No abstract

Three digital earth models were designed and constructed during SEAM Phase II to study exploration challenges at the scale of modern land seismic surveys. Although built as generic models, each was based on one or more related geologic type areas. The Barrett model represents the seismic anisotropy of complex laminated and fractured shale reservoirs, based on the Woodford and Eagle Ford formations and set below a stratigraphic overburden and near surface of a North American midcontinent basin. The Arid model features the extreme property contrasts of desert terrains in a 500 m thick near surface that juxtaposes hard carbonate bedrock and soft sediments filling karsts, typical of the Saudi Arabian Peninsula. The Foothills model contains sharp surface topography and alluvial fan-like sediments above complex fold-and-thrust structures based on the compressive tectonics of the Llanos Foothills of South America. All three models were built in workflows that combined automated steps with a large measure of manual model building, which represents the current state of the art in geologic modeling for large-scale geophysical simulations. The Barrett and Arid models each contain about 1.5 billion grid cells representing regions 10 × 10 × 3.75 km in physical size. The Foothills model has about 2 billion cells representing a region about 14.5 × 12.5 × 11 km. Full elastic-wave simulations with these models were run for a combined total of about 170,000 shots, usually with millions of recorded channels per shot, generating several petabytes of seismic data in standard and novel shot-receiver geometries. Selected shots from these simulations show that large, detailed earth models can reproduce features of land seismic surveys that continue to challenge the best modern seismic data processing and imaging techniques.

A B S T R A C TIn seismic interpretation and seismic data analysis, it is of critical importance to effectively identify certain geologic formations from very large seismic data sets. In particular, the problem of salt characterization from seismic data can lead to important savings in time during the interpretation process if solved efficiently and in an automatic manner. In this work, we present a novel numerical approach that is able to automatically segmenting or identifying salt structures from a post-stack seismic data set with a minimum intervention from the interpreter. The proposed methodology is based on the recent theory of sparse representation and consists in three major steps: first, a supervised learning assisted by the user which is performed only once, second a segmentation process via unconstrained 1 optimization, and finally a post-processing step based on signal separation. Furthermore, since the second step only depends upon local information at each time, the whole process greatly benefits from parallel computing platforms. We conduct numerical experiments in a synthetic 3D seismic data set demonstrating the viability of our method. More specifically, we found that the proposed approach matches up to 98.53% with respect to the corresponding 3D velocity model available in advance. Finally, in appendixes A and B, we present a convergence analysis providing theoretical guarantees for the proposed method.

Local tomography is interactive, ray-based, residual-interval-parameter analysis for updating background anisotropic velocity parameters. The method operates directly on image gathers generated by anisotropic curved-ray Kirchhoff time migration. A locally 1D, spatially varying, vertical transversely isotropic model is assumed. The background anisotropy parameters are the instantaneous ͑interval͒ vertical compression velocity V P and the two Thomsen anisotropy parameters, ␦ and . The interval velocity ␦ is updated from short-offset reflection events, and is updated from available long-offset data. The medium parameters are updated from the top down both vertically and by layers, one parameter at a time. The picked residual-anisotropy parameters correspond to the residual-moveout ͑RMO͒ curves that best fit the migrated reflection events. The method is based on splitting the contribution to the computed RMO at a given point into two parts: from overburden residual parameters and from the actual picked residual parameter. This approach allows for direct residual-interval-parameter analysis to be applied in the same way we perform the commonly used residual-effectiveparameter analysis. The local tomography enables a controlled interactive estimation of the long-wavelength anisotropy parameters. The reliable anisotropy parameters estimated by the local approach are used as a background ͑guid-ing͒ model for a global tomography. This makes it possible to successfully apply a global constrained inversion that is performed simultaneously for all parameters of all output intervals using detailed RMO information.