Soil Science Society of America Journal 66:206-211 (2002)
© 2002 Soil Science Society of America
DIVISION S-5 - PEDOLOGY
Soil Investigations using Electromagnetic Induction and Ground-Penetrating Radar in Southwest Tennessee
Daniel J. Inman,
Robert S. Freeland*,
John T. Ammons and
Ronald E. Yoder
Biosystems Engineering and Environmental Sciences, The Univ. of Tennessee, P.O. Box 1071, Knoxville, TN 37901-1071
* Corresponding author (rfreelan{at}utk.edu)
 |
ABSTRACT
|
|---|
Conducting precise soil investigations rapidly and nonintrusively is of great interest to soil scientists and engineers. As such, comparisons of ground-penetrating radar (GPR), electromagnetic induction (EMI), and traditional soil survey techniques were made on loessial soils in southwest Tennessee. The objectives of this study were to: (i) conduct a complete soil morphological, chemical, and physical characterization by methods; (ii) conduct a nonintrusive soil investigation by GPR and EMI; and (iii) compare the results from a traditional investigation with the nonintrusive investigation. The soils were located on an upland position. Parent material was loess–alluvium–Tertiary sand. Measured loess thickness ranged from 90 to 144 cm and measured alluvium thickness ranged from 82 to 151 cm. Soil morphological and physical properties in the upper 130 cm of nine pedons were analyzed statistically and grouped by pedon sample site. For an unbiased assessment, all GPR and EMI data were analyzed independently by pedon sample site, and data were grouped on the basis of similarities. Groupings of sites were compared by Kappa statistics. At the subgroup level, all sites were classified as Ultic Hapludalfs. Groupings of sites based on soil morphology and physical data had strong agreement 
with groupings of sites based upon GPR data, which were first targeted by a precursory EMI survey.
|
INTRODUCTION
|
|---|
SINCE 1990, researchers from The University of Tennessee Agricultural Experiment Station (TAES) have been working on a long-term water-quality monitoring project at the Ames Plantation in Southwest Tennessee (Fig. 1)
. Studies have demonstrated that preferential flow paths control the subsurface movement of water and solutes (Yoder et al., 1998). These preferential flow paths seem to be ephemeral and influenced by soil morphology (Freeland et al., 1997). Efforts to delineate these paths by means of traditional soil sampling techniques have been inconclusive. Moreover, the fact that these flow paths are ephemeral suggests that random or grid sampling would be inadequate for correct characterization. The use of a more inclusive soil mapping protocol by EMI in tandem with GPR has been proposed to delineate these preferential flow paths. As a first step towards this goal, comparisons of EMI and GPR data interpretations with traditional soil investigations were required.
The ease and speed with which subsurface information can be acquired by EMI make the technology an ideal precursory investigative tool to target intensive GPR surveys. When integrated with a global positioning system (GPS), copious amounts of data can be acquired rapidly, georeferenced, and plotted by commercially available software programs (Kitchen et al., 1996). Studies show that using EMI and GPR together will reduce the amount of time spent in the field and will improve the accuracy and cost effectiveness of the survey (Cannon et al., 1994; Doolittle and Collins, 1998). As an initial reconnaissance tool, EMI can help pinpoint anomalies and delineate electrically homogeneous and heterogeneous patterns within a field, and thus direct a more-focused GPR survey.
Once a mapping protocol using both EMI and GPR is established (which may be site-specific), soil information could be acquired rapidly. This could then be applied to both field-scale and regional-scale high intensity soil maps. To be sure, it is unlikely that any mapping protocol utilizing geophysical tools will in the near future make traditional soil morphological observations obsolete. Nevertheless, these technologies could increase greatly the precision of soil maps by providing more accurate soil-mapping unit delineations and additional unit inclusions. By reducing labor and costs, soil maps could be kept more current (Doolittle, 1982; Schellentrager et al., 1988).
Electromagnetic induction has proven to be an effective tool for obtaining subsurface information, especially in areas having saline soils (Geonics Ltd., 1998). This technology allows for the detection of lateral changes in subsurface properties. As with GPR, success of an EMI survey depends upon the local soil morphological, physical, and chemical properties (McNeill, 1980). Ammons et al. (1989) used EMI to separate Natraqualf and Ochraqualf map units in West Tennessee, suggesting that the EMI may be an appropriate tool for rapid soil delineation in the loess derived soils of West Tennessee. In addition to effectively delineating soil-mapping units, EMI can in some instances estimate depths and thickness of soil horizons (Doolittle et al., 1994; Geonics Ltd., 1995; Kitchen et al., 1996). Inferences drawn from EMI data can be misleading when the underlying soil is composed of multiple layers having dissimilar materials (Doolittle and Collins, 1998).
For the past two decades, GPR has been used extensively for soil investigations. However, the effectiveness of GPR is largely dependant upon soil properties and conditions. In general, the most favorable results will be produced on dry, coarse-textured soils with low amounts of expandable clays (Doolittle and Collins, 1995; Smith and Jol, 1995). Under favorable soil conditions, GPR is well suited for estimating depth to soil horizons, approximating depth to both claypans and fragipans, and mapping lithologic discontinuities (Collins et al., 1986; Collins and Doolittle, 1987; Collins et al., 1989; Rebertus and Doolittle, 1989; Doolittle et al., 1994; Freeland et al., 1997; Doolittle et al., 2000). These and other studies have established that both EMI and GPR can acquire soil information under a variety of soil conditions. However, results often vary upon site conditions and characteristics. Additional studies are needed to establish the benefits of the use of the two technologies as an integrated soil-surveying tool.
The objective of this study was to compare the results from a traditional soil investigation to a nonintrusive investigation using GPR and EMI in tandem on loessial soils in southwest Tennessee. Results would determine the usefulness of EMI as a precursory investigative tool that could be used to direct the more costly and time-consuming GPR surveys at this site.
|
MATERIALS AND METHODS
|
|---|
Research Site
This study was conducted at the Ames Plantation (N 35°13'39'' W 89°22'32'') in southwestern Tennessee (Fig. 1). Located in the Coastal Plain Region of Tennessee [Major Land Resource Area (MLRA) 134], soils in this area are formed in variable depths (<1–2 m) of loess overlying ancient terrace deposits underlain by Tertiary-aged sand deposits. The site is located on a regularly mowed nonproduction field that has been mapped as Memphis Silt Loam (fine-silty, mixed, active, thermic Typic Hapludalf) (USDA, 1964). This site is in close proximity to several instrumented watersheds where intensive soil investigations had been conducted previously.
Electromagnetic Induction Survey
Apparent electrical conductivity (ECa) is the depth-weighted average of electrical conductivity within a given column of soil reported in milliSiemens per meter (mS/m), and can be measured by an electromagnetic induction meter. Factors affecting soil conductivity include soil moisture content, salinity, and temperature (Freeland, 1989). Coarser soils, in general, tend to be less electrically conductive than finer soils.
The EMI used in this study was a Geonics EM31-MK2 (Geonics Ltd., Mississauga, ON, Canada)1. The effective depth of measurement is dependant upon the intercoil spacing of the instrument, transmission frequency, and dipole orientation. Optimal effective depths of observation for the EM31-MK2 are approximately 3 and 6 m in the horizontal and vertical dipole directions, respectively. The unit has 10, 100, and 1000 mS/m conductivity range settings, with a measurement resolution of 0.1% full scale (Geonics Ltd., 1995).
Three surveys of ECa were conducted across a 1-ha field, which were spaced over varying seasonal soil moisture conditions (17 March 1999 avgECa = 4.0 moist; 25 May 1999 avgECa = 4.8 wet; 14 March 2000 avgECa = 2.7 dry). Measurements were taken on a quasi-grid pattern at approximately every 3 m (Fig. 2)
. Readings were recorded with the instrument at 
1-m height in the horizontal dipole orientation. Global Positioning System data were taken at each measurement point with a Trimble Ag GPS 132 (Trimble Navigation Ltd., Sunnyvale, CA). Data from the EMI and GPS were merged into ArcView 3.2 (Environmental Systems Research Institute, Inc., Redlands, CA), and maps of ECa were produced.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 2. (a) EMI sample points acquired (1947 points/ha with some overlap) and (b) corresponding conductivity map—dry soil conditions.
|
|
Ground-Penetrating Radar Survey
Ground-penetrating radar detects changes in the electromagnetic properties of earthen materials. Output from the GPR is dependent on the attenuation and propagation of radio waves through a material. As the GPR antenna is pulled across the ground surface, pulses of radio waves are transmitted into the subsurface. The pulses reflect off soil boundaries having dissimilar dielectric properties, with greater and more abrupt dissimilarities producing greater reflection amplitudes. Reflected waveforms returning from the subsurface are recorded as a function of two-way travel time in nanosecond increments.
The GPR system used in this study is a GSSI Subsurface Interface Radar (SIR) System 10-A (Geophysical Survey Systems, Inc., New Salem, NH). This system is comprised of a system unit, monitor, keyboard, and antenna. For this survey, a 200-MHz antenna (GSSI Model 5106) was used. An all-terrain vehicle (ATV) was used to transport the GPR hardware and to tow the antenna.
A survey using the GPR was performed over a 30.5- by 30.5-m2 section of the field, which was selected by ECa maps of the entire field (Fig. 2). The survey plot was selected to be traversing user-defined ECa demarcations, and away from the field perimeter. Six survey transects were overlaid on the plot to form a 3-by-3 grid. Parallel transects were spaced at 15.25 m. The GPR antenna was pulled adjacent to the transect, on both sides traveling in the northwesterly and southeasterly orientations. Additional GPR surveys were taken perpendicular to these orientations (Fig. 3) . A calibration/reference GPR survey was taken in the near proximity of the survey plot, and is designated transect A-B (Fig. 3). Data from the GPR were processed by GSSI RADAN software (Geophysical Survey Systems, Inc., New Salem, NH), which included ground normalization and horizontal scaling. For interpretation, the 16-bit radar data were color enhanced to 256 colors by Adobe Photoshop (Adobe Systems Incorporated, San Jose, CA).
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 3. Study site showing locations of Soil Cores 1 to 9, GPR transect grid, and locale of GPR calibration/reference transect A-B.
|
|
Traditional Soils Investigation
Nine pedons (Sites 1–9) were sampled on a 3-by-3 grid overlaid on the 30.5- by 30.5-m2 site used in the GPR survey (Fig. 3). Sampling depths ranged from 2.5 to 3 m. A Giddings GSRP-S-M hydraulic probe (Giddings Machine Co., Ft. Collins, CO) fitted with a 7.62-cm-diam. core was used to extract each pedon. Pedons were sampled and described in the field according to Soil Survey Staff (1993). Standard soil survey laboratory methods were used to determine particle size, cation exchange capacity, exchangeable bases, and base saturation (Soil Survey Staff, 1996). Total elemental analysis was determined by a modified aqua-regia hydrofluoric acid procedure (Ammons et al., 1995). Extracts were analyzed by inductively coupled argon plasma-emission spectroscopy. Bulk density was determined by the nonpolar liquid method (Smith, 1957). All field data and notes were adjusted to the laboratory data. Data from the total elemental analysis were used to help identify major discontinuities. Pedons were classified according to Soil Taxonomy (Soil Survey Staff, 1999).
Data Analysis
From the upper 130 cm, soil physical properties including percent sand, silt, and clay; bulk density; particle density; and total porosity were analyzed by pedon sample site by Least Significant Difference (LSD) mean separation (SAS Institute, 1998). This procedure determined what soil properties were different statistically between sampled pedon sites. These properties were further analyzed by Ward's Minimum Variance Clustering Analysis (SAS Institute, 1998). On the basis of this analysis, sampled pedon sites were subgrouped within each horizon for the Ap, Bt1, Bt2, and BC1 horizons. From these results, a frequency table tallying how many times each pedon sample site was grouped with every other pedon sample site was produced. Final overall groupings from the traditional soil investigation were based upon a user-defined frequency 
3 (max. = 4). Groupings produced by the traditional soil data using Ward's Minimum Variance Clustering Analysis were compared with groupings formed from the nonintrusive investigation using Kappa statistics (SAS Institute, 1998).
In order to maintain a blind study, engineers from The University of Tennessee's Agricultural and Biosystems Engineering Department conducted all GPR and EMI data collection and analysis independent of the traditional soil analysis. Ground-penetrating radar data from the upper 130 cm were analyzed visually and independently by two team members. Data were subjectively grouped on the basis of imaging similarities within the upper 130 cm of the soil (Fig. 4)
and within the surrounding proximity of each pedon sample site.
|
RESULTS
|
|---|
All pedon sample sites were located on upland positions. Parent material was loess–alluvium–Tertiary sand. Slope of the pedon sample area was <2% and elevation was approximately 165 m above sea level. Morphology of Site 1 is provided as a representative descriptive profile of all pedons sampled (Table 1). Two taxons were identified (Table 2) from Soil Taxonomy (Soil Survey Staff, 1999). At the nine pedon sample sites, loess thickness ranged from 90 to 144 cm, and alluvium thickness ranged from 82 to 151 cm (Table 2). Bulk density measurements were high in the Bt1 horizons of all sites, likely from surface compaction. From Table 3, only average percent sand and silt in the upper 130 cm were significantly different between sites (P 
0.05). The chemical properties that were determined by laboratory methods were found not to be significantly different.
From the ECa data, a distinct zone of higher ECa readings trending from northeast to southwest was seen (Fig. 2b). This trend in higher ECa readings had a weak negative relationship with surface topography (Yoder et al., 2000). This pattern was repeatable across seasonal soil moisture conditions, uniformly shifting only in amplitude. In general, the same regions had higher ECa readings in relationship to the same regions of lower ECa readings, shifting in amplitude with changes of seasonal soil moisture conditions. Further analysis using regression statistics revealed that ECa had a strong positive linear relationship with percent sand (r2 = 0.76, P = 0.002) (Table 4). When ECa was regressed against all of the other soil physical and chemical properties measured by laboratory methods, these relationships were not statistically significant at the P 0.05 level of probability. The sand percentage is probably due to the presence of discontinuous elluvial bodies that were present in and around the loess–alluvium interface (Table 1). Higher ECa readings, however, are likely a result of a slight perching of water at this interface.
Visual subjective interpretation of the GPR scans at and in the near proximity of the pedon sample sites by two independent team members allowed the nine pedon sites to be placed into three groups (Groups I, II, and III) on the basis of the image similarities (Fig. 4). Group I (Sites 2 and 5) had heterogeneous dielectric properties in the upper 130 cm. Group II (Sites 6, 8, and 9) had somewhat of a transitional zone between homogeneous and heterogeneous dielectric properties above the loess–alluvium interface (upper 130 cm). Group III (Sites 1, 3, 4, and 7) had relatively uniform GPR samples above the loess–alluvium interface (upper 130 cm).
Subgroups of pedon sample sites by horizon Ap, Bt1, Bt2, and BC are reported in Table 5. Final overall groupings using the soil physical data were determined on the basis of a frequency 3 (a value selected from the GPR image interpretation groupings). Final overall groupings were (i) Sites 2 and 7; (ii) Sites 5, 6, and 8; and (iii) Sites 1, 3, 4, and 7. Site 7 was grouped twice, whereas Site 9 was not grouped. Pedon site groupings of the GPR data (I, II, and III) compared well with the pedon site groupings using the soil physical data (1, 2, and 3) as shown in Table 6.
|
CONCLUSIONS
|
|---|
Using traditional soil investigation methods, we identified two soil taxons. A strong positive relationship with percent sand (r2 = 0.76, P = 0.002) existed with EMI values. Using EMI boundary demarcations to target a GPR survey to traverse regions of user-defined classes of homogenous conductivity, we identified three groups of GPR patterns by subjective visual interpretation. Pedon sample site groupings statistically clustered from soil physical data within the upper 130 cm compared well with pedon site groupings classified visually by independent GPR interpretations 
.
This study highlights the usefulness of EMI as a broad precursory investigative tool that can be used to direct and focus the more costly and time-consuming GPR surveys. For this research site, using the two technologies together is a promising survey technique for obtaining high intensity, nonintrusive, spatially continuous soil information.
|
ACKNOWLEDGMENTS
|
|---|
Support for this project was provided by the United States Department of Agriculture-Water Resources Assessment Protection Program of the National Research Initiative, "Predicting Offsite Subsurface Migration of Agrochemicals—Noninvasive Surveying," contract no. 99-35102-8273, and by the Trustees of the Hobart Ames Foundation, Ames Plantation, Grand Junction, TN.
|
NOTES
|
|---|
1 Mention of a trade name, proprietary product, or specific equipment does not constitute a guarantee or warranty by the authors or The University of Tennessee and does not imply approval of a product to the exclusion of others that may be suitable. 
Received for publication December 6, 2000.
|
REFERENCES
|
|---|
- Ammons, J.T., M.E. Essington, R.J. Lewis, A.O. Gallagher, and G.M. Lessman. 1995. An application of modified microwave total dissolution technique for soils. Commun. Soil Sci. Plant Anal. 26:831–842.
- Ammons, J.T., M.E. Timpson, and D.L. Newton. 1989. Application of aboveground electromagnetic conductivity meter to separate Natraqalfs and Ochraqualfs in Gibson County, Tennessee. Soil Surv. Horiz. 30:66–70.
- Cannon, M.E., R.C. McKenzie, and G. Lachapelle. 1994. Soil salinity mapping with electromagnetic induction and satellite based navigation methods. Can. J. Soil Sci. 74:335–343.
- Collins, M.E., and J.A. Doolittle. 1987. Using ground penetrating radar to study soil microvariability. Soil Sci. Soc. Am. J. 51:491–493.[Abstract/Free Full Text]
- Collins, M.E., G.W. Schellentrager, J.A. Doolittle, and S.F. Shih. 1986. Using ground penetrating radar to study changes in soil map unit composition in selected Histosols. Soil Sci. Soc. Am. J. 50:408–412.[Abstract/Free Full Text]
- Collins, M.E., J.A. Doolittle, and R.V. Rourke. 1989. Mapping depth to bedrock on a glaciated landscape using ground penetrating radar. Soil Sci. Soc Am. J. 53:1806–1812.[Abstract/Free Full Text]
- Doolittle, J.A. 1982. Characterizing soil map units with the ground penetrating radar. Soil Surv. Horiz. 23:3–10.
- Doolittle, J.A., K.A. Sudduth, N.R. Kitchen, and S.J. Indorante. 1994. Estimating depth to claypans using electromagnetic inductive methods. J. Soil Water Conserv. 49:552–555.
- Doolittle, J.A., and M.E. Collins. 1995. Use of soil information to determine application of ground penetrating radar. J. Appl. Geophys. 33:101–108.
- Doolittle, J.A., and M.E. Collins. 1998. A comparison of EM induction and GPR methods in areas of karst. Geoderma 85:83–102.
- Doolittle, J.A., G. Hoffmann, P. McDaniel, N. Peterson, B. Gardner, and E. Rowan. 2000. Ground-penetrating radar interpretations of a fragipan in Northern Idaho. Soil Surv. Horiz. 41:75–82.
- Freeland, R.S. 1989. Review of soil moisture sensing methods using soil electrical conductivity. Trans. ASAE 32:2190–2194.
- Freeland, R.S., R.E. Yoder, and J.T. Ammons. 1997. Observing preferential flow paths with ground penetrating radar. Tenn. Agric. Sci. Issue 181, Winter 1997.
- Geonics Ltd. 1995. EM31 – MK2 Operating Manual. Mississauga, ON, Canada, October 1995.
- Geonics Ltd. 1998. Applications of Electromagnetic Methods—Soil Salinity. Mississauga, Ontario Canada, July 1998.
- Kitchen, N.R., K.A. Sudduth, and S.T. Drummond. 1996. Mapping of sand deposits from 1993 Midwest floods with electromagnetic induction measurements. J. Soil Water Conserv. 51:336–340.
- McNeill, J.D. 1980. Electromagnetic terrain conductivity measurements at low induction numbers. Geonics Limited. Technical Note TN-6, Mississauga, ON, Canada.
- Rebertus, R.A., and J.A. Doolittle. 1989. Landform and stratigraphic influences on variability of loess thickness in northern Delaware. Soil Sci. Soc Am. J. 53:843–847.[Abstract/Free Full Text]
- SAS Institute. 1998. SAS system for elementary statistical analysis. SAS Inst., Cary, NC.
- Schellentrager, G.W., J.A. Doolittle, T.E. Calhoun, and C.A. Wettstein. 1988. Using GPR to update soil survey information. Soil Sci. Soc Am. J. 52:746–752.[Abstract/Free Full Text]
- Smith, D.G., and H.M. Jol. 1995. Ground penetrating radar antenna frequencies and maximum probable depths of penetration in Quaternary sediments. J. Appl. Geophys. 33:93–100.
- Smith, R.M. 1957. Soil structural evaluation with a non-polar liquid. Soil Sci. 84:311–322.
- Soil Survey Staff. 1993. Soil Survey Manual. USDA Handbook No. 18. U.S. Gov. Print. Office, Washington, DC.
- Soil Survey Staff. 1996. Soil survey laboratory methods manual. Soil Survey Investigation Report No. 42 version 3.0. National Soil Survey Center, Lincoln, NE.
- Soil Survey Staff. 1999. Keys to soil taxonomy. U.S. Gov. Print. Office, Washington, DC.
- USDA, SCS. 1964. Soil Survey of Fayette County, Tennessee. U.S. Gov. Print. Office, Washington, DC.
- Yoder, R.E., D.C. Yoder, C.R. Mote, M.E. Essington, J.T. Ammons, T.C. Mueller, and D.D. Tyler. 1998. Subsurface Water Movement on an Agricultural Watershed. Paper presented at the ASCE 1998 Int. Water Res. Engineering Conf., Memphis, TN. 3–7 Aug. 1998. ASCE, Reston, VA.
- Yoder, R.E., R.S. Freeland, J.T. Ammons, and L.L. Leonard. 2000. Mapping agricultural fields with GPR and EMI to predict offsite movement of agrochemicals. In Proceedings of the Eighth International Conference on Ground Penetrating Radar (GPR 2000). The University of Queensland. 22–26 May 2000. Gold Coast, Australia.
This article has been cited by other articles:
|
|
|
|
 
E. Segal, S. A. Bradford, P. Shouse, N. Lazarovitch, and D. Corwin
Integration of Hard and Soft Data to Characterize Field-Scale Hydraulic Properties for Flow and Transport Studies
Vadose Zone J.,
August 1, 2008;
7(3):
878 - 889.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
L. Cockx, M. Van Meirvenne, and B. De Vos
Using the EM38DD Soil Sensor to Delineate Clay Lenses in a Sandy Forest Soil
Soil Sci. Soc. Am. J.,
June 29, 2007;
71(4):
1314 - 1322.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. Hornung, R. Khosla, R. Reich, D. Inman, and D. G. Westfall
Comparison of Site-Specific Management Zones: Soil-Color-Based and Yield-Based
Agron. J.,
March 2, 2006;
98(2):
407 - 415.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
B. J. Allred, J. D. Redman, E. L. McCoy, and R. S. Taylor
Golf Course Applications of Near-Surface Geophysical Methods: A Case Study
Journal of Environmental & Engineering Geophysics,
March 1, 2005;
10(1):
1 - 19.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
B), particle density (
in the upper 130 cm.