Sigsbee2B Public Distribution. Created Nov. 9, 2002 by the SMAART Joint Venture.
Sigsbee2B is a 2D model & acoustic finite difference synthetic dataset with constant density.
Sigsbee2B is very similar to Sigsbee2A. See description of Interval velocity model below.
The data contained within this distribution remains the property of SMAART whose members are BHPBilliton Petroleum, BP, and ChevronTexaco Exploration & Production Technology Company. Anyone may use the data and publish the results under the conditions described in the Data Release Agreement.
The Sigsbee2B distribution consists of the following files
filesize (bytes) filename
130,651 data_release_agreement.pdf
7.677 sigsbee2b_readme.txt (this file in ascii text format)
34,816 sigsbee2b_readme.doc (this file in Word format)
38,400 sigsbee2b_journal_announcement.doc
951,254,160 sigsbee2b_nfs.segy (shot gathers for the no-free-surface case)
951,254,160 sigsbee2b_fs.segy (shot gathers for the free-surface case)
16,149,444 sigsbee2b_stratigraphy.segy (true stratigraphic interval velocity)
10,762,452 sigsbee2b_migration_velocity.segy (migration interval velocity)
16,149,444 sigsbee2b_reflection_coefficients.segy (P-wave reflectivity)
213,561 sigsbee2b_migration_velocity.jpg
245,888 sigsbee2b_stratigraphy.jpg
335,994 sigsbee2b_reflection_coefficients.jpg
317,044 sigsbee2b_stratigraphy_index.jpg (stratigraphy qualitative)
365,901 sigsbee2b_zero_offset_fs.jpg (zero offset trace section / free-surface)
315,609 sigsbee2b_zero_offset_nfs.jpg (zero offset trace section / no-free-surface)
314,983 sigsbee2b_zero_offset_mp.jpg (zero offset trace section / difference: fs-nfs)
1,565,696 sigsbee2b.ppt (PowerPoint file including jpg’s)
All floating point data are in IBM real format.
True stratigraphic interval velocity model and P-wave reflectivity
Number of velocity traces = 3201 from X= 10000 to X= 90000 ft
Velocity trace increment = 25 ft.
Velocity depth sample = 25 ft
Total Depth of Model: 30000 ft
Salt Velocity: 14800 ft/sec
Water velocity = 4920 ft/sec
Background Sediment Velocity: V = Vo + kZ; Vo = 5000 ft/sec; Z = depth below water bottom;
V = 5000 + 0.3 (Z - Z_seafloor).
There is no density model.
The reflection interfaces are all the result of velocity contrasts, which fluctuate within +/-300 ft/sec relative to V(z) linear velocity function. This produces reflectivities up to about +/- 3%, enough to produce reflections, but not sufficient to destroy the basic gradient macro model which is used in depth migration (see "Migration Velocity Model" below).
Sigsbee2A model was designed to test the effects of illumination on imaging without complications due to multiples or other noise. Therefore Sigsbee2A was created without a free surface (actually absorbing top surface) so no free-surface multiples. It also has a dim seafloor, therefore weak interbeds between salt and seafloor. Seafloor velocity was 5120 below water at 4920 producing a reflection coefficient of (5120 - 4920) /(5120 + 4920) = 0.02, so it's just like any other siliciclastic reflector.
Sigsbee2B model was designed to add realistic multiples (specular and diffracted) to the Sigsbee2A model to test imaging with both illumination and noise problems. Sigsbee2B has a thin (one node thick), fast (9000 ft/s) veneer at the seafloor to give a bright reflection event, since density is constant in the model and therefore cannot provide the actual impedance contrast. Water (4920 ft/s) over a 9000 ft/s halfspace gives a reflection coefficient of about 4/14 or about 0.28. However, since the fast layer is only one node thick, the actual reflection coefficient is a bit smaller, but still within the expected range for real data, in other words greater than 0.20. Sigsbee2B should thus have more realistic free surface and internal multiples, but their amplitude will not be quite the same as a model generated with variable density.
Migration interval velocity model
Number of velocity traces = 2132 from X= 10025 to X= 89938 ft
Velocity trace increment = 37.5 ft.
Velocity depth sample = 25 ft
Total Depth of Model: 30000 ft
The "Migration Velocity" model replaces the true sediment velocities with the background gradient of V=5000 + 0.3(Z-Zseafloor). This smoothed velocity function is probably as good as you would get from real data and hence will give a more realistic image. For migration algorithms that require some smoothing, it may be better to use this smooth velocity model instead of the exact stratigraphic interval velocity. For migration algorithms that do not require smoothing, using the exact velocity may produce some noticeable improvements in the areas of the very deep synclines on top of salt. However, this degree of velocity resolution may not be achievable on real data.
Sigsbee2B Computation Information:
Acoustic Finite Difference (constant density)
Grid size: dX = dZ = 25 feet
Dominant Frequency ~ 20 Hz, Max frequency = 40 Hz.
Source and receiver depth = 25 ft below water surface.
Single source point, 75-foot receiver array.
Source and receiver ghosts included in the recorded wavelet. For the free-surface case, these ghosts are generated internally by the finite difference algorithm. For the no-free-surface case, the ghosts were explicitly generated by adding an extra source and extra receiver in their mirror locations with opposite sign. The true absorbing boundary was about 5000 feet above sea level, call this datum z = -5000. Then, near sea level (z=0), we put mirror shots, one at +h and the other at -h (h is the nodal spacing), negated the excitation source function of the shot at -h, and fired them simultaneously. This produces a pretty faithful (at around the 5% level) imitation of surface ghosting, valid for all take-off angles. The same thing was done for each receiver position. Receivers at +h were superposed with the negative of receiver recordings at -h. So you get the actual ghost effect without the wholesale free-surface reflection (which was necessary for us when subtracting these wavefields from those contaminated with real free-surface multiples). The only wave path to be incorrectly modeled is the direct wave from source "horizontally" to receiver. A quick sketch shows that, because there is one (common) ghost in this case instead of two, the calculated amplitude is exactly twice the real expected amplitude. Other than this the approach gives very good results.
A 56 ms negative static shift was applied, which positions the maximum wavelet amplitude at the theoretical arrival time (wavelet no longer causal).
Sigsbee2B Acquisition Information:
Number of shots = 496, from X = 10925 to X = 85175 ft
FFID = 495-0
Shot interval = 150 ft
Minimum offset = 0 ft
Maximum offset = 26025 ft (end-on)
Max. number of offsets per shot = 348
Max. number of traces in cmp gather = 87
Receiver interval = 75 ft
CMP interval = 37.5 ft
CDP range = 25-2069(1)
Record length = 11.992 seconds
Sample rate = 8 ms
Source depth below sea-level = Receiver depth below sea-level = 25 ft
Remark: Source is pulling the cable. Starting at the far right side of the model
(X=90000 ft) the source location walks through the cable until the full cable
length is reached (26025 ft).
Imaging Suggestions
Maximum depth to image = 31000 ft
Depth step ~ 25 ft (this was the resolution on the input seismic modeling finite difference grid)
Aperture: 25000 ft on each side of an image point should be sufficient.
The Smaart JV hope you will find the data useful for current and future research.