Global optimization of data quality checks on 2-D and 3-D networks of GPR cross-well tomographic data for automatic correction of unknown well deviations Page: 4 of 4
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
was tested with a +1 ns and -1 ns shift in the time-picks of
the two wells.
Table 1. Inverted well deviation angles compared to the angle of
deviation used to generate synthetic data (see figure 3). Each
comparison is repeated with static shifts in time-zero to estimate
the sensitivity to uncorrected static errors.
Zero Time Well Deviations Angles
Shift (ns) 00 2.5 -50 7.50 5.0K
no shift 0.294 2.519 -4.686 9.804 1.592
2 0.295 2.400 -4.415 9.273 1.531
-2 0.292 2.651 -4.994 10.408 1.658
-1 and +1 2.070 4.300 -6.462 7.988 -0.191
The results show that the technique is precise for low
angles of deviation, and tolerant of constant data errors,
with decreasing precision at larger deviation angles and
more complicated geometries. However, the technique is
sensitive to changes in time picks between tomography data
sets (eg. -1 and +1 ns). Our experience is that shifts of this
nature are rare, and we acquire travel times in free space
before and after each tomography data set to ensure that
any drift or change in time-zero has not occurred.
Application to field data
Here we are attempting to use the well deviation inversion
technique to correct data collected at the Department of
Energy's Integrated Field Research Challenge (IFRC) site
near Rifle, Colorado. At the site we have a network of
GPR tomography data sets consisting of 12 wells
interconnected by 16 2-D inverted tomograms. Initial
inspection of neighboring tomograms showed significant
discontinuities in velocity and structure. Because of the
tight controls on other data errors that could lead to these
sudden shifts, deviated wells were suspected and motivated
this study. Subsequently 9 of the 12 wells have been
logged for deviation to compare with our inversion results.
The 2-D map view plots of the inversion estimated
deviations and logged deviations are shown figure 4.
0.1 - -
-0.3 -0.2 -0.1
While, the inverted deviations are nearly twice as large as
the recorded deviations, both data sets show a systematic
drift to the west. The lack of better agreement may come
from the model trying to correct data quality issues that
stem from a series of small errors not obvious in the data
quality control methods.
Discussion and Conclusions
This study has shown promise for automatic
correction of well deviations in GPR tomography data. The
analysis of synthetic data shows that very precise estimates
of well deviation can be made for even with constant data
errors. However, the analysis of the synthetics and the
application of the method to a large network of field data
show that the technique is sensitive to varying data errors
between neighboring tomograms. We are investigating
more sophisticated models including deviation bends, time-
zero shifts, and anisotropy. Our current attempts to deal
with these additional complexities are complicated by their
strong correlation with our QC measures, leading to
incorrect solutions. We need QC measures that uniquely
indicate particular data problems in the presence of noise
The simplicity of our model will not remove all artifacts of
deviation from the inverted tomograms. This residual
deviation may be corrected with the method proposed of
Cordua et al. (2008) and Cordua et al. (2009). Their work
on accounting for correlated data errors during inversion,
using a non-diagonalized covariance matrix, demonstrated
the ability to remove small spatially correlated data errors
without losing resolution. We feel that this method is a
perfect complement to our approach in that we are
attempting to remove only large general trends and leave
small spatially correlated data errors untouched to avoid
corrupting real information. The eventual hope is that data
errors in large networks of GPR tomography data sets can
be mitigated to allow for 3-D monitoring of subtle changes
in subsurface properties during field experiments.
We would like to thank Susan Hubbard and Jonathan Ajo-
Franklin for their ideas and support on this project.
Funding for the study was provided by the US Department
of Energy Contract DE-AC02-05CH11231 to the LBNL
Sustainable Systems SFA.
Figure 4. Plot of well deviations estimated from the inversion
method (gray) and the plot of well deviations derived from
deviation logs (black).
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Sassen, D. S. & Peterson, J. E. Global optimization of data quality checks on 2-D and 3-D networks of GPR cross-well tomographic data for automatic correction of unknown well deviations, article, March 15, 2010; Berkeley, California. (https://digital.library.unt.edu/ark:/67531/metadc1015623/m1/4/: accessed March 26, 2019), University of North Texas Libraries, Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.