Search

Lateral changes in velocity about faults can give rise to fault shadow (FS) zones on time-migrated data volumes, which can result in structural interpretation artifacts in the fault trap reservoir. To address this issue, we have adopted a new reconstruction method of FS distortion structures based on a deep learning fully connected network (FCN). We use the 3D stratigraphic dip attributes to quantitatively delineate the extent of the FS zone. Then, we train a model to construct a nonlinear trend surface based on the structures of the stratigraphic reflectors that fall outside of the shadow zone. Finally, we use this nonlinear trend surface to compensate for the distorted structure within the FS zone. We calibrate our method using synthetic data and find that the method can accurately recover structural data within the FS distortion zone. We then test the effectiveness of our workflow by applying it to recover real FS distortion structures in the Pearl River Mouth Basin of the South China Sea. The results confirm that our method significantly reduces drilling depth errors in the FS zone. Compared to the traditional polynomial fitting method, the multilayer, multiparameter, and flexible nonlinear activation function of FCN is more capable of reconstructing nonlinear geologic structures in the FS zone. We find the FCN-based geologic reconstruction method to be efficient and effective for exploring potential structures in the FS zone and thereby in avoiding the risks of structural failure.

We have developed a calibrated, simple time-lapse seismic method for estimating saturation changes from the CO2CO2-storage project at Sleipner offshore Norway. This seismic method works well to map changes when CO2CO2 is migrating laterally away from the injection point. However, it is challenging to detect changes occurring below CO2CO2 layers that have already been charged by some CO2CO2. Not only is this partly caused by the seismic shadow effects, but also by the fact that the velocity sensitivity for CO2CO2 change in saturation from 0.3 to 1.0 is significantly less than saturation changes from zero to 0.3. To circumvent the seismic shadow zone problem, we combine the time-lapse seismic method with time-lapse gravity measurements. This is done by a simple forward modeling of gravity changes based on the seismically derived saturation changes, letting these saturation changes be scaled by an arbitrary constant and then by minimizing the least-squares error to obtain the best fit between the scaled saturation changes and the measured time-lapse gravity data. In this way, we are able to exploit the complementary properties of time-lapse seismic and gravity data. © 2017 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

Internal multiple reflections have different propagation paths than primary reflections and surface-related multiples; thus, they can complement illumination where other waves are unavailable. Consequently, migration of internal multiples may provide vital images in the shadow zone of primary reflections. We have developed an approach to image structures with internal multiples. With this method, internal multiples can be separated and used as data for reverse time migration. Instead of applying the imaging condition directly to the source and receiver wavefields, we decomposed the extrapolated source and receiver waves into up- and downgoing waves. Then we applied the imaging condition to upgoing source and receiver waves to form an up-up (UU) image and also to downgoing source and receiver waves to form a down-down (DD) image. The image of the true reflector and the artifacts behave differently in the DD and UU images. A similarity analysis is conducted to the UU and DD images to separate true images from artifacts. Numerical examples with simple velocity models are used to demonstrate how to construct UU and DD imaging conditions and eliminate migration artifacts. Finally, we test the 2D SMAART JV model with salt structures for migration using internal multiples.

Earthquake and exploration seismology share a common foundation in elastic wave propagation. Most of what we know about earthquakes is known from the elastic waves they emit, while seismic methods in exploration are all about using scattered elastic waves to image earth structure. Seismologists have been using elastic waves to probe the earth since the very earliest instruments. After all, it was early seismic measurements that led to the spherical shell model of earth structure (crust, mantle, and core) taught to grade school students.

We apply least-squares reverse-time migration (LSRTM) to a three-dimensional ocean-bottom node dataset. When applying to field datasets, a simple data-fitting objective function in LSRTM is insufficient. The Deimos field has complex salt structure that obscures part of the image deep down. We use a target-oriented data-reweighting scheme to emphasize deeper parts of the image near the salt. To make the algorithm more robust, we utilize the subsurface-angle domain to filter some of the noise in the image space. With the above tools to improve the robustness of LSRTM, we found that LSRTM provide a better image than conventional RTM method with better relative amplitude balance, higher resolution, and improved illumination near the shadow zone. We also found that the image calculated using joing-LSRTM of primary and mirror signals is superior than conventional single-mode imaging.

In addition to vertical time shifts commonly observed in time-lapse seismic images, horizontal displacements are apparent as well. These apparent horizontal displacements may be small relatively to seismic wavelengths, perhaps only 5 m at depths of 5 km, but they consistently suggest an outward lateral expansion of images away from a compacting reservoir. It is well known that apparent vertical displacements are caused mostly by a decrease in seismic wave velocities above compacting reservoirs. Those same velocity changes contribute to horizontal displacements. This contribution can be computed from the velocity changes that, in turn, can be estimated from measured vertical displacements. Horizontal displacements computed in this way are similar to those measured, and this similarity suggests that horizontal as well as vertical displacements may be largely due to velocity changes.

Seismic imaging need not be synonymous with or rely on the presence of reflectors in the Earth. Much can be gleaned from the nonreflected wavefield. For example, direct waves map out smooth velocity variations in crosswell seismic tomography, wide-angle refracted waves play a crucial role in full waveform inversion (Hole et al., 2005), and surface waves provide unmatched sensitivity to near-surface shear-wave velocity structure. Guided waves exist in both cracks (Korneev, 2008) and boreholes, the latter referred to as tube waves. The famous Biot slow wave is itself a guided-wave phenomenon akin to a tube wave (Norris, 1987). Surface-wave dispersion has a long history in seismology and was the first seismic characteristic to be subjected to an automated inversion procedure (Dorman and Ewing, 1962).

Structural imaging of sedimentary reflectors is improved within the shadow zone underlying steeply dipping base salt structure by using energy which has reflected postcritically from the base of salt. This concept is tested with both ray traced and wavefield synthetics, and then applied to real subsalt data, with encouraging results. The possibilities and limitations of this approach are discussed and evaluated. Improving subsalt illumination consists of a series of marginal assaults on the problem. This talk presents one such assault

The interpretation of salt geometry and the determination of sediment velocities are critical aspects of velocity model building for 3D salt and subsalt imaging. Here we investigate several factors that limit the resolution of this process. Using a simple synthetic model of tabular salt we first investigate velocity resolution below salt, and the effect this has on determining depths of reflectors below salt. We show for this model, that depth resolution due to velocity error drops by roughly a factor of three as one moves from sediments outside salt to sediments below salt. Second, we investigate the effect of salt mispositioning on the migrated image, particularly below the lateral edge of the salt. We find that lateral salt mispositioning can enhance the shadow zone beneath the lateral edge of salt, and lead to increased migration artifacts.

The shallow near surface attenuates seismic energy more than is generally appreciated. Within a few feet of the surface, seismic wave velocities drop as low as 100 feet per second for P waves. In addition, the quality factor Q for seismic propagation drops to values as low as 0.3 in unconsolidated, uncompacted soils. As a result, the attenuation length (the distance in which amplitude decreases by l/e) can be as short as one foot for 100 Hz P waves. Attenuation of upcoming seismic reflections and of downgoing airborne noise by the top few feet of soil is so large that it can be advantageous to plant geophones in shallow holes. These “buried” geophones are more effective at detecting high-frequency seismic reflections than are conventional surface geophones.

The effect of depth and lateral velocity variations on the width of a seismic beam propagating through the earth is investigated. The model used is unusual and provides an interesting alternative insight to the conventional models appearing in the geophysical literature. Some basic calculations show that, at frequencies and depths of interest in oil exploration, a significant lateral shift in the center of the seismic beam occurs in addition to the expected broadening of the seismic beam.

During the past 8 years, the U.S. Geological Survey has conducted an extensive investigation on the use of vertical seismic profiles (VSP) in a variety of seismic exploration applications. Seismic sources used were surface air guns, vibrators, explosives, marine air guns, and downhole air guns. Source offsets have ranged from 100 to 7800 ft. Well depths have been from 1200 to over 10,000 ft. We have found three specific ways in which VSPs can be applied to seismic exploration. First, seismic events observed at the surface of the ground can be traced, level by level, to their point of origin within the earth. Thus, one can tie a surface profile to a well log with an extraordinarily high degree of confidence. Second, one can establish the detectability of a target horizon, such as a porous zone. One can determine (either before or after surface profiling) whether or not a given horizon or layered sequence returns a detectable reflection to the surface. The amplitude and character of the reflection can also be observed. Third, acoustic properties of a stratigraphic sequence can be measured and sometimes correlated to important exploration parameters. For example, sometimes a relationship between apparent attenuation and sand percentage can be established. The technique shows additional promise of aiding surface exploration indirectly through studies of the evolution of the seismic pulse, studies of ghosts and multiples, and studies of seismic trace inversion techniques. Nearly all current seismic data‐processing techniques are adaptable to the processing of VSP data, such as normal moveout (NMO) corrections, stacking, single‐and multiple‐channel filtering, deconvolution, and wavelet shaping.

A zone of sands embedded in shale acts as a filter, both in reflecting energy back to the surface and in transmitting energy to reflectors below them. For a single layer of sand, the reflection filter is periodic--reflecting no energy at some frequencies and more than either of the two individual interfaces at other frequencies. Separating the sand zone into two parts by inserting a thin layer of shale results in reflection filters which differ greatly from one another. The particular filter curve generated depends upon the location of the shale layer. A sand zone filters reflections from interfaces below the zone in a manner complementary to the reflection filter. Where the most energy is reflected, the least is transmitted; conversely, where the least energy is reflected, the most is transmitted. The models considered in this report could easily give rise to high-amplitude reflections; but, unless the amplitudes were very high, there would be little filtering of deeper reflections. However, for very high-amplitude reflections and narrow-band data, little energy would be transmitted and a shadow zone would result. For very high-amplitude shallow reflections and broad-band data, a low-frequency shallow reflection would cause high-frequency deep reflections; a high-frequency shallow reflection wouldmore » cause low-frequency deep reflections. (17 figures)« less

Vibration impulses of variable frequency and duration have been generated by means of an electrically excited vibrator and the resulting seismic waves recorded at the ground surface along a 200‐ft traverse. The first arrivals were refractions from the water table and a deeper clay‐siltstone interface, and these checked with the results of a standard refraction survey. The amplitudes of displacement of the refracted waves varied in each case with approximately the inverse square of distance; the critical distance was marked by a discontinuity of amplitude. Two later impulsive arrivals recorded within 50 msec of the first were interpreted respectively as a transformed reflection from 85 ft depth and an ordinary compressional reflection from 200 ft depth. A dispersive Rayleigh wave gave an independent estimate of the shear velocity and thickness of the surface layer. Air‐coupled waves of frequencies 70.8 cps and 330 cps were recorded and have been related to the first‐ and third‐mode Rayleigh waves respectively.

Air‐coupled seismic waves of low frequencies were excited by aircraft‐generated sonic booms and detected by a hydrophone at moderate depth and a geophone on the sea floor. The experiments were conducted on the continental shelf off New Jersey.

This paper presents a multichannel method for enhancing the signal-to-noise ratio, when the noise is purely random. We show that under some assumptions the contribution of each wave may be individualized. The quality of the results is sensitive to the estimation of the interspectral matrix from the initial traces. The proposed method is general: according to the type of document (shot point or CDP gather, stacked seismic section), we indicate how the interspectral matrix is best estimated.

Solving for the physical properties of the subsurface from seismic observations using nonlinear inversion requires the solution of the forward problem, the simulation of seismic waves. I formulate the equations required to simulate seismic wave propagation in an arbitrarily inhomogeneous anisotropic elastic medium using the method of finitedifferences. Density and all of the twenty-one elastic moduli may vary at every grid-point on a Cartesian mesh. An implementation is described on the finegrain parallel Connection Machine supercomputer. Each grid-point of the three-dimensional volume is assigned its own processor thereby harnessing the natural parallelism of physics. Results are presented of modeling anisotropic seismic waves in three dimensions through a homogeneous model with twentyone non-zero elastic moduli. The complexity of the wave propagation suggests that standard methods of seismic processing break down in regions of strong anisotropy. However, with the massively parallel computer revolution and advances in nonlinear inversion, we may soon enter a new era where all seismic phenomena are taken into account including those related to strong anisotropy thereby providing useful new information about rock properties.