Abstract The latest information about the SEG Advanced Modeling (SEAM) Corporation and its projects is available at http://www.seg.org/SEAM/ . The SEG is a not-for-profit professional society with more than 30,000 members and the mission “to advance the science and technology of applied geophysics.” One of SEG’s several volunteer committees is the Research Committee, with membership composed of interested individuals from corporate research centers, technology start-ups, national laboratories, and universities. During the mid-1990s, this committee organized a volunteer effort to create two 3D earth models and simulated seismic data sets — the SEG/EAGE Salt and Overthrust Models ( Aminzadeh et al. 1997 ). The computing facilities were provided for free, and the project took substantially longer to complete than anticipated. The result was a data set that spurred much successful research in seismic imaging and that still today occasionally shows up in academic research papers. SEG’s only regret was that this was a “one-off’ activity. Computing technology has advanced. Seismic acquisition and imaging technology has advanced. There are many big challenges (not only deepwater subsalt) that would benefit from collaborative earth modeling and geophysical simulations. In 2005, the SEG Research Committee recommended the concept of SEAM to SEG’s Executive Committee. The intent was to establish a repeatable approach in which one modeling/simulation effort would be followed by another. The focus of each project would be driven by industry need. A research consortia would be formed in which modeling challenges would be broadly shared. Participants would pay a set amount during specific finite-duration projects.

Abstract The SEAM Phase I Subsalt Earth Model is a 3D representation of a deepwater Gulf of Mexico salt domain, complete with fine-scale stratigraphy and fluid-filled reservoirs. The model measures 35 km east-west by 40 km north-south by 15-km depth. A pictorial overview of the model is shown in file PhIEarthModelReport_Day_ AM2009.ppt, which was presented at the SEAM workshop in October 2009. SEAM’s mission is to produce earth models and simulations on those models to advance the state of the art in multiple disciplines over a multiyear period. Phase I pushed the limits of model building. The model is not only large, but it also is highly complex with realistic faults, overturned beds, overhanging salt, density diffractors to generate diffracted multiples, turbidite fans, and braided stream channels reservoirs. Some compromises had to be made due to limitations of model-building tools available. (See “Model Construction Methodology” for details.) There was some limited progress in this area during Phase I, but much more could be done. We tried to make the best model we could for the SEAM project, given the limitations in hardware, software, and time available. The following sections point out the areas where compromises were made and where there are open areas for development. There is a joke about cabinetmakers that goes something like this: The difference between a professional and an amateur is that the professional does not point out his mistakes in a project. If our goal as an industry is to learn and improve, then we need

Abstract At the start of the SEAM project, a general consensus was that a finite-difference (FD) method or related method would be the most efficient to create a synthetic seismic data set from the SEAM geologic model. This was not a requirement, and SEAM was open to alternative simulation approaches. As with all methods, FD has its specific limitations and costs associated with a particular simulation. The SEAM organization tried to anticipate many of these when choosing the geologic model size and survey specification so that the simulation could be performed in a reasonable amount of time and cost. The main limitation is that the model is sampled at discrete points along a regular grid. Thus, it is not possible to capture jumps in material properties at arbitrary points. This leads to time shifts in output data when comparing results of simulations that use finer or coarser spatial sampling. These time shifts are a major cause of differences as discussed in “Benchmark Comparisons” and related references. The discretization of a continuous model also has implications for the solution. In short, the spatial wavelength has to be adequately sampled (cf. the Nyquist criterion). Inadequate spatial sampling leads to what is called numerical dispersion, i.e., different frequencies propagate at different velocities. The numerical dispersion is different for FD schemes of different accuracy order. A higher-order scheme in general is less affected by numerical dispersion, which allows the use of a larger grid size for a given amount of dispersion; however, higher-order schemes require.

Abstract The Acquisition Design Committee was responsible for integration between all technical committees and proposing the seismic acquisition plan. They began work in September 2007 by examining the trade-offs between the type of simulations, number of shots, number of receivers, size of computation aperture, quality of numerical calculations (fidelity), and frequency content. They discussed many possible simulation and acquisition scenarios: variable density acoustic, elastic, multicomponents, ocean bottom receivers, 3D (vertical seismic profiling [VSP]), and 4D. Early estimates of acquired data volume ranged from 100 to 2000 terabytes (TB) of data. As with field acquisition, there are economic and practical (operational) limitations, as well as the geophysical considerations as discussed in “Design Considerations.” Ray-trace modeling, 2D-acoustic modeling, and limited 3D modeling were carried out to establish some of the geophysical parameters and to understand the time and cost implications for the project, as summarized in “Acquisition Design Modeling.” The preliminary results in the PPT files and published articles referenced in this chapter provided a baseline for the initial test runs. The acquisition and simulation specifications changed over time as more details were developed by the technical committees, as more knowledge was gained about the capabilities and limitations of numerical modeling and data storage, and through discussions with the simulation vendor once they were under contract; finally, they were approved by the SEAM Management Committee. The final specifications, listed in Appendix B of Chapter 5 , are somewhat different than those in the various PowerPoint presentations referenced in this chapter and the initial vendor contracts.

Abstract Numerical simulations were done by Tierra Geophysical, now part of Landmark Graphics Corporation. SEAM did not specify a specific approach to perform numerical simulations. Instead, SEAM developed quality requirements for the simulations and allowed vendors to meet those requirements using any numerical approach, whose specifics may remain unknown to SEAM. Hence, this chapter does not attempt to describe the specifics of the simulation computation. Rather, it focuses on run parameters so that the data can be reproduced; artifacts from postsimulation filtering to apply the SEAM far-field wavelet; global amplitude scaling process; use of ghost source and receivers so that absorbing upper surface data have the same wavelet as the free surface data; the degree of variation in the data caused by factors such as a change to the simulation code, use of multiple compute clusters, and elimination of numerical problems that occurred on a very small percentage of the shots; and problems that occurred, and how they were identified and corrected. There are significant connections between this and Chapter 3 for topics of source wavelet, trace scaling, absorbing upper surface, and verification. Chapter 3 also covers vendor qualification procedure, requalification using the production model, and verification of production simulations. The SEG-Y header definitions that were actually used are listed in Appendix A . The simulation specifications as actually implemented are listed in Appendix B . These evolved over time in response to knowledge gained and as more details were developed by the technical committees,

Abstract SEAM contracted with Nexus Geosciences Inc. (NGI), now part of WesternGeco, to conduct quality control (QC) evaluation of all data delivered by the simulation vendor. The steps in the QC process were developed by the Execution Committee and approved by the Management Committee (see Appendix A , “Description of QC Services”). The data flow diagram in Figure 4 of in Chapter 1 illustrates the relationships of all the vendors, SEAM committees, and individuals involved. The QC vendor received compressed data from the simulation vendor on USB disk. The data were decompressed, and the QC process was initiated. Once data were passed through the QC process, the USB disks were sent to the storage vendor. During the decompression, traces that comprise the classic data sets were extracted. These were sent to Stew Levin, who stored and compiled these data and compressed them for long-term storage. The quality-assurance protocol was designed to provide a thorough but timely assessment of data quality and model performance and to identify potential problems for correction. The procedure contains a number of steps to Test quality of the numerical simulations, including evaluation for excessive amplitudes that may indicate numerical problems. Ensure shots and traces are in their expected and correct locations. Prepare fold maps. Ensure no corruption of data during compression and loading of storage media that were used to transport data between simulation, QC, and data storage vendors. Extract zero-offset traces, depth-migrate them, and overlay on the model. Raw zero.

Abstract One of the painful lessons learned from the earlier SEG/EAGE 3D modeling work was that provision needs to be made quite early in the process for long-term storage and distribution so that the data that are generated don’t become inaccessible within a few years. For example, the old SEG/ EAGE 3D model shot records are sitting on D2 tapes in the basement of the Tulsa SEG headquarters. It is now cheaper for someone to regenerate those shot records from scratch on a moderate Linux cluster than it is to try to recover the data from those outmoded tapes. In line with the overall SEAM business model to put out for bid contracts for the various aspects and phases of the project, the Storage and Distribution Committee was formed to research commercial storage options, estimate budget impact, and help draft bid specifications. The first step was to assess the amount of data that would need to be stored. As the model and acquisition design were constantly in flux, this proved a challenge. At various points, proposed data acquisition schemes ranged from the low hundred terabytes to more than two petabytes. Clearly, as the volume of data would be a large component in the cost of storage and distribution, the committee assembled a list of companies that would have the capability and potential interest in storing and distributing SEAM data. These covered a range of business models: traditional E&P data storage and transcription, PC data backup and restore, and corporate data center.

Abstract “SEAM is a collaborative industrial research effort dedicated to largescale, geophysical numerical simulation projects. The projects are designed to provide the geosciences exploration community with earth models and simulated data that represent significant geophysical challenges of high business value to the petroleum resource industry. The Phase I project produced a deepwater subsalt earth model designed to capture as much physics and realism as possible in a 3D model that was relevant to oil and gas exploration. The 3D model covers a 40 õ 35 õ 15 km area and includes a complex salt intrusive in a folded Tertiary basin. The primary deliverable was the seismic data set of variable density acoustic simulations consisting of 200 TB of uncompressed traces for over 60,000 shots. Also delivered to the participants were several smaller compressed subsets of these data (“classic” data sets) intended for easier handling, simpler distribution to third parties, and easier comparison of imaging tests results. This report covers how the prime objectives of Phase I were met. Details are outlined in chapters on Model Development, Numerical Design and Vendor Qualification, Acquisition Design, Production Simulations, Quality Control, and Data Storage and Distribution.”

Abstract Deep-water sedimentation is currently a major focus of both academic research and industrial interest. Recent studies have emphasized the fundamental influence of seafloor topography on the growth and morphology of submarine ‘fans’: in many turbidite systems and turbidite hydrocarbon reservoirs, depositional system development has been moderately to strongly confined by pre-existing bounding slopes. This publication examines aspects of sediment dispersal and accumulation in deepwater systems where basin-floor topography has profoundly affected deposition, and the associated controls on hydrocarbon reservoir architecture and heterogeneity. The papers herein offer a global perspective which is wide-ranging in terms of both approach and location, including contrasting case studies of outcrop, subsurface, modern and experimental systems.

Abstract Four main controls (tectonics, climate, sedimentary characteristics and processes, and sea-level fluctuations) commonly interact with each other and do so at varying intensities. This results in a wide variety of basin types and shapes, timing of transport within the sequence stratigraphy framework, transport and depositional processes, grain size ranges, and distribution of sediment within a basin. Two major end members of turbidite systems can be recognized: coarse-grained/sand-rich and fine-grained/mud-rich. Coarse-grained fans typically belong to active margin settings. They prograde gradually into a basin and show a decrease in thickness and grain size in the downflow direction. The sediment source is near the coastline, and the turbidite basins are commonly small to medium in size. The fine-grained fans occur on passive and active margins, prograde rapidly into a basin, and deposit most of the input sand in the distal fan as oblong sheet sands. Tectonically confined basins normally have their sediment source nearby, and therefore will be filled with coarse-grained fans. Most of the open (unconfined) basins are medium to large in size, have their sediment source far from the coast, and therefore lose the coarser fractions during continental transport. Diapirically controlled basins are small- to medium-sized confined basins that have a fine-grained turbidite fill, but may not reveal the bypassing of the majority of the sand to the outer fan because of the abundance of sediment transport to the basin.

Abstract Two end-member classes of sediment distribution systems on topographically complex slopes are distinguished here: (a) cascades of silled sub-basins , and (b) connected tortuous corridors . In the first scenario a process of filling and spilling of successive silled sub-basins down a slope occurs. For each sub-basin a sill tends to hinder further downslope flow of at least the basal sandy portions of sediment gravity flows until deposition reduces the relief sufficiently to allow spill down-slope. Spill is associated with incision in the sill. In the connected tortuous corridors scenario, flows avoid bathymetric obstacles, but follow a (laterally confined) continuous tortuous path down the slope. Without complete three-dimensional imaging of slope architecture it can be possible to incorrectly infer from two-dimensional profiles a cascade of silled sub-basins model. Thus flow paths in adjacent apparent subbasins can be connected out of the plane of section. Convergent thinning and convergent baselap stratal patterns occur in both scenarios, but only in the silled sub-basin case do such patterns occur against closing frontal slopes. For a given complex slope morphology, dominant controls on fill patterns and reservoir architecture are (a) the history of sediment supply character, and (b) rates of structure growth relative to rates of smoothing of topography by erosional and depositional processes. Two particularly important aspects of sediment supply are (i) flow volumes relative to scales of receiving spaces, and (ii) flow properties (in particular, transported grain size distribution, flow thickness and flow concentration), these controlling depositional gradients and the equilibrium profiles to which slopes tend to grade.

Abstract Two sets of scaled laboratory experiments were performed to examine the effect of flow volume, flow density and grain-size distribution on the transport efficiency of turbidity currents. The experiments employed two sediment analogues (ballotini and silica flour) intended to model medium- to coarse-grained sand and mud respectively. In the first set of experiments each parameter was varied to examine its effect upon deposit geometry. Increases in the initial flow density, volume and proportion of fines had the effect of increasing the amount of sediment that was transferred to the floor of the experimental tank by the turbidity currents. Increase of each of these parameters has a characteristic effect on the three-dimensional geometry of the deposit: the deposits of large-volume flows are elongate, and those of fines-rich flows are broad. Increase of flow density increases the initial potential energy of the flow, thus increasing the runout distance; increase of the initial density beyond a sediment concentration of 13% by mass results, however, in a reverse of the geometrical trend of deposit elongation, possibly because of turbulence suppression at high densities. Increase of flow volume also increases the initial potential energy, and reduces the rate of velocity decrease due to gravitational spreading. Increase in the proportion of fines leads to maintenance of negative buoyancy, as the fine fraction remains suspended until the flow has virtually come to rest; it also decreases the settling velocity of the coarser fraction and thus delays its sedimentation. The second set of experiments was performed to investigate the influence of flow efficiency on the interaction of turbidity currents with topography. A single arcuate obstacle was placed in the path of the flows. In successive experiments flow efficiency was increased by progressively increasing the proportion of fines (silica flour). Both the proportion of sediment reaching the obstructing topography and the proportion of it able to surmount the topography increased as flow efficiency increased. Thus flow efficiency may determine whether or not an enclosed basin hosts deposits whose geometry has been affected by the confinement, and may also determine the relative effectiveness of the topography in confining inbound turbidity currents, and thus trapping their sediment load.

Abstract The modern sandy Golo turbidite system (500 km 2 ) is located in a confined basin on the eastern margin of Corsica. The Golo turbidite system is fed by a single river, which supplies coarse sand derived from active weathering of the neighbouring mountains. The late Quaternary deposits have been imaged using a closely spaced grid of 1000 km of sparker seismic-reflection profiles (line spacing close to 1.6 km, vertical resolution of 2m). The turbidite system is composed of four non-coalescent fans that were at times active simultaneously and of two small deposits onto the slope. The resulting sedimentation pattern is characterized by stacked turbidite deposits. At a regional scale, there is a continuum of fan morphologies and geometries from south to north. The use of both seismic and sedimentary facies, together with mapped seismic geometry of sedimentary bodies, allowed definition of four architectural elements: (1) submarine valley (canyon and gully), (2) sandy channel, (3) muddy levee, and (4) sandy lobe. Some of these architectural elements can be recognized at a scale that is comparable to outcrop examples. Features such as progressive lateral migration and avulsion, or complex longitudinal evolution (progradation and retrogradation), can also be accurately described. Despite the active tectonics along the studied margin, the main variations in sedimentation appear to be controlled by eustatic changes, pre-existing seafloor topography, and sediment source characteristics. The general pattern of sedimentation is controlled by the influence of a confining slope, leading to the predominance of aggradation and to specific morphology and architecture of sedimentary bodies.

Abstract Terraces have been frequently observed and described along turbidite valleys. Many interpretations have been aimed at determining the origin of these structures, including a tectonic origin, succession of infilling and incision processes, channel-wall slumps, or inner levee aggradation. The Zaire submarine valley presents a complex structure with multiple terraces bordering a deep incised meandering thalweg. The detailed analysis of the morphology, the seismic structure and the recent sedimentation (in cores) along the Zaire upper-fan valley show that terraces are inner levees confined within the incised valley. Many terraces correspond to the infilling of abandoned meanders, and aggrade by deposition of turbidite sequences due to current overflows. The major process affecting the initiation and the development of terraces inside the valley is the vertical incision of the thalweg, simultaneously with meander migration.

Abstract Three major controlling factors affect turbidite deposition in foredeep basins: tectonics in the source area, tectonics in the belt-basin system, and variations of sea-level (local or global). These factors are expected to have different effects on the volume, grain size, provenance and distribution of clastic sediments during the evolution of the basin. The interplay of these factors is investigated for the latest Oligocene-Middle Miocene Northern Appennines Foredeep turbidite wedges by means of turbidite-system-based lithostratigraphy and field mapping, integrated with nannoplankton biostratigraphy and sedimentary petrography. Almost all recognized turbidite systems, unless tectonically truncated, show an overall stacking pattern formed by a lower sand-rich, thickly bedded stage ( depocentre stage ) passing upward into mud-rich, thinly bedded stages, eventually grading up to mostly mudstone units ( abandonment stage ). This rhythmically repeated pattern is interpreted as the result of the abrupt switching on and off of coarse-grained input, coupled with an alternating increase/decrease of depositional rate recorded in all detected systems. Biostratigraphy makes it possible to distinguish the switching-off of turbidite systems due to depocentre migration (a new system is switched on basinward) from that due to a regional decrease of clastic input. Sandstone petrography records the compositional variation related to tectonically induced source reorganization. In the latest Oligocene-Middle Miocene NAF foredeep wedges, this integrated dataset allows us to recognize: two different phases of source tectonics in the latest Oligocene and the middle Burdigalian; two major episodes of basin tectonics and related depocentre shift in the latest Oligocene and the Langhian, plus a minor middle Aquitanian phase; and three intervals of reduced regional turbidite deposition during the Late Aquitanian, Middle Burdigalian and Early Serravallian, possibly linked to sea-level rises.

Abstract Seabed faulting can have a significant impact on the routeing and behaviour of gravity currents depositing sand on deepwater basin floors. The Neogene El Cautivo Fault in the Tabernas-Sorbas Basin, SE Spain, is a rare example of a fault that demonstrably propagated through to the seabed during turbidite deposition, allowing the interplay between deepwater sedimentation and tectonics to be explored. The fault is associated with a wide (up to 350 m) gouge zone that varies significantly in thickness along its length, reflecting upward expansion towards the original seabed and progressive burial as fault activity ceased. Kinematic and stratigraphic evidence indicate that the fault was a dextral oblique strike-slip fault that accommodated an area of deeper ponded bathymetry (a ‘mini-basin’) and accelerated subsidence on its southern flank. Active faulting controlled the routeing of turbidity currents (revealed by changing provenance across the structure), rates of seabed deformation (resulting in differential subsidence and ‘growth’ of the stratigraphy), and the behaviour of the ponded currents (producing distinctive bipartite beds when deposition was in localized ponded depressions). The seabed expression of the fault varied from a forced fold, which warped the surface causing local wedging and onlap in the vicinity of the structure, and an unstable scarp that locally collapsed. The fault gouge fabrics and vein arrays are consistent with faulting of soft, water-rich sediment close to the seabed.

Abstract Basin-floor fans and slope fans present major differences in their internal architecture related to changes in: (1) margin morphology, (2) relative sea-level change, and (3) sediment supply. These variations are illustrated in the outcrops of the Pab Sandstone in Pakistan. The Pab Sandstone third-order sequence was deposited on the Indo-Pakistani margin during the Upper Maastrichtian. Uplift of the margin induced erosion on the shelf, incision of submarine canyons on the slope and the development of a sand-rich, high-efficiency basin-floor fan extending over hundreds of kilometres on the basin floor. During transgression, sediment accumulated in backstepping shoreface deposits on the shelf, and a minor mud-rich slope fan was deposited in the basin. Finally, a sand-rich braided delta prograded across the shelf, feeding a sand-rich slope fan where it reached the shelf margin. This slope fan was of more limited lateral extent. The Lower Pab basin-floor fan shows the effects of flow funnelling and confinement due to a canyon incised into the slope. It consists mainly of channel complexes deposited by superconcentrated density flows to low-density turbidity currents. In contrast, the Upper Pab slope fan shows little confinement and low transport efficiency. It consists of tabular lobes, aggrading mid-fan channels and conglomeratic channels in the upper fan. The low transport efficiency of the gravity flows probably explains the low degree of organization of the slope fan.