Kansas Geological Survey, Open-file Report 2005-56
by
Richard D. Miller and Julian Ivanov
KGS Open-file Report 2005-56
Final Report to
U.S. Army Engineer R&D Center
Geotechnical and Structure Laboratory
Vicksburg, Mississippi
Oct. 10, 2005
This applied research project evaluated the potential of a variety of seismic methods to characterize the condition of levee cores constructed in the 1970s as part of the International Boundary and Water Commission (IBWC) program in south Texas. Preliminary studies of levee cores in certain areas uncovered evidence of cracking in the expansive clays locally mined and used during construction of the core. Cracking of this nature is likely the result of more than eleven years of drought in south Texas and would increase the overall permeability and leak potential of the levees. This suggestion was made based on analysis of four different data sets: abnormally low conductivity determined by both airborne and surface geophysical surveys, abnormally high levels of grout intake during borehole plugging operations, and core samples intact when first removed from the ground and placed in plastic containment vessels showing marked shrinkage and visible cracking after one year in controlled storage.
Five levee sites were selected based on airborne geophysics and physical inspection to represent the range of conditions expected in levee cores during extended periods of drought in this area of south Texas. Lithology at each of these sites varied in sand and clay concentrations and types. Core materials for each levee site were locally mined at various locations within the river valley and therefore each possessed different physical properties as evident in core drill samples and electrical properties. Miles of airborne EM and LIDAR acquired in a continuous fashion over the levees in this area were instrumental in identifying and classifying each of these five very diverse sites.
Seismic methods have proven marginally successful identifying anomalies in levees on a few occasions. Most of these studies have focused on direct wave analysis, targeting areas with reduced seismic velocities. Lower seismic velocities are usually indicative of less strength or softer materials. Therefore, anomalously low velocities for a particular levee could be an early indicator of failure potential. Testing at each of these five sites was more extensive than any earthen structure study currently available in the scientific literature. The testing included compressional and shear first-arrival analysis (classic refraction, turning-ray tomography, and through-levee tomography), multi-channel surface-wave analysis, and vibration harmonics analysis. Tests were conducted both on the levee crest and at equivalent locations along the levee toe, with expanded studies at sites identified as good candidates for ponding experiments.
Tests were designed to evaluate both body waves and surface waves using well-documented methodologies specifically adapted to the levee problem. Due to the shallow depths of investigation, reflection was not considered a viable technique and therefore tests specifically designed to evaluate reflected arrivals were not undertaken. Seismic data were recorded using both horizontally polarized source and receivers and vertical source and receivers. Shots for the 2-D surveys were recorded at stations along the lines of receivers. A 3-D tomography experiment was conducted using shots on one side of the levee face recorded by receivers on the adjacent side. Data quality was method dependent, but in general most recorded data were good, possessing excellent signal-to-noise ratios and good-to-poor signal bandwidth and range of recorded frequency. Seismic velocities (compressional and shear) were estimated from measurements of first-arrival time/offset distances and inversion of surface-wave phase velocities as a function of frequency.
These investigations targeted seismic velocities, both absolute and relative (changes). Seismic velocities of levee materials were estimated and compared both site to site and within specific sites. A unique study of surface-wave phase velocities was conducted observing phase variations in the expected (for consistent material characteristics) uniform wavetrain at and near resonance (resonance in this case is controlled by levee height and surface-wave velocity of the materials: wavelength). This surface-wave study was conducted in hopes of identifying anomalous zones where changes in phase velocity might be indicative of reduced or increased material strength. Seismic velocities were measured based on travel time between adjacent sets of receivers.
Body-wave propagation characteristics are unique to the material through which the seismic energy is traveling. Shear velocity is generally accepted as a relative measure of material strength or stiffness. Compressional velocity is a measure of both the rock matrix and pore materials. Therefore, increases in shear velocity will generally indicate stronger materials, while increases in compressional velocity in unconsolidated materials is a good indicator of increased saturation.
Compressional-wave velocities were for the most part within a "reasonable" range for this setting; however, shear-wave velocities were estimated to be significantly higher than expected based on both levee materials and equivalent compressional-wave velocities. Shear velocities were consistently measured with a Vp/Vs ratio around 2, which is generally more characteristic of consolidated rocks. Ratios for unconsolidated fill materials such as these are generally expected to fall in the 3 to 5 range. This higher-than-expected ratio could result in measuring mode-converted shear rather than the primary direct shear arrival. It is also possible this higher-than-expected shear velocity could be real and related to these earth materials and the mechanical compaction used to construct these levees.
Estimates of shear velocity using both refraction tomography and slope intercept methods provided shear velocities that were unrealistically high and with offset-dependent arrival patterns extremely consistent with the faster compressional-wave arrivals. Calculating shear-wave velocity from inverted surface waves was strongly dependent on bandwidth and percentage of higher-mode energy recorded. During the first survey, ground conditions were not conducive to producing and/or recording broadband surface waves. Therefore, no confident shear-wave velocity sections were produced. On the second trip near-surface conditions had sufficiently changed to allow sufficient broadband surface wave that a 2-D shear wave profile could be produced for the levee core.
Velocity anomalies within the levee were detected at each of the three Retamal levee sites. Distribution and range of values for these anomalies are consistent with variations in material types used during construction and the construction process itself. It is not clear that velocity information alone will be sufficient to identify areas with a high density of cracks, which could be present as a result of the dewatering during drought of the expansive clays used in some places during core construction. However, it does seem likely that reduction in the material stiffness of the levee core could be used to identify failure risk areas with a relatively high resolution. Discontinuities in the levees associated with cracks seem to interfere with the otherwise uniform propagation of surface waves through the levee. These disturbances, once fully understood, could provide relatively accurate locations of weak zones within the core material.
Problems and pitfalls associated with using seismic techniques to estimate velocities intended to help characterize levee competence do exist and require significant attention to detail and understanding of the seismic-wavefield arrival patterns (t-x) and significance of the spectral properties of each mode. In particular, mode converted shear-wave energy can lead to completely incorrect conclusions. Interpreting the propagation irregularities in surface-wave energy is not clearly understood and, therefore, is not yet ready for use as a routine tool in interrogating levees. It must also be kept in mind that the geometry of the levee and the proximity of its basal contact with native earth can result in refracted first arrivals dominating the majority of close-offset traces where direct waves are normally expected.
Rapid, precise seismic methods for identifying areas worthy of further investigation could be developed for specific levee geometries and construction materials. Monitoring is by far the most confident and accurate application for seismic techniques on levees. Consideration must be given for changes in skin conditions due to seasonal variations in moisture. At the five sites studied on the Retamal and Main Levees, LRGV compressional-wave velocity estimations were most accurate for all conditions using refraction tomography. Shear-wave-velocity survey data were contaminated with mode-converted energy and therefore difficult to use to estimate material characteristics. Changes in near-surface conditions between the first and second survey resulted in an increase in recorded surface-wave bandwidth and therefore reasonably confident shear-wave velocity estimations within the levee. This change in surface conditions did not seem to change the arrival patterns observed on data recorded to capture first-order shear-wave first arrivals.
Infiltration of water into the levee skin was identified on seismic data during the ponding experiment conducted during the second site visit at site 2 (oxbow lake site). Notable changes in both compressional and shear velocity can be associated with the infiltration of water dammed against the south levee face. Compressional-wave data suggest percolation of water into the native river valley sediments beneath the levee. Shear-wave velocity change was rapid, occurring at the very beginning of the simulation, and was isolated to one area within the pond. The isolated nature of the infiltration on the shear data could be related to a fracture/crack system opened as a result of the years of drought and dewatering of the core. An alternate possibility is a possible material inconsistency resulting from construction practices and locally mined core material.
Considering the observations from the ponding experiment and five-site study, it is clear that the seismic tool can be used during flood events to detect more permeable areas where infiltration is active and the potential exists for failure. The most effective use of this tool would be as a monitoring system, where a baseline survey is acquired for all suspect areas, then during a flood event repeat surveys are run using differencing techniques to detect weak points pre-failure. Complications from mode conversions and near-surface dependent propagation characteristics will limit the use of this tool in some settings until more advanced processing capabilities have been developed. Clearly, more information is present in the seismic wavefield than we currently have the capability to meaningfully extract. Optimized future use of this tool will depend to some degree on acquisition of baseline data sets that will allow full wavefield processing once the methods have been fully developed. Current research in these areas is active and incrementally moving forward with providing solution to many problems encountered on this study.
Read the PDF version (8.4 MB)
Report cover as a PDF (6.8 MB)
Kansas Geological Survey, Geophysics
Placed online July 13, 2006
Comments to webadmin@kgs.ku.edu
The URL for this page is http://www.kgs.ku.edu/Geophysics/OFR/2005/OFR05_56/index.html