S41B-01 INVITED
Illuminating Asset Value through New Seismic Technology
The ability to reduce risk and uncertainty across the full life cycle of an asset is directly correlated to creating an accurate subsurface image that enhances our understanding of the geology. This presentation focuses on this objective in areas of complex overburden in deepwater. Marine 3D seismic surveys have been acquired in essentially the same way for the past decade. This configuration of towed streamer acquisition, where the boat acquires data in one azimuth has been very effective in imaging areas in fairly benign geologic settings. As the industry has moved into more complicated geologic settings these surveys no longer meet the imaging objectives for risk reduction in exploration through production. In shallow water, we have seen increasing use of ocean bottom cables to meet this challenge. For deepwater, new breakthroughs in technology were required. This will be highlighted through examples of imaging below large salt bodies in the deep water Gulf of Mexico. GoM - Mad Dog: The Mad Dog field is located approximately 140 miles south of the Louisiana coastline in the southern Green Canyon area in water depths between 4100 feet to 6000 feet. The complex salt canopy overlying a large portion of the field results in generally poor seismic data quality. Advanced processing techniques improved the image, but gaps still remained even after several years of effort. We concluded that wide azimuth acquisition was required to illuminate the field in a new way. Results from the Wide Azimuth Towed Streamer (WATS) survey deployed at Mad Dog demonstrated the anticipated improvement in the subsalt image. GoM - Atlantis Field: An alternative approach to wide azimuth acquisition, ocean bottom seismic (OBS) node technology, was developed and tested. In 2001 deepwater practical experience was limited to a few nodes owned by academic institutions and there were no commercial solutions either available or in development. BP embarked on a program of sea trials designed to both evaluate technologies and subsequently encourage vendor activity to develop and deploy a commercial system. The 3D seismic method exploded into general usage in the 1990's. Our industry delivered 3D cheaper and faster, improving quality through improved acquisition specifications and new processing technology. The need to mitigate business risks in highly material subsalt plays led BP to explore the technical limits of the seismic method, testing novel acquisition techniques to improve illumination and signal to noise ratio. These were successful and are applicable to analogue seismic quality problems globally providing breakthroughs in illuminating previously hidden geology and hydrocarbon reservoirs. A focused business challenge, smart risk taking, investment in people and computing capability, partnerships, and rapid implementation are key themes that will be touched on through out the talk.
S41B-02 INVITED
Next Generation Seismic Imaging; High Fidelity Algorithms and High-End Computing
The rich oil reserves of the Gulf of Mexico are buried in deep and ultra-deep waters up to 30,000 feet from the
surface. Minerals Management Service (MMS), the federal agency in the U.S. Department of the Interior that
manages the nation's oil, natural gas and other mineral resources on the outer continental shelf in federal
offshore waters, estimates that the Gulf of Mexico holds 37 billion barrels of "undiscovered, conventionally
recoverable" oil, which, at 50/barrel, would be worth approximately 1.85 trillion. These reserves are very difficult
to find and reach due to the extreme depths. Technological advances in seismic imaging represent an
opportunity to overcome this obstacle by providing more accurate models of the subsurface.
Among these technological advances, Reverse Time Migration (RTM) yields the best possible images. RTM is
based on the solution of the two-way acoustic wave-equation. This technique relies on the velocity model to
image turning waves. These turning waves are particularly important to unravel subsalt reservoirs and delineate
salt-flanks, a natural trap for oil and gas. Because it relies on an accurate velocity model, RTM opens new frontier
in designing better velocity estimation algorithms.
RTM has been widely recognized as the next chapter in seismic exploration, as it can overcome the limitations of
current migration methods in imaging complex geologic structures that exist in the Gulf of Mexico. The chief
impediment to the large-scale, routine deployment of RTM has been a lack of sufficient computer power. RTM
needs thirty times the computing power used in exploration today to be commercially viable and widely usable.
Therefore, advancing seismic imaging to the next level of precision poses a multi-disciplinary challenge.
To overcome these challenges, the Kaleidoscope project, a partnership between Repsol YPF, Barcelona
Supercomputing Center, 3DGeo Inc., and IBM brings together the necessary components of modeling,
algorithms and the uniquely powerful computing power of the MareNostrum supercomputer in Barcelona to
realize the promise of RTM, incorporate it into daily processing flows, and to help solve exploration problems in a
highly cost-effective way. Uniquely, the Kaleidoscope Project is simultaneously integrating software (algorithms)
and hardware (Cell BE), steps that are traditionally taken sequentially. This unique integration of software and
hardware will accelerate seismic imaging by several orders of magnitude compared to conventional solutions
running on standard Linux Clusters.
http:www.3dgeo.com/news/news_20061120.html
S41B-03
Seismic imaging using curvelets
We show that with curvelets the leading-order approximation (in angular frequency, horizontal wavenumber, and migrated location) to Kirchhoff depth-migration becomes a simple transformation of the coordinates of the curvelets in the data, combined with amplitude scaling. This transformation is calculated using map migration, which uses the local slopes provided by the curvelet decomposition of the data. We verify the accuracy of the method using numerical examples for homogeneous media. These examples indicate that using the leading-order approximation only provides a good approximation to common-offset migration when rays do not diverge beyond the spatial support of a curvelet, but looses accuracy when they diverge beyond this support. This shows the need for correction beyond leading order, even for homogeneous media when the data contains diffracted waves. We proceed to show how correction beyond the leading order can be accounted.
S41B-04
Enhanced Seismic Imaging of Turbidite Deposits in Chicontepec Basin, Mexico
We test, as postprocessing tools, a combination of migration deconvolution and geometric attributes to attack the complex problems of reflector resolution and detection in migrated seismic volumes. Migration deconvolution has been empirically shown to be an effective approach for enhancing the illumination of migrated images, which are blurred versions of the subsurface reflectivity distribution, by decreasing imaging artifacts, improving spatial resolution, and alleviating acquisition footprint problems. We utilize migration deconvolution as a means to improve the quality and resolution of 3D prestack time migrated results from Chicontepec basin, Mexico, a very relevant portion of the producing onshore sector of Pemex, the Mexican petroleum company. Seismic data covers the Agua Fria, Coapechaca, and Tajin fields. It exhibits acquisition footprint problems, migration artifacts and a severe lack of resolution in the target area, where turbidite deposits need to be characterized between major erosional surfaces. Vertical resolution is about 35 m and the main hydrocarbon plays are turbidite beds no more than 60 m thick. We also employ geometric attributes (e.g., coherent energy and curvature), computed after migration deconvolution, to detect and map out depositional features, and help design development wells in the area. Results of this workflow show imaging enhancement and allow us to identify meandering channels and individual sand bodies, previously undistinguishable in the original seismic migrated images.
S41B-06
Efficient calculation of the Green function in time-lapse seismic studies using boundary- integral representations
There are many applications where one is interested in calculating the response of a wavefield to a local perturbation in a medium. For example, in time-lapse seismic monitoring extensive modelling of the seismic response to local perturbations in the medium is often needed to determine the feasibility of detecting changes due to CO2 sequestration or production of hydrocarbons. We show the connection between a convolution-type acoustic reciprocity theorem of wavefields in two media with different medium parameters and the Lippmann-Schwinger integral equation. This connection leads to a boundary-integral representation of the full Green function in a perturbed medium between any two points outside of the perturbation and two points, one of which is inside and one of which is outside the perturbed area. This integral contains the impulse reponses due to both monopole and dipole sources located on the bounding surface of the perturbation only. If the bounding surface can be covered with sufficiently fewer sources than the number of sources in the acquisition geometry, the boundary-integral representations we present allow efficient calculation of the full Green function due to the local perturbation. There is no constraint on the magnitude of the perturbation. We verify numerically the accuracy of these representations for the simple case of wave propagation in one dimension, and discuss its potential use in time- lapse seismic studies such as monitoring of CO2 sequestration, hydrocarbon reservoirs, or nuclear- waste storage sites.
S41B-07 INVITED
Model Order Reduction for Wave Propagation
Large-scale 3-dimensional elastic wave propagation simulations are still very demanding, even with modern parallel computing, as witnessed by recent simulations of an earthquake in the Los Angeles Basin (Terashake). Thus, our ability to run many such simulations with sources in different positions to predict the effect of possible earthquakes or for tomographic imaging is very limited. Model order reduction is a technique to extract natural modes from a limited number of simulations of a dynamical system and use them to obtain a much smaller dimensional system that still retains enough accuracy. The technique has been employed successfully in many different applications but little has been done for wave propagation problems. We present some preliminary results by using the Proper Orthogonal Decomposition (SVD) on a set of snapshots of some high fidelity calculations.
S41B-08
Site Response in the Northern San Francisco Bay, California
Santa Rosa, California sustained unexpectedly high damage from the 1906 San Francisco earthquake (M7.8) and the 1969 Santa Rosa Earthquake sequence (M5.6 and M5.7). At the nearby city of Napa, ground motion was also unexpectedly high during the 2000 Yountville earthquake (M5.2). This history of unexpectedly strong ground motion, in combination with high metropolitan populations in Santa Rosa (over 450,000) and Napa (over 100,000), and their close proximities to Holocene active faults such as the Rodgers-Healdsburg fault system, West Napa Fault, and Green Valley fault, make estimation of site response in these regions particularly important. We estimate site response at 24 sites in this region by inverting seismograms of local events (radius < 100km) for source, site, and propagation characteristics. Regional seismicity was recorded from January, 2004, to September, 2006. We inverted spectra of 33 earthquakes for P-waves and 25 earthquakes for S-waves, covering a range of hypocentral distances and azimuths. Seismic cone penetration testing (SCPT) was done at three stations in Santa Rosa and three stations in Napa to determine near surface S-wave velocities and constrain absolute site amplification. High site response is estimated from the inversion in the city of Napa, in agreement with slower near-surface S-wave velocities from SCPT. Response is estimated to be lower at adjacent sites outside of the valley. Site response is also high in the city of Santa Rosa, although lower than Napa. High response in Santa Rosa is in agreement with faster near-surface S-wave velocities from SCPT.