Data Processing Methods for 3D Seismic Imaging of Subsurface Volcanoes: Applications to the Tarim Flood Basalt

Summary

Three-dimensional (3D) reflection seismology is a powerful method for imaging subsurface volcanoes. By using industrial 3D seismological data from the Tarim Basin, we illustrate how to extract the sills and the conduits of subsurface volcanoes from seismic data cubes.

Abstract

The morphology and structure of plumbing systems can provide key information on the eruption rate and style of basalt lava fields. The most powerful way to study subsurface geo-bodies is to use industrial 3D reflection seismological imaging. However, strategies to image subsurface volcanoes are very different from that of oil and gas reservoirs. In this study, we process seismic data cubes from the Northern Tarim Basin, China, to illustrate how to visualize sills through opacity rendering techniques and how to image the conduits by time-slicing. In the first case, we isolated probes by the seismic horizons marking the contacts between sills and encasing strata, applying opacity rendering techniques to extract sills from the seismic cube. The resulting detailed sill morphology shows that the flow direction is from the dome center to the rim. In the second seismic cube, we use time-slices to image the conduits, which corresponds to marked discontinuities within the encasing rocks. A set of time-slices obtained at different depths show that the Tarim flood basalts erupted from central volcanoes, fed by separate pipe-like conduits.

Introduction

The aim of most of the industrial seismic imaging projects in sedimentary basins is to explore for hydrocarbon reservoirs. In recent years, hydrocarbon exploration has expanded to basins containing large amounts of igneous rocks because many of the volcanogenic basins have considerable oil and gas reservoirs. However, because of the interface of igneous rocks in the volcanogenic basins, seismic data processing presents a series of challenges induced by various intrusions, such as reduced energy transmission, intrinsic attenuation, interference effects, refraction and scattering¹. Therefore, oil field companies are focusing their efforts on reducing such a "negative impact" on seismic imaging^2,3,4.

Igneous bodies within sedimentary basins are easily identified by two dimensional or 3D seismic reflection imaging due to the large acoustic impedance contrast with the encasing rocks^1,5,6. This method can provide spectacular images of both vertical and horizontal structures of the volcanic plumbing systems^{7,8,9,10,11,12,13}. However, the strategies of imaging subsurface volcanos are very different from that of oil and gas explorations^8,14,15. This has limited the use of industrial seismic data in studies of subsurface volcanoes, apart from a few successful cases^10,15,16. In this paper, we report detailed procedures of seismic data processing, which are customized for interpretation of subsurface volcanoes. We process two seismic cubes, TZ47 and YM2 (Figure 1), to show how to visualize the buried igneous bodies in the Tarim flood basalt¹⁷.

Protocol

NOTE: The data processing procedures include: synthetic seismogram calculation, synthetic-real seismic trace correlation, and geo-body extraction. Below are the step-by-step details of each procedure.

1. Calculation of Synthetic Seismogram

1. Calculate the acoustic impedance at each interval of the down-well logging curve.
   NOTE: Acoustic impedance is the product of 'seismic wave velocities' and 'density' (ρ*ν)). The data are often averaged to sampling intervals larger than 1 ft, in order to reduce the computation time and aliasing.
2. Calculate the reflection coefficients (R₀) at each interface by using the acoustic impedance calculation:
   
   where ν₁ and ν₂ are the averaged velocities of the layers below and above the interface, respectively; ρ₁ and ρ₂ are the corresponding averaged densities.
   1. If the well does not intersect the igneous bodies, use nearby wells that have intersected the target rocks to obtain the parameters (velocity, density, etc.).
3. Chose a wavelet that has an amplitude and phase spectrum similar to that of the nearby seismic data.
4. Convolve the synthetic wavelet with the reflection series for the entire well survey and generate a synthetic seismic trace. The final simulated seismic trace T(t) can be described by the convolutional model as below:
   
   where R₀(t) is the reflection coefficient, w(t) is the wavelet and n(t) is the noise.
5. If the frequency of the seismic data has large variations throughout the whole well, re-compute the synthetic seismic trace using a wavelet with a different phase and a dominant frequency at different depth intervals.
   1. Repeat the process if the match between the synthetic trace and the seismic data is not satisfactory.
6. Perform the calculation with the provided software (e.g., Petrel E&P Software Platform).
   1. Start the software. Select File | Open Project | and then select the demo research project tlm (users can select their own desired projects). The project should contain well data, wired log, well tops, seismic cube, and interpretation surface in the research area.
   2. Click on Home | Windows | 2D Windows | 3D Windows to open two display windows to show the data sets according to user's preference.
   3. In the "Wells Tree of Input Pane", right click the desired well. Open the Settings window of the well and select the Time tab to create a new time log. Select Velocity Function, then select DT data in the new time log. Click the OK button to close settings window. A new one-way time log is automatically created and will be shown in the "Wells Tree of Input Pane".
      NOTE: A one-way time log is a time-depth relationship of this well. Wired log domains can be transformed to time domains and be shown in the time domain window.
   4. Activate an existing 3D Window by clicking the displayed window. If there is no 3D Window displayed, create a new 3D Window by clicking Home | Windows | 3D Windows. Select TWT in the toolbar of the 3D Window to show the 3D Window in time domain.
   5. Select representative wired logs (such as 'GR', 'DT', or 'RT') in the Wells Tree to show them in the 3D Window; at the same time, select the seismic profile in the 'seismic' tree of the 'input' pane to show them in the same 3D Window.
   6. Use the Manipulate Plane tool in the toolbar of the 3D Window to adjust the location of the profile to intersect the well; the user will see that the wired log has been transformed to the time domain and displayed with the seismic profile in the same 3D window.
   7. Click Seismic Interpretation | Seismic Well Tie | Seismic Well Tie Process. Choose Integrated Seismic Well Tie in the type of study row, and add desired well in the Well row. Choose calibrated one-way time log as time-depth relationship in the TDR row of the input tab, choose seismic cube in the seismic row. Choose any log in the RC calculation method.
   8. Click Launch Wavelet Toolbox to create a Ricker wavelet to apply in this process. Click OK and a new well section window and synthetic seismogram display will be created.

2. Correlate the Synthetic Traces with the Real Seismic Reflectors

1. Use an automated correlation application, like Seismic Well Tie in the platform, to adapt the resulting synthetic trace to the vertical scale of the seismic section.
2. Adjust the synthetic seismogram to increase the overlapping of high amplitude reflectors of the synthetic trace and real trace.
3. Adjust the synthetic seismogram and the real trace repeatedly. When the overlapping trace reaches the maximum, the interpreter has reached the "best fits" between the obtained synthetic seismogram and real traces.
   1. Repeat the process until the correlations reach the desired level.
4. Perform the correlation with the provided software.
   1. Activate the window created in step 1.6.3, which is the one-way time log automatically created from the acoustic log.
      NOTE: This automatically created 'one-way time log' is not perfectly correlated with the real seismic reflectors. The users should calibrate the correlations between the one-way time log and the real seismic reflectors.
   2. To calibrate their correlations, choose a continual and representative reflector that is intersected by the well. Then manually adjust the depth of the well log. For example, to adjust the depth of the DT log, right click the One-Way Time Log in well tree | select the Calculator tool | then add a small time increment (for example, 10 ms) by typing 'DT=DT+10' in the input dialog of the Calculator tool.
   3. If the '10 ms' increment is too large or too small, change the increment to another time (can be negative value) in the 'calculator' tool. Check the correlation between the well log and the selected seismic horizon repeatedly and then adjust the time increment repeatedly, until the correlation is perfectly calibrated.

3. Extraction of Basaltic Sills

1. Pick 2 high-amplitude reflectors encasing the target sills.
   NOTE: Most intrusions are expressed in seismic data as tuned reflection packages, whereby the reflections from the upper and lower intrusion contacts cannot be distinguished. Tuning occurs when the vertical intrusion thickness is between λ/4 and λ/8 (λ is seismic wavelength)¹⁹. Therefore, sills are shown as a set of strong reflections in the seismic section, and their apparent thickness is false.
2. Extract probes between the horizons corresponding to the two high-amplitude reflectors.
   NOTE: There are different tools based on the rendering technique that can help the interpreters better visualize the targets, such as "box probes", "surfaces probes" and "well probes". However, for identification of contacts between the sills and encasing strata, the best tool is "surface probe". ("Surface probe, etc." are terms in 'Petrel' software. The software users should be familiar with these terms).
3. Remove the areas surrounding the geological objects of interest by changing the Voxel connectivity opacity threshold value. Set the default threshold value to 20%. The visualization method of "opacity rendering" is used here to display the result of the extraction of basaltic sills ( Figure 2C).
   NOTE: There are high amplitude reflections along the surface between igneous rock and sedimentary rock because of their significant difference in acoustic impedance. Make the low amplitude parts transparent to highlight the shape of the igneous bodies.
4. As the value for isolation can be higher than 20 - 30%, change the value with small increments to make sure all important igneous bodies are not lost; the larger the value, the higher the risk of losing the volume of the real igneous bodies.
5. Perform the operation with the provided software.
   1. Click the Seismic Interpretation pane, click Insert a Horizon Probe. A probe will be added in the geobody interpretation probes tree of input pane. Double click the added horizon probe and a pop-up window will appear.
   2. Click the Horizons tab in the pop-up window and choose two seismic surfaces that isolate the zone of sills. Click OK to apply the operation.
   3. Check the newly added probe in the geobody interpretation probes tree shown in the input pane. A seismic cube will then appear in the 3D window.
   4. Double click the probe and choose the Opacity tab. A histogram of seismic amplitude will be shown in the tab. Use the left mouse button to draw a line in the histogram to control the opacity of the seismic cube. The low amplitude parts of the tube should be invisible and the high amplitude parts will be left.
   5. Adjust the histogram repeatedly until the desired shape of the interested geobody is achieved.

4. Extraction of the Feeding Conduits

1. Choose continuous and high energy reflection horizons at different depths beneath the surface lava flow.
2. Do time slicing along the selected horizons, to find out discontinuities corresponding to the vertical conduits.
3. Adjust the Two-Way Time (TWT) repeatedly, to achieve the best imaging of the discontinuities of the conduits.
   NOTE: Seismic data cannot image vertical structures well, so better images from amplitude volumes and variance volumes are chosen by comparing clearness at different travel times.
4. Try different slicing techniques, and then choose which can better image the discontinuities.
   NOTE: Different tools can be used here, such as variance body slicing. Its theoretical basis is the similarity between each seismic section and adjacent seismic traces in the seismic data. Another tool, the variance cube, is a new data body processed by the conventional seismic data, which is helpful for the identification of changes in the structure and lithology, plane combination of the fault, etc.²⁰
5. Plot the slices at different travel times or depths into a 3D space.
6. Perform the operation with the provided software.
   1. Double click Volume Attributes in the geophysics tree of the processes pane. Check Structural Methods in the category column and Variance in the attribute column. Select the seismic cube to input box and adjust the other parameter in the parameter tab. For better reading performance, check the box in the realize column. A variance cube is created in the seismic tree of the input pane.
   2. Right click the variance cube and click Insert Time Slice Intersection to show more horizontal intersections in the 3D window. Use the Manipulate Plane tool in the toolbar of the 3D window to adjust the location of the slices to optimize the display of conduits.
   3. Right click the seismic amplitude cube and click Insert Time Slice Intersection to show more horizontal intersections in 3D window. Do the same operation as step 4.6.2 to adjust the location of the slices to optimize the display of conduits.

Representative Results

We demonstrate the usefulness of the techniques described above by applying them to 2 types of igneous bodies, horizontal sills and vertical volcanic conduits. Extraction of the sills is conducted by using the opaque rendering technique, and interpretation of the volcanic conduit is performed by using slicing technique.

Extraction of sills

Industrial drilling wells have intersected many sills in the Yingmai-2 area from the Northern Tarim Basin¹⁷, but the 3D distribution of the sills remains unclear. In order to interpret the sills, we process 3D seismic data from a seismic cube in this area. Firstly, we identify the horizons related to the presence of the sills in the seismic cube by correlating the synthetic seismograms with the seismic cross-sections (Figure 2A). Then we insert surface probes (Figure 2B) into the horizons to constrain the lateral extent of the sills. Finally, we use opacity rendering to extract the geo-bodies of sills (Figure 2C) from the seismic cube. We find that the sills turn to separated lava lobes at the distal end, which indicates that the flow direction is from the dome center to the rim of the dome (Figure 2C).

Interpretation of volcanic conduit

Following the steps detailed in section 4, we obtain six time slices at different depths in the original seismic cube (Figure 3A). Variance body time slices are also shown (Figure 3B). We choose different slicing depths for the variance body time slices because the best resolution of this method is achieved at depths different from that in the original seismic cube. It is clear that the conduits can be imaged by the time slicing technique.

Figure 1: Sketch Geological Map of the Tarim Continental Flood Basalt¹⁸ and the Location of the Seismic Cubes.
1. Tarim block; 2. Desert; 3. Major fault; 4. Flood basalt; 5. Seismic cubes. Please click here to view a larger version of this figure.

Figure 2: Procedures of Extracting the Geo-bodies of Basaltic Sills Encased in Sedimentary Strata.
A. Correlation between the synthetic seismogram (green bars around the drilling well) and the seismic cross section; B. Surface probes along the horizon of sills; C. The extracted geo-bodies of the sills, which is located above the dome center (colored with gray scale); D. A typical dolerite drilling core sample from the TZ47 area. Please click here to view a larger version of this figure.

Figure 3: Three-dimensional Structure of the Conduits.
Imaged by time slice of the original seismic cube (A) and of the re-computed coherent body seismic cube (B). Please click here to view a larger version of this figure.

Discussion

Here we demonstrate 2 methods for illustrating the morphology and structure of the plumbing system of buried basaltic volcanoes; one is opacity rendering, the other is time slicing.

The opacity rendering method is suitable for geo-bodies that have continuous and near horizontal interfaces with the encasing strata. With this method, one can extract the 3D morphology of magma lobes. Normally, flow directions should be along the long axis of the magma lobes. It is also important that the surface horizons have high reflection coefficients (R₀). If the R₀ is too low at the interface, interpreters will not be able to insert surface probes to the target horizons. For example, sonic velocity of basaltic sills is around 5500 m/s, and carbonates have similar velocity of 6,000 m/s¹². Thus, the reflection coefficient at the sill-carbonate contacts would be too low to be identified by surface probes. When using this technique, precise knowledge of the velocities of the target rocks is required. If the velocity data are not available or not properly estimated, application of this method to seismic cubes will be highly limited.

The time slicing method can apply to geo-bodies that have no continuous and horizontal surfaces. When the igneous intrusions have sonic velocities very different from the encasing rock (in most cases, higher than the encasing rock), interpreters can use the time slicing technique to image the boundaries between the intrusions and the surrounding rocks. If the wall rock have similar sonic velocities, it is also very difficult to identify the igneous intrusions from the country rocks.

Materials

Name	Company	Catalog Number	Comments
The Petrel E&P software platform	Schlumberger		software version:2014