WORK PLAN : 3-D Subsurface Seismic Characterization of Environmental Remediation Site OU2, Hill Air Force Base, Utah

Alan Levander and Colin Zelt

Department of Geology & Geophysics MS-126, Rice University, Houston, TX 77251-1892

713-527-6064 / 713-527-4757

713-285-5214 (fax)


This project is funded by Department of Energy to conduct a proof of concept 3-D high resolution seismic survey at Operable Unit 2 (OU2), an active environmental remediation site at Hill Air Force base, Utah. The experiment is to be done with Health and Safety oversight provided by Duke Engineering and Services, and in cooperation with chemical engineers from Rice University (Hirasaki and Miller, Principal Investigators for the AATDF Surfactant/Foam Flood project conducted at OU2). The lack of adequate aquifer structure characterization resulted in a well pattern for the Surfactant/Foam Flood test that was not optimum. Interpretation of the test results speculates on the aquifer structure beyond the pattern. Duke Engineering is now engaged in remediation efforts including an extensive well field (~40 wells). Two-dimensional seismic characterization of this site by Rice geophysicists in 1998 identified the buried paleo-channel topography in the shallow clay layer and showed that it varies in structure over short spatial scales. The bottom of the paleochannel in the clay layer acts as a collection point for DNAPLs (in particular, trichloroethylene, TCE). The low spots in the paleo-channel were not known with enough precision to locate the well field so that all of the contaminants could be treated. The 3-D seismic survey is designed to map the characteristics of the buried paleo-channel in three dimensions to aid future remediation efforts. The 3-D survey is a feasibility study to determine whether conventional high resolution and advanced seismic imaging techniques can be used to characterize this and other very shallow environmental cleanup sites with the resolution required to sensibly locate a well field. The survey data will be used as one of the testbed datasets for advanced seismic processing methods being developed by the PIs and another Rice professor (Symes, Computational and Applied Math), which have been funded by DOE (October, 1997).


OU2 at Hill Air Force Base in north central Utah is and has been the target of cleanup efforts to remove DNAPLs from the shallow subsurface. The DNAPLs were originally disposed of in two trenches, and have subsequently descended through the porous surface fill, coming to rest in low spots of a paleo-channel at the top of an impermeable clay layer, at a depth of approximately 8-16m depth (Figure 1) . Geologically the surface material consists of porous and permeable Pleistocene silts, sands, gravels, and clays which comprise the Provo formation, and which overlie the Pleistocene Alpine formation, a porous, but largely impermeable clay, silt, and sand layer. The site is over a "geologically complex slump feature in which distinct variations in geology and hydrologically occur within tens of feet both areally and vertically1". There are three water tables in the area, the most shallow of which is at about 25 feet depth, lies above the Alpine formation, and is recharged by surface runoff. The two deeper water tables (over 300 feet below the ground surface) are used for residential and industrial water sources. Contaminants from the shallowest layer have been found in local streams and seep springs, which has led the Air Force to trench around the waste disposal site and install an impermeable bentonite barrier to contain the waste.

The contaminant collection points are in structural lows found along a paleo-channel cut in the Alpine formation which coincidentally is below the two trenches used for waste disposal. The structural lows in the Alpine formation should be mapped in three-dimensions with a resolution of 1.0 m or finer to provide sufficient data on potential DNAPL pools. The Air Force would also like to have the location of the two waste disposal trenches identified. The site was investigated with a ground penetrating radar (GPR) survey in August and September 1997, and with 2-D seismic profiling in July 1998 by Rice University geophysicists. The proof of concept 3-D seismic survey will complement the previous investigations, greatly expanding our knowledge of the subsurface and addressing the issue of why test remediation wells were unable to completely rid the subsurface of TCEs.

Previously a number of geological, geotechnical, and geophysical investigations have been undertaken at the site, including hydrologic properties measurements, and resistivity, electromagnetic conductivity, self potential, ground penetrating radar, seismic refraction surveys, and the 1998 seismic reflection/refraction survey. The latter is of most interest as background for this proposal.

We visited Hill AFB in October 1997 to review the site for suitability for seismic operations and to ascertain what permitting and training requirements needed to be met to allow us access to work on the Base. Unfortunately the onset of winter precluded field operations in 1997 and early 1998. We conducted the 2-D seismic experiment in a one week period in August, 1998 (Figure 2).

Unlike the previous seismic investigations at OU2, we found that near surface velocities were very low, and are generally less than the air velocity (330 m/s). Low velocities appear to extend to the top of the clay layer, but are poorly constrained between about 4m depth and the base of the alluvium on top of the clay layer. Three refraction models are presented in our Interim Report to DOE (June, 1999) with velocities of 230-300 m/s at the surface. Velocity at the base of the surface layer is poorly constrained and we have tested velocities of ~350m/s, ~700m/s, or 900m/s. The velocity at the top of the substrata could be 1,180m/s to 1,500 m/s, depending on the velocity of the overburden.


The seismic survey we will conduct will image the base of the aquifer with a resolution of about 0.5 meters vertically and laterally. It is unlikely that seismic measurements will image DNAPLs directly, however, they should be capable of mapping out small scale variations (~0.5 m) in depth and lateral extent of the base of the aquifer, and therefore should be useful for identifying locations at which DNAPLs will pond. The 3-D measurements will provide an exact subsurface description of the acoustic impedance contrast at the overburden-clay boundary, eliminating the need for and inaccuracy in kriging between well-bore samples.

Work Plan

We are planning a 3-D high resolution near-vertical to wide-angle seismic reflection swath over the OU2 site at Hill AFB. The swath will be ~40 m in width, crossing the known location of the channel (Figure 1). In three weeks time we should be able to conduct a survey covering an area of 80 by 40m2 to 120 by 40m2, which will be sampled with ~.5m resolution. We are using a practical 3-D experiment design which we have modified to take account our manpower resources, and our need for wide-angle measurements for advanced processing. The experiment was designed in conjunction with the Land Acquisition Group of Western Geophysical.

We would also like to drill a well in the area outside of the contaminated region to conduct a "checkshot" survey to get reliable velocities below 4m depth. We do not have funds for drilling a well, but we can arrange for instrumentation for the check shot survey if Hill AFB is willing to pay for the cost of the drilling.


We will have sixteen to twenty people in the field. Six people will be assigned to 4 seismic sources, two each to each of the two cabled seismographs, 4 to the individual seismographs, and 1-5 to data downloading and processing. The project will be managed by Alan Levander and/or Colin Zelt, both professors at Rice University who have considerable experience with this type of surveying. A number of other personnel also have seismic field experience (Igor Morozov, Diana Dana, Kara Hackwith from Rice, and a number of UT El Paso staff and students). The remaining people are Rice graduate or undergraduate students or students from other schools. Inexperienced people will be paired with more experienced members of the field crew.

Instrumentation & Seismic Sources

We will employ two or three portable cabled seismograph systems recording with high frequency geophones, and 400 individual seismographs planted at 35cm intervals along each profile. Two of the cabled systems are owned by the PASSCAL program of IRIS, and are 60 channel recorders. We are discussing adding a 96 channel system from the University of Wyoming. The 400 individual seismographs are owned by the University of Texas at El Paso, and are loaned to the PASSCAL instrument pool. Our 3-D recording swath will thus consist of 520 to 616 vertical component channels, depending on involvement by Wyoming.

Source points will be at 35cm intervals, yielding 20 fold data. In 1998 we tested three different sources to determine which provides the best quality data in terms of frequency (resolution) and energy: (1) a portable weight drop system, (2) a single shot .22-caliber rifle shooting .22 short and .22 magnum bullets, and (3) a Seisgun source which is a modified 8 gauge shotgun which shoots black powder blanks. We found that the .22 magnum rounds produced the best data for our purposes. This source has a frequency spectrum from about 100-300Hz, and has good signal penetration

We propose to use 4 of the .22 magnum rifles. They will be fired into small holes about 3 inches deep. The holes are made with a steel rod and hammer.

One of us (AL) has examined soil samples from the Provo and Alpine formations from OU2 available at the Chemical Engineering Department at Rice. Visual examination leads us to believe that these two distinct lithologies should be adequate to produce a sizable seismic reflection signal, i.e., the Provo and Alpine lithologies appear to be different enough in mechanical properties and mass density to provide an adequate impedance contrast to produce reflection signals. Another likely seismic target in the shallow subsurface is the top of the water table in the Provo formation. These conclusions were validated by excellent reflection signals in the 1998 2-D survey (Figure 2).

To adequately characterize the subsurface for remediation efforts at this site requires a resolution of about 0.33 m. Seismic resolution laterally and vertically of l/4 to l/2 can be achieved in a properly processed seismic section, where l is the wavelength at the center of the seismic band. Assuming a target velocity of 350 m/s (i.e. the mean surface layer velocity), requires a source with a center frequency of 350 Hz. Although this is rather high, we demonstrated that we can generate signals with a frequency in the 100-300Hz range with the .22 magnum to image shallow structures. Resolution then will be about 0.44m, somewhat worse than the desired value. There is not much that we can do about this with the sources available to us at present. Since the vertical travel path to the target for a 200Hz signal is only about 12 wavelengths roundtrip, the signal from a source directly over the target is partly obscured by surface borne noise fields (Rayleigh and acoustic waveguide modes). We shot a broad range of source-receiver offsets to spatially separate the noise and reflecting wavefields and have processed the 1998 data to remove the effects of non zero source-receiver offset (Figure 2).


The seismic lines will be spatially referenced to existing AATDF wells in the OU2 DNAPL source zone and other base reference points already surveyed (see accompanying figures). For 3-D experiments accurate navigation data is essential, and we are trying to arrange for a commercial surveying crew to obtain GPS locations of survey reference points. An alternative source for the GPS locations would be from the Air Force or Duke Engineering.


The seismic crew will not be handling hazardous chemicals, or subsurface contaminants. We operate from the surface using noninvasive remote sensing technologies. Geophones have spikes which are driven 4 inches into the ground, and the seismic sources require small holes 3 to 6 inches deep, but otherwise the seismic acquisition is completely noninvasive.

The only potentially dangerous aspect to the seismic program are the .22 magnum rifle seismic sources. Being sensitive to this problem we minimize the danger by assigning four qualified and experienced people to handle the 4 rifles, 1 person to handle the ammunition for the rifles, and 1 person to command the source group. No other members of the field crew touch the sources. The firearms are kept unloaded until the time interval in which they are fired, with the ends of the barrels positioned into the hole in the ground. The ammunition handler provides a single round to each gun handler. The source commander issues the order to fire. The .22 caliber sources are bolt-action rifles from which we remove the bolt except when the guns are in use. The source commander keeps the bolts when the guns are not in use, separating the guns from their firing mechanisms.


We will have a seismic processing package in the field with us which is capable of producing first order analysis and interpretations in the field. We intend to have preliminary 2-D interpretations available by the time we leave Hill AFB, although 3-D processing and final interpretations will require considerably longer. HAFB will receive three copies of the final report when it is completed and retains the right to review and edit any historical information related to OU2. Rice University will send two copies of the final report to EPA VIII (Sandra Bourgeois) and UDEQ (Mo Slam).


We are constrained to perform the field work at HILL AFB from July 6 to July 31, 1999; or at specific time later in the summer or fall in 1999, because of instrument scheduling conflicts. The first day will be used for orientation, on-site training and testing of the sources and other equipment. The remaining time will be used to collect the seismic data in the 3-D corridor and prepare a preliminary interpretation as mentioned above.

Figure 1. Proposed location of 3-D seismic survey. White box shows the proposed 3-D survey area. Solid black lines are 1999 2-D profiles. Triangles are wells.

Figure 2. Seismic profiles from 1998 experiment A) Line 1 with interpreted channel shown in green. B) Line 2 with interpreted paleo-channel shown in green. (Compare to Figure 3).

Figure 3. Seismic velocity models from 1998 experiment for Line 2. The boundary between layers 1 and 2 for 9 different models of the line 2 shotgun data. Three colors indicate velocities assumed for the base of layer 2; solid line indicates that a laterally varying velocity at the top of layers 1 and 2 was determined; dotted line indicates that a laterally varying velocity at the top of layer 1 was determined but a velocity of 1500 m/s at the top of layer 2 was assumed; dashed line indicates that the best constant velocity was determined for the top of layer 1 (the surface) and a velocity of 1500 m/s at the top of layer 2 was assumed. All models were obtained using a regularized inversion with flatness constraints so that the lateral velocity variations and boundary topography would be as flat (constant) as possible.