet Transform in Hardrock Seismic Exploration

We present feasibility of applying Discrete Curvelet Transform (DCT) to pre-stack data acquired in the hardrock environment using the Lalor dataset. Two algorithms were developed to help reduce random and coherent noise at various stages of the processing sequence and to obtain a high-quality seismic volume representative of the subsurface. First we demonstrate the ability of DCT to separate coherent events by employing it for ground-roll attenuation. Secondly we tackle the problem of velocity model building for pre-stack time migration (PSTM). We design an automatic workflow for gathers conditioning and moveouts picking which require only minimal human interaction. Robustness of our method is demonstrated by examples convincing that DCT can be effectively used to solve variety of problems encountered during processing and imaging of the seismic data acquired in the ore exploration context.

Full-waveform Inversion of Conventional Vibroseis Data Using Preconditioning Focused on Low Frequency Enhancement

Despite the popularity that full-waveform inversion (FWI) has gained in recent years, its application to Vibroseis data is still challenging, due to the problematic low-frequency content. We present the results of acoustic frequency-domain FWI of Vibroseis data with sweep starting at 6 Hz, acquired with standard 10 Hz geophones in 2010 in south-east Poland. 4.5 Hz matching filter and curvelet denoising in the frequency domain are used to enhance the low-frequency content of the data. This, together with dense sampling of the frequencies in the first inversion group, resulted in a geologically plausible P-wave velocity model, which correctly reproduces the data, including the far-offset arrivals and wide-angle reflections.

Toward a robust workflow for deep crustal imaging by FWI of OBS data: The eastern Nankai Trough revisited

Crustal-scale imaging by the full-waveform inversion (FWI) of long-offset seismic data is inherently difficult because the large number of wavelengths propagating through the crust makes the inversion prone to cycle-skipping. Therefore, efficient crustal-scale FWI requires an accurate starting model and a stable workflow minimizing the nonlinearity of the inversion. Here, we attempt to reprocess a challenging 2D ocean-bottom seismometer (OBS) dataset from the eastern Nankai Trough. The starting model is built by first-arrival traveltime tomography (FAT), which is FWI-assisted for tracking cycle skipping. We iteratively refine the picked traveltimes and then reiterate the FAT until the traveltime residuals remain below the cycle-skipping limit. Subsequently, we apply Laplace-Fourier FWI, in which progressive relaxation of time damping is nested within frequency continuation to hierarchically inject more data into the inversion. These two multiscale levels are complemented by a layer-stripping approach implemented through offset continuation. The reliability of the FWI velocity model is assessed by means of source wavelet estimation, synthetic seismogram modeling, ray-tracing modeling, dynamic warping and checkerboard tests. Although the visco-acoustic approximation is used for wave modeling, the synthetic seismograms reproduce most of the complexity of the data with a high traveltime accuracy. The revised FWI scheme produces a high-resolution velocity model of the entire crust that can be jointly interpreted with migrated images derived from multichannel seismic data. This study opens a new perspective on the design of OBS crustal-scale experiments amenable to FWI; however, a further assessment of the optimal OBS spacing is required for reliable FWI.

Crustal-scale Imaging from Ultra-long Offset Node Data by Full Waveform Inversion - How to Do It Right?

We present feasibility of applying full-waveform inversion (FWI) to densely sampled node data for better imaging of complex targets being beyond the reach of typical towed-streamer surveys. We attempt to process the SFJ-OBS ultra-long offset data acquired in the Eastern Nankai Trough region by first-arrival traveltime tomography (FATT) and state-of-the-art frequency-domain acoustic FWI in order to obtain high resolution crustal velocity model. Our efforts were focused on mitigating cycle-skipping and developing robust quality control workflow for this extremely challenging dataset. The starting FATT model was built in an iterative mode with first-arrival picks refinement after initial FWI. Subsequently, we develop a hierarchical, multiscale layer-stripping inversion strategy with simultaneous opening of the inverted offset range and relaxation of the damping constants, coupled with tuning of the pre-whitening of the diagonal pseudo-Hessian. As a result, the cycle-skipping is kept under control and the reliability of the final model manifests by a good data fit (quantified by the time shifts between real and synthetic data calculated using Dynamic Image Warping) and very consistent source wavelets. FWI model shows several prominent geological features and ties well with a PSDM image build from towed-streamer data acquired along the same profile.

Full-waveform Inversion Applied to Regional Vibroseis Profile from SE Poland

We show a case study of full-waveform inversion (FWI) applied to a regional 240 km long seismic profile POLCRUST located in south-east Poland. The experiment is unique because it portrays different tectonic deformation styles over a relatively short distance. It was designed for deep reflection imaging, but we decided to use FWI to provide a high-resolution P-wave velocity model of the shallow (down to 4 km) subsurface. The acquisition parameters, namely the use of 10 Hz geophones and Vibroseis sweeps starting at 6 Hz, made it necessary to design a non-standard data preconditioning workflow for enhancing low frequencies including match filtering and curvelet denoising. Final model was validated using multiple procedures, such as comparison with borehole data and independently processed Kirchhoff prestack depth migration. Velocity profiles from our model match the check shot-data and the anomalies in our model coincide very well with the events in migration, imaging the Carpathian Thrust or small natural gas reservoirs in the Miocene sediments of the Carpathian Foredeep.

Curvelet-based Gather Conditioning for Effective Depth Imaging of Legacy Seismic Data -Case Study from Central Poland

Here we demonstrate a case study of depth imaging applied to legacy data (shot in 70s and 80s) from Central Poland with a strong overprint of salt tectonics. We use a novel, curvelet-based approach to condition the low-fold gathers in order to improve the performance of the autopicker and subsequent tomographic model updates. Superior results are obtained when a proper conditioning of the gathers is done before running autopicker for tomography. Our 2D Discrete Curvelet Transform based conditioning algorithm run in a two-step mode on the common offset sections and on the depth-slices seems to improve the performance of the autopicker and thus provide more reliable input to grid tomography. Additionally, in case of legacy data, such conditioning acts as a trace regularization.

Strategies for Complex Thresholding in Curvelet-based Data Denoising

Discrete Curvelet Transform (DCT) introduce minimal overlapping between coefficients representing signal and noise in the curvelet domain, hence being well-suited for data denoising. However, appropriate and optimal weighting of these coefficients remain challenging and the most important stage of curvelet-based noise attenuation. Setting one threshold level for all coefficients may be insufficient for optimal noise attenuation, hence we focus on more complex, scale- and angle-adaptive approach to thresholding. We find empirically that adjusting threshold levels according to certain frequency bands and dips gives results superior to global thresholding. We demonstrate our noise attenuation approach by applying DCT to 3D post-stack seismic data and conditioning input data for 2D full-waveform inversion.