GM-SYS Profile: Density Analysis from Seismic Velocity Field Using Forward Modeling: A Case Study from the Black Sea
In a marine environment with no wells to tie to, it is often difficult to interpret lithologies and general rock properties from seismic data alone – difficult, but not impossible. One can convert seismic stacking velocities to a density model and tie the seismic data to the gravity data by using the GM-SYS® Profile™ extension to Oasis montaj™.
We quickly constructed a relatively simple density model consistent with seismic data in an area of the Black Sea. We applied a simple velocity/density conversion to seismic stacking velocities from a 500 km-long marine seismic line. The available data included a seismic depth section (Figure 1); 14,000 time-velocity pairs from 506 shot points (Figure 2); bathymetry-topography; and the observed Bouguer gravity profile.
Figure 2: Sample of Stacking Velocity Information
We first used the Dix equation (Dix, 1955) to convert the stacking velocities to interval velocities and then used these interval velocity-time pairs to calculate depth. We applied Gardner’s standard coefficient and exponent (ρ=0.23 v 0.25 where ρ is density and v is velocity in ft/sec) in Gardner’s equation (Gardner, et. al., 1974) to the interval velocities, ultimately resulting in 14,000 final density-versus-depth points that we used to make the density-depth grid. One could, alternatively, adjust the Gardner parameters to one’s particular situation, switch to the Nafe-Drake (Ludwig, et. al., 1970) velocity-density relationship at some point, or use some other variation.
Next, we loaded the depth-density pairs into a database (Figure 3). Note that depths in meters are negative below sea level. We set the distance channel to be the "X" channel and the depth channel to be the "Y" channel and then gridded the density channel at a 500 meter grid spacing. This takes the usual X-Y grid (and map) and converts it to a cross-section view of X-Z.
Figure 3: Density-depth pairs loaded into a database. The red profile is bathymetry and the green symbols are the density values at their calculated depths.
The resulting density cross section (Figures 5 and 6) shows a number of interesting features. The densities across the upper portion of the section are consistent, but there is a density inversion in the middle of the section. Most of the upper part of the section is extremely low density with the "normal" density versus depth beginning at about 8,000 meters. Even the density values at a depth of 14 kilometres (at the bottom of the image) appear to be very consistent. Furthermore, even at the scale in Figure 5, which makes it impossible to see any detail, there is an indication that the basement is denser in some areas than others.
Figure 4: Segment of the contoured density-depth grid from the middle of the seismic line (no vertical exaggeration). The square gridlines are 2,000 meters on each side, providing over 2,000 density points in this image.
Figure 5: Full-length section at VE 1:1 The black rectangle is the area drawn in Figure 4; this has the same color table.
Next we contoured the density grid at 0.1 gm/cc and then loaded a bitmap image of the seismic depth section into an Oasis montaj map under the density contours (Figure 6). One can quickly stretch the bitmap group to line up with shot point-depth references by eye; if one wants to be more precise, one can use a two-point "warp" procedure to register the bitmap.
Figure 6: Sample of the density contours on top of the seismic amplitude depth section. VE 1:1
In Figure 6, the yellow interpreted seismic horizon around 4,500 m below sea level correlates well with the top of the density inversion (the closed 2.3 g/cc contour). The section below is 2.2 and 2.1 g/cc until one reaches the 2.3 g/cc layer at about 8,000 meters. Below 8,000 m (tighter contours on the image) there is a more "normal" density with depth relationship.
To construct the model, one would export the density map with contours displayed as bitmap image. Next one would build a starting model from the bathymetry/topography and import the bitmap. Finally, one would build block boundaries along the contours in the bitmap image and assign densities manually to each block. Users can simplify the density boundaries according to their own judgment (we built a relatively small model with only 25 blocks). The result is the model in Figure 7.
Figure 7: Example of a portion of a model built by copying contours on bitmap
Figure 8 shows the density model with the seismic depth image loaded as a backdrop. Some of the blocks look a little jagged due to the vertical exaggeration of about 10:1. The calculated response is the thin black line in the upper panel and the observed Bouguer gravity is the heavier line in the upper panel. Note that the observed curve clearly indicates a regional response which we interpret as crustal thinning in the eastern Black Sea, directly beneath the large basin clearly shown on the seismic section.
Figure 8: Model constructed from density contours with seismic depth section shown as backdrop with VE 10:1.
Finally, we dealt with the crustal thinning by introducing a "Moho" horizon into the model and inverting on the depth to that horizon to fit the regional gravity (Figure 9). The blue line in the upper panel is the response of just the "Moho" horizon. Again, the thin black line is the calculated response of the entire model including the "Moho" and the sedimentary section.
Figure 9: Model with “Moho” horizon and calculated response shown in blue. VE 7.3:1
We don’t suggest that our "Moho" horizon is the actual Moho interface. It is just an expedient device to remove a regional; however, it does have some resemblance to the actual geologic situation.
The high velocity zones shown in pink below 11 km (near the middle of the seismic section in Figure 7) are likely real and contribute to the medium-wavelength gravity highs. They appear to correlate with these anomalies. Depending on the goal of the project, one may want to model these features, particularly if magnetic data are also available for the section.
Conclusion
We were able to quickly convert seismic stacking velocities to a density model using over 14,000 time-velocity pairs, and tie the seismic data to the gravity data. This density model is particularly useful in areas without other geologic constraints (e.g., wells or outcrops) to help interpret lithologies. Including the seismic data in the model prior to removing a regional gravity field results in a more realistic model than attempting to determine the regional field first.
MBL, Inc.
Challenge
Interpreting lithologies and general rock properties from seismic data in a marine environment without wells.
Solution
Convert seismic stacking velocities to a density model and tie the seismic data to the gravity data with Oasis montaj and the GM-SYSŪ Profile extension.
More Resources