CHAPTER TWENTY-TWO: SEISMIC PETROPHYSICS 2 Editing/Modeling Logs Case Histories

Table of Contents
22.00: Introduction To This Chapter
22.01: Seismic Modeling With Modeled Logs
1. Fluid Replacement in Reef
2. Fluid Replacement in Gas Sand
3. Editing with Trend Lines
4. Editing Sonic Log
5. Fisher – Good Editing
6. Faust, Smith, Wyllie Editing
7. Layer Replacement on Reef
8. Editing by Regression
22.02: In Conclusion
22.03: Exercises for Chapter Twenty-Two
22.04: Bibliography for Chapter Twenty-Two

Continue to Chapter Twenty-Three

Publication History: This Chapter formed part of Chapter Ten of Volume Two of The Log Analysis Handbook, a self published series of course notes covering geological and geophysical aspects of log analysis. First published in 1978, revised 1985, and 1993. Completely revised and re-organized for this electronic edition Sep 2002. A portion of this Chapter was also published as "Determination of Seismic Response Using Edited Well Log Data" by E. R. Crain and J. D. Boyd, CSEG, October 1979. ** Best Paper Award, CSEG, 1979**

CHAPTER TWENTY-TWO: SEISMIC PETROPHYSICS 2 Editing/Modeling Logs Case Histories

22.00 Introduction To This Chapter
This Chapter contains Case Histories for the methods described in Chapter Twenty-One. These show how logs are edited, processed, and transformed for use in geophysical applications. The primary aim of this Chapter is to show how to create a synthetic seismic trace from log data that accurately represents the seismic response of the subsurface. If the synthetic seismogram is reasonable, then calibration of seismic depth sections, seismic inversion results, attribute analysis, and AVO modeling will be successful.

Chapter Twenty-Three deals with how these transformed logs and synthetic seismograms are used to calibrate inverted seismic sections, seismic attribute interpretations, and amplitude versus offset studies..

22.01 Case Histories: Log Editing and Modeling
Log editing for seismic purposes is an interpretative judgment. Consequently, you run the risk of editing out valid data. The risk is worthwhile, in order to avoid inaccurate results.

Figures 22.01 and 22.02: Case History - Swan Hills Reef
This is a Swan Hills reef section in the Rosevear area of Alberta with significant gas filled porosity. Figure 22.01 contains the log analysis results and seismic results (acoustic impedance and reflection coefficients) on a highly compressed depth scale. Formation tops are shown and the modeled interval is marked.

FIGURE 22.01: Reflection coefficient, acoustic impedance, and log analysis before and after gas model - depth scale

FIGURE 22.02: Seismic traces, acoustic impedance, and log analysis before and after gas model - time scale

The model merely replaced the mud filtrate seen by the logs with a mixture of gas and formation water. The model results, on a two way time scale, are shown on Figure 22.02. The shaded area on the acoustic impedance curve shows the difference between log recorded values and the modeled values. Reflection coefficients and peak amplitude on the synthetic are about 40% higher after modeling. The modeled values more closely represent the formation as seen by the seismic impulse, and this is confirmed by the actual seismic data.

This example prepared by the author and published in "Determination of Seismic Response Using Edited Well Log Data" by E.R. Crain and J.D. Boyd at CSEG Annual Symposium, October 1979. The model uses the log response equations for sonic and density data and a pseudo-travel-time for gas. The pseudo travel-time method may over estimate the gas effect, but this can be controlled by reducing the gas effect to match the real seismic reflection amplitude.

The bright spot caused by the gas is a characteristic of some reefs in this area. It is interesting to consider what the reflection would be like if the porosity was at the top of the reef instead of in the middle. The acoustic impedance of the gas filled porosity is almost the same as the overlying shale.

There would be no reflection at the top of the carbonate, and the base of the porosity would be mapped as the carbonate top. Such cases undoubtedly exist and models clearly demonstrate why they might not be found by seismic interpretation.

Figure 22.03: Case History - Gas and Water Sand
The second example illustrates a synthetic seismic section derived from a single well in the Arctic Islands. The well contains gas in a thick porous sandstone. The object of the model section was to determine if water bearing sands could be distinguished from gas sands, and what critical sand thickness was required before the interpreter could be sure that the sand was present.

FIGURE 22.03: Seismic model comparing gas and water bearing sands of different thicknesses

Since the geology of the area, as well as log character, suggest that the sand is eroded from the top at an unconformity, we selectively removed 10 feet at a time from the top of the sand and made a synthetic trace for each case. Both a water and a gas model were used. The sand was originally 80 feet thick.

The sand being modeled is between 810 and 830 milliseconds. It is evident from these plots that a gas sand 40 feet thick gives rise to about the same seismic response as an 80 foot water sand, and that no seismic event can be expected if the sand is wet and less than 60 feet thick, or gas bearing and less than 30 feet thick. These results are corroborated by the seismic data and other wells in the area.

Prior to making these models, two dry holes had been drilled based on bright spot analysis on seismic sections. The abandonments cost $15,000,000 each in 1977 dollars. After the models were made, it was clear that bright spots were not sufficient criteria for defining gas prospects in this area, and that better geological control was also needed.

Many more models could be made, and often are made, during the course of a project. Various wavelets at varying frequencies are often needed to narrow down the possible choices before modeling is even attempted. The model parameters or wavelet may have to be adjusted to obtain a better match, and since this is a modeling problem, there may be more than one model which will adequately match the seismic data. This example prepared by the author in 1977 using the seismic modeling module of the LOG/MATE software package..

Figures 22.04 and 22.05: Case History - Editing with Trend Data
This example presents a severe, but plausible, simulated editing situation for a seismic modeling job. This example is taken from "Log Editing in Support of Detailed Seismic Studies", by Brian E. Ausburn, SPWLA, 1977. A brief excerpt of his paper discusses the editing process as follows:

FIGURE 22.04: Original and edited logs

Interval 3000 - 5520: a sand/shale sequence, most shales are washed out, yet the majority of sands remain in gauge. The combination of hole enlargement and shale alterations results in apparent shale velocities and densities lower than actual formation values.

This judgment is often based on trend curves developed from other wells in the area, where hole conditions for the equivalent stratigraphic section are superior. In this simulated example, the correction trend is substantiated by observed shale velocity values in sections where the hole has no wash out, as at 4960-5220 and 5370-5220.

If a mechanical reason for hole washout cannot be determined, the substitution value should be considered carefully. It is possible the zone washed out because it is different from the nearby equivalent zones.

Shale alterations and washouts also affect the density log. Hole problems can be subdivided into rugosity and enlargements. Rugosity is sufficient to cause density log errors with little hole enlargement. Both are corrected by applying modeling techniques described later.

Interval 5520-6900: a mixed section of limestone, sandstone, shale and salt. With the exception of a few obvious hole washouts noted on the caliper in shale sections, the majority of editing is related to the salt section.

Note that the density log has been edited from 6790-6900 even though the hole is in gauge throughout that section. Due to a different electron/bulk density relationship for salt, from most other sedimentary rocks, the apparent log bulk density of salt is not the true bulk density. By contrast, the sonic response in the good hole salt section substantiates the validity of the sonic edits in the washed out salts.

Figure 22.25 compares synthetic seismograms that would be obtained from raw and edited log data. As predicted from the severity of edits shown on Figure 21.29, the differences in synthetic seismograms are significant.

FIGURE 22.05: Synthetic seismogram before and after editing

Note that the shallow section of the synthetic from unedited data has more character than from edited data. These apparent events represent only the observed contrasts in the erroneous information and not any real acoustic impedance contrast in the subsurface.

The last major event from the unedited set occurs at approximately 2.030 seconds, while on the edited version this event occurs at approximately 1.875 seconds. This difference reflects the significance of the uphole velocity in achieving depth/time ties between seismic and wellbore data. In this case, the too slow velocity observed on the uphole portion of the raw log could make a difference of 500 to 1,000 feet in the location of the deeper events. Raw log data generally makes a synthetic too long, but invasion in gas zones and vuggy porosity can make it too short.

Figure 22.06: Case History - Editing Sonic Log With Resistivity
This is an example of manual editing using the resistivity and other log curves in a well to remove spikes and skips. When skips can be convincingly identified by their characteristic square wave shape, the resistivity and neutron curves often provide guidance for the edit.

FIGURE 22.06: Editing with resistivity as a guide

More sophisticated math can be used, as described earlier, to transform the resistivity to a porosity and then to a sonic travel time, or more directly from resistivity to travel time. Knowing the math helps us make quick and simple edits by eye without resorting to complex computer models.

Figures 22.07 to 22.10: Case History - Fisher Good Editing
This example shows the effect of the Fisher-Good method on noisy sonic and density data. The illustrations show:
1. original and edited sonic
2. original and edited density
3. synthetic from original sonic and density
4. synthetic from edited sonic and density

FIGURE 22.07: Original and edited sonic log

FIGURE 22.08: Original and edited density log

FIGURE 22.09: Synthetic from original logs

FIGURE 22.10: Synthetic from edited logs

The close match of the edited logs to the original (where no editing was required) demonstrate the calibration of parameters. Rational logs are generated where bad hole effects seriously degrade both original sonic and density. The synthetic seismogram has no spurious reflections and horizon time picks are reliable.

The method is described in more detail in "An economic approach to sonic error corrections: The EASElog process"; Fischer, J.G., Good, W.F.; p. ??, SEG 1985. This example was prepared by Harold Ryan using the seismic modeling module of the author's LOG/MATE software package.

Figures 22.11 to 22.13: Case History - Faust and Smith Editing
These examples show the results of using Wyllie time average (response equation), Faust, and Smith editing techniques to four US cases. The log plots show:
1. gamma ray log
2. resistivity log
3. computed lithology
4. original sonic and Wyllie edited sonic
5. original sonic and Faust edited sonic
6. original sonic and Smith edited sonic
7. error between original sonic and Wyllie edited sonic
8. error between original sonic and Faust edited sonic
9. error between original sonic and Smith edited sonic

FIGURE 22.11: Faust, Smith, and Wyllie editing - onshore Gulf Coast, Texas

FIGURE 22.12: Faust, Smith, and Wyllie editing - offshore Gulf Coast, Texas (upper) and Fort Worth Basin, Texas (lower)

FIGURE 22.13: Faust, Smith, and Wyllie editing - East Texas

It is clear that reasonable sonic, and also density if needed, can be generated by these techniques. The error traces help identify the places to concentrate effort. Places where the error is small identify reasonable good original log data and where parameters can be refined by successive iterations. These illustrations are from "In search of the well tie: what if I don't have a sonic log?"; Adcock,S.; SEG Leading Edge, p. 1161-1164, Dec 1993.

Figure 22.14: Case History - Layer Replacement on a Reef
Modeling is not new. This example dates from 1962, and illustrates the result of replacing shale with a reef buildup. The wavelet is fairly low frequency by today's standards, but matched the seismic resolution of the day.

FIGURE 22.14: Layer replacement in a Devonian Reef

The reef is thinned from its maximum thickness down to zero to see what the seismic signature looks like for each case.

We have found in foreign work that the operators have not always had the advantage of re-acquiring or re-processing older data, so interpreters are obliged to use lower frequency data. It is important to match the synthetic frequency content to the seismic available.

Figures 22.15 to 22.17: Case History - Multiple Regression Editing
Here are three US examples from Patchett's work showing:
1. effect of sonic edits on synthetic
2. creation of synthetic sonic and density, with resulting synthetic seismogram
3. comparison of regression derived synthetic with real data

FIGURE 22.15: Log editing by multiple regression - original and edited logs and seismogram trace

FIGURE 22.16: Synthetic sonic and density log from multiple regression

FIGURE 22.17: Comparison of synthetic to real seismic section

Other examples are contained in the original paper and are well worth reviewing. See "Automatic editing of travel time and density logs", Patchett,J.G.; CWLS Trans, 1991.

22.03: In Conclusion
This Chapter has covered most of the procedures that can be done to a log to make it useful as a calibrating tool for seismic modeling and inversion. How to use this data is covered in the next Chapter.

The various techniques and case histories show how important it is to edit and model the log data before you use it for seismic purposes. Don't waste your time, or risk a meaningless interpretation, by skipping this step. Please bear in mind that the log analysis math and rules given in this Chapter are a very tiny subset of the science of petrophysics, so use some common sense and good judgment in its use. Find an expert if you need one.

22.04 Exercises For Chapter Twenty-Two

Exercise 22.01 - Quality Control Synthetic Seismogram
1. Locate and edit the spurious log data on the following illustrations. What do you think caused the logs to "go crazy"? What kind of artifact did this data create on the synthetic seismogram? (15 marks)

FIGURE 22X.01A: Original log data for Exercise 22.01

FIGURE 22X.01B: Synthetic seismograms from original log data

2. Review the edited logs and synthetics for this example shown in the illustrations below. Comment on two way time differences between "before" and "after" models, synthetic seismogram reflection amplitude and phase differences, and any other visual aspects of this example that you might want to explain to team members and cohorts on this project. (15 marks)

FIGURE 22X.01C: Edited log data for Exercise 22.01

FIGURE 22X.01D: Revised synthetic seismograms for Exercise 22.01

3. Compare the revised synthetic seismograms with the recorded seismic data in the illustration below. Compare at both Well 1 and Well 2. Which well location does it match the best? Which synthetic best matches the frequency content of the recorded data? (10 marks)

FIGURE 22X.01E: Original seismic data for Exercise 22.01

4. If the original log data and synthetic seismogram were sold to you by a commercial service bureau, what would you say to the service representative after you reviewed the log plots and seismogram? (10 marks)

22.04: Bibliography for Chapter Twenty-Two
See Chapter Twenty-One

ABOUT THE AUTHOR

E. R. (Ross) Crain, P.Eng. is a Consulting Petrophysicist and a Professional Engineer with over 35 years of experience in reservoir description, petrophysical analysis, and management. He has been a specialist in the integration of well log analysis and petrophysics with geophysical, geological, engineering, and simulation phases of oil and gas exploration and exploitation, with widespread Canadian and Overseas experience.

His textbook, "Crain's Petrophysical Handbook on CD-ROM" is widely used as a reference to practical log analysis. Mr. Crain is an Honourary Member and Past President of the Canadian Well Logging Society (CWLS), a Member of Society of Petrophysicists and Well Log Analysts (SPWLA), and a Registered Professional Engineer with Alberta Professional Engineers, Geologists and Geophysicists (APEGGA)