DUE: Friday, Dec. 2, 2011
In this exercise you'll gain some experience with interpreting 3D seismic volumes. We'll use OpendTect, a free, open-source, multiplatform 3D interpretation package. The data set will be the F3 data volume, which is in the North Sea off the Netherlands.
OpendTect is installed on the computers in the Mac lab (1004) -- HOWEVER, this exercise requires use of the latest version of OpendTect (v.4.01l ). If the version in the lab is older, I recommend you download a recent copy of the software yourself here,. You can run the new version on your own computer or laptop, or on the lab Mac.
NOTE: While using OpendTect, I strongly recommend that you get in the habit of frequently saving your work, in case the software crashes (this has happened to me occasionally). To save your session, go to "Survey" -> "Session" -> "Save..."
To conduct this case study, follow these steps. Portions to hand in as part of your assignment are colored in red.
1. Download the demo data set. To begin, be sure you have the demo data set on your computer. The best way to do this is by copying the data from a portable disk drive in class (it's >6 Gb in size), which you should have done before beginning this exercise. Alternatively, you can download the data set from the OpendTect webiste by going to this link and following instructions for downloading the torrent. Be sure to put the data set in the "ODData" directory in your home directory, or wherever else you've set as the "Data Root" directory for OpendTect. (The first time you open OpendTect, it will prompt you for a data directory. You just select the folder where you want this to reside, and the software creates a directory called "ODData" in that location.)
2. Open OpendTect and load the demo data set. To open the data set, go to the "Survey" -> "Select/Setup" menu. Click on the Demo_OpendTect project, then click "OK." You will see a sketch of the axes of a 3D volume.
3. Show data: an inline, a crossline, and a time slice. To visualize the seismic data, do the following. For an inline, click on the word "Inline" on the left menu, then "Add" in the pull up menu. The software will add an inline plane in the middle of the volume (e.g., inline #127). To show the data on that plane, control-click where it says "<right-click>", then "Select Attribute" => "Stored Cubes" => "All lines" (the latter is the name of the seismic data volume). You'll see the data. You can change the color palette by clicking at the top right (the default is "Seismics").
Repeat this process for one crossline, and one time slice, of your choice.
At this point you may want to spend some time getting used to moving around in the window -- e.g., rotating the volume, zooming, and using the edit arrow (top left) to change the time slice, inline, or crossline in the view.
4. Bright spot detection and visualization. This exercise follows pages 24-28 of the Introduction to OpendTect, which can be downloaded here (click on "Training Manual).
A prominent bright spot is visible on inline 250 of the data set. Here you'll develop an energy attribute to highlight the bright spot, and then you'll render it in 3D. Follow the Training Manual instructions on pages 24-28 to develop an energy attribute. (Note: unless you have a two-button mouse, you can skip the color-bar editing exercise on p. 27 -- there appears to be a bug in the Mac OS version of the code... control-clicking on the color bar does not bring up the appropriate menu.) Be sure to choose your preferred time gate for the energy attribute by following the instructions on p. 28. (NOTE: on my Mac, I had to click on "Calculate" twice to get it to work.)
When creating the seismic output (following p. 29 of the Manual), choose the following volume subselection: inline 100-400, crossline 750-1150, Z range 300-600.
Make an image showing the bright spot in 3D, using the "chimney" color scale. Make another image showing the bright spot body along with an inline and crossline seismic line (in greyscale). Describe the shape of the main bright spot body, and, using the intersecting seismic sections, speculate on what controls the shape.
5. Display a tracked horizon. The demo data set comes with several tracked horizons, called e.g. "Demo 5 -> FS7". To display horizons, click on the word "Horizon" in the Tree scene window, then "Load" -> "[horizon name]" -> "OK." The horizon should appear as a surface, with two-way traveltime ("Z values") color-coded. Notice that you can also see the picks themselves as a red line plotted on the data.
6. Pick a new horizon. You can learn how to pick a new horizon in the Training Manual, pages 42-47.
We will pick a new horizon -- let's call it the "blue" horizon -- near the base of the section. This event is a strong trough (colored "red" when the seismic data is plotted with the "Seismics" color scale) which you can find by setting your inline plane to 425 and your crossline to 431; the reflection is the trough centered at about 1628 ms TWTT. To begin picking this horizon, click on "Horizon" in the elements window at the left, then "New...". This will bring up the "Tracking Setup" dialog box. Make sure you pick "Min" under "Event Type" for best results -- this means the auto-tracker will follow a trough (minimum amplitude), which is what we want here. Fill out each panel in this dialog, and be sure to choose the "Stop" option under the Event tab. Begin making picks (or "seeds", as they're called here) by clicking on the "Create seed" icon at the bottom. (NOTE: The horizon-tracking icons at the bottom of the screen are only active if you click on the horizon you wish to track in the tree panel. If you find that you can't activate one of those icons by clicking on it, double-check to make sure you've highlighted the horizon in the tree.)
To begin tracking, you'll first have to set up a "track area" -- that is, the subvolume within which the tracking algorithm will make its calculations. (The seismic volume is too big to track horizons simultaneously throughout the entire volume -- OpendTect will crash if you try to do that.) View the tracking area by clicking on the "Show track area" icon on the bottom panel. You should see a volume box. You can resize this box by dragging on the corners, or move it by clicking and dragging in the center. I recommend keeping the track area subvolume at the default size when it first comes up. The tracking procedure is then a repeated sequence of seeding some picks in the subvolume, clicking on the "Auto-track" icon, checking your tracks for accuracy, then moving the track area subvolume to a new (but slightly overlapping) location. (Don't get discouraged; this will go pretty fast once you get the hang of it.)
Make some picks on the horizon throughout the volume (i.e., on several different inlines and crosslines). The more seeds you pick, the better constrained the horizon tracking will be.
Tips:
• The picking will only be activated when you have the horizon highlighted in the "elements" window, and when you are in "interact" mode (not "view" mode).
• Be careful where you click whenever you have the "Create seed" icon active. It is easy to click somewhere thinking you were in "view" mode, and thus winding up with bad seeds (so to speak). If that happens, you'll have to delete the seeds as described above, or in section 3.3.2 of the online documentation.
• Be sure to save your horizon before quitting a session. (Save it by control-clicking on the horizon, clicking "save" or "save as" if you want to rename it.) Saving a session will NOT save your horizon!
• When you reload, revise seeds, and retrack a previously saved horizon, you may notice a weird, blocky appearance to the horizon. Don't panic; this is just a display bug. To get back to the proper display, do the following: (1) Save the horizon. (2) Remove the horizon from the tree. (3) Load the saved horizon. It should now look fine.
• When you reload a previously saved horizon, you have to turn "tracking" back on if you wish to revise the tracks. Control-click on the horizon name in the tree, click "tracking", then "start tracking."
When you've picked the event on several crosslines and inlines, you are ready to auto-track the horizon. (NOTE: this would be a good time to save your session.) First, be sure that your tracking box includes your seeds. To track the volume, click on the "Auto-track" icon on the bottom panel. You should see a horizon everywhere in the volume. To visualize the depths of the horizon, control-click on the "Z values" attribute under the New Horizon, and select "Select Attribute" -> "Z values". You should see a color-code depth surface. Repeat for other sub-volumes until you've filled the volume.
Turn in a plot of your first tracked horizon, with the TWTT ("Z value") color-coded.
At this point, you'll need to apply some QC by looking for obvious artifacts -- i.e., things like holes or linear "streaks." To fix those, go to an inline or crossline in the area in question and look at the tracked horizon. Usually you can see a cycle skip or other artifact on the cross-section that clarifies the artifact. Just make some more picks ("seeds") on the horizon in these new cross-sections and re-track the horizon. NOTE: After re-tracking a horizon, to see the current Z-values, you must control-click on the "Z values" attribute and select "Select Attribute" -> "Z values". It appears that you must do this every time you change the track. Your finished horizon map should look something like this.
Turn in a plot of your final tracked horizon, with the TWTT ("Z value") color-coded.
7. Calculate an isopach map. Now let's calculate an isopach map (map of "thickness", in TWTT) between the blue horizon and the Top Vineyard. To do this, control-click on the blue horizon and select "Calculate isopach..." Follow the instructions and create an isopach. Note that the isopach is another good way to find artifacts in your horizons.
Turn in a plot of your isopach between blue-Demo 1 --> MSF4, with the TWTT ("Z value") color-coded. Where is the unit between these two horizons thickest and thinnest? What controls the thickness of the isopach?
8. Calculate a horizon-based similarity map. It is often desirable to plot a particular seismic attribute along a horizon, rather than on a horizontal time slice. First, you'll define a similarity attribute by doing the following: click on the "Edit 3D Attributes" button at the top left of the screen, then click on "Open Attributes Set" at the top left of the attributes window. In the pull-down window at the top right, select "Similarity." Click on "select" button next to "input data", click on "Stored", and select the input data to be "4 Dip Steered Median Filter." Next, under the "Steering" pull-down window, select "Full." A "Steering Data" box will appear -- click on "Select..." next to this box, and choose "2 Steering BG Detailed". Finally, enter a name for your new attribute in the "Attribute Name" box (let's call it "Similarity Steered"), and click on "Add as New."
Next, we'll apply this new attribute to a horizon. First, load the "Demo 6 -> FS8" horizon (control-click on "Horizon" in the tree and select "Load"). Then, control-click on the Demo 6 horizon and select "Add Attribute." Control-click on the new, blank attribute and assign your attribute by selecting "Select Attribute" -> "Attributes 3D" -> "Similarity Steered" (or whatever you called your coherency attribute when you defined it). This will calculate on the fly (this will take a couple of minutes). (NOTE: you will probably wish to save the calculated horizon attribute for future use, so that you don't have to wait for the computer to re-calculate it every time. Do this by control-clicking on the Similarity attribute in the tree, then selecting "Save Attribute" and "OK".
The similarity plot will highlight many fascinating features that are not otherwise obvious on the seismic data. Experiment with a few different color scales to produce a plot that you like (I recommend gray scale, magic, red-white-blue, or red-white-black).
Turn in a single plot showing Similarity along the FS8 horizon. Using crossing seismic sections, interpret the geological origins of the following features: (1) the bright lines of low similarity in the NE corner of the volume; (2) the circular feature on the northern edge of the volume, between crosslines 800-1000; and (3) the curved string-like feature that goes through inline 520/crossline 672.
Next, add the volume-rendered image of the bright spot you created in step #4 above to the image of the horizon-based similarity. Add some transparency to the energy attribute so that you can see the similarity horizon beneath it. (Add transparency by control-clicking on your energy attribute in the tree, then clicking on "Properties" and using the transparency slider.)
Turn in this plot. What are the geological controls on the distribution of this bright spot?
10. Create a coherency iso-surface. Iso-surfaces are ways to visualize structures in a 3D volume based on contouring a surface of some attribute (e.g., similarity). Click on "Volume" in the left-hand panel and select "Add." Control-click in the blank attribute name, then select "Select Attribute" -> "Attributes" -> "Similarity Steered" (or whatever you called your similarity volume calculate in step #8 above). You can see the volume by clicking on "Volren" (short for "volume rendering"). Note that you can move this block around, by clicking on the volume name in the left panel, then dragging the volume. Focus in on the area around the circular feature on the northern edge of the volume. (I recommend sizing your volume to include inlines 625-750, crosslines 810-930, and times 430-620.)
To add an iso-surface, control-click on your similarity volume name and select "Add" -> "Iso surface". You will see a histogram of the values of the similarity attribute, and a box at the bottom where you can select the value to contour. Set this somewhere around 0.75 and hit OK. This will show volume elements of that iso-surface (to see these clearly, make sure that you have turned off the "Volren" option).
Turn in an image of your iso-surfaces around the circular feature What can you say about the 3D shapes of the low coherency zones? What could these structures possibly represent, geologically speaking?