www.elsevier.nlrlocaterjappgeo
Use of four-dimensional ground penetrating radar and advanced
visualization methods to determine subsurface fluid migration
Ralf Birken
), Roelof Versteeg
1Lamont-Doherty Earth ObserÕatory of Columbia UniÕersity, 61 Route 9W, Palisades, NY 10964-8000, USA
Received 13 October 1998; received in revised form 26 February 1999; accepted 15 March 1999
Abstract
Ž . Ž .
Four-dimensional 4D or time-lapse three-dimensional 3D ground penetrating radar surveys can be used to monitor and image subsurface fluid flow. This information can be used to create a model of hydrogeological properties. The massive amount of data, which is present in and can possibly be generated from 4D GPR data sets, precludes a manual interpretation. Consequently, 4D data sets have to be processed and visualized in a way that extracts models and allows for data visualization in a semi-automatic way. The principles behind such an approach are applied to the Borden data set, which is used to demonstrate how advanced visualization can assist in the interpretation of raw and processed data. In the Borden data set, changes in reflectivity between different time-steps unveil areas of fluid migration in three dimensions. The combination
Ž .
of these reflectivity changes between different combinations of the 3D subsets of the 4D data set is used to create a model of hydrogeological properties. While this model does not yield a quantitative description of porosity, permeability or hydraulic conductivity, it is a qualitative proxy for a combination of these properties. q2000 Elsevier Science B.V. All rights reserved.
Keywords: GPR; Modeling; Visualization; Hydrogeology; Monitoring; DNAPLS
1. Introduction
One of the most complex problems in earth sciences is that of the determination of fluid flow behavior such as flow pathways, flow ve-locity andror hydraulic conductivity in a het-erogeneous subsurface. Information on fluid
Ž .
flow behavior both present and predicted is
)Corresponding author. Tel.: q1-914-365-8327; fax:
q1-914-365-8150; e-mail: [email protected]
1
E-mail: [email protected].
needed for the determination of the size and shape of groundwater recharge areas, the assess-ment of contamination impact of spills and plumes, and the planning and assessment of remediation efforts. For all these problems, it is recognized that a good knowledge of the
three-Ž .
dimensional 3D values of hydraulic conductiv-ity is essential, yet most often, the only knowl-edge available on the hydraulic conductivity is that deduced from a number of sampling wells and pumping test experiments. These
experi-Ž .
ments only give a poor bulk approximation of the hydraulic conductivity, and consequently,
0926-9851r00r$ - see front matterq2000 Elsevier Science B.V. All rights reserved.
Ž .
the predictions of fluid flow behavior based on these approximations are more often than not unsuccessful.
There is thus a strong need for a method, which can generate a reliable subsurface model of hydraulic conductivity. Recent Ground
Pene-Ž .
trating Radar GPR work has demonstrated the feasibility of surface and borehole radar to
de-Ž
tect fluid movement Brewster and Annan, 1994;
.
Lane et al., 1996 . Research conducted in time-dependent monitoring of oil and gas reservoirs, using multiple instances of 3D seismic data sets, has demonstrated the feasibility of extracting a time- and space-dependent model of subsurface fluid flow. From this a model of subsurface,
hydrogeological properties can be derived
ŽAnderson et al., 1991, 1994 . Thus, four-di-.
Ž .
mensional 4D or time-lapse 3D GPR could be the method to provide models of hydrogeologi-cal properties. The extraction of these models from the 4D data and the visualization of 4D radar data are the topic of this paper.
2. 4D-seismic and 4D-GPR
4D-seismic visualization and analysis tech-niques were pioneered by the 4D Technologies Research Group at the Lamont-Doherty Earth
Ž
Observatory of Columbia University Anderson
.
et al., 1991, 1994 and have by now become well adapted in the oil industry. 4D-seismic technology uses multiple 3D seismic data sets
with the aim of monitoring changes in
impedance caused by drainage and migration of hydrocarbons and other subsurface fluids. The goal of this effort is to enhance reservoir man-agement.
The principle of 4D-GPR is the same as that of 4D-seismic: the same survey is repeated at different times, and as the solid fraction of the earth will not change noticeably between mea-surements, the change between data must be a result of changes in the subsurface fluid phase. The motivation behind 4D GPR is in a way
similar to that of 4D seismic: we want to ob-serve the results of contaminant migration or the effectiveness of remediation methods.
If we want to use 4D seismic or GPR, an imported component of these methods is the verification, using numerical models, that what is observed actually corresponds to a fluid mi-gration. Through a collaborative effort with Rice University, we currently have access to a
two-Ž .
dimensional 2D finite-difference time-domain
ŽFDTD-TE modeling code with similar charac-.
teristics as the code described by Xu and
Ž .
McMechan 1997 . This code allows arbitrary electrical conductivity, permittivity and mag-netic permeability variations on a grid, and has an option to include relaxation mechanisms. This code is a modification of a code written by
Ž .
Bergmann et al. 1996 which is in itself a
modification of a visco-elastic modeling code
ŽRobertsson et al., 1994 . The code has absorb-.
ing boundaries, is second-order in time and fourth-order in space and in its current incarna-tion uses a Ricker source wavelet whose center frequency can be specified at run time.
This code was used in its current incarnation
Žin which it has a uniform radiation pattern to.
simulate a simplified 4D experiment in which a
Ž .
saline tracer ´ s81, ss1000 mSrm is re-r
Ž
leased in a low conductivity medium ´ s5.58, r
.
Ž . Ž .
Fig. 1. Ten time steps a to j of numerical modeling of a time-lapse radar simulation. A spill of saline tracer is simulated as seen by a 300-MHz GPR antenna.
These results indicate that from real 4D data, we could in theory, determine information re-lated to the fluid phase and to the controlling
Ž
hydrogeological parameters hydraulic
conduc-.
tivity, permeability . Implementing this, and ac-tually extracting information on subsurface fluid
Ž
flow from GPR data is fairly complex Sander,
.
1994; Sander and Olhoeft, 1995 . Some of this
complexity arises because collecting the
‘‘same’’ geophysical survey is less trivial than it sounds.
Depending on the method used, a positioning
Ž .
trace spacing requirements needed for the com-parison of features in 4D. Ideally, the line spac-ing should be of the order of the trace spacspac-ing to allow for optimum visualization of subsur-face structures in 3D, but in reality, the trace spacing is mostly much smaller than the line spacing. For a 200-MHz GPR survey with a cross-line dip of 208 and a subsurface velocity of 0.07 mrns, we need cross-line spacing of 60 cm to avoid spatial aliasing. As for most 4D surveys, these kinds of parameters are not at-tainable and will degrade the data analysis.
The main complexity of analyzing 4D radar data is of course the processing and visualiza-tion. These are discussed using the 200-MHz and 500-MHz Borden data sets.
3. The Borden data set
Ž .
The Borden data set collected in 1991 is the most carefully collected and comprehensively described 4D GPR data set on which data have
Ž
been published Brewster and Annan, 1994;
.
Sander, 1994; Sander and Olhoeft, 1995 . This data set, consisting of a closely spaced grid of 2D common offset profiles of 200 MHz and 500 MHz GPR data over a controlled spill was
Ž .
collected at the Canadian Forces Base CFB Borden in July 1991. At this base, an imperme-able cell was constructed within a saturated sand aquifer with surface dimensions of 9=9 m2 and with a depth of 3.3 m. 770 l of
Tetra-Ž .
chloroethylene TCE , a dense non-aqueous
Ž . Ž
phase liquid DNAPL Pankow and Cherry,
.
1995 , was released in the center of the cell. The medium in the cell in which the TCE was released is 3.3 m of a medium to fine-grained beach sand underlain by 3 m of medium stiff clay. A full description of the release
experi-Ž .
ment is given in Brewster and Annan 1994
Ž .
and Brewster et al. 1995 and references therein. Eight East–West lines and eight North–South lines of GPR data were collected along profiles on an orthogonal grid with 1 m gridspacing. To
characterize the background of the site, data were collected before the spill and then at 32
Ž .
times after the spill up to 1000 h post-spill . Initial times between data collection was 8 h, while later in the experiment, the times between data acquisition increased to about 40 h. For this paper, parts of the 200-MHz and the 500-MHz data sets were analyzed and both are used to illustrate different processing and visualiza-tion steps.
4. Data analysis: processing, data differenc-ing and data visualization
While the obvious link between the applica-tions of 4D software in a seismic and a radar environment facilitates the analysis of 4D radar, there are some complicating differences. These are due to the differences in size of the data and the number of data sets and data redundancy. Typically, in seismic, we deal with a fairly
Ž . Ž
small three to four number of large several
.
100’s–1000 GB surveys. In radar, the cost of surveys is relatively minimal, so we can have
Ž .
tens of relatively small 100’s–1000’s MB sur-veys, for example, 33 in case of the 200-MHz Borden data set.
Since we can afford to collect multiple data sets with GPR, we can thus combine multiple different subtraction mechanisms. In effect, in
Ž .
the case of n data sets, we have Ý ny1
differences available. For 3 data sets, this is 3; for 5, it is 10; and for 33, this is 528. Note that this assumes that we are only looking at simple
Ž
differences between two surveys i.e., first
.
derivatives . If we start looking at higher
deriva-tives, the amount of difference data sets goes up accordingly. The interpretation of these 528 dif-ferences can obviously not be done using any kind of manual interpretation, and we thus need an integrated environment for processing and interpretation. This implies that we need some kind of both data reduction and automatic anal-ysis. Even at a fairly modest 20 MB size per survey, this would produce 10 GB of data using first derivatives. Examining all these data in a profile by profile mode is of course infeasible. We thus need to pre-process our data with the objective to prepare the data so that each survey becomes ‘‘similar’’ before differencing and vi-sualizing the data.
Ž .
Fig. 3. Data difference between pre-spill and post-spill 500 MHz data for line E5 at 150 h. Comparison is shown for four
Ž . Ž . Ž . Ž .
Data pre-processing is done using a combina-tion of in-house and commercial processing and interpretation packages. One of the core efforts of 4D data acquisition work is to ensure that the surveys are located at the same place and use the same recording times. Therefore, it is re-quired to remove any shifts in recording time
Ždepth and space navigational errors among. Ž .
the surveys. The time differences could have been introduced by changes in instrument
pa-Ž .
rameters different t , coupling changes, in-0 strument drift, etc. Spatial shifts could have been introduced by inaccurate navigation. A
Ž
first time correlation of the data using first
.
arrivals is fairly straightforward. However, it
was found that the exact spatial location for the data is actually a very complex issue due to the need for high precision.
Both the 200-MHz and the 500-MHz Borden data set, which are used here to illustrate the 4D approach, were carefully controlled and there was no need for spatial coordinate transforma-tions andror corrections.
For the data differencing and visualization, we use the same package which is being used by Lamonts 4D seismic group. This commer-cially available software package Advanced
Vi-Ž .
sual System AVS is a UNIX workstation-based visualization programming environment. It en-ables users to analyze, manipulate and display
Ž . Ž .
Fig. 4. 3D visualization of 200 MHz Borden East profiles. Pre-processed a pre-spill data and b data taken 8 h after spill.
large volumes of complex data, including 2D and 3D images, 3D graphics and multidimen-sional numeric data. AVS utilizes a graphical
Ž .
user interface GUI which allows the user to directly interact with program input and output
ŽIrO parameters. The user constructs a data.
Ž .
analysis network Fig. 2 . The data network performs not only data IrO, but also parts of the data processing and visualization. As the connection between modules is taken care of implicitly in this package, research can concen-trate on processing and visualization instead of code maintenance efforts. The 4D GPR data set
Ž .
of the Borden experiment 33 repeated surveys takes up 46 MB for each of the two profile
directions. Each of these files is loaded into AVS and visualized in a matter of seconds on a SUN ULTRA30 workstation using a Creator3D graphics card.
The most important function of the 4D-GPR package is the computation of similarities and differences among regions between the data sets. The differencing process is currently applied as a simple subtraction operator between the dif-ferent data sets. Future plans include the incor-poration of region growing in the data sets and subsequent subtraction of the regions between different data sets. It should be noted that straightforward subtraction is an inherently un-stable operation due to the wave character of
Ž . Ž .
Fig. 5. 3D visualization of 200 MHz Borden East profiles. Difference between pre-spill data Fig. 4a and data collected a
Ž . Ž .
Ž
our data i.e., if we are off by 1r2 wavelength,
.
our subtraction operator will fail . This of course is the need behind more complex and stable subtraction operations.
Fig. 3 illustrates the perils of not optimizing the time shift on a 500-MHz profile, by illustrat-ing the differences between usillustrat-ing an optimum
Ž . Ž
shift Fig. 3c and shifts which are 1 Fig. 3a
. Ž .
and d or 2 Fig. 3b from the optimum. For the optimum shift, there is no spurious energy intro-duced, while for even a shift of a few samples, we introduce ‘‘false’’ differences.
Ž
Once an optimum shift is determined which
.
can be different for each profile , difference volumes can be created. Fig. 4 shows how this
is done using the 200-MHz data set. All vol-umes shown in Figs. 4 and 5 have been muted after the first arrival and later than 110 ns, based on the known area of interest from
previ-Ž .
ous work Brewster and Annan, 1994 . Fig. 4a displays the pre-processed 200 MHz raw data containing the eight East profiles collected be-fore the spill and Fig. 4b 8 h after the spill started. Note the strong reflection appearing in an area of the upper North–East corner of the cut-out in Fig. 4b. The resulting difference be-tween the two displayed data sets is shown in Fig. 4c, enhancing the differences so that they are much more pronounced. In Fig. 5, three more difference volumes are shown for 25, 36
and 45 h after the spill began illustrating how
Ž .
the spill strong reflection appears to move downwards and to spread out in North–South direction along a layer close to the surface. The 3D difference plots create a time-series showing the expansion of the DNAPL throughout the
cell. As a matter of fact, within AVS, it is possible to run through all 32 difference vol-umes like a movie which illustrates the move-ment of the DNAPL much better than can be shown in Figs. 4 and 5. Running AVS on the computer further allows cutting out of different
Ž .
Ž . Ž . Ž .
cubes and examining selected plains in arbitrary direction.
5. Creation of a 3D proxy model for hydroge-ological properties
Difference visualization is extremely useful, but limited in that it does not give us informa-tion on the hydrogeological properties of the medium. However, there is obviously informa-tion on hydrogeological properties contained in the difference data. One way to extract this information is by using the change in subsurface reflectivities between different difference cubes. Doing so creates a proxy model for hydrogeo-logical properties, which is an amalgam of a number of different properties, but definitely not a model for one single property. To illus-trate this, the 500 MHz Borden data set is used. The 200 MHz data show essentially the same image, but provide less resolution compared to the 500-MHz data.
The difference cubes for the Borden data set allows us to visualize the change in subsurface reflectivity. This change is linked to the migra-tion of the DNAPL. In the Borden experiment, a constant hydraulic head was maintained so that the only driving force for the DNAPL migration is gravity, which makes the DNAPL sink. The parameters influencing the path of the DNAPL are a combination of connectivity and hydraulic conductivity or permeability. Thus, a model of where the DNAPL has gone can be extracted from the GPR data and provide an estimate for these combined parameters.
First, the 3D difference between the pre-spill
Ž
data and the data at repeated acquisition time 5,
.
14, . . . 340 h are taken. After that subtraction, two 4D cubes of data for each repeated acquisi-tion time, one for the eight North–South pro-files and one for the eight East–West propro-files
ŽFig. 4a are available. These cubes represented.
the optimum difference and so the optimum representation of where the DNAPL moved. These two cubes were merged into one cube.
Note that this is an inherently unsatisfying effort
Ž .
as the line spacing 1 m is very large compared
Ž .
to the trace spacing 5 cm . For certain times,
Ž .
this 4D cube will have high amplitudes at
places where the DNAPL is. However, as the DNAPL migrates, these places shift as a func-tion of time.
Our 3D proxy model volume was created by assigning, to each point in the 3D cube, the maximum amplitude value that was associated
Ž .
with that point in any of the 4D cubes Fig. 6 . This operation creates a map, which indicates ‘‘where and how much of the DNAPL passed through’’.
The net result of this effort is a 3D cube
Ž
which can be either visualized with AVS Fig.
.
7 or simply sliced to see the change in
hydro-Ž .
geological properties Fig. 8 . Note that both figures provide similar information. The red color represents areas with hydraulic properties that provide preferential pathways for the DNAPL. The blue color represents areas of hydraulic properties that are no pathways for the DNAPL during the experiment. Fig. 8a indi-cates the injection point of the DNAPL, which occurs at 70 cm depth. At subsequent depths, the spreading out and the distribution in the channels of the DNAPL can be observed. Fig. 7
Žwhich illustrates the same data shows both the.
Ž .
spreading in the channel Fig. 7a and the fur-ther descent of the DNAPL to the bottom of the
Ž
cell through a system of vertical channel Fig.
.
7b . This information conforms with
observa-Ž
tions from earlier analysis Sander, 1994;
Brew-.
ster and Annan, 1994; Brewster et al., 1995 — however, it was arrived at without any manual interpretation of any of the underlying data, and it illustrates the power of our processing and visualization effort in making sense of the data.
6. Conclusions and further work
explore the vast amount of information, a semi-automatic data processing and interpretation scheme is needed. Examples from the Borden data set show both the need for and the feasibil-ity of this approach. In addition to visualizing differences, we can also exploit these differ-ences to arrive at proxy models for hydrogeo-logical properties.
While these results are encouraging, there are still many unanswered questions associated with using 4D GPR methods in a field setting. Ques-tions on acquisition, data calibration, repeatabil-ity and survey frequency, as well as questions on what we really do observe and can extract in terms of subsurface fluid flow and hydrogeolog-ical properties need to be investigated. We hope to answer these questions in our ongoing re-search.
Acknowledgements
We would like to acknowledge the extremely fine and high-quality work that has been done by the University of Waterloo and others in-volved in the Borden Experiment in collecting and documenting this data set. We feel that this data set is a major asset for the geoscience community in the study of time-dependent fluid flow behavior. We will continue to use this data set for further analysis and testing. Further we like to thank Bill Symes and his colleagues at Rice University for making his 2D FDTD-TE code available to us. This research is supported by the EPA under grant number R 825209-01-0. Roelof Versteeg also acknowledges the support of the Schlumberger foundation.
References
Anderson, R., Cathles, L.M., Nelson, H.R., 1991. Data cube depicting fluid flow history in Gulf Coast
Sedi-Ž .
ments. Oil and Gas Journal 89 44 , 60–65.
Anderson, R., Flemings, P., Losh, S., Austin, J., Wood-hams, R., 1994. Gulf of Mexico growth fault drilled, seen as oil, gas migration pathway. Oil and Gas Journal
Ž .
94 23 , 97–104.
Bergmann, T., Robertsson, J.O.A., Holliger, K., 1996. Numerical properties of staggered finite-difference so-lutions of Maxwell’s equations for ground-penetrating radar modeling. Geophysical Research Letters 23, 45– 48.
Brewster, M.L., Annan, A.P., 1994. Ground penetrating monitoring of a controlled DNAPL release: 200 MHz
Ž .
radar. Geophysics 59 8 , 1211–1221.
Brewster, M.L., Annan, A.P., Greenhouse, J.P., Kueper, B.H., Olhoeft, G.R., Redman, J.D., Sander, K.A., 1995. Observed migration of a controlled DNAPL release by
Ž .
geophysical methods. Groundwater 33 6 , 977–987. Lane, J.W., Jr., Haeni, F.P., Placzek, G., Wright, D.L.,
1996. Use of borehole radar methods to detect a saline tracer in fractured crystalline bedrock at Mirror Lake, Grafton County, New Hampshire. GPR’96, Sixth Inter-national Conference on Ground Penetrating Radar, Sendai, Japan, pp. 185–190.
Pankow, J.F., Cherry, J.A., 1995. Dense Chlorinated Sol-vents and Other DNAPLS in Groundwater. Waterloo Press.
Robertsson, J.O.A., Blanch, J.O., Symes, W.W., 1994. Viscoelastic finite difference modeling. Geophysics 59, 1444–1456.
Sander, K.A., 1994. Characterization of DNAPL move-ment in saturated porous media using ground penetrat-ing radar. MSc, Colorado School of Mines.
Sander, K.A., Olhoeft, G.R., 1995. 500 MHz ground pene-trating radar data collected during an intentional spill of tetrachloroethylene at Canadian Forces Base Borden in 1991. USGS Digital Data Series DDS-25.
Xu, T., McMechan, G.A., 1997. GPR attenuation and its numerical simulation in 2.5 dimensions. Geophysics 62
