Experimental Observation of the Seismic Effect of Variations in the Thickness of the Water-Table Waveguide

Publication Date


Type of Culminating Activity


Degree Title

Master of Science in Geophysics



Major Advisor

John R. Pelton


In order to improve the clarity of seismic reflection images of the upper few decameters of the Earth's crust, it is first necessary to document the characteristics of wave phenomena which interfere with reflections from shallow interfaces. The focus of this study is on guided waves which propagate in the waveguide defined between the Earth's surface and the water table. Guided waves in the water-table waveguide may dominate the optimum window on high-resolution reflection seismograms in some geological environments and cause severe interference with shallow reflections. The objective here is to deduce the importance of guided waves relative to other sources of interference through the results of a simple field experiment. Repeated fixed-source multichannel seismic records (noise tests) were acquired over a sixteen-month period (November 1990 through March 1992) in an area of southwest Boise with a seasonal variation of water-table depth (historical evidence suggests a typical variation from 1-2 m in the autumn to 8-10 m in the spring). Temporal variations in the noise tests were expected as the water table dropped and the waveguide thickened, and these were compared with predictions from an appropriate model for wave propagation in a water-table waveguide.

The Mossy Cup experiment (named after a nearby street) was conducted in a flat field underlain by fluvial and lacustrine sediments to a depth of at least 150 m. Considerable effort was expended to constrain all identifiable acquisition parameters so that temporal variations in the noise tests could be attributed to changes in subsurface conditions. In particular, all geophone stations were fixed in concrete and each station housed the same geophone throughout the experiment. A custom downhole seismic gun was designed, constructed, and tested to fire electrically detonated shotgun shells under water within a steel-cased borehole. Source loads were obtained commercially and remained constant to within the loading tolerances maintained by the manufacturer. Proper utilization of the new seismic gun required control devices to center the gun inside the borehole and to consistently place the chamber at the same depth. Although these devices were not finalized until the final six months of the experiment, the data for that period still span a substantial change in water-table depth (4.2 m to 6.8 m) and were used exclusively in the final interpretation.

Each Mossy Cup noise test contains all of the basic features expected of a high-resolution reflection seismogram recorded in the Boise River valley: subdued Rayleigh-type surface waves, ground-coupled air waves (shallow source test only), and reflections from moderate depths (60-200 m). As is typically the case for the Boise area, waveform complexities within the optimum window at early arrival times preclude the identification of individual reflections originating from shallow depths. Of particular interest within the optimum window is a band of seismic energy of 30-ms duration which follows the first arrival and shows a steady increase in frequency with increasing offset (peak: frequency near 150 Hz at a source-receiver distance of 14 m and near 300 Hz at 124 m). Three mechanisms are proposed to account for this band: (1) guided waves propagating in a water-table waveguide with lateral variations in thickness and/or velocity, (2) dispersion of guided waves, and (3) scattering from near-surface heterogeneities. Seismic refraction data acquired at the Mossy Cup site suggest that the water-table waveguide thins in the same direction as the observed increase in frequency, thereby supporting the first possibility. However, there is insufficient evidence to rule out any of the three proposed mechanisms. so that all three should be considered viable explanations of the variable frequency band following the first arrival.

For the purposes of this thesis, the simplest realistic model of the water-table waveguide is a uniform elastic layer (unsaturated sediments) over a uniform elastic half space (saturated sediments), with a substantial P-velocity contrast (at least 2: 1) between the layer and the half space. As the model waveguide thickens, pulses corresponding to upgoing reflections from the bottom of the waveguide are spaced at greater intervals, resulting in pulse separation which should be an easily identifiable waveform characteristic to support the occurrence of guided waves in real data. Examination of the Mossy Cup noise tests in both the time and frequency domains does indicate significant temporal variations, but these may be the result of changes in near-surface conditions (such as moisture content in the soil), and there is no consistent evidence for pulse separation. Unfortunately, the lack of observed pulse separation does not mean that guided waves in the water-table waveguide were not generated at the Mossy Cup site. The difficulty stems from source-induced vibration of the steel casing in the source well. The vibration caused the source signature to be much longer than expected (40-ms duration), and this could have easily masked the evidence for guided waves. In summary, the Mossy Cup experiment was not successful in determining the relative importance of guided waves in causing the observed waveform complexities in the optimum window. However, the experiment did establish the proper methodology and much of the infrastructure for future waveguide studies. Of particular importance is the new electrical downhole seismic gun developed for the Mossy Cup experiment which has proven to be an inexpensive, reliable, and versatile impulsive source for shallow seismic studies using a variety of wet and dry boreholes (the new source is described in great detail in the Appendix).

Files over 30MB may be slow to open. For best results, right-click and select "save as..."