Dissertations / Theses on the topic 'Seismic processing and interpretation'

Extracting fault, unconformity, and horizon surfaces from a seismic image is useful for interpretation of geologic structures and stratigraphic features. Although interpretation of these surfaces has been automated to some extent by others, significant manual effort is still required for extracting each type of these geologic surfaces. I propose methods to automatically extract all the fault, unconformity, and horizon surfaces from a 3D seismic image. To a large degree, these methods just involve image processing or array processing which is achieved by efficiently solving partial differential equations.

For fault interpretation, I propose a linked data structure, which is simpler than triangle or quad meshes, to represent a fault surface. In this simple data structure, each sample of a fault corresponds to exactly one image sample. Using this linked data structure, I extract complete and intersecting fault surfaces without holes from 3D seismic images. I use the same structure in subsequent processing to estimate fault slip vectors. I further propose two methods, using precomputed fault surfaces and slips, to undo faulting in seismic images by simultaneously moving fault blocks and faults themselves.

For unconformity interpretation, I first propose a new method to compute a unconformity likelihood image that highlights both the termination areas and the corresponding parallel unconformities and correlative conformities. I then extract unconformity surfaces from the likelihood image and use these surfaces as constraints to more accurately estimate seismic normal vectors that are discontinuous near the unconformities. Finally, I use the estimated normal vectors and use the unconformities as constraints to compute a flattened image, in which seismic reflectors are all flat and vertical gaps correspond to the unconformities. Horizon extraction is straightforward after computing a map of image flattening; we can first extract horizontal slices in the flattened space and then map these slices back to the original space to obtain the curved seismic horizon surfaces.

The fault and unconformity processing methods above facilitate automatic flattening and horizon extraction by providing an unfaulted image with continuous reflectors across faults and unconformities as constraints for an automatic flattening method. However, human interaction is still desirable for flattening and horizon extraction because of limitations in seismic imaging and computing systems, but the interaction can be enhanced. Instead of picking or tracking horizons one at a time, I propose a method to compute a volume of horizons that honor interpreted constraints, specified as sets of control points in a seismic image. I incorporate the control points with simple constraint preconditioners in the conjugate gradient method used to compute horizons.

Seismic images and the geologic information they provide contribute significantly to our understanding of the earth's subsurface. In this thesis, I focus on methods relevant for constructing and interpreting seismic images, including methods for velocity estimation, seismic imaging, and interpretation, which together address key aspects of reflection seismic data processing. Specifically, I propose improved methods for semblance-based normal-moveout velocity analysis, for seismic imaging by least-squares migration, and for the automatic extraction of geologic horizons.

To compute a seismic image, an estimate of the subsurface velocity is needed. One common method for constructing an initial velocity model is semblance-based normal-moveout (NMO) velocity analysis, in which semblance spectra are analyzed to identify peaks in semblance corresponding to effective NMO velocities. The accuracy of NMO velocities obtained from semblance spectra depends on the sensitivity of semblance to changes in velocity. By introducing a weighting function in the semblance calculation, I emphasize terms that are more sensitive to velocity changes, which, as a result, increases the resolution of semblance spectra and allows for more accurate NMO velocity estimates.

Following velocity analysis, a seismic image of the subsurface is computed by migrating the recorded data. However, while velocity analysis is an important step in processing reflection seismic data, in practice we expect errors in the velocity models we compute, and these errors can degrade a seismic image. Instead of minimizing the difference between predicted and observed seismic data as is done for conventional migration, I propose to minimize the difference between predicted and time-shifted observed data, where the time shifts are the traveltime differences between predicted and observed data. With this misfit function, an image computed for an erroneous velocity model contains features similar to those obtained using a more accurate velocity.

Once a seismic image is computed, a common task in interpreting the image is the identification of geologic horizons. As an alternative to manual picking or autotracking, I propose methods to automatically and simultaneously extract all horizons within an image. To extract geologic horizons, a seismic image is unfaulted and unfolded to restore horizons to an undeformed, horizontal state from which they can be easily identified and extracted.

Detection of Morrow A sandstones is a major problem in the exploration of new fields and the characterization of existing fields because they are very thin and laterally discontinuous. The present research shows the advantages of S-wave data in detecting and characterizing the Morrow A sandstone. Full-waveform modeling is done to understand the sandstone signature in P-, PS- and S-wave gathers. The sandstone shows a distinct high-amplitude event in pure S-wave reflections as compared to the weaker P- and PS-wave events. Modeling also helps in understanding the effect of changing sandstone thickness, interbed multiples (generated by shallow high-velocity anhydrite layers) and sidelobe interference effect (due to Morrow shale) at the Morrow A level.

Multicomponent data need proper care while processing, especially the S-wave data which are aected by the near-surface complexity. Cross-spread geometry and 3D FK filtering are effective in removing the low-velocity noise trends. The S-wave data obtained after stripping the S-wave splitting in the overburden show improvement for imaging and reservoir property determination. Individual P- and S-wave attributes as well as their combinations have been analyzed to predict the A sandstone thickness. A multi-attribute map and collocated cokriging procedure is used to derive the seismic-guided isopach of the A sandstone.

Postle Field is undergoing CO₂ flooding and it is important to understand the characteristics of the reservoir for successful flood management. Density can play an important role in finding and monitoring high-quality reservoirs, and to predict reservoir porosity. prestack P- and S-wave AVO inversion and joint P- and S-wave inversion provide density estimates along with the P- and S-impedance for better characterization of the Morrow A sandstone. The research provides a detailed multicomponent processing, inversion and interpretation work flow for reservoir characterization, which can be used for exploration in other parts of the world as well.

Today’s petroleum industry demand an ever increasing amount of compu- tational resources. Seismic processing applications in use by these types of companies have generally been using large clusters of compute nodes, whose only computing resource has been the CPU. However, using Graphics Pro- cessing Units (GPU) for general purpose programming is these days becoming increasingly more popular in the high performance computing area. In 2007, NVIDIA corporation launched their framework for developing GPU utilizing computational algorithms, known as the Compute Unied Device Architec- ture (CUDA), a wide variety of research areas have adopted this framework for their algorithms. This thesis looks at the applicability of GPU techniques and CUDA for off-loading some of the computational workload in a seismic shot modeling application provided by StatoilHydro to modern GPUs. This work builds on our recent project that looked at providing check- point restart for this MPI enabled shot modeling application. In this thesis, we demonstrate that the inherent data parallelism in the core finite-difference computations also makes our application well suited for GPU acceleration. By using CUDA, we show that we could do an efficient port our application, and through further refinements achieve significant performance increases. Benchmarks done on two different systems in the NTNU IDI (Depart- ment of Computer and Information Science) HPC-lab, are included. One system is a Intel Core2 Quad Q9550 @2.83GHz with 4GB of RAM and an NVIDIA GeForce GTX280 and NVIDIA Tesla C1060 GPU. Our sec- ond testbed was an Intel Core I7 Extreme (965 @3.20GHz) with 12GB of RAM hosting an NVIDIA Tesla S1070 (4X NVIDIA Tesla C1060). On this hardware, speedups up to a factor of 8-14.79 compared to the original se- quential code are achieved, confirming the potential of GPU computing in applications similar to the one used in this thesis.

This thesis develops and tests 3D Fast Marching Method (FMM) algorithm and apply these to seismic simulations. The FMM is a general method for monotonically advancing fronts, originally developed by Sethian. It calculates the first arrival time for an advancing front or wave. FMM methods are used for a variety of applications including, fatigue cracks in materials, lymph node segmentation in CT images, computing skeletons and centerlines in 3D objects and for finding salt formations in seismic data. Finding salt formations in seismic data, is important for the oil industry. Oil often flows towards gaps in the soil below a salt formation. It is therefore, important to map the edges of the salt formation, for this the FMM can be used. This FMM creates a first arrival time map, which makes it easier to see the edges of the salt formation. Herrmann developed a 3D parallel algorithm of the FMM testing waves of constant velocity. We implemented and tested his algorithm, but since seismic data typically causes a large variation of the velocities, optimizations were needed to make this algorithm scale. By optimising the border exchange and eliminating much of the roll backs, we delevoped and implemented a much improved 3D FMM which achieved close to theoretical performance, for up to at least 256 nodes on the current supercomputer at NTNU. Other methods like, different domain decompositions for better load balancing and running more FMM picks simultaneous, will also be discussed.