USF Libraries
USF Digital Collections

Ground penetrating radar response to thin layers

MISSING IMAGE

Material Information

Title:
Ground penetrating radar response to thin layers examples from Waites Island, South Carolina
Physical Description:
Book
Language:
English
Creator:
Guha, Swagata
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
numerical models
laminations
high-resolution
radar wave
subsurface exploration
Dissertations, Academic -- Geology -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Thin layers (layers that are not resolvable in terms of GPR wavelengths) are very common in sedimentary deposits. To better understand ground penetrating radar (GPR) wave behavior in sequences of thin layers with contrasting electromagnetic parameters, 1D FDTD simulations are run for simple layer distributions. Laminated (mm-scale) sequences can produce reflected energy with 10-20% of the amplitude of reflections from equivalent isolated contacts. Amplitude spectra from laminae packages are shifted toward higher frequencies. Such spectral shifts in radar profiles may potentially be used as indicators of fine-scale laminations. A comparative study of GPR records and models generated from core data from Waites Island, South Carolina, a Holocene barrier island, suggest that magnetite-rich laminae contribute significantly to radar profiles, but that some features in the radar traces cannot be associated with lithologic changes seen in vibracores.
Thesis:
Thesis (M.S.)--University of South Florida, 2005.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Swagata Guha.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 86 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001606113
oclc - 60455668
notis - AKF9239
usfldc doi - E14-SFE0000593
usfldc handle - e14.593
System ID:
SFS0025283:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001606113
003 fts
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 050509s2005 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000593
035
(OCoLC)60455668
9
AKF9239
b SE
SFE0000593
040
FHM
c FHM
090
QE26.2 (ONLINE)
1 100
Guha, Swagata.
0 245
Ground penetrating radar response to thin layers
h [electronic resource] :
examples from Waites Island, South Carolina /
by Swagata Guha.
260
[Tampa, Fla.] :
University of South Florida,
2005.
502
Thesis (M.S.)--University of South Florida, 2005.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 86 pages.
3 520
ABSTRACT: Thin layers (layers that are not resolvable in terms of GPR wavelengths) are very common in sedimentary deposits. To better understand ground penetrating radar (GPR) wave behavior in sequences of thin layers with contrasting electromagnetic parameters, 1D FDTD simulations are run for simple layer distributions. Laminated (mm-scale) sequences can produce reflected energy with 10-20% of the amplitude of reflections from equivalent isolated contacts. Amplitude spectra from laminae packages are shifted toward higher frequencies. Such spectral shifts in radar profiles may potentially be used as indicators of fine-scale laminations. A comparative study of GPR records and models generated from core data from Waites Island, South Carolina, a Holocene barrier island, suggest that magnetite-rich laminae contribute significantly to radar profiles, but that some features in the radar traces cannot be associated with lithologic changes seen in vibracores.
590
Adviser: Kruse, Sarah E.
653
numerical models.
laminations.
high-resolution.
radar wave.
subsurface exploration.
690
Dissertations, Academic
z USF
x Geology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.593



PAGE 1

Ground Penetrating Radar Response to Thin Layers: Examples from Waites Island, South Carolina by Swagata Guha A thesis submitted in the partial fulfillment of the requirements for the degree of Masters of Science Department of Geology University of South Florida Major Professor: Sarah E. Kruse, PhD Peter J. Harries, PhD Eric A. Oches, PhD Date of Approval November 12, 2004 Keywords: subsurface exploration, radar wave high-resolution, laminations, numerical models Copyright 2004, Swagata Guha

PAGE 2

Acknowledgement Completion of this thesis involved a great degree of patience for which I am indebt to my major professor, Dr Sarah Kruse. I would like to thank my committee members, Dr Peter Harries and Dr Eric Oc hes for their words of encouragement and support. Sincere gratitude goes to Dr Eric Wr ight of the Marine Science Department of Coastal Carolina University for his assist ance in collecting fiel d data and providing valuable information. Special thanks to Dr Richard Davis and Dr Ping Wang for their time and help during the sediment analysis procedure. A ssistance rendered by Dr John Jaeger and his students at University of Florida, Gainsville is also acknowledged. I am greatly thankful for the invaluable help from my fellow students, Steve Guggino, Jennifer Krock, Josh Anderson, David Tidwell, Alyssa Saint John a nd Arnell Harison during various stages of this project.

PAGE 3

Table of Contents List of Tables ii List of Figures iii Abstract v Introduction 1 Ground Penetrating Radar in Sediments 3 Wave Responses to Thin Layers (thin beds) 4 Models of GPR response to thin layers 8 Method 8 Models and Results 9 Internal layering within packages 13 Laminae 16 Summary and Discussion of model results 21 Case Study: Waites Island, South Carolina 23 Geologic setting 23 Data collection and processing 23 Laboratory Methods 31 Core sediment analysis 31 Layered model simulating core laminations 37 Model parameters 37 Analysis of models 39 Pattern of returns 39 Attenuation 40 Spectra 40 Conclusions 52 References 53 Appendices 57 Appendix 1 58 Appendix 2 63 Appendix 3 64 Appendix 4 68 i

PAGE 4

List of Tables Table 1. Physical parameters used in model simulations 39 ii

PAGE 5

List of Figures Figure 1a. Reflection Event 1 and Event 2 from a 50 cm thick bed 6 Figure 1b. Reflection events distinguishable for thick layer and indistinguishable for thin layer in synthetic GPR traces 6 Figure 1c. Reflections from beds with decreasing thickness with respect to dominant wavelength 7 Figure 2. Single layer model 10 Figure 3. Reflections from a single layer 10 Figure 4. Change in trace form with increasing thickness (surrounding medium with r = 20) 11 Figure 5. Change in trace form with increasing thickness (surrounding medium with r = 4) 12 Figure 6. Change in reflection form of different layer thicknesses at different frequencies 12 Figure 7. Model with internal layerings with 30cm package 14 Figure 8a. Reflections from multiple layers (total thickness=30cm) at 100 MHz with fixed interlayer spacing (=1cm) 14 Figure 8b. Reflections from multiple layers (total thickness=30cm) at 200 MHz with fixed interlayer spacing (=1cm) 15 Figure 8c. Reflections from multiple layers (total thickness=30cm) at 500 MHz with fixed interlayer spacing (=1cm) 15 Figure 9. Schematic representation of model suites 16 Figure 10. Reflections from a sequence of layers with fixed layer thickness (1mm) and varying interlayer spaces at different frequencies 19 Figure 11. Averaged amplitude spectra for different models at 100 MHz (fig.11a), 200 MHz (fig.11b) and 500 MHz (fig.11c). 21 iii

PAGE 6

Figure 12. Location map: Waites Island 25 Figure 13. Cross-section of Waites Island along GPR transect (LINE 1) 26 Figure 14. Reflection profiles at core site WI1 27 Figure 15. Reflection profiles at core site WI2 28 Figure 16. CMP at site WI1 29 Figure 17. CMP at site WI2 30 Figure 18. Core sections of WI1 33 Figure 19. Core sections of WI2 34 Figure 20. Magnetic susceptibility, porosity and gamma bulk density values for core section (WI1 A) 35 Figure 21. Heavy mineral layers containing magnetite in core section, WI1 (A) 35 Figure 22. Grain size distribution for core section WI2 (A) 36 Figure 23. Schematic representations of Model 1 and Model 2 38 Figure 24. Comparison of real and model traces for site WI1. 25aMHz, 25b 100MHz 42 Figure 25. Comparison of real and model traces for site WI1. 26aMHz, 26b 100MHz, 26c 50MHz 44 Figure 26. Expotential trend of attenuation in real and model data for site WI1 at different frequencies. 46 Figure 27. Expotential trend of attenuation in real and model data for site WI2 at different frequencies. 49 Figure 28. Layer model apparent resistivity and layer thicknesses estimation 50 Figure 29. Averaged amplitude spectra of GPR traces at two different sites (close to WI1 and WI2) along LINE1. 51 iv

PAGE 7

Ground Penetrating Radar Response to Thin Layers: Examples from Waites Island, South Carolina Swagata Guha ABSTRACT Thin layers (layers that are not resolvable in terms of GPR wavelengths) are very common in sedimentary deposits. To better understand ground penetrating radar (GPR) wave behavior in sequences of thin layers with contrasting electromagnetic parameters, 1D FDTD simulations are run for simple layer distributions. Laminated (mm-scale) sequences can produce reflected energy with 10-20% of the amplitude of reflections from equivalent isolated contacts. Amplitude spectra from laminae packages are shifted toward higher frequencies. Such spectral shifts in radar profiles may potentially be used as indicators of fine-scale laminations. A comparative study of GPR records and models generated from core data from Waites Island, South Carolina, a Holocene barrier island, suggest that magnetite-rich laminae contribute significantly to radar profiles, but that some features in the radar traces cannot be associated with lithologic changes seen in vibracores. v

PAGE 8

Introduction Geophysical methods can aid stratigraphic interpretation of sedimentary depositional environments by providing information on bed geometry and internal structures within the deposits. The stratigraphic interpretation depends on the resolution capabilities of the geophysical technique being employed. Ground penetrating radar (GPR) is one such non-invasive, portable device that can provide relatively high-resolution, near-continuous shallow subsurface profiles. GPR is similar to seismic exploration in its approach, except that it utilizes electromagnetic waves rather than acoustic waves. Radar reflections produced are due primarily to variations in the dielectric properties of the ground; GPR wavelengths, and hence spatial resolution, are on the order of centimeters to meters. Because of the wavelength dimensions, GPR can be used to delineate subsurface geology in various sedimentary environments. Facies analysis of deltaic and coastal deposits, and recognition of internal structures of braided-river deposits, boundaries of ancient channel fills and river courses are some of the examples of radar stratigraphic applications for the reconstruction of depositional environments (Jol and Smith, 1991, 1992, 1997, Heinz et al, 2003, Woodward et al, 2003). Coastal studies have incorporated this geophysical technique for various purposes, including demarcation of ancient and modern shorelines (Neal et al, 2003), sediment budget estimation (Van Heteren et al 1996), description of facies (Baker, 1991, Gawthorpe et al, 1993), mapping of internal structures of sand dunes (Harari, 1996), and understanding the development of coastal landforms (Jol et al,1996, 2003, Nishikawa et al, 2000, Daly et al, 2002, Neal et al, 2002, 2003, Smith et al, 2003). GPR has also proved to be a useful tool for estimation of water content in soils (Greaves et al, 1996, Hagrey et al, 2000, Schmalz et al, 2002) and detection of the water table in different environments (Van Overmeeren, 1998). Most of the above-mentioned investigations have involved the stratigraphic interpretation of GPR profiles combined with field observations. Fewer studies have 1

PAGE 9

been devoted to understanding the physical causes of reflections in sediments. Among these, Olhoeft (1980s), Knoll and Knight (1994), Van Dam (2002 and more), and Moore et al. (2004) have addressed the relationships between sedimentary and electromagnetic properties that are responsible for reflection patterns in sands and aeolian sediments. There remain many settings where radar responses to sediments are poorly understood. In particular, in many, if not most, sedimentary environments, bed thicknesses are much smaller than the radar wavelengths used. For reflecting strata whose thicknesses are 1/4 th or 1/3 rd of the radar wavelength, identification of reflections off the top of the strata and off the bottom as separate events in the GPR record is plausible (Fig. 1a). However, for thinner layers it is impossible to assign a single reflection event to each contact (Fig. 1b). The GPR response acquired from a sequence of thin beds is an interference pattern produced by a complex interaction of the propagating radar wave with these layers. In this thesis an attempt is made to better understand the nature of GPR responses to thin sedimentary layers, and to determine whether more information about thin layers can be derived from the frequency and attenuation characteristics of radar traces. The relationship between sedimentary layer patterns and radar response is examined through several suites of simple numerical models. The numerical models are aimed at answering the following questions: 1. What sort of thin-layer patterns typical of sedimentary environments will yield radar returns of significant amplitude? 2. What are the effects of very thin beds (laminae) on radar returns? Can any information regarding the nature of thin beds be derived from the frequency and attenuation patterns of radar records? Finally, a comparative study of radar records and cores from a sandy barrier island setting has been conducted, adding to the relatively small volume of literature on the physical causes of radar reflections in sediments. 2

PAGE 10

Ground Penetrating Radar in Sediments Ground penetrating radar (GPR) systems emit electromagnetic waves from a transmitting antenna. Part of this electromagnetic energy is scattered when the incident waves encounter an inhomogeneity in the subsurface, and wave energy reflected back up to the surface is detected by a receiving antenna. Radar wave velocities are determined largely by the permittivity (dielectric constant) and magnetic permeability of the medium; wave attenuation primarily by the electrical conductivity of the medium and the wave frequency; and reflection coefficients at contacts by the contrast in the three electromagnetic propertiespermittivity, magnetic permeability, and conductivity. Both vertical and horizontal resolution limits in radar profiling are function of the radar wavelength, which is primarily controlled by antenna frequency and ground permittivity. Higher frequency antennas generate shorter wavelength radar pulses, from which finer-scale features can be resolved. However, stronger attenuation of high frequency pulses means that their depth of penetration is more limited. Ground penetrating radar wave propagation is discussed in greater detail in Appendix 1. In sediments in general, the petrophysical relationships between electromagnetic properties and the geologic and hydrogeological properties of strata are complex and relatively poorly understood. Laboratory studies (e.g. Olhoeft,1994, Knoll and Knight, 1994, van Dam and Schlager, 2002, Schon, 1996), theoretical models (e.g. Topp, 1985), and comparisons of radar records with cores and outcrops (e.g. Jol and Smith, 1991, 1996, Baker,1991, Wood, 1990, van Heteren, 1998, Vandenberghei and van Overmeeren,1999, Harari, 1996, van Dam and Sclanger, 2002) show that reflections in sediments are primarily controlled by changes in i) porosity, ii) mineralogy and iii) pore water content. Appendix 2 gives an overview of the physical causes of radar reflections encountered in different sedimentary environments. In many settings, the contacts between sedimentary strata (e.g. sands, muds, peat, organic-rich zones with variations in porosity, mineralogy, and/or water content) are often very closely spaced. Sedimentary layers, designated as beds (thickness > 1 cm) or laminations (thickness < 1cm) (McKee and Weir, 1953) show a wide range of variability in their mode of occurrences (given in Appendix 3) in coastal, lacustrine, fluvial, and 3

PAGE 11

aeolian environments. Imaging the depths of interest in geologic applications (meters to tens of meters), however, typically requires antenna frequencies low enough that the radar wavelengths are on the order of tens of cms or more. Thus, it is common in sedimentary studies that GPR wavelengths are considerably greater than the thickness of many layers. Commonly, the radar record represents the combined response to many thin layers (e.g. van Dam, 2003), and there is difficulty in identifying a one-to-one correspondence between an event in a radar profile and a contact observed in core or nearby outcrop. Wave response to thin layers (thin beds) Thin bed is a term widely associated with seismic exploration, especially for reservoir characteristics. By definition, reflections obtained from the top and bottom of a thick bed are distinguishable in time, whereas, for an isolated thin bed separate responses from the top and bottom interfaces are not resolvable since the reflections overlap in time (e. g. Knapp, 1990). The resultant waveform from a thin bed is a single pulse which is an approximate time derivative of the source pulse (e g Widess, 1973). For a bed of thickness t, w(t) and w(t + t) are the wavelets of reflection from the top and bottom interfaces of the bed, the composite wavelet is given by (e. g. Knapp, 1990): w(t) w(t + t) t dw/dt Widess (1973) showed for a Ricker wavelet incident upon thin beds, the amplitude of the composite reflection is linearly proportional to bed thickness and inversely proportional to the wavelength. The relation is given by: A d 4Ab/ b where A d is the amplitude of the composite wavelet, A is the amplitude of reflection for a thick bed, b is bed thickness, b is the dominant wavelength in the bed. Widess (1973) demonstrated with the above expression, that for a bed in a homogeneous medium, the practical thick vs. thin resolution limit is attained when bed thickness is approximately 1/8 th of the dominant wavelength () i.e. the top and bottom of beds with thicknesses less than /8 th cannot be distinguished as separate events (fig.1c). Clearly, GPR studies in sedimentary settings often deal with bed thicknesses that fall in the thin bed category. Furthermore, in most settings thin beds are not isolated, but 4

PAGE 12

contained within strata composed of multiple thin beds or laminae. The wave response to such a package will be an interference pattern determined by the spacing of the individual lamina and their reflection coefficients. Examples of marine seismic investigations of sequences of thinly bedded show that strong reflections may be recorded even when individual layers are indistinguishable and reflected energy is a resultant interference pattern. Mayer (1980) showed that seismic reflectors observed in profiles of pelagic carbonate sediments did not corresponding to individual layers but a composite outcome of interferences from several small layers. Knapp (1990) studied seismic wave response to Upper Pennsylvanian cyclothems (alternating thin layers of limestone and shale) and arrived at the conclusion that a high frequency (> 500 Hz and a wavelength of 6m) would be required for better resolution of thin beds Radar waves are expected to behave in similar way, in which one-to-one correlation of geologic horizons to GPR reflectors is not expected in many settings. There are relatively few examples, however, that examine radar wave response and the nature of the waveforms generated from a sequence of thin beds. Van Dam (2002) showed for an aeolian deposit, with 100 and 200 MHz antennas radar responses were largely interference patterns. Kruse and Jol (2003) modeled GPR interactions to large (1-2 m) and small-scale (cm) strata in a deltaic depositional setting, and found that graded sequences could not be distinguished from ones bounded by thin beds. To determine whether information on thin beds could be extracted from radar records in sedimentary environments, several hypothetical scenarios are examined with the aid of numerical modeling. 5

PAGE 13

0 20 40 60 80 100 120 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Reflections from a single layerTime (ns)Amplitude (mV) Event 1 Event 2 Fig. 1a Reflection Event 1 is from top and reflection Event 2 is from the bottom of a 50 cm thick layer. 0.8cm 19cm 40cm 1.6cm 2cm 8cm 12cm 16cm 25 Time (ns) 55 Fig. 1b Reflection events distinguishable for thick layer (e. g. 40 cm) and indistinguishable for thin layer (e.g. 8 cm) in synthetic GPR traces. 6

PAGE 14

Fig.1c Reflections from beds with decreasing thickness with respect to dominant wavelength, (Widess, 1973) 7

PAGE 15

Models of GPR response to thin layers Method Radar wave propagation through thin layers is simulated with finite-difference time-domain (FDTD) codes. These techniques are commonly used for GPR data analysis (e.g. Cassidy and Murdie, 2000, Wang and Mc Mechan, 2002, Teixeira et al, 1998, Kruse and Jol, 2003). FDTD methods allow the user to simulate radar propagation through strata and features with variable permittivity, magnetic permeability, and conductivity. The signal recorded at a hypothetical receiving antenna can be compared against those observed in the field. The FDTD methods propagate electromagnetic waves through a grid, solving a finite difference approximation to Maxwell's equations for electric and magnetic fields. The codes used here follow the classic staggered grid formulation of Yee (1966). All models described in this study are one-dimensional (1D), and hence contain the following implicit assumptions: (1) the transmitted energy is a vertically-traveling plane wave normally incident on the surface; (2) there is no offset between the transmitter and receiver positions; (3) subsurface layers are horizontal, smooth, and laterally homogeneous. Furthermore it is assumed that there is no lateral or vertical anisotropy of physical parameters (electrical conductivity ( ), relative permittivity/dielectric constant, ( r ), relative magnetic permeability, ( r ) which are further defined in Appendix 1 ). The 1D approximation fails to account for the dipole nature of typical commercial GPR antennas and attendant radiation pattern, geometrical spreading, effects associated with non-vertical incidence on interfaces, as well as any 3D complexities in the subsurface. Despite the obvious shortcomings, 1D modeling offers the advantages of rapid computations and a basic understanding of radar wave propagation behavior from studying simple scenarios, and is often used as a first step in GPR data interpretation (Bano, 1996, Kruse and Jol, 2003). For the purposes of this study--to examine the first-order characteristics of GPR returns from finely laminated sequences of thin layersD 8

PAGE 16

modeling is adequate, and the benefits of 1D speed and simplicity outweigh the shortcomings. 2D and 3D modeling, beyond the scope of this study, are needed to fully address the limitations associated with the 1D approximation in this context. As the models developed here are aimed at interpretation of GPR response to beds as thin as 1 mm scale, the cell size in the finite difference grid (dx) used in the runs in this study is set to 0.25 mm and the time interval (dt) is 0.5 ns. These grid parameters statisfy the Courant stability criteria and the recommendation that cells be smaller than 1/10 the wavelength (Kunz and Luebbers 1993). As the thinnest layers in our models are 1 mm, a cell dimension of 0.25 means that individual layers are comprised of at least 4 Yee cells. All models assume a tri-lobed pulse similar to the one transmitted by the pulseEkko 100 GPR systems (Sensors & Software Inc.). Models and results Two suites of layered models have been analyzed to better understand the response of radar waves of frequencies typically used in geological studies (50-400 MHz) to thin (mm-decimeter scale) layers of dimensions observed in sedimentary settings. In each suite, alternating layers with different relative permittivity, magnetic permeability and conductivity values form the layered models. Prior to discussion of multi-layered models, the resolution of isolated layers under conditions potentially observed in coastal deposits is addressed. For example, Figure 2 shows a 50 cm thick layer of r = 35, r =1.2 and = 4mS/s sandwiched between layers of lower r values of 20, r = 1 and = 1mS/s. This could represent, for instance, a magnetite-enriched layer or a lower porosity layer within a package of saturated sand. The radar wave velocities are ~ 0.05 m/ns through the slower and more conductive central layer and 0.06 m/ns through the surrounding sand. Figure 3 shows a pulse with center frequency 100 MHz traveling through the package. The pulse on the left is the downgoing pulse as it crosses the air-ground boundary in the model. As the wave front encounters the top of layer 2, a fraction of the energy is reflected back and the reflection is recorded as Event 1. Energy reflected back from the bottom of layer 2 appears as 9

PAGE 17

Event 2. Because events 1 and 2 can be distinguished from one another, this layer is not considered a thin layer. 1m1m50cmSinglelayerLayer1Layer3Layer2 Fig. 2 Single-layer model 0 20 40 60 80 100 120 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Reflections from a single layerTime (ns)Amplitude (mV) Event 1 Event 2 Fig. 3 Reflections from a single layer For the case where layer 2 is thin enough that the two events cannot be distinguished, the amplitude and form of the composite reflection follow the thin bed description of Widess (1973); reflection amplitude decreases with layer thickness. Figure 4 shows that for the hypothetical magnetite-enriched layer within saturated sand, the thick-thin bed transition occurs at ~20 cm thickness for a 100 MHz pulse. In this case, the wavelength of the radar wave within layer 2 is ~50 cm, so the thick-to-thin bed transition for this radar scenario occurs where layer thickness is 2/5 central wavelength, slightly greater than the 1/8 wavelength criteria described by Widess (1973). 10

PAGE 18

It is worth noting that it is the wavelength and hence velocity in layer 2, rather than the surrounding medium, that determines its resolution. Figure 5 shows, for example, a scenario identical to figure 4 excepting a faster ( r = 4) surrounding medium. The thin-thick bed transition thickness occurs again at ~20 cm. As the thin layer threshold is based on the ratio of the layer thickness to pulse wavelength, the thin layer resolution limit will decrease with increasing antenna frequency (decreasing wavelength). For the magnetite-rich layer sandy layer considered here, Figure 6 shows the thin bed resolution limits versus antenna frequency for those frequencies commonly used in geologic surveys. Even at 500 MHz, mm-scale laminae commonly seen in sedimentary environments fall well below the thin-bed limit. 0.8cm 1.6cm 2cm 8cm 12cm 16cm 19cm 40cm 25 Time (ns) 55 Fig. 4 Change in trace form with increasing layer thickness (surrounding medium with r = 20) 11

PAGE 19

1.5cm 0.5cm Fig. 5 Change in trace form with increasing layer thickness (surrounding medium with r = 4) 50 100 150 200 250 300 350 400 450 500 0 5 10 15 20 25 30 35 40 frequency (MHz)layer thickness (cm) Fig.6 Change in reflection form of different layer thicknesses at different frequencies. A transition thickness is observed at 39 cm for 50 MHz, 19.5 cm for 100 MHz, 10 cm for 200 MHz and 4.2cm for 500 MHz 30cm 16cm 5cm 15cm 19cm 17 Time (ns) 60 12

PAGE 20

Interna not found in isolation, but whose total thickness may itself be on the scale of the radar wavelength. s addresses the conditions under which internal the d ge ern; the trace appears similar in form to that of a single n about the s l layering within packages In practice in many sedimentary settings, thin layers are within packages The first suite of multi-layer model layering within packages could be resolved from radar records. Figure 7 shows such a multiple layered model. Radar returns are modeled for a 30 cm thick package imaged with 100, 200, and 500 MHz pulses. The internal layers (dark zones on Fig 7) have properties of the isolated layer in the case considered above: r = 35, r =1.2 and = 4 mS/s. The interlayer spacing and background has r = 20, r =1 and = 1 mS/s. In themodels, the interlayer spacing is kept fixed at 1 cm while the layer thicknesses are variefrom 1cm to 6 cm, 9 cm and 14.5 cm. For 100 MHz pulses incident on the package, (Figure 8a), the 30 cm packaitself constitutes a thin layer, and the presence of internal layering produces no additionalsignature in the overall reflection patt 30 cm thick layer. The amplitude of the returns from the package decreases somewhat with decreasing internal layer thickness. This is presumably due to increasingly destructive interference as layer thickness decreases, as well as a reduction in the bulk difference in material properties between the package and the surrounding medium for models with thinner layers. The net result is that packages with finer internal structure will have lower amplitude returns. For the model scenario here, however, these differences are subtle enough that they might be difficult to resolve in practice. For higher frequency pulses, e.g. 200 MHz (fig. 8b) and 500 MHz (fig. 8c) incident on the same model packages, the total package itself does not constitute a thin layer (reflections from top and bottom can be distinguished), and some informatio internal structure can clearly be resolved. In both cases, energy returns from within "thick bed" package are significant (i.e. returns from within the package have amplitude~ 50% or more of the package boundary reflections) for all models except the uniform (30 cm) layer and the models in which both layers and interlayers are 1 cm in thickness. Only the 500 MHz returns from packages with thicker internal layers show peaks that canbe correlated with individual interlayers. In summary, for the velocities considered here 13

PAGE 21

layering on scales of a few cm to 10s of cms can yield interference patterns with amplitudes of 50% or more of those from thick bed contacts. For layering on 1-cm scale, returns from within the package are much smaller (~2.1% of the return from the top of the bed). 1m1m 30cmMultiplelayersLayersInterlayerspacing Fig. 7 Model with internal layerings with 30 cm package 30 35 40 45 50 55 60 65 -0.1 -0.05 0 0.05 0.1 0.15 Time (ns)Amplitude (mV) Reflections from multiple layers (total thickness=30cm) at 100MHz with fixed interlayer space (=1cm ) layer thickness=1cmlayer thickness=6cmlayer thickness=9cmlayer thickness=14.5cmlayer thickness=30cm Fig. 8a 14

PAGE 22

30 35 40 45 50 55 60 -0.1 -0.05 0 0.05 0.1 0.15 Time (ns)Amplitude (mV)Reflections from multiple layers (total thickness=30cm) at 200MHz with fixed interlayer space (=1cm) layer thickness=1cmlayer thickness=6cmlayer thickness=9cmlayer thickness=14.5cmlayer thickness=30cm Fig 8b 30 35 40 45 50 55 60 -0.1 -0.05 0 0.05 0.1 0.15 Time (ns)Amplitude (mV)Reflections from multiple layers (total thickness=30cm) at 500MHz with fixed interlayer space (=1cm) layer thickness=1cmlayer thickness=6cmlayer thickness=9cmlayer thickness=14.5cmlayer thickness=30cm Fig. 8c 15

PAGE 23

Laminae The models above indicate that for the "saturated sand" scenario returns from cm-scale layering should be low in amplitude, an order of magnitude smaller than returns from equivalent thick bed contacts. We note, however, that in the above models, the low-amplitude response in the 1-cm layer stems in part from geologically implausible assumption that all layers are exactly 1 cm thick. In fact, GPR surveys in sandy coastal environments in many cases do yield coherent returns in settings where only mm-thick laminae, separated by mms or cms, are visible in cores or trenches (e.g. Hand, 1998, Dog Island thesis; Moore et al. (2004). To better understand conditions under which laminae could generate a significant radar response, a suite of layered models simulating sedimentary laminations is generated. In this suite the "magnetite-rich" (dielectric constant = 35, relative magnetic permeability = 1.2 and conductivity of 4 mS/m) layer thickness is kept constant at 1mm and the "saturated sand" (20, 1 and 1mS/m) interlayer spacings are varied. In each model, the interlayer thicknesses are set tsian distribution about a fixed mean. odels were run with means of 2 mm, 5 mm, and 50 mm, with standard deviations set to half of the means (1 mm, 2.5 mm and 25 mm respectively). To satisfy model constraints the Gaussian interlayer thickness distributions were then rounded to the nearest 0.25 mm, and set o a Gaus M no thinner than 1 mm (fig. 9). Fig.9 Schematic representation of model suites A mm interlayer space; B 5 mm interlayer space; C mm interlayer space. Dark zones represent layers. The interlayer spaces follow a Gaussian distribution and are not constants as shown in fi gure. 16

PAGE 24

The amplitudes and frequency characteristics of the returns from the layered models are examined for pulses with center frequencies of 100, 200 and 500 MHz. Sample traces are shown in fig 10a, 10b and 10c. Peak amplitudes of the interference patterns returns from the laminated zone s are on the order of or greater than the amplitudes of the 1cm models of the suite above, despite the fact that the amplitude of the return from each single lamina would be an order of magnitude smaller. The Gaussian distribution of laminae generates periods of constructive interference not simulated with fixed spacing cm-scale models, and it is clear that the assumed distribution of interlayer (in this case interlaminar) thicknesses influences the interference patterns. The frequency spectra of the traces are computed using a standard Fast Fourier Transform (fft) algorithm in Matlab. The ffts were run in each case for a time window of 22ns that spans energy returning from the layered package. The spectra, shown in Figure 11, have been averaged for three realizations of each model. correlated with the fact that reflected energy is being generated in response to the multiplrter s A spectral shift towards higher frequencies is observed in the returns for all three models (Figure 11a, 11b, 11c). When compared with the original input pulse, moreenergy is being returned to the surface at higher frequencies. This can possibly be e thin layers. The higher frequency components of the pulse correspond to showavelengths. Since wavelength is inversely proportional to the reflection amplitude(following the equation A d 4Ab/ b Widess, 1973), the higher frequency componentshould perceive correspondingly greater reflection coefficients. 17

PAGE 25

30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 time (ns)amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude100 MHz 200 MHz 500 MHz Fig 10a 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 time (ns)amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude100 MHz 200 MHz 500 MHz Fig.10b 18

PAGE 26

30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 time (ns)amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude 30 35 40 45 50 55 60 65 70 75 80 -0.02 -0.01 0 0.01 0.02 amplitude100 MHz 200 MHz 500 MHz Fig.10c Fig. 10 Reflections from a sequence of layers with fixed layer thickness (1mm) and varying interlayer spaces at different frequencies. 10a interlayer space =2 mm (mean), 10b interlayer space = 5 mm (mean), 10c interlayer space =50 mm (mean) 19

PAGE 27

0 20 40 60 80 100 120 140 160 180 200 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Averaged amplitude spectra of sequence with Gaussian distribution of interlayers at 100 MHzfrequency (MHz)amplitude fixed layer thickness =1mm, mean interlayer space =2mmmean interlayer space=5mmmean interlayer space=50mmoriginal pulse Fig.11a 50 100 150 200 250 300 350 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Averaged amplitude spectra of sequence with Gaussian distribution of interlayers at 200 MHzfrequency (MHz)amplitude fixed layer thickness =1mm, mean interlayer space =2mmmean interlayer space=5mmmean interlayer space=50mmoriginal pulse Fig. 11b 20

PAGE 28

100 200 300 400 500 600 700 800 900 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Averaged amplitude spectra of sequence with Gaussian distribution of interlayers at 500 MHzfrequency (MHz)amplitude fixed layer thickness =1mm, mean interlayer space =2mmmean interlayer space=5mmmean interlayer space=50mmoriginal pulse Fig.11c Fig.11 Averaged amplitude spectra for different models at 100 MHz (fig.11a), 200 MHz (fig.11b) and 500 MHz (fig.11c). Summary and discussion of model results Single layer models : The GPR resolution limit for isolated thin, K=35 beds within saturated sands (K=20) is examined, following the form of the seismic analysis of Widess (1973). Considering antennas typically employed in geologic studies, resolution limits decrease from ~40 cm (50 MHz) to 4 cm (500 MHz). Package with internal layering (30 cm package, cm-scale layering) : When the package thickness itself is at the thin bed resolution limit (100 MHz case), finer-scale internal layering tends to simply reduce overall reflection amplitude. When the package is thick (200 and 500 MHz examples) internal layers may be directly resolved or appear as an interference pattern. Returns from internal layering drop off dramatically in amplitude when interlayer and layers are both thin and similar in dimension, in this case on the order of 1 cm. This suggests that for significant amplitude returns in an 21

PAGE 29

environment characterized by the velocities assumed here, there must be variability in the layering, or layering on scales of ~5 cm or more. Fine laminae (separated by mm to cm): These sets of models deal with multiple thin layers, 1 mm in thickness, with mean interlayer spaces of 2 mm, 5 mm, and 50 mm. Reflection amplitudes from packages of laminae, can be as high as 10 or 20% of the equivalent thick bed reflection. Amplitudes are higher for the models with mean interlaminar spacing 5 and 50 times the lamina thickness, than for mean interlaminar spacing two times the lamina thickness. Thus, irregularity in laminar and interlaminar thicknesses tends to increase amplitudes. Models of returns from lamina packages show spectral shifts toward higher frequencies. Such a spectral shift in portions of a trace may thus perhaps be an indicator of fine-scale (mm to cm scale) layering within a unit. The models show, as expected, that higher frequency waves have stronger amplitude returns from lamina packages. Thus in theory, criterion for identification of the presence of laminations in a sequence calls for a comparative analysis at both low and high frequencies. This may be difficult in practice, however, as intrinsic attenuation increases with increasing frequency, and this effect will probably outweigh lamina effects. 22

PAGE 30

Case study: Waites Island, South Carolina Geologic setting Waites Island is located on the northeastern coast of South Carolina and forms a part of the Grand Strand barrier island system (fig.12). This 5 km long and 0.5 km wide Holocene island (Wright et al, 2001) has a seaward beach and dune system. Moving from the shore toward the mainland, it has five distinctive depositional environments: namely, washover / barrier island, salt marsh, back-barrier intertidal, back-barrier subtidal and fresh water peat (Adam et al, 2001), (fig.13). As part of a larger study of the stratigraphic development of Waites Island, extensive subsurface imaging was done using GPR. In this thesis, we examine profiles across two sites in the northern part of the island, where cores were taken. Data collection and processing Two sets of GPR profiles were collected using a PulseEKKO 100 system manufactured by Sensors and Software Inc. The first, in May 2002, was a long transect (LINE 1) run across the island, perpendicular to the shoreline, with 100 MHz antennas (Fig.12). Readings were collected at fixed time intervals, and traces were correlated with position every 5 meters. Trace positions were then linearly interpolated between the known values, and the profiles resampled to a uniform spacing of 0.25 m. The same core sites (fig. 12) were then revisited in February, 2003 when shorter profiles and CMPs centered over the two core sites were collected, both parallel and perpendicular to the shoreline, with 50 MHz, 100 MHz and 200 MHz antennas. Transmitters with voltages of 400 V (200 MHz and 100 MHz) and 1000 V (50 MHz) were used. Data processing was done with the aid of PulseEKKO version 4.22 software to remove noisy traces from the original records. Other processing steps involved merging data files and applying dewow filters and AGC gains to the data. 23

PAGE 31

Conversion of the two-way travel time to depth was done by the conventional CMP velocity analysis (Appendix 1). The best-fitting velocities were found to be 0.12 m/ns for the unsaturated zone and 0.07 m/ns was obtained for the saturated zone. The saturated and unsaturated zones are demarcated on the GPR reflection profiles on the basis of the position of the water table (fig. 14) The GPR profiles span locations where vibracores were collected in August, 2002 at sites WI1 and WI2 (on fig.12). Detailed descriptions of cores are given in Appendix 4. Site WI1 is located in the dunes of a ridge/dune system formed parallel to the shore and WI2 is located in a low in the dunes on the landward side of the barrier island (Eric Wright, personal communication). Further coring was done at the same sites in August, 2003. These cores were acquired for analysis in a Multi-Sensor Core Logger (MSCL) to determine sediment bulk density. Lengths of core sections were approximately 1m. A Schlumberger resistivity survey was conducted at site WI2 to obtain the ground apparent resistivity and hence conductivity informaion. Reflection profiles and CMPs Reflection profiles at site WI1 (fig. 14) show two prominent reflectors, R1 and R2, at depths of 2.5 m and 5 m respectively. R1 represents the water table whereas R2 is interpreted as reflections from a marsh layer, correlatable with the top of the marsh layer in core section WI1 (D) in fig.18. At site WI2, profiles show the water table as the single strong reflector (fig.15). These reflection events are also identified in the CMPs (fig.16 and fig17). The presence of reflections other than those mentioned above is noticeable in profiles at both sections. These additional reflection events cannot be associated with any conspicuous lithologic changes in the core sediments. To better understand the nature and characteristics of the radar reflections at Waites Island, the vibracore sediments were analyzed in the laboratory. 24

PAGE 32

Fig. 12 Infrared aerial photo of Waites Island showing cross-island GPR transect (LINE 1) and core sites WI1 and WI2 on LINE 1. 25

PAGE 33

Fig.13 Waites Island cross section along LINE 1 showing different depositional environments as interpreted from vibracores and GPR data (Eric Wright, personal communication) 26

PAGE 34

Fig. 14 Reflection profiles at core site WI1 27

PAGE 35

Fig.15 Reflections profiles from core site WI2 28

PAGE 36

separation (m)time (ns) 1 2 3 4 5 6 0 20 40 60 80 100 120 140 160 180 (a) separation (m)time (ns) 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 -20 0 20 40 60 80 100 120 140 160 180 (b) separation (m)time (ns) 1 2 3 4 5 6 7 8 9 -20 0 20 40 60 80 100 120 140 160 180 (c) Fig. 16. CMP at site WI1. (a) at 200 MHz, (b) at 100 MHz and (c ) at 50 MHz 29

PAGE 37

separation (m)time (ns) 1 2 3 4 5 6 0 20 40 60 80 100 120 140 160 180 (a) separation (m)time (ns) 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 0 20 40 60 80 100 120 140 160 180 (b) separation (m)time (ns) 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 0 20 40 60 80 100 120 140 160 180 (c) Fig.17 CMP at WI2 (a) at 200 MHz, (b) at 100 MHz and (c) at 50 MHz 30

PAGE 38

Laboratory Methods Core sediment analysis The split sections of cores (WI1 and WI2) were photographed (photos obtained from Marine Science at Coastal Carolina University, E. Wright, personal communication). Figures 18 and 19 show the split-core sections. The presence of finer layers of heavy minerals or beach laminations (e.g. Komar, 1974), in different portions of each of the cores is the most noticeable heterogeneity in the sediment. Section WI1 (A) of WI1 (collected in August, 2003) contains extensive layers (mm scale) of heavy minerals alternating with fine sand from a depth of 45 cm to 110 cm (fig. 18b). Sediments of core section WI1 (A) were analyzed for bulk density and porosity information by running it in the Multi-Sensor Core Logger at the University of Florida, Gainesville. Data sampled had a sampling interval of 0.5 cm. The gamma bulk density has a slightly high value from 50 cm to 80 cm of the core and a gradual drop is seen for core sediments after 80 cm (fig. 20). However, no other significant change is observed in the overall data. The corresponding porosity values increase from 80 cm to the base of the core (fig. 20). Grain-size analysis was done for samples of core section WI2 (A) using the sieve analysis method. Grain size distribution patterns are the same for all samples (fig.22). Heavy minerals formed the finer particles in all samples. To obtain a better image of the lamination patterns in WI1 (A), a lacquer peel was made from which the layer thicknesses were measured. The individual heavy mineral enriched layers are approximately 1 mm in thickness. At places they are very closely spaced forming a zone with a thickness of 2-5 cm (18b). The lacquer peel helped in the measurement of individual thin heavy mineral layer and their interspaces which would otherwise be difficult to discern from the split core. A Bartington susceptibility meter was used to acquire the magnetic susceptibility of sediment samples from 40 cm to 110 cm of the core where the heavy mineral laminations are in abundance. A sample volume of 8 cc was taken from the central portion of the core starting from 40 cm and ranging to 110 cm for measuring the susceptibility. Samples were collected every 2 cm. 31

PAGE 39

The susceptibility data as given in figure 22 shows an appreciable match with the lamination patterns obtained from the lacquer peel. Some samples of the concentrated heavy minerals from WI1 (A) at depths of 65-70 cm and 105-110 cm were separated using non-toxic heavy mineral separation liquid, Sodium Polytungstate. The primary objective of this separation was the identification of iron-oxide-bearing minerals such as magnetite from other heavy minerals (e.g. zircon, apatite). The presence of magnetite (dielectric constant value of 33.7 (Schon,1996) can serve as a prominent reflecting surface when forming a lamina Deposition of magnetite along lamination boundaries in dune sands causing GPR reflections has been noted by Harari (1996). Baker (1991) also correlated GPR reflections from heavy mineral layers in a barrier island sand deposit. On the WI1 (A) core, samples taken from 55-60 cm and 85-90 cm from the same core were treated with a strong bar magnet which resulted in separation of magnetite particles (fig. 23). Thus GPR returns from this setting are likely generated by the magnetite-bearing laminae, in addition to the water table, marsh contact, and other variations not seen by eye or in the porosity structure. 32

PAGE 40

Fig. 18 (a) Core sections of WI1, (b) layer patterns obtained from lacquer peel of WI1 (A). Core description given in Appendix 4. 33

PAGE 41

Fig. 19 Core sections at site WI2 34

PAGE 42

Fig. 20 Magnetic susceptibility, gamma bulk density and porosity values for core section WI1 (A) Heavyminerals(withmagnetite)undermicroscopeinbeachlaminationslaminae CoreWI1(A):70-80cmsectionFig. 21 Heavy mineral layers containing magnetite in core section WI1 (A) WI1(A)0123405060708090100110core length (cm)magnetic susceptibility (SI) WI1(A)0.60.650.70.750.8406080100core length (cm)porosity WI1 (A)1.31.41.51.61.7406080100core length (cm)gamma bulk density (g/cc) 35

PAGE 43

Fig. 22 Grain size distribution for core section WI2 (A). Samples collected at intervals of 5 cm as represented by rectangles on core diagram. 36

PAGE 44

Layered models simulating core laminations To test the hypothesis that the Waites Island radar profiles could in fact be produced by the fine-scale layering and other contacts seen in the cores (as well as the water table), two sets of models have been designed. Model 1 simulates the lamination patterns as obtained from the lacquer peel and photographs for core WI1, and model 2 the core WI2 based on descriptions of core photos. Fig. 23 gives the schematic presentations of the layered models. Models were required to (1) assume geologically plausible values of permittivity, magnetic permeability, and conductivity; (2) have contacts at observed depths; (3) fit the observed average saturated and unsaturated zone velocities; and (4) have bulk conductivity on the order of 1-5 mS/m (as observed in the field). Since the 1D model does not account for attenuation of energy due to geometric spreading, a factor of 1/r 2 where r is the distance traveled and, dttvr)( has been applied to the model traces. Model parameters Model parameters that satisfy the constraints are given in Table 1. As most petrophysical parameters (e.g. moisture content, organic content, iron-oxide content, grain shape and orientation) of the Waites core are unknown, these models are non-unique in the sense that many models could be generated that yield equivalent or better fits to the observations. Without better knowledge of core characteristics, fine-tuning of the models to better fit observed traces is not warranted. Here discussion is focused simply on the quality of these simple models, and addressing the question: can the basic features of the radar traces be explained from contacts and layer patterns visible in cores? 37

PAGE 45

Fig. 23 Schematic representation of Model 1 (A) and Model 2 (B) 38

PAGE 46

Table 1. Physical parameters used in model simulations Model 1 K (dielectric constant) r (relative magnetic permeability) S/m (conductivity) unsaturated sand 3 1 0.001 saturated sand 20 1 0.001 magnetite 15 1.1 0.004 mud 17 1 0.002 marsh 6 1 0.003 Model 2 K S/m unsaturated sand 6 1 0.001 saturated sand 20 1 0.004 magnetite 15 1.1 0.003 Analysis of models GPR traces were selected from the records at the two study sites. Plots of individual traces at different frequencies are compared to traces produced by the models as given in fig. 24 and fig. 25. Analysis focuses on the saturated zone, for which the water table reflection provides a convenient reference signal. In these models the water table is simulated as a 20 cm-thick contact across which electromagnetic properties change linearly. Pattern of returns: The 60 cm package of laminations present above the water table does not yield significant distinctive returns in either observed or model records. Radar wavelengths are longer in the faster unsaturated zone, and detectable returns would 39

PAGE 47

be confined to a short time window between direct arrivals and the water table reflections. Reflections from within the saturated zone for require the presence of changes that are not incorporated in Model 1. (Some of the observed returns may represent diffractions). The saturated zone of WI1 lacks heavy mineral laminations, but the mud and marsh layers include sporadic mud lenses and organic materials. Perhaps the muds lenses and organics are part of coherent layers on a larger scale. Model 2 is a laminated section (laminations present above and below water table) consisting of sand and magnetite layers (mm and cm scale). The 200 MHz model result shows reflection patterns similar to the GPR record (Figure 25a). The presence of thin magnetite layers is sufficient to explain the reflections seen in the 200 MHz record. However, comparing the lower frequency GPR responses at 100 MHz and 50 MHz at this site points out that lithologic changes on a larger scale than the laminations exist that are not perceptible in the core data, particularly in the deeper parts of the record. Attenuation: For both sites, exponential decay curve fits of the observed and model trace envelopes yield higher attenuation coefficients for the models. This discrepancy may be due to: (a) unrealistically high conductivities in the models; (b) poor modeling of the water table reflector; (c) underestimation of the true attenuation due to noise late in the record; and/or (d) underestimating contrasts (reflection coefficients) between layer types. Better estimates of the layer contrasts could be derived from amplitude-versus-offset analysis of the data (e.g. Kruse and Jol, 2003), which is beyond the scope of this work. Explanation (a) is unlikely, as the Schlumberger resistivity survey at site WI2 (fig. 28) shows a conductivity of 5 mS/m; at radar frequencies the effective terrain conductivity should be higher, rather than lower. Spectra: Comparison of spectra from models and data are of little utility, as there is not a good constraint on the spectra of the downgoing pulse in the observations. However, one can compare spectra of 100 MHz traces from sites with known laminations (near WI2) with those from sites without (near WI1). From model results of the previous section, it is infered that an upward shift in the amplitude spectrum should be associated with returns from fine layered packages will result with increasing frequencies. Traces from the 100 MHz (LINE 1) cross-island transect have been selected. Amplitude spectrum of traces consisting of reflections from fine layers in the saturated zone at site 40

PAGE 48

WI2 (Figure 29b) show a spectral shift relate to that from the saturated non-laminated zone of WI1 (Figure 29a). Thus, spectral analysis of traces may be a useful measure to identify presence of thin layers in coastal deposits. 41

PAGE 49

10 20 30 40 50 60 70 80 90 100 -0.2 -0.1 0 0.1 0.2 amplitude 10 20 30 40 50 60 70 80 90 100 -0.2 -0.1 0 0.1 0.2 time(ns)amplitudeoriginal trace model trace water table water table Fig.24a 20 30 40 50 60 70 80 90 100 -0.05 0 0.05 amplitude 20 30 40 50 60 70 80 90 100 -0.05 0 0.05 amplitudeoriginal trace model trace water table water table Fig.24b Fig.24 Comparison of real and model traces for site WI1. 24aMHz, 24b 100MHz 42

PAGE 50

10 20 30 40 50 60 70 80 -0.2 -0.1 0 0.1 0.2 amplitude 10 20 30 40 50 60 70 80 -0.2 -0.1 0 0.1 0.2 time(ns)amplitudeoriginal trace model trace water table water table Fig. 25a 20 30 40 50 60 70 80 -0.2 -0.1 0 0.1 0.2 amplitude 20 30 40 50 60 70 80 -0.2 -0.1 0 0.1 0.2 time(ns)amplitude original tracemodel trace water table water table Fig. 25b 43

PAGE 51

30 40 50 60 70 80 90 100 -0.2 -0.1 0 0.1 0.2 amplitude 30 40 50 60 70 80 90 100 -0.2 -0.1 0 0.1 0.2 time(ns)amplitudeoriginal trace model trace water table water table Fig.25c Fig. 25 Comparison of real and model traces for site WI1. 25a MHz, 25b 100 MHz, 25c 50 MHz 44

PAGE 52

Averaged envelope of traces at site WI1 at 200MHzy = 1408.4e-0.043x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+03020406080100time (ns)amplitude Averagedenvelope Expon. (Averagedenvelope) Envelope of model trace at site WI1 at 200MHzy = 878.47e-0.0809x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+03020406080100time (ns)amplitude Model traceenvelope Expon. (Modeltrace envelope) Fig.26a 45

PAGE 53

Averaged envelope of traces at site WI1 at 100MHzy = 10943e-0.0385x0.00E+002.00E+034.00E+036.00E+038.00E+031.00E+04050100150time (ns)amplitude Avereged envelope Expon. (Averegedenvelope) Envelope of model trace at site WI1 at 100 MHzy = 16318e-0.0519x0.00E+002.00E+034.00E+036.00E+038.00E+031.00E+041.20E+041.40E+041.60E+04050100150time (ns)amplitude Model traceenvelope Expon. (Model traceenvelope) Fig.26b Fig. 26 Expotential trend of attenuation in real and model data for site WI1 at different frequencies. 46

PAGE 54

Averaged envelope of traces at site WI2 at 200MHz y = 2246.3e-0.0647x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+033.50E+030204060time (ns)amplitude Averaged envelope Expon. (Averagedenvelope) Envelope of model trace at site WI2 at 200MHzy = 3813.4e-0.0808x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+030204060time (ns)amplitude Model traceenveloope Expon. (Modeltrace enveloope) Fig.27a 47

PAGE 55

Averaged envelope of traces at site WI2 at 100MHzy = 1780.1e-0.0316x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+03020406080time (ns)amplitude Averaged envelope Expon. (Averagedenvelope) Envelope of model trace at site WI2 at 100MHzy = 3332.1e-0.0466x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+033.50E+03050100time (ns)amplitude Model traceenvelope Expon. (Modeltrace envelope) Fig. 27b 48

PAGE 56

Averaged envelope of traces at site WI2 at 50MHzy = 5224.1e-0.021x0.00E+005.00E+021.00E+031.50E+032.00E+032.50E+033.00E+033.50E+034.00E+034.50E+03020406080100120time (ns)amplitude Averaged envelope Expon. (Averagedenvelope) Envelope of model trace at site WI2 at 50MHzy = 1108.3e-0.0368x0.00E+002.00E+024.00E+026.00E+028.00E+021.00E+031.20E+031.40E+031.60E+031.80E+032.00E+03020406080100120time (ns)amplitude Model trace envelope Expon. (Model traceenvelope) Fig.27c Fig. 27 Expotential trend of attenuation in real and model data for site WI2 at different frequencies. 49

PAGE 57

Schlumberger resistivity at WI21101001000110100ab/2 (m)apparent resistivity (ohm-m) Apparent resistivity: 195.15 ohm-m (layer1) 64.382 ohm-m (layer 2) 74.812 ohm-m (layer 3) 198.5 ohm-m (layer 4) Fig 28 Layer model apparent resistivity and layer thicknesses estimation (program used ENVIRT6). 50

PAGE 58

0 50 100 150 200 250 300 0 1 2 3 4 5 6 7 8x 104 frequency (MHz)amplitudeAveraged amplitude spectra of GPR traces at site WI1 at 100MHz Fig.29a 0 50 100 150 200 250 300 0 1 2 3 4 5 6 7 8x 104 frequency (MHz)amplitudeAveraged amplitude spectra of GPR traces at site WI2 at 100MHz Fig.29b Fig. 29 Averaged amplitude spectra of GPR traces at two different sites (close to WI1 and WI2) along LINE1. 51

PAGE 59

Conclusions Thin beds are common in GPR surveys in sediments, but extracting information on thin beds from GPR profiles has not been widely discussed. From simple models of radar wave propagation, and comparative analysis of GPR records and core sediment records from a sandy barrier island, the following conclusions are drawn: 1) The presence of multiple thin beds, such as magnetite-enriched laminae in coastal sedimentary deposits, can produce low-amplitude but detectable GPR returns at the commonly-used frequencies of 100 to 500 MHz. The returns cannot be directly correlated with individual laminations. Return amplitudes will depend on the distribution of layer and "background interlayer" thicknesses, and as the difference between layer and interlayer thickness increases. Layering within packages with package thickness near the thin bed limit would be difficult to detect. 2) Spectral analysis of both model and real traces indicate that the presence of mm-scale laminations is accompanied by a spectral shift towards higher frequencies. Spectral shifts may serve as an indicator of local zones of thin laminations. 3) Comparison of observations with models based on contacts seen in cores in Waites Island, SC, suggests that thin magnetite-rich layers are important contributors to the radar signal, especially at 200 MHz, but do not in themselves explain all the major features of the radar record. Contacts not noted in cores, perhaps subtle variations in porosity, are required to explain the basic pattern of returns and attenuation characteristics of the Waites GPR profiles. 52

PAGE 60

References Annan, A. P. (2001). Ground penetrating radar workshop notes. Sensors & Software Mississauga, Ontario. Bano, M. (1996). "Modeling and inverse Q imaging of ground penetrating radar waves in 1 and 2D." Geophysical Research Letters 23(22): 3123-3126. Ellefsen, K., J. (1999). "Effects of layered sediments on the guided wave in crosswell radar data." Geophysics 64(6): 1698-1707. Gawthorpe, R., L., Collier, R., E., LI., Alexander, J., Bridge., J., S., Leeder, M., R. "Ground penetrating radar: application to sandybody geometry and heterogeneity studies." Geological Society special publication 73: 421-432. Greaves, R., J., Lesmes, D., P., Lee., J., M., Toksoz, M., N. (1996). "Velocity variations and water content estimated from multi-offset, ground-penetrating radar." Geophysics 61(3): 683-695. Hagrey, S., A., Muller, C. (2000). "GPR study of pore water content and salinity in sand." Geophysical Prospecting 48: 63-85. Hand, J. (1998). Geologic History of Modern Morphodynamics of Dog Island, Franklin County, Florida. Department of Geology Tampa, University of South Florida: 132. Harari, Z. (1996). "Ground Penetrating Radar (GPR) for imaging strtigraphic features and groundwater in sand dunes." Journal of Applied Geophysics 36(1): 43-52. Heinz, J., Aigner, t., Ed. (2003). Three-dimensional GPR analysis of various Quaternary gravel-bed braided river deposits (southwestern Germany) Ground Penetrating Radar in Sediments. London, The Geological Society Special Publication. Irving, J. D. and R. J. Knight (2003). "Removal of wavelet dispersion from ground-penetrating radar data." Geophysics 68(3): 960-970. Jol, H., Lawton, D., C., Smith, D., G. (2003). "Ground penetrating radar: 2-D and 3-D subsurface imaging of a coastal barrier spit, Long Beach, WA, USA." Geomorpholoy 53: 165-181. 53

PAGE 61

Jol, H., M., Lawton, Smith, D., G., Meyers, R., A. (1996). "Digital ground penetrating radar (GPR): A new geophysical tool for coastal barrier research (examples from the Atlantic, Gulf and Pacific Coasts, U.S.A)." Journal of Coastal Research 12(4): 960-968. Jol, H., M., Smith, D., G. (1991). "Ground penetrating radar of northern lacustrine deltas." Canadian Journal of Earth Sciences 28: 1939-1947. Jol, H., Smith, D., G. (1995). "Ground penetrating radar surveys of peatlands for oilfield pipelines in Canada." Applied Geophysics 34: 109-123. Jol, H., Smith, D., G., Lawton, D., C. (1996). Ground penetrating radar: High resolution stratigraphic analysis of coastal and fluvial environments GCSSEPM Foundation 17th Annual Research Conference. Keary, P., Brooks, M. (1991). An Introduction to Geophysical Exploration London, Blackwell Sc. Publications. Knapp, R. W. (1990). "Vertical resolution of thick beds, thin bedsand thin-bed cyclothems." Geophysics 65: 1183-1190. Knight, R., J., Nur, A. (1986). "Dielectric enhancement due to the presence of thin gas pockets." Transactions of the SPWLA twenty-seventh annual logging symposium 27: 1-JJ11. Knoll, M. D. K., R (1994). Relationships between dielectric and hydrologic properties of sand-clay mixtures. Proceedings of the fifth International Conference on Ground Penetrating Radar, Waterloo Center for Groundwater Research,, Kitchener, Ontario, Canada. Komar, P., D. (1976). Beach Processses and Sedimentation Englewood Cliffs, New Jersey, Prentice-Hall, Inc. Kruse, S. K. J., H.M., Ed. (2003). Amplitude analysis of repetitive GPR reflections on a Lake Bonneville delta, Utah Ground Penetrating Radar in Sediments. London, Geological Society. Kunz, K. S. L., R. J. (1993). The Finite Difference Time Domain Method for Electromagnetics Boca Raton, CRC Press. Mayer, L. A. (1980). "Deep sea carbonates: physical property relationships and the origin of high-frequency acoustic reflectors." Marine Geology 38: 165-183. Neal, A., Pontee, N., I., Pye, K., Richards, J. (2002). "Internal structure of mixed-sand-and-gravel beach deposits revealed using ground-penetrating radar." Sedimentology 49: 789-804. 54

PAGE 62

Neal, A., Richards., Pye, K. (2003). "Sedimentology of coarse-clastic beach-ridge deposits, Essex, southeast England." Sedimentary Geology Nishikawa, T., Ito, M. (2000). "Late Pleistocene barrier-island development reconstructed from genetic classification and timing of erosional surfaces, paleo-Tokyo Bay, Japan." Sedimentary Geology 137: 25-42. Noon, D., A., Stickley, G., F., Longstaff, D. (1998). "A frequency-independant charahterisation of GPR penetration and resolution performance." Journal of Applied Geophysics 40: 127-137. Olhoeft, G. R. (1994). Petrophysical causes of electromagnetic dispersion 5th International Conference on Ground Penetarting Radar, Kitcener, Ontario. Olhoeft, G. R. (1998). Electrical, magnetic and geometric properties that determine ground penetrating radar performance 7th Internatioal Conference on Ground Penetrating Radar, Lawrence, Kansas, USA, The Univ. of Kansas. Pettijohn, F. J. (1984). Sedimentary Rocks Delhi, CBS Publishers and Distributors. Schmalz, B., Lennartz, D., Wachsmuth, D. (2002). "Analyses of soil water content variations and GPR attribute distribuions." Journal of Hydrology 267: 217-226. Schon, J. H. (1996). Physical properties of rocks: fundamentals and principles of petrophysics. Handbook of geophysical exploration K. T. Helbig, S., Pergamon. 18: 583. Smith, D., G., Jol, H., M. (1997). "Radar structure of Gilbert-type delta, Peyto Lake, Banff National Park, Canada." Sedimentary Geology 113: 195-209. Tercier, P., Knight., R, Jol, H. (2000). "A comparison of the correlation structure in GPR images of deltaic and barrier-spit depositional environments." Geophysics 65(4): 1142-1153. Tomer, M. D., Boll, J., Kung, J. S. (1996). "Detecting illuvial lamallae in fine sand using ground-penetrating radar." Soil Science 161(2): 121-129. Topp, G. C. D., J.L (1985). "Measuremnet of soil water content using time-domain reflectometry (TDR): a field evaluation." Soil Science Society of American Journal 49: 19-24. van Dam, R. L. (2001). Causes of Ground-Penetrating Radar Reflections in Sediment. Amsterdam, Vrije Universiteit. van Dam, R. L., Schlager,W., Dekkers, M.J. & Huisman, J.A. (2002). "Iron oxides as a cause of GPR reflections." Geophysics 67: 536-545. 55

PAGE 63

van Dam, R. L., van Den Berg, E.H., van Heteren, S., K., C., Kenter (2002). "Influence of organic matter in soils on radar-wave reflection: sedimentological implications." Journal of Sedimentary Research 72: 341-352. van Dam, R. L. S., W. (2000). "Identifying causes of ground-penetrating radar reflectionsusing time-domain reflectometry and sedimentological analyses." Sedimentology 47: 435-449. van Heteren S., FitzGerald, D., M., Barber, D.C., Kelley, J., T., Belknap, D., F. (1996). "Volumetric analysis of a New England barrier system using ground-penetrating-radar and coring techniques." Journal of Geology 104: 471-483. van Overmeeren, R., A (1998). "Radar facies unconsolidated sediments in the Netherlands: A radar stratigraphy interpretation method for hydrogeology." Journal of Applied Geophysics 40: 1-18. Widess, M. B. (1973). "How thin is a thin bed?" Geophysics 38(6): 1176-1180. Woodward, J., Ashworth, P.J., Best, J.L., Sambrook Smith, Ed. (2003). The use and application of GPR in sandy fluvial environments: methodological considerations Ground Penetrating Radar in Sediments. London, Geological Society Special Publications. Yee, K. S. (1966). "Numerical solution of initial boundary problems involving Maxwell's equations in isotropic media." Institute of electrical and Electronics Engineers Transactions on Antennas and Propagation AP 14: 302-309. 56

PAGE 64

Appendices 57

PAGE 65

Appendix 1 Radar wave propagation The properties of the ground determine the key aspects of the radar wave propagation, namely (1) its velocity and (2) its attenuation. For high frequencies or very low conductivity, the above equation becomes /cv where c = electromagnetic wave velocity in vacuum (3 x 10 8 m/s) and the properties of the medium the wave travels through are = magnetic permeability and = dielectric permittivity. So both permittivity and magnetic permeability play important roles in determining the velocity. Electromagnetic waves attenuate with increasing distance from the transmitter due to spherical spreading of energy, scattering from heterogeneities and dissipation of energy into subsurface materials (Annan and Davis, 1989). The attenuation of the downgoing pulse due to the latter can be expressed by A= A 0 e -z where A is the amplitude of the pulse at some distance z from the source, A 0 is the pulse amplitude at a given distance z 0 and /5.0 (Van Dam, 2000). Reflection coefficient, resolution, depth of penetration Reflection coefficient determines the amplitude response of the reflected signal from layers of different dielectric properties. For normal incidence, reflection coefficient 1212ZZZZR Z 1 and Z 2 are impedances of lower layer and upper layer respectively. (Impedance, )/(jjZ where 1j f 2 angular frequency, expressed in radians/s). Physical parameters influencing radar wave propagation Dielectric constant, electric conductivity and magnetic permeability are three parameters influencing radar wave propagation. Dielectric constant: Of the greatest significance among the three, is relative permittivity. Relative permittivity ( r ), also referred to as dielectric constant (K), is defined as the ratio of material permittivity (), to that of vacuum ( 0 = 8.85 x 10 -12 F/m). 58

PAGE 66

Appendix 1 (continued) Dielectric constants for earth materials are strongly influenced by water content, and hence by both degree of saturation and porosity of the material. Table 1 shows dielectric constant values for some geologic materials. While air has a K value of 1, water has K 80. So an increase in water content is associated with a corresponding increase in permittivity. A porous material with greater water content than a nonporous substance will show a greater K value. Experiments show that bulk dielectric constant is also influenced by the pore internal geometry and pore fluid configuration (Endres and Knight, 1992), and thus mixing formulas that account for these effects are required for full understanding of dielectric constants. Clay content, which influences water retention capacity, also governs dielectric properties (Schon, 1996). Electrical conductivity According to Archies equation (Archie, 1942), bulk conductivity () of porous materials varies with water content. cwnmsa where = porosity, w = pore water conductivity, c = soil grain surface conductivity, a = constant between 0.4 to 2, m=constant between 1.3 to 2.5, s= degree of saturation, and n = constant ~ 2 Archies Law shows that high conductivities in geologic materials may be due to both grain size effects (e.g. clays) or pore water influence (e.g. saline water). For most materials, the conductivity is assumed to be isotropic although it may vary in anisotropic materials such as along bedding planes in stratified rocks (Telford, 1984). Magnetic permeability It is given by )1(0k where 40 x 10 -7 H/m is the magnetic permeability for vacuum and k is magnetic susceptibility. For most GPR purposes, magnetic permeability does not hold a major influence as compared to the electrical effects. However, the presence of iron oxides e.g magnetite in soils or rocks will result in changes in magnetic susceptibility. 59

PAGE 67

Appendix 1 (continued) Table 1. Dielectric constant, conductivity, velocity and attenuation of some geologic material (from Annan and Davis, 2002) Material K (mS/m) v (m/ns) a (dB/m) Air 1 0 0.30 0 Fresh water 80 0.5 0.033 0.1 Sea water 80 3 x 10 3 0.01 10 3 Dry sand 3-5 0.01 0.15 0.01 Saturated sand 20-30 0.1-1.0 0.06 0.03-0.3 Limestone 4-8 0.5-2 0.12 0.4-1 Shales 5-15 1-100 0.09 1-100 Silts 5-30 1-100 0.07 1-100 Clays 5-40 2-1000 0.06 1-300 Granite 4-6 0.01-1 0.13 0.01-1 Ice 3-4 0.01 0.16 0.01 GPR resolution and depth of penetration Resolution involves the detection of reflection events separately from one another in temporal and/or spatial domains and needs to be considered in both the vertical and horizontal directions. Vertical resolution Vertical resolution implies distinction between reflections from closely spaced contacts, one overlying the other, e.g. reflections occurring from the top and bottom surfaces of a sedimentary layer. It depends on the pulse width and is generally considered to be of the dominant wavelength (Kearey and Brooks, 1991). Because resolution increases with decreasing dominant wavelength, it increases with increasing frequency. However, with increasing depth higher frequencies undergo greater attenuation due to energy loss and hence vertical resolution generally decreases with increasing depth (Irving and Knight, 2003). 60

PAGE 68

Appendix 1 (continued) Horizontal resolution Horizontal resolution expresses GPRs ability to distinguish reflection events from closely spaced targets at equal depths. It depends on the Fresnel zone, which is the interface from where the reflected energy within half the transmitted wavelength undergoes constructive interference to enhance the resultant reflection. The Fresnel zone width is given by the following equation: )2(zw for z >> where w = width, = wavelength of source and z =reflector depth (Kearey and Brooks, 1991). The Fresnel equation governs horizontal resolution because reflectors separated by a distance less than the Fresnel zone width cannot be distinguished as separate events in the reflection records (Kearey and Brooks, 1991). Horizontal resolution decreases with depth as the Fresnel zone width increases with depth, but also decreases with increasing depth of investigation due to the dispersion effects (loss of higher frequencies) discussed for vertical resolution. Depth of penetration Because terrain conductivities are frequency-dependent, and generally increase with increasing frequency, as discussed above, depth of penetration is largely dependent on the transmitter frequency. There is thus a tradeoff--higher frequency will limit the depth of investigation but will result in a better resolution. Fig.1 gives a relationship between different antenna frequencies and probable depth of penetration in unconsolidated sediments (Smith and Jol, 1995). 61

PAGE 69

Appendix 1 (continued) Fig.1 Comparison of different antenna frequencies with maximum probable depths of penetration in Quarternary unconsolidated sediments. Interpolation of points along best fit line indicates maximum robable depths of investigation of 66m and 18m for 12.5 MHz and 400MHz respectively (Smith and Jol, 1995). 62

PAGE 70

Appendix 2 Sedimentary Environment Reflectors Reference Coastal Barrier Island Spit Beach ridge Lacustrine deltas Fluvial deposits Aeolian deposits Sand-silt contact Freshwater-brackish water contact Peat, heavy mineral layers Sand-mud interface, lag deposits Cheneir deposit (sand-shell, sand-silt contacts) Mud-peat contact Sand-peat contact Peat-clay contact Silt, gravel, peat, water table Heavy minerals Sand-silt contacts, organic content, iron-oxide (goethite with high water retention capacity) content Soil lamellae Jol et al, 1996 Baker, 1991 Van Heteren et al, 1998 Daly et al, 2002 Neal et al, 2002 Jol and Smith, 1991 Van Heteren, 1997 Wood, 1990 Vandenberghe and van Overmeeren, 1999 Harari,1996 van Dam and Schlager, 2002 Tomer et al, 1996 63

PAGE 71

Appendix 3 Layer patterns in sedimentary environments Delineation of sedimentary features has been a prime objective of stratigraphic research. Due to its efficiency as a high-resolution subsurface exploring tool, GPR has been greatly employed to these studies. In this section, sedimentary bed geometry in some depositional settings is discussed as a prelude to the following sections that will deal with GPR response to diverse layer patterns. Beds and laminations A primary depositional feature of sedimentary succession is layers or stratification which is defined by mineralogical and textural differences of a layer from its adjacent ones. Classified on the basis of their morphology, sedimentary layers are designated as beds (thickness > 1cm) or laminations (thickness < 1cm). McKee and Weir (1953) proposed further subdivision of beds and laminations on the basis of their thickness. A bedded sequence itself may be composed of several beds that are defined by bedding planes. Pettijohn (1984) described the following four classes of such sequences in terms of uniformity of bed thickness and their lateral continuity: 1. Beds equal or subequal in thickness, laterally uniform in thickness and continuous 2. Beds unequal in thickness, laterally uniform and continuous, 3. Beds unequal in thickness, laterally variable in thickness but still continuous 4. Beds unequal in thickness, laterally variable and discontinuous. Excluding the rarely occurring massive bed, most beds are characterized by their internal structures and their orientation. The most common internal structure within beds is laminations. These interlaminations, formed by mineralogical and textural heterogeneity of components, can be repetitive in nature. They often display either a parallelism or non-parallelism to the bounding surfaces, forming cross-laminations in the latter case. These laminations, found mostly in fine grained sands and silts, show thickness of 0.5 mm to 1.0 mm (Pettijohn, 1984). Gradational changes within a bedded sequence result in graded bedding as seen in Bouma cycle of turbidites. 64

PAGE 72

Appendix 3 (continued) Figure (from Collinson and Thompson, 1989) below shows different layer patterns encountered in sedimentary environments. Beds and laminations in different sedimentary environments Layer thickness is largely a function of the grain size, which in turn, is controlled by the nature of the depositional processes responsible for the formation of layered sequences. Hence it is expected that the pattern of layers deposited by different processes will exhibit some uniqueness depending on the depositional settings. An overview of the mode of occurrence of sedimentary layers is given below. Aeolian deposits: Sand dunes of aeolian origin are often internally characterized by sets of cross-beds. These cross-beds can be of several types e.g. tabular-planar, wedge-planar, and convex upwards, depending on their geometry and orientation. Associated with the beddings are small scale (cm Appendix 3 (continued) 65

PAGE 73

scale) internal stratifications which, in many cases, are defined by the presence of heavy minerals. Bounding surfaces, forests, bottomsets, onlap and overlap features in aeolian sand dunes have been well documented with ground penetrating radar. Fluvial deposits: Cross-bedding and laminations are the dominant internal structures in fluvially deposited sandstones. Sandstone units can extend vertically for meters forming successive sequences of thick beds. Parallel laminations, defined by the grain diameter, occurring parallel to flat bed surfaces also occur in both aqueous and aeolian sand deposits. Fine laminations in shale with thicknesses in the range of 0.5mm to 1mm occur as alternations of 1) coarse and fine grains e.g. silt and clay, 2) light and dark of organic layers, 3) calcium carbonate and silt (Pettijohn, 1984). Glacio-lacustrine deposits: The most significant layered sequence of glacio-lacustrine deposits is that of varves. Formed on glacial lake bottoms, varves show a typical couplet structure and can attribute to extensive laminated sequences when deposited for a long duration of time. These cyclic layers, representing seasonal changes, show very consistent layer configuration, the individual layer thicknesses varying from mm scale in most instances to cm scale and forming parallel laminations. Coastal deposits: Dunes and ripple marks are the primary sedimentary structures in coastal environments. Coastal sand dunes, like aeolian sand dunes, also show internal cross-stratification. Fine-scale layering is observed in beach laminations formed by alternating laminae of light (quartz and feldspar) and dark minerals (heavy minerals). Best seen in cross-sections, beach laminations are 1 to 20mm thick (Komar, 1976) and may extend laterally upto 25m (Thompson, 1937). 66

PAGE 74

Appendix 3 (continued) Volcaniclastic deposits: Among the pyroclastic/volcaniclastic deposits, accumulation in layered sequence takes place in case of ash falls and base surges. Ash falls, with well-defined bedding, show a decreasing trend of bed thickness away from its source. Base-surge deposits show internal laminations and cross-laminations. Statistical analysis of layered sedimentary sequence Stratigraphic sequences defined by rhythmic or repetitive lithounits is not an infrequent occurrence in nature. These units follow a definite pattern of spatial distribution e.g. they may be of equivalent thicknesses or may be equally spaced. A statistical approach is often taken to describe this lithostratigraphy. Mutual dependence of layers in vertical sequences has been studied in details by applying Markov chain analysis. Statistical study of layer thickness has been sparse. In recent years varve thickness analysis and time series analysis of cyclic lithounits have been carried out. A Gaussian distribution pattern of layer thickness can be expected as in the histogram below showing frequency of varve layer thickness (mm scale) from Lake Elk (data obtained from Eric Oches (personal communication) 0 1 2 3 4 5 6 7 8 9 10 0 100 200 300 400 500 600 Layer thicknessFrequency 67

PAGE 75

Appendix 4 Above core section is equivalent to core WI1 (A). Core descriptions obtained from Eric Wright, (personal communication), Coastal Carolina University. 68

PAGE 76

Appendix 4 (continued) WI1contd. 050100 0-1.79m:DUNE0-1.16m:lightyellowishbrowngradationaltolightolivebrownfinesands;rootedattopandbecominglaminateddownward.0.95-1.04m:thinangledlaminaewithhollowburrow(1.02)1.04-1.18m:lowangledthicklaminae1.18-1.79m:lightyellowishbrownfinesands;thickplanarlaminaewithheavybedsprominentattop.1.39m:horizontalwoodyroot1.57-1.69m:angledlaminae1.79-1.89:INTERDUNE1.79-1.87m:greenishgraylaminatedfinesandwithorganic-richlensesmorecommonattop;thin(<1cm)oxidizedlayerformingsharpcontactattop.surfacesediment 69

PAGE 77

Appendix 4 (continued) WI 1 contd. 70

PAGE 78

Appendix 4 (continued) WI1 contd. 71

PAGE 79

Appendix 4 (continued) WI2Length(m):2.02m 050100 0-0.34m:darkolivegrayveryfinesandwithcommonveryfineheavies0-0.15m:fewwiryroots/rootpieces0.15-0.18m:largefleshyroots,smallerroots0.34-0.40m:lightgrayfinesandwithcommonveryfineheavies0.40-1.00m:darkolivegrayveryfinesandwithcommonveryfineheavies0.52m:burrow(lens)ofdifferentcoloredsand0.57-0.75m:contortedbedding0.63m:sharplyangledlaminae0.75-0.85m:prominentcurvedlaminae0.75-2.02m:laminae(laminationsmostlyfaint)0.90-0.92m:bedofheavies 72

PAGE 80

Appendix 4 (continued) 73

PAGE 81

Appendix 4 (continued) 1.00-1.27mfinesandswithshellfragments1.07mwholecocinashell1.12mbedoffinesandswithabundantshellfragments1.25-1.27mthickbedofsandswithabundantshellfragments1.27-2.00msandswithoutshellfragments;distinctbedding(duetosedsizeorpresenceofheavies?)1.38mburrowfilledwithmuddysandsBeds1.27-1.48m1.48-1.52m1.52-1.57m1.57-1.62m1.62-1.63m1.63-1.68m1.68-1.69m1.69-1.72m1.72-1.74m1.74-1.79m1.79-1.84m1.84-1.87m1.87-1.90m1.90-1.92m1.92-1.94m1.94-1.96m1.96-2.00m1.98mBorrow 74

PAGE 82

Appendix 4 (continued) 2.00-2.41mfinedarksandslotsofheaviesthroughoutnoshellsorshellfragments(distortedbeddingorburrows)2.39-2.41mrootsandfillerfromthesurface(ifneededthecorecatchersedsareinaplasticbagmarked0208wi19,inthewarehouse) 75

PAGE 83

Appendix 4 (continued) 76

PAGE 84

Appendix 4 (continued) depth (m) 6 7 8 9 10 11 -2 -1 0 1 2 3 4 Reflection profile at site WI 2 at 200MHz Reflection profile at site WI2 (August, 2003) 77

PAGE 85

Appendix 4 (continued) Beach lamination in dune outcrop at Waites Island. depth(m)distance (m) 8 10 12 14 16 18 -4 -3 -2 -1 0 1 2 Reflection profile of dune outcrop in Waites Island 78

PAGE 86

Appendix 4 (continued) Interference patterns observed in reflection profile over a heavy mineral enriched trench, in the beach in Jekyll Island, Georgia 79