Reprinted from the Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP99 - Oakland, CA). Environmental and Engineering Geophysical Society, Wheat Ridge, CO, 723-732, 1999.

SURFACE AND BOREHOLE SEISMIC CHARACTERIZATION
OF THE BOISE HYDROGEOPHYSICAL RESEARCH SITE


Lee M. Liberty, William P. Clement, and Michael D. Knoll
Center for Geophysical Investigation of the Shallow Subsurface (CGISS)
Boise State University
1910 University Dr.
Boise, ID 83725

© The Environmental and Engineering Geophysics Society

ABSTRACT

We conducted borehole to borehole, borehole to surface, and surface seismic experiments to optimize data acquisition parameters, obtain a seismic velocity model, and to characterize seismic stratigraphic units in a shallow aquifer at the Boise Hydrogeophysical Research Site. The aquifer consists of coarse (cobble-and-sand) fluvial deposits underlain by clay at 18-21.5 m depth. We acquired data with a downhole seismic source (sparker), a sledge hammer source, a hydrophone string, a borehole geophone, and a surface geophone array to compare seismic signal quality and to place limitations on each seismic method. To fully characterize seismic reflections, the source-receiver geometry is an important parameter for both borehole and surface experiments. Direct arrivals and the presence of seismic reflections suggest a correlation between porosity changes and seismic velocities at the site. We have defined four seismic stratigraphic horizons that correlate with surface and borehole ground penetrating radar results, geophysical logs, and lithologic logs. These results provide an initial framework for hydrologic modeling.

INTRODUCTION

Imaging subsurface boundaries in the upper 20 m using seismic methods in a coarse-grained fluvial environment is often difficult. The Boise Hydrogeophysical Research Site (BHRS) is a research wellfield containing 18 cored boreholes (4 inch diameter) that extend through a coarse-grained cobble-and-sand aquifer and terminate in a clay unit (for additional site details, see Barrash and others, 1999; Clement and others, 1999a). Numerous tests are underway to thoroughly characterize the three-dimensional distribution of lithologic, hydrologic, and geophysical parameters at the site. The overall goal of the project is to develop methods for mapping variations in permeability by combining information from hydrologic and non-invasive geophysical techniques. Large-scale permeability changes often occur at lithologic boundaries, and seismic velocity contrasts often occur at these same boundaries. To investigate the seismic character at this site, we conducted a series of borehole and surface experiments with different source-receiver geometries.

Many seismic methods have been used to characterize shallow sites, including crosswell tomography (e.g., Hyndman and Harris, 1996), vertical seismic profiling (VSP) methods (e.g., Michaels, 1998; Milligan and others, 1997), and surface seismic reflection methods (e.g., Bachrach and Nur, 1998; Birkelo and others, 1987). Generally, crosswell seismic data are analyzed to generate a smoothed velocity map between boreholes based on first arrival picks. However, reflections in these data can identify discrete seismic boundaries. VSP data are often used to define an interval velocity profile at a borehole, but again reflections may also appear. Surface seismic reflection data is widely acquired to locate seismic boundaries across a site. Often these data suffer from poor resolution in the upper 20 m due to large velocity gradients (Miller and Xia, 1998) and interference from near-surface effects, including ground roll, guided waves, and the air wave (e.g., Steeples and Miller, 1998). At the BHRS, we integrated surface and borehole seismic methods to locate seismic boundaries and better constrain the hydrologic models. In particular, we show that an optimum window approach, similar to one used in surface seismic imaging (Hunter and others, 1984), is necessary in crosswell and VSP seismic surveys to best locate all reflections. We used the identified reflections to better constrain the results of our transmission experiments. The results of our study suggest that combined borehole and surface transmission and reflection seismic methods can provide valuable information about the structure and properties of the upper 20 m in coarse-grained sedimentary aquifers. We will use these results and survey methods to optimize the design of future seismic experiments.

CROSSWELL SURVEYS

The objectives of our crosswell seismic surveys are to estimate velocities from direct, first arrivals and to locate reflections that may appear after the first arrivals. We acquired several crosswell data sets using a downhole sparker source and a 36-channel, 0.5 m spaced hydrophone string to determine velocity gradients and seismic boundaries in the upper 20 m of the BHRS. The sparker source is a high frequency, broad band source (center frequency at 2000 Hz) that is highly reproducible and capable of propagating energy to distances greater than 20 m (Rechtien and others, 1993). We recorded each experiment with a 60-channel Geometrics RX 60, 24-bit seismograph at a 0.25 ms sample rate. Although the center frequency of the sparker source is above the recording limits of our seismograph (2000 Hz Nyquist frequency), the antialias filter allowed us to record high quality data below 1400 Hz.

Figure 1 shows crosswell seismic results from wells A1-C2, a well pair separated by approximately 8 m. Distinct direct arrivals appear in the data from all downhole, source-receiver pairs (traces) below the water table, thus providing a good data set for tomographic inversion (e.g., Clement and others, 1999b). To provide an initial velocity model for analysis, we extracted a level run from the tomography data set. A level run is compiled by selecting common depth source-receiver pairs. We picked first arrival times and then calculated the velocity model by dividing the distance between the source-receiver pairs by the picked arrival time. The water table depth was 2.4 m below land surface at the time of these experiments. We did not process source-receiver pairs in the unsaturated zone due to poor source and receiver coupling. Also, we did not measure velocities in the basal clay due to lack of ray coverage. Results from the level run (Figure 1) show interval velocities between 2400-2900 m/s with at least two distinct low velocity intervals (horizons A and B) at 7.0 m and 13.5 m depth. These low velocity intervals match regions where relatively porous cobble dominated intervals occur in each well. Above and below these two intervals, relatively tight (less porous) cobble zones are observed.

Figure 1 shows a comparison between the 1-D level run velocity model and porosity estimates derived from well logging (neutron logs). We see correlations between high velocity intervals and low porosity values, and between low velocity intervals and high porosity values, as expected (e.g., Marion and others, 1992). Although the porosity logs from wells A1 and C2 are not identical, both logs correlate inversely with the level run velocity model. The correlation between porosity and velocity and the variation in porosity between the two boreholes suggest that a 2-D velocity model may better fit the seismic data between boreholes A1 and C2, and that the BHRS is likely not uniform in seismic velocities across the site.

To test the accuracy of the initial velocity model, we compared the observed seismograms with an acoustic finite difference model (Figure 1). The synthetic seismograms were generated using the level run velocity model and a minimum phase-shifted Ricker wavelet with the same observed frequency (1200 Hz) as the data. The first arrival times from the synthetic data closely match the observed arrival times, but small travel time differences do occur. Two-dimensional velocity effects probably account for these slight time differences. Although not presented here, we can use the derived 1-D velocity model and first arrival picks to calculate a 2-D velocity model using a variety of inverse techniques (e.g., Aldridge and Oldenburg, 1993; Clement and others, 1999b) and perhaps predict inter-well porosity values from these data.

The velocity model suggests that two distinct low velocity intervals appear between wells A1-C2. However, assumptions of straight ray paths and first-arrival picking uncertainties can artificially smooth the velocity model or lead to incorrect velocity values. To confirm the existence of distinct seismic stratigraphic layers, we search for reflections in the seismic data. To confidently identify reflections using crosswell seismic methods, we must select the source-receiver geometry that allows us to distinguish reflections from energy that arrives parallel to the direct arrival energy. Therefore, we must examine a range of shot and receiver depths.

Four seismic reflection packages are identified in both the observed and calculated gathers (Figure 1). Arrivals are associated with the water table at 2.4 m depth, two low-velocity intervals (horizons A and B), and the basal clay unit at roughly 19 m depth. This latest arriving reflection from the clay is not distinct from the first arrival energy on the record from the shot at 18.3 m depth. However, as the shot depth decreases, the upcoming reflected energy separates from the first arrivals and the reflection from the basal clay is observed. Also, as the shot depth decreases, we see both upcoming and downgoing reflection paths, thus increasing the number of raypaths (while also increasing interferences) that we can model. When the seismic source is near the top of the aquifer (3.3 m depth), only the downgoing arrivals appear. A change in apparent frequency is also observed as the shot decreases in depth. Constructive interference can, in part, explain this effect; as the shot depth decreases, the downgoing reflections from the large velocity gradient at the water table merge with the first arrival energy.

Reflections more clearly separate in time when the source is near the aquifer base and energy reflects downward from the overlying horizons. When the source is near the aquifer base, the unsaturated zone and upcoming seismic energy least influence the reflected arrivals. This suggests that the optimum source-receiver geometry for identifying reflections measurable above the aquifer base (more than a few traces) is when the source is deep in the borehole and the geophones are at and above this depth in the adjacent borehole. But, if a seismic horizon is near the base of the borehole, we must change the experiment to image the reflections from the top of the seismic unit. It should also be noted that tube wave energy can affect the optimum reflection window. By observing shot records from many depths, we can find the optimum shot depth to image each reflection package and better define the seismic facies.

Seismic reflections appear on both the observed shots and calculated models where velocity gradients exceed a few hundred m/s and at interfaces greater than 2 m thick. Separating reflections between the top and bottom of the two low-velocity units is difficult (Figure 1, horizons A and B) because the recorded frequencies and the high velocities limit our resolution. If we could record higher frequencies using a seismograph with a faster sample rate, these reflections may more clearly separate in time.

We reproduced the crosswell seismic experiment in different boreholes (B1-C1) separated by 5.1 m using a single, clamped, 3-component, 14-Hz geophone to determine if pressure sensitive hydrophones respond differently than velocity sensitive geophones (Krohn and Chen, 1992). Shot records from the saturated zone contain arrivals similar to the crosswell data acquired with the hydrophone string, including distinct direct arrivals and the four reflections (from the water table, horizons A and B, and the basal clay). The frequency content of the geophone and hydrophone located at the same depth is similar, but large amplitude tube wave arrivals appear at later times with the clamped borehole geophone (Figure 2). An obvious difference between the hydrophone and the borehole clamped geophone data is the change in polarity of the direct arrival from energy originating from above the geophone compared to energy arriving from below the geophone. This response is expected from a velocity sensor, where first motion energy arriving from above the receiver responds with a positive voltage change and first motion energy arriving from below the geophone responds with a negative voltage change. When the clamped geophone is above the water table, the signal quality deteriorates (Figure 2). The most notable changes are the "ringy" nature of the recorded signal and the lower apparent center frequency which decreases from approximately 1000-1400 Hz when the source-geophone pairs are at 10-18 m depth, to approximately 350-500 Hz in the unsaturated zone. Although the frequencies do appear to decrease when the source and receiver are just below the water table, the most notable change occurs when the source or receiver is located above the water table.

VSP SURVEYS

We calculated interval velocities and located reflections by acquiring both VSP and reverse VSP (RVSP) data sets. We acquired the RVSP data with the sparker seismic source and surface geophones (buried 0.2 m). With this configuration, we can examine seismic velocities from the unsaturated zone by measuring delays in travel time between the borehole clamped geophone and the surface geophone (Figure 2). Reflections are difficult to distinguish in RVSP data because the wavefield energy does not separate, as we observed in the crosswell seismic data. A travel-time difference of 6-7 ms is noted between the geophone clamped at 3 m depth in well C1 and the surface geophone adjacent to well C1. Although the raypaths are not identical, we estimate an average velocity of 400-450 m/s for the unsaturated zone.

We acquired a standard VSP data set (Figure 3) at the BHRS using a sledge hammer source at the surface and a 36-channel hydrophone string. We measured interval velocities within each borehole by subtracting the difference in travel time between first arrival times for adjacent geophone pairs. We acquired three offset VSP data sets in borehole B1 to compare the calculated interval velocities to the crosswell seismic velocities and also to determine if reflections appear in the VSP data. A low amplitude first arrival is observed on shallow receivers in the 0.35 m offset VSP. This phase is not observed in the 1 m and 2 m offset VSP data set. We suspect the near-vertical incidence VSP records this slower arrival (velocity of 750-850 m/s) due to the near-vertical radiation pattern of the sledge hammer source. For the 1 m and 2 m offset VSPs, arrival times from the water table (2.4 m) appear before the arrival times from shallower and deeper hydrophone depths. Arrivals appear first at the water table because travel times from head wave arrivals are faster than direct arrivals, thus providing a reverse travel time for the sensors above the water table (energy is arriving first from below the sensor). A slight delay in travel time for the 2 m offset VSP appears and is likely due to a trigger delay in the recording seismograph.

To examine reflections in the VSP data, we flattened the first arrivals on all VSP records, normalized the amplitudes of each trace to a window surrounding the first break arrival, removed this first-arrival energy with a median filter, then unflattened the data. These steps remove the downgoing energy that is parallel to the first arrival (Figure 3b). A number of reflections are more obvious after removing the downgoing energy. On all three records, a coherent reflection is observed near the bottom of the borehole. Although tube waves are often generated from the bottom of the borehole (Milligan and others, 1997), the interpreted depth of the arrival correlates with the basal clay depth as well as the results from the borehole to borehole studies. Additional reflections appear on the 1 m and 2 m offset VSP data (at 13.5 m depth) that are not observed in the vertical incidence record. We suspect tube waves interfere on the normal incidence gather. As the source moves further away from the borehole, the tube wave energy arrives at later times. We do not observe a distinct reflection from the upper low velocity interval defined from the crosswell seismic data (horizon A, Figure 1) likely due to insufficient spatial coverage to confidently identify the arrival. We might have been able to resolve this horizon better if we had acquired data with more dense spatial coverage. As observed in the crosswell reflection data, we can alter the source and receiver geometry to optimize signal quality and better locate the observed reflection packages.

SURFACE SEISMIC SURVEYS

A surface seismic experiment is often the least expensive seismic method to deploy at a site, especially a site without well control. The problem with surface seismic data in coarse-grained environments is often the energy trapped in the unsaturated zone due to the large velocity contrast between the unsaturated (250-700 m/s) and saturated zones (2500-3000 m/s). We acquired a surface seismic data set with a sledge hammer source and twenty-four 10 Hz buried geophones to identify marker seismic horizons without the aid of a borehole. Coherent energy appears on these records (Figure 4), however, reflection energy does not clearly separate from other coherent phases on this shot record. We generated a synthetic seismogram (Figure 4) using an acoustic finite difference routine to compare with the observed results. We used a velocity model derived from the crosswell, VSP, and surface seismic data. Results show a large amplitude reflection and multiple from the water table, and additional low-amplitude, subhorizontal reflections that correlate with the two intermediate cobble units, and the top of the clay. Confident identification of reflections below the water table is difficult on the observed and synthetic data. When surface wave energy is added to the modeled data, the optimum window (Hunter and others, 1984) for identifying reflections is greatly reduced to the region between the direct and refracted arrivals.

As we have shown, we can record coherent first arrival energy with both borehole and surface seismic sources and with both borehole and surface geophones. When we place both the source and receivers at the surface, though, large amplitude surface waves, refractions, and wave guide effects interfere with the reflections. Even standard refraction data are difficult to analyze because the air wave arrives before the direct or refracted arrivals, thus decreasing the confidence of correlating the appropriate phases. Also, VSP results showed that source energy radiation patterns may generate weak direct arrivals at locations offset from the shot. Seismic reflections may appear in the surface seismic data, but imaging these reflecting horizons with confidence in a processed, stacked seismic section is difficult due to the small, true-fold window (Liberty and Knoll, 1998).

CONCLUSIONS

We have successfully acquired seismic data at the BHRS in a shallow cobble-and-sand fluvial aquifer. The seismic data correlate with lithologic logs, geophysical logs, and ground penetrating radar results (Peterson and others, 1999). These results provide an initial framework for hydrologic modeling. We have identified four reflectors that are associated with the water table, two intermediate cobble units, and the top of the clay. Both first arrivals and reflections from crosswell data can resolve seismic boundaries that are greater than 2 m apart. Crosswell and VSP first arrivals provide velocity values while the reflections locate seismic velocity boundaries at this site. Analysis of seismic methods suggest:

1) The source-receiver geometry in crosswell and VSP reflection acquisition is critical to identify reflecting boundaries.

2) Radiation patterns may affect the ability of surface and borehole sensors to record near-surface direct energy.

3) To accurately calculate near-surface velocities using VSP methods, close source-to-borehole offsets are necessary, but offset VSP data better distinguish tube wave energy from the reflected energy.

4) Although resolution (frequency) decreases with VSP methods compared to crosswell methods, we can still identify reflections at significant velocity boundaries with wavefield separation methods.

5) Surface seismic data analysis indicates that having sources and receivers above the water table often creates large amplitude coherent arrivals that interfere with reflection and refraction energy. Although reflections may appear on the surface seismic records, confidence in mapping each boundary is reduced.

ACKNOWLEDGEMENTS

This project is supported by U. S. Army Research Office grant DAAH04-96-1-0318. Cooperative arrangements with the Idaho Transportation Department, the U. S. Bureau of Reclamation, and Ada County allow development and use of the BHRS, and are gratefully acknowledged. We also gratefully acknowledge the use of instrumentation supplied by a grant from the M. J. Murdock Charitable Trust. Contribution no. 0087 of the Center for Geophysical Investigation of the Shallow Subsurface at Boise State University.

REFERENCES

Aldridge, D. F., and D. W. Oldenburg, 1993, Two-dimensional tomographic inversion with finite-difference traveltimes: Journal of Seismic Exploration, v. 2, p. 257-274.

Bachrach, R. and Nur, A, 1998, High-resolution shallow seismic experiments in sand, Part 1: Water table, fluid flow, and saturation: Geophysics, v. 63, p. 1225-1233.

Barrash, W., T. Clemo, and M. D. Knoll, 1999, Boise Hydrogeophysical Research Site (BHRS): Objectives, design, initial results: Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP99-Oakland, CA), Environmental and Engineering Geophysical Society, Wheat Ridge, CO, submitted.

Birkelo, B.A., Steeples, D.W., Miller, R.D., and Sophocleous, M.A., 1987, Seismic reflection study of a shallow aquifer during a pumping test: Ground Water, v.25, p. 703-709.

Clement, W. P., M. D. Knoll, L. M. Liberty, P. R. Donaldson, W. Barrash, and J. R. Pelton, 1999a, Geophysical surveys across the Boise Hydrogeophysical Research Site to determine geophysical parameters of a shallow, alluvial aquifer: Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP99-Oakland, CA), Environmental and Engineering Geophysical Society, Wheat Ridge, CO, submitted.

Clement, W. P., L. M. Liberty, and M. D. Knoll, 1999b, Reverse VSP's and crosshole seismic tomography while coring: Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP99-Oakland, CA), Environmental and Engineering Geophysical Society, Wheat Ridge, CO, submitted.

Hunter, J.A., Pullan, S.E., Burns, R.A., Gagne, R.M., and Good, R.S., 1984, Shallow seismic reflection mapping of the overburden-bedrock interface with the engineering seismograph-some simple techniques: Geophysics, v. 49, p. 1381-1385.

Hyndman, D. W., and Harris, J. M., 1996, Traveltime inversion for the geometry of aquifer lithologies: Geophysics, v. 61, p. 1728-1737.

Krohn, C.E. and Chen, S.T., 1992, Comparisons of downhole geophones and hydrophones: Geophysics, v. 57, p. 841-847.

Liberty, L.M. and Knoll, M., 1998, Time-varying fold in stacked seismic reflection data: a new quality control procedure for shallow high-resolution applications: Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP98-Chicago, IL), Environmental and Engineering Geophysical Society, Wheat Ridge, CO, p. 745-751.

Marion, D., Nur, A., Yin, H., and Han, D., 1992, Compressional velocity and porosity in sand-clay mixtures: Geophysics, v. 57, p. 554-563.

Michaels, P., 1998, In situ determination of soil stiffness and damping: Journal of Geotechnical and geoenvironmental engineering, p. 709-719.

Miller, R.D. and Xia, J., 1998, Large near-surface velocity gradients on shallow seismic reflection data: v. 63, p. 1348-1356.

Milligan, P. A., J. W. Rector III, and R. W. Bainer, 1997, Hydrophone VSP imaging at a shallow site: Geophysics, v. 62, p. 842-852.

Peterson, J. E. Jr., E. L. Majer, and M. D. Knoll, 1999, Hydrogeological property estimation using tomographic data at the Boise Hydrogeophysical Research Site: Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP99-Oakland, CA), Environmental and Engineering Geophysical Society, Wheat Ridge, CO, submitted.

Rechtien, R.D., Hambacker, K.L., and Ballard, R.F., 1993, A high frequency sparker source for the borehole environment: Geophysics, v. 58, no. 5, p. 660-669.

Steeples, D.W. and Miller, R.D., 1998, Avoiding pitfalls in shallow seismic reflection surveys: Geophysics, v. 63, p. 1213-1224.