Oluwatosin Akinpelu C 201009 PhD Thesisghgh

7/26/2019 Oluwatosin Akinpelu C 201009 PhD Thesisghgh

1/295

Ground Penetrating Radar Imaging of Ancient Clastic

Deposits: A Tool for Three-Dimensional OutcropStudies

by

Oluwatosin Caleb Akinpelu

A thesis submitted in conformity with the requirementsfor the degree of Doctor of Philosophy

Geology DepartmentUniversity of Toronto


2/295

Ground Penetrating Radar Imaging of Ancient Clastic

Deposits: A Tool for Three-Dimensional Outcrop Studies

Oluwatosin Caleb Akinpelu

Doctor of Philosophy

Geology DepartmentUniversity of Toronto

2010

Abstract

The growing need for better definition of flow units and depositional heterogeneities in petroleum

reservoirs and aquifers has stimulated a renewed interest in outcrop studies as reservoir analogues

in the last two decades. Despite this surge in interest, outcrop studies remain largely two-

dimensional; a major limitation to direct application of outcrop knowledge to the three

dimensional heterogeneous world of subsurface reservoirs. Behind-outcrop Ground Penetrating

Radar (GPR) imaging provides high-resolution geophysical data, which when combined with two

dimensional architectural outcrop observation, becomes a powerful interpretation tool. Due to the

high resolution, non-destructive and non-invasive nature of the GPR signal, as well as its

reflection-amplitude sensitivity to shaly lithologies, three-dimensional outcrop studies combining

two dimensional architectural element data and behind-outcrop GPR imaging hold significant

promise with the potential to revolutionize outcrop studies the way seismic imaging changed

basin analysis.


3/295

architectural-element mapping with GPR imaging to obtain three dimensional architectural data

from outcrops.

Case studies from a variety of clastic deposits: Whirlpool Formation (Niagara Escarpment),

Navajo Sandstone (Moab, Utah), Dunvegan Formation (Pink Mountain, British Columbia),

Chinle Formation (Southern Utah) and St. Mary River Formation (Alberta) demonstrate the

usefulness of this approach for better interpretation of outcrop scale ancient depositional

processes and ultimately as a tool for refining existing facies models, as well as a predictive tool

for subsurface reservoir modelling. While this approach is quite promising for detailed three-

dimensional outcrop studies, it is not an all-purpose panacea; thick overburden, poor antenna-

ground coupling in rough terrains typical of outcrops, low penetration and rapid signal attenuation

in mudstone and diagenetic clay- rich deposits often limit the prospects of this novel technique.


4/295

Acknowledgments

I sincerely appreciate the support and guidance of Dr. Andrew Miall throughout the

duration of my graduate studies and especially in seeing this research project to

completion; thank you for giving me an opportunity of a lifetime.

I am particularly grateful to Dr. Nick Eyles for releasing the Ground Penetrating Radar

equipment from his research group throughout the data collection phase of my research;

your guidance in the early years helped set me in the right direction. To Ms. Lynn

Slotkin, I appreciate your assistance from the inception of my graduate program; even

those blunt e-mails were motivational. The painstaking review of the thesis and

suggestions by Dr. Rebecca Ghent and Dr. Uli Wortmann are also well-appreciated.

GPR Field assistance by Thomas Meulendyk helped avoid several field seasons of futile

data collection efforts; I am thankful for helpful tips on GPR data collection and

processing. Field assistance and technical support from Tudorel Ciuculescu remains

matchless; I am highly indebted to you for those long summer days at the outcrops.

Gerald Bryant, I owe a few chapters of my thesis to your assistance, accommodation and

field guidance on all my Utah field trips; I am very thankful for your help.

I am also grateful for funding provided by Ontario Graduate Scholarship, Natural

Sciences and Engineering Research Council, American Society of Petroleum Geologists


5/295

To Toyin and Elizabeth who bore the brunt of my dedicated effort to complete this

research project and for the countless hours of field assistance and proof-reading; I owe

the completion of this research project to you.


6/295

Table of Contents

Abstract ii

Acknowledgement iv

Table of contents vi

List of Tables viii

List of Figures ix

List of Appendices xiii

Chapter 1: Research Problem, Objectives and Significance 1

Chapter 2: Research Methodology: 9

Pre-survey Assessment 13

GPR Survey Planning 18

GPR Data Acquisition and Recording 28

GPR Data Processing 29

GPR Interpretation Methodology and Radar Stratigraphy 47

Chapter 3: Case Studies:


7/295

Shinarump Conglomerate, Hurricane Mesa (Utah) 137

St. Mary River Formation (Monarch, Alberta) 153

Chapter 4: Summary, conclusion and recommendation for future research 170

References 179

Appendix 1: Conductivity, relative permittivity and radar velocity data in sediment 219

Appendix 2: GPR instrumentation 220

Appendix 3: List of published GPR studies on sediments 227

Appendix 4: GPR Data Processing Software and Display 279

Appendix 5: GPR profiles from study locations 280


8/295

List of Tables

Table 1 GPR data sheet 28

Table 2 Radar facies chart 61

Table 3 Whirlpool Sandstone lithofacies 75

Table 4 Whirlpool Sandstone Radar Facies 82

Table 5 Navajo Sandstone Radar Facies 105

Table 6 Dunvegan Formation Radar Facies 128

Table A1 Typical radar propagation data in sediment 219

Table A3 List of Published GPR studies on Sediments 227


9/295

List of Figures

Figure 1: Relative resolution of different typical geophysical, geological

and remote sensing measurement 2

Figure 2: Outcrop mapping combined with GPR profile 10

Figure 3: GPR measurement setup 11

Figure 4: GPR antennas size with decreasing antenna frequency 13

Figure 5: Various modes for antenna deployment 22

Figure 6: Single transmitter-receiver common offset survey mode 24

Figure 7: Single transmitter-receiver common midpoint survey mode 27

Figure 8: Artificial structure on downstream accretion element createdby elevation variation along the profile line 31

Figure 9: GPR profile before and after de-wow filtering 32

Figure 10: GPR profile before and after background removal 33

Figure 11: GPR profile before and after deconvolution 35

Figure 12: GPR profile before and after amplitude gain 37

Figure 13: GPR Profile before and after diffraction migration 38

Figure 14: Navajo sandstone outcrop and 100 MHz antenna GPR line 45


10/295

Figure 17: Analyses of published GPR studies on sediments 51

Figure 18: GPR Image of a fluvial channel from Dunvegan Formation outcrop 54

Figure 19: Radar facies from Fraser (B.C) and Niobrara River, Nebraska 55

Figure 20: Examples of inclined radar facies 57

Figure 21: Examples of reflection-free radar facies 58

Figure 22: Example of horizontal radar facies 59

Figure 23: Example of hyperbolic reflection 60

Figure 24: Radar facies chart of presumed characteristic reflection

patterns from various sedimentary environments 62

Figure 25: Generalized stratigraphic chart for the Silurian system in New York 69

Figure 26: Early Silurian paleogeographic map of the Niagara Region 70

Figure 27: Location of the Whirlpool sandstone outcrop at Artpark, New York 71

Figure 28: Outcrop photo of the Whirlpool sandstone at Artpark, New York 72

Figure 29: Raw, unprocessed GPR data from the Whirlpool Sandstone 73

Figure 30: Artpark vertical section summarizing the main outcrop observations 74

Figure 31: Location of GPR lines at Artpark State Park, New York 77

Figure 32: Annotated Whirlpool outcrop photo and GPR lines 79

Figure 33: GPR line orthogonal to the Whirlpool Sandstone outcrop at Artpark 81


11/295

Figure 37: 3D GPR image of the ancient Navajo dune complex 90

Figure 38: Generalized stratigraphy of the Navajo Sandstone 94

Figure 39: Map of Navajo Sandstone location within Utah 96

Figure 40:Map showing field locations of radar profiles at Big Mesa, Utah 98

Figure 41: Outcrop photo of Navajo Sandstone central dune giant cross-bed

compared with smaller cross-bed set at Big Mesa near Moab, Utah. 100

Figure 42: Outcrop photo showing damp eolian interdune deposits overlainby flat-bedded fresh 101

Figure 43: Outcrop photo and GPR profile at Big Mesa, Utah 103

Figure 44: Radar profiles and perspective views of 3D GPR surveys at Big Mesa 107

Figure 45: Outcrop photo from Moab (Utah) revealing damp phase interdunedeposit from the Carmell Formation 110

Figure 46: Paleogeography during deposition of Dunvegan Formation 118

Figure 47: Lithostratigraphic relationships between the Dunvegan Formation

and the underlying Shaffesbury and overlying Kaskapau formations 119

Figure 48: Map location of Dunvegan Formation outcrop and GPR lines 121

Figure 49: Map showing location of GPR lines at Dunvegan outcrop 123

Figure 50: Dunvegan annotated outcrop photomosaic and GPR line 130

Figure 51: Dunvegan annotated outcrop photo and GPR line 131

Figure 52: Dunvegan annotated outcrop photomosaic and GPR line 133

Figure 53: GPR line and 3D schematic of Dunvegan depositional model Pink


12/295

Figure 56: Location of Shinarump Conglomerate outcrop at Hurricane Mesa 142

Figure 57: Location of Shinarump Conglomerate outcrop at Hurricane Mesa

showing GPR line 144

Figure 58: Photomosaic of the Shinarump conglomerate outcrop 148

Figure 59: Uninterpreted and interpreted GPR line at Huricane Mesa 149

Figure 60: Combined Shinarump GPR profile and outcrop Photomosaic 151

Figure 61: St. Mary River Formation Upper Cretaceous Paleogeography 155

Figure 62: Alberta Cretaceous Stratigraphic chart showing St. Mary River 156

Figure 63: Location of St. Mary River formation outcrop 158

Figure 64: Location of GPR lines at St. Mary River formation outcrop 160

Figure 65: Outcrop photo showing St. Mary River Formation channelsandstone 161

Figure 66: Outcrop photo showing St. Mary River Formation crevassesplay sandstone 162

Figure 67: Radar image of St.Mary River Formation ribbon channel sandstone 163

Figure 68: Radar profile showing ribbon channels and horizontal reflectors 164

Figure 69: Uninterpreted and interpreted St. Mary River Formation GPR profile 165

Figure 70: Block diagram depicting depositional model of St. Mary RiverFormation at the study site near Monarch, Alberta 168

Figure 71: Three-dimensional GPR data display from Navajo Sandstone

interdune deposit at Big Mesa Utah 175


13/295

Figure A4: Mala Geoscience Rough Terrain Antenna (RTA) 224

Figure A5: Mala Geoscience shielded antennas 225

Figure A6: GSSI monostatic and bistatic 100 MHz antennas 226

Figure A7: Analyses of 150 published GPR studies on sediments 278

Figure A8: GPR profiles from Artpark (New York, USA) 280

Figure A9: GPR profiles from Big Mesa, near Moab (Utah, USA) 281


14/295

List of Appendices

Appendix 1: Radar propagation data in sediment 219

Appendix 2: GPR Instrumentation 220

Appendix 3: List of published GPR studies on sediments 227

Appendix 4 GPR Data Processing Software and Display 279

Appendix 5: GPR profiles from study locations 280


15/295

Chapter 1

Research Problem, Object and Significance

1.1 Introduction

The need for better prediction of reservoir architecture and distribution of lithological

heterogeneity in aquifer characterization, hydrocarbon primary field development and

enhanced recovery projects is driving a burgeoning interest in outcrop-based studies of

ancient clastic deposits both as analogues for subsurface reservoirs and as a tool for

refining existing facies models. This research is driven by the need to account for thevariability and architectural complexity in subsurface clastic reservoirs, especially for the

purpose of petroleum resource evaluation and development, where conventional seismic

resolution is too poor to unambiguously delineate sandbody geometry or map their

internal heterogeneity.

Subsurface interpretation of reservoir architecture currently relies on seismic, well logs

and core data which are rarely closely spaced enough to permit unequivocal

interpretation. Practical limitations of most sources of high-resolution data rarely permit

the close spacing required for unambiguous interpretation (Figure 1). Advances in

seismic data acquisition, data processing and three-dimension data visualization in the

last two decades have significantly improved the resolution of reservoir information


16/295

and fluid content in reservoirs. In spite of these advances, predicting sub-seismic scale

heterogeneities in reservoirs is still a daunting challenge for many primary fielddevelopment, enhanced oil recovery and aquifer characterization projects.

Figure 1: Chart showing trade-off between the relative resolution of the informationobtained using different typical geophysical, geological and remote sensing measurementacquisition approaches and the relative scale of the investigations for which thoseacquisition geometries are typically used. (Image taken from Rubin and Hubbard, 2005).

The development of affordable digital imaging technologies as well as advances in

photogrammetry, aerial photography, LiDAR scanning, satellite imagery and positioning

have significantly improved outcrop studies. Increased spatial accuracy from satellite and

laser positioning systems provides access to geostatistical and geospatial analyses that


17/295

captured and brought to the geological laboratory via LiDAR scanning for further

measurements and analysis; bedding orientations and dips can then be measured at thelaboratory, considerably saving time spent doing fieldwork.

Despite these developments, outcrop studies remain largely two-dimensional, as most of

these novel techniques are not penetrative; they only reveal outcrop-face sedimentary and

stratigraphic features that can only be inferred beyond the outcrop-face with much

difficulty. The need for a more reliable tool of reconstructing fluvial architecture from

outcrops led to Mialls (1985) proposition of architectural element analysis following

the pioneering work of Allen (1983). Architectural elements are associations of

genetically related facies, which have some environmental significance. In depositional

systems, they are packets of genetically related strata, which define depositional elements

larger than individual bedform and smaller than channels. In outcrops, they are defined

by grain size, bedform composition, internal sequence, and most importantly, their

external geometry. This mapping technique has not only been applied to ancient fluvial

deposits as proposed by Miall (1985); it has been successfully applied on outcrops of

diverse depositional origin (Kocurek, 1981; Nio and Yang, 1993; Clark and Pickering,

1996). While there have been many outcrop studies with geometry and lithofacies of

channel and channel-belt deposits determined in outcrops described as analogues for

subsurface strata (Tyler and Finley, 1991; Mjos and Walderhaug, 1993; Dreyer, 1993a,

1993b; Dreyer et al., 1993; Robinson and McCabe, 1997; Willis and Gabel; 2003), the


18/295

The need to address the two dimensional limitation of outcrop studies motivated the

recent interest surge in Ground Penetrating Radar (GPR) imaging. However, due to theoperational difficulties of GPR imaging on outcrops and the susceptibility of GPR signals

to diffraction resulting in noise-laden images, very few outcrop-based GPR studies

(Appendix 3) have been successfully conducted (Corbeanu et al., 2001; Szerbiak et al.,

2001; Lee et al, 2009). Consequently, much of the current GPR effort has focused on

modern environment and Quaternary sediments (Beres and Haeni ,1991; Jol and Smith,

1991; Gawthorpe et al 1993; Huggenberger et al., 1994; Beres et al., 1995; Bridge et

al.,1995,1998; Van Overmeeren 1998; Bristow et al.,1999,2000b; Baker et al., 2001;

Corbeanu et al., 2001, 2002; Nobes et al.,2001; Hammon et al.,2002, Hornung and

Aigner 2002; Best et al, 2003; Cardenas and Zlotnik, 2003; Heinz and Aigner 2003;

Skelly et al., 2003, Woodward et al., 2003) as radar profiling is more convenient in

modern environments usually of less rugged terrain and modern sediments are less prone

to diagenetic alterations, which often make sedimentological interpretation of GPR data

from ancient clastic deposits a Herculean task. Although there has been an improvement

in our understanding of many ancient deposits via GPR studies of modern sediments, the

varied and unpredictable preservation potential of facies units in their modern setting

limits the usefulness of these studies. It is the preserved ancient record that is ultimately

what reservoir geologists are interested in. Miall (2006) sounded a caveat in this regard

and underscored the continued role of three-dimensional studies of ancient outcrop


19/295

direct visual observation of rock types and their spatial arrangement, thereby providing

invaluable insight into reservoir complexities for subsurface modelling (Geehan, 1993).Key components of significant relevance to reservoir characterization include facies

types, macro-scale lithological heterogeneity, channel dimensions, sandbody geometry

and their stacking pattern in relation to local and regional controls. These outcrop-scale

features control fluid-flow behavior in petroleum reservoirs, and aquifer and outcrop

analogs have been shown to provide valuable insight into reservoir architecture

(Grammer et al, 2004). While outcrop studies typically provide valuable data on vertical

dimensions of architectural elements within a depositional system, rarely do they yield a

three-dimensional picture of potential reservoir distribution especially from the

standpoint of aerial dimensions. Knowledge of the internal architecture is generally

limited because the exposures are either strike-oriented, dip-oriented, or at an orientation

oblique to these directions. These limitations have significantly weakened the usefulness

of outcrop studies as analogs for subsurface clastic reservoirs in the past; successful

application of GPR imaging on outcrop will provide an opportunity to validate current

clastic facies and architectural models that are mostly actualistic and conceptual

constructs.

The purpose of this research is four-fold:

to demonstrate that sedimentological and stratigraphic data can be obtained from

b d fil d di l h id l h ld i h i i


20/295

to demonstrate the effectiveness of GPR imaging in addressing the limitations of

two dimensional outcrop studies of ancient non-marine sandstones in a mannerthat has not been successfully explored before,

to develop a field methodology for outcrop-based GPR surveys, and

to provide a reliable framework for stratigraphic interpretation of GPR images as

a guide for future research.

By incorporating ground penetrating radar data with outcrop architectural element

analysis, data that are often difficult to unambiguously obtain, such as local paleoflow

variability, channel belt width, channel dimension, sinuosity, sandbody connectivity,

stacking patterns and lithological heterogeneity can now be readily obtained. Successful

application of GPR imaging on ancient clastic deposits will have far-reaching

significance; being able to obtain data such as true channel dimensions, channel

sinuosity, incised valley dimensions and their internal geometry will not only provide

hard data to validate or dispel many of the current clastic facies models based mostly on

studies of modern environment and Quaternary deposits; it will also testactualistically-

derived empirical equations (often used to predict channel belt width in the subsurface)

that relate maximum channel depth, channel width, and channel-belt width (Collinson,

1978; Lorenz et al., 1985, 1991; Fielding and Crane, 1987; Bridge and Mackey, 1993b).

Two features of GPR imaging make the technology exceptionally suitable for addressing


21/295

Although GPR imaging has been used intermittently for outcrop studies for over 25

years, the technology was not widely embraced as GPR instrumentation and dataprocessing technology was in its infancy and many of the GPR profiles generated were

not thoroughly processed to remove extraneous reflections required to ensure

unequivocal sedimentological and stratigraphic interpretation (Pratt and Miall, 1993;

Stephens, 1994). In the last few years, there have however been significant advances in

GPR instrumentation, data processing and display. Recently developed GPR systems are

lightweight, portable, robust and digital. In addition, the fact that GPR data can be viewed

and test-processed in real time during surveys ensures that both quality and results can be

assessed in the field. These advances as well as the availability of GPR processing

applications with two and three-dimensional visualization and interpretation aids have

enhanced rapid acquisition of continuous, shallow subsurface profiles that are ideally

suited for the investigation of sediments, shallow stratigraphy and sedimentary

architecture from both a 2-D and 3-D perspective.

Outcrop studies will undoubtedly benefit more from full-resolution three-dimensional

GPR surveys which produce 3D volumes that can be sliced and rotated to reveal features

such as map views of channel geometry, dimensions, sinuosity, anabranching and

temporal changes in depositional patterns. However, most GPR surveys are still two-

dimensional due to operational difficulties of 3D GPR surveying on outcrops. The few

three-dimensional surveys conducted to date are limited to small areas because of the


22/295

outcrop analogs has been demonstrated with 2-D profiles from earlier studies (Meyers et

al., 1996; Lunt and Bridge, 2004) as well as with fence diagrams constructed fromstaggered GPR profiles obtained in different directions behind outcrop face, ground-

truthed with outcrop sedimentological and architectural element mapping.


23/295

Chapter 2

Research Methodology

2.1 Introduction

The reliability of GPR outcrop based reservoir analogue studies depends on the qualityof outcrop data and GPR survey as well as subsequent processing of the GPR data. Many

of the earlier outcrop studies utilizing GPR (Pratt and Miall, 1993; Stephens, 1994; White

et al, 2004; Zeng et. al. 2004) were plagued with noise making it difficult to correlate

outcrop observations to radar reflectors. This is attributed to susceptibility of

electromagnetic signals to diffractions resulting from fractures as well as diagenetic

cementation in rocks (Smith et al., 2006). Diffraction occurs at discontinuities of

reflectors (such as fractures) and objects whose dimensions are small compared to the

transmitted electromagnetic wavelength. Figures 2 A and B illustrate the prevalence of

diffraction noise in earlier outcrop GPR studies.

Advances in GPR instrumentation and data processing over the last decade have

significantly improved GPR data quality. The field methodology for this research

involves obtaining photomosaics of laterally extensive sandstone exposures

supplemented by GPR profiles behind the mapped outcrop face. The photomosaics serve

a dual purpose: they act as base maps for detailed mapping of bounding surfaces and


24/295

Figure 2: (A) An example of outcrop face mapping of major bounding surface combinedwith GPR profile behind the cliff face. (Image taken from Zeng et al, 2004) (B) Ground-penetrating radar traverse measured about 100 m behind a dip-parallel outcrop of theupper Frewens sandstone body. (Image taken from White et al., 2004); note the difficultyin correlating GPR reflections with outcrop bounding surfaces.

Architectural-element analysis involves delineating major bounding surfaces as well as

recognition of suites of geometric elements characterized by distinctive facies

B

20 m

A

Diffraction noise

Diffraction noise


25/295

fluvial and eolian deposits) and unraveling their three-dimensional architecture (Miall,

1996)

Acquisition and processing of GPR data are the more laborious phases of the research

involving selection of the appropriate GPR system as well as efficient field survey and

data processing procedures. GPR is an electromagnetic method that, in many ways, is

similar to seismic reflection survey. A transmitting antenna radiates an electric pulse into

the ground that, under ideal conditions, behaves kinematically similar to an acoustic

wave. The pulse is transmitted, reflected, and sometimes diffracted by features that

correspond to changes in the dielectric properties of the earth. The waves that are

reflected and diffracted back toward the earths surface may be detected by a receiving

antenna, amplified, digitized, displayed, and stored for further analysis (Figure 3)


26/295

sometimes, sequence boundaries. The transmitter and receiver units are usually separate,

so the survey design is flexible. Data can be collected either in continuous or step mode.

In continuous mode, the antennas are dragged over the surface, whereas in step mode, the

antennas are placed on the ground in steps progressively along the profile. GPR surveys

usually proceed by recording a trace at each of a large number of survey points along a

line (or over a grid) with a fixed transmitter-receiver offset. As in reflection-seismic data,

GPR profiles may be plotted directly, or combined and plotted as a volume. The main

differences are that the scale of a GPR survey is about three orders of magnitude smaller

than that of a reflection seismic survey, and the resolution is correspondingly higher,

making the technology suitable for outcrop scale imaging.

The choice of antenna frequency is a tradeoff between the depth of target beds and the

intended resolution of the survey; the lower the frequency of the antenna, the poorer the

resolution but the greater the depth of penetration (Davis and Anan 1989; Jol, 1995).

Currently commercially available frequencies are between 12.5 and 1200 MHz, the time

sample increment is usually about 1 ns, the propagation velocities are one-quarter to one-

half that of light in a vacuum, and the average depth of penetration ranges from 5 to 40m,

depending primarily on the electrical conductivity and water content of the subsurface

materials. Generally, the lower the frequency, the bulkier the GPR equipment (Figure 4) ;

this implies that low frequency antenna (below 100 MHz) ideal for regional-scale

geological imaging may require more than one person to handle during surveys especially


27/295

Attenuation of GPR signals in rocks increases with water saturation and decreasing grain

size; wet and clayey rocks significantly attenuate radar signals thereby reducing depth of

penetration.

Figure 4: GPR antennas get bulkier with decreasing antenna frequency. (Image takenfrom http://www.sensoft.ca/products/pulseekko/pulseekkopro.htmll; - accessed August6, 2010).

2.2 Pre-survey Assessment

While GPR has proven to be an invaluable tool for high-resolution imaging in outcrop-


28/295

cut; it is easier to rule out situations where georadar is totally unsuitable than to state with

certainty that ground radar imaging will be successful. The following criteria must be

thoroughly evaluated before embarking on a GPR survey program:

2.2.1 Depth of Formation of Interest

How deep is the formation of interest? is perhaps the most important question in planning

GPR surveys. Many geological formations that are potential targets for GPR imaging are

located beneath several metres of overburden (for instance, many outcrops of Whirlpool

Sandstone and St. Mary River Formation) or beneath formations that might significantly

attenuate transmitted radar signals. The lowest frequency antenna currently available for

GPR surveys is 12.5 MHz (Figure 4) and this is limited to less than 50 metres penetration

depth in most clastic rocks, even in dry, clay-free lithologies. In addition, the attenuation

characteristic of the overburden and target medium is critical to penetration depth of

radar signals. A conservative rule-of-thumb is that radar will be ineffective if the actual

target depth is greater than 50% of the maximum range. A rough guide to penetration

depth D is

where conductivity, is in mS/m. This equation is not universal but is applicable when

(2.1) (Annan, 2004)


29/295

In geological applications, the parameters which have the greatest influence on the

electrical properties, and hence on penetration depth, velocity and reflector characteristics

are: (i) water saturation, (ii) clay content and (iii) pore water salinity. Dry, clay-free

sediments generally have higher velocities of propagation and lower attenuation

coefficients than wet strata. Depth of penetration is therefore highest in dry, porous

sediments (e.g. limestone, where probing depths may theoretically be up to 100 m) andlowest in water-saturated clay where depth of penetration may be less than 1 m (Cook

1975). In some cases, sand-rich formations can have diagenetic clays not obvious in

outcrop and could result in severe signal attenuation; hence, the need for pre-survey

assessment.

2.2.2 Contrast in electrical properties required toreflect or scatter a detectable amount of energy

GPR investigates the subsurface by making use of electromagnetic signals, which

propagate into the subsurface. Because reflections of radar signals are generated by

changes in electromagnetic impedance (contrast in electric and magnetic properties)

across the lithologies that the signals travel through in the same way seismic reflections

are generated across lithological contacts with acoustic impedance changes, detectable

reflections are expected in geological formations with contrasting lithologies. When an

electromagnetic wave propagates through the ground and encounters a surface where the

l t i d/ ti ti f th d h t f it ill b


30/295

are usually weak, although occasionally, magnetic properties can affect radar responses.

This is often the case when imaging diagenetically altered formations containing iron

oxides; hence, it is important to be cognizant of such magnetic effects as they can

significantly distort interpretation. Transmission of electromagnetic signals through rocks

is controlled by three parameters: dielectric permittivity, electrical conductivity and the

magnetic permeability. Dielectric permittivity is the ability of a material to store andrelease electromagnetic energy in the form of electric charge or its ability to restrict the

flow of free charges or the degree of polarization exhibited by a material under the

influence of an applied electric field. Conductivity is the ability of a material to pass free

electric charges under the influence of an applied field; this in rocks typically occurs via

dielectric conduction (in resistive lithologies which require the atoms to slightly polarize

to produce displacement currents) and electrolytic conduction (dominant in moist or wet

lithologies). Magnetic permeability describes how intrinsic atomic and molecular

magnetic moments respond to a magnetic field. Magnetic permeability in most clastic

rocks is very low and its effect on electromagnetic signal transmission can be ignored.

Both dielectric permittivity and electrical conductivity are strongly dependent on water

content (which is largely dependent on lithology); hence radar reflections patterns are

closely linked with lithological variability in rocks. For outcrops with minimal

lithological heterogeneity, it is sometimes efficient to conduct a preliminary small scale

survey at the outcrop to determine optimal target depth, resolution and appropriateness of


31/295

2.2.3 Noise sources that could preclude the use of GPR

Recorded GPR signals during surveys can be masked by electromagnetic signals fromnearby sources such as a radio transmitter or an electric power line located near the

survey site; in such cases, external signals may saturate the sensitive receiver electronics.

Radio transmitters are potential sources of interference and powerful radio signals can

overwhelm receiver electronics. Mobile phones are increasingly becoming a ubiquitous

form ofinterference; proximity to metal objects can also be disastrous for GPR survey.

Reflections can come from objects away to the side (sideswipe) and may be very strong if

metallic reflectors are involved. Surface features can produce strong sideswipe resulting

from substantial radiation of energy along the ground/air interface if ground conductivity

is high. Shielded antennas are very useful in such situations and are available in the 100

MHz frequency range and above. The shields are about the same size as the antenna and

use absorbing material to damp out the undesired signals. At lower frequencies, antenna

size and portability makes shielding impractical (as antenna size increases with

frequency); these considerations probably explain why there are no commercially-

available shielded GPR systems for antenna frequencies below 100 MHz. Practical

limitation of portability also explains why most shielded GPR systems consisting of both

transmitter and receiver antenna elements are housed in a single housing; hence, bistatic

antennas useful for common midpoint surveys are rarely shielded.


32/295

surveying equipment to the target outcrop. Difficulty in accessing outcrops is one of the

major factors explaining why there has been limited application of GPR imaging to

solving three dimensional outcrop problems. Although some can be reached by short

hikes, many outcrops ideal for GPR surveys can only be reached by ATVs (All Terrain

Vehicles) or helicopters; hence, accessibility to outcrops should be thoroughly evaluated

before embarking on GPR surveys.

2.3 GPR Survey Planning

Once the feasibility of GPR survey at an outcrop has been ascertained, proper design of

the survey as well as optimizing data acquisition to meet expectations and honor surface

constraints is the next challenge of outcrop-based GPR studies and requires thorough

planning.

In planning for GPR surveys on ancient sedimentary deposits, the following factors

critical to obtaining high-fidelity profiles must be considered:

Choosing the right GPR system there is currently a wide range of

commercially available GPR systems to choose from (see Appendix 2); it is

therefore important to choose a GPR system suitable for the purpose of the study.

Stratigraphic studies requiring time-depth conversion via common mid-point

surveys (data collection by moving the antennas progressively apart at an equal

offset from a central survey point) which gives the flexibility of calibrating radar


33/295

project is also critical; this is a trade off between the depth of the geological

formations being studied and the resolution required in the study. Low frequency

antennas (below 100 MHz) are better-adapted for low resolution deeper target

(20-50 metres deep) surveys while high frequency antennas (above 100 MHz) are

better-suited for high resolution shallow (less than 20 metre deep targets). As a

rule of thumb, the vertical resolution is theoretically one-quarter of thewavelength = v/f, where v is the velocity of the electromagnetic wave in the

rock (see Appendix 1 for typical velocities in different lithologies) and f is the

center frequency of the GPR antenna. It is also important to note that the

bulkiness of most GPR systems increase with decreasing antenna frequency

(Figure 4); this should be considered when planning GPR surveys as such low

frequency systems may require two or more people to handle.

Surface Elevation Uneven topography is characteristic of most outcrop top.

GPR surveys require flat outcrop tops or centimeter-scale topographic survey over

the GPR transects or across the area covered by the 3D grid if three-dimensional

GPR survey is being planned. This is a major challenge in conducting GPR

surveys on outcrops. Collection of topographic data is an essential component of

GPR survey especially where the outcrop surface is not flat and this could gulp a

significant chunk of survey time. Improper incorporation of topographic data into

the GPR lines may result in inaccurate velocities for static correction, which often


34/295

and GPS. In areas with subdued topography (less than 5 m change in elevation),

laser levels provide a fast and accurate method for collecting topographic data. In

more complex terrains, a total station provides greater flexibility and higher

accuracy; optical levels are also ideal but increase survey time. Where GPS units

are being considered for GPR surveys, Differential GPS units with centimeter-

scale accuracy are required. GPS surveys significantly reduce GPR survey time

especially in three-dimensional surveys as GPS measurements obviate the need

for a survey grid. In addition, with GPS measurements, real world locations of

GPR profiles can be easily displayed in Google Earthor imported into digital

topographic maps. It is however important to note that GPS receivers have

difficulty working under thick vegetation and require post processing. Some high-

resolution GPS tools also have compatibility problems with GPR units; if possible

GPS unit should be connected to the GPR equipment and tested before setting out

on field surveys. Most of the GPR surveys for this research were conducted on

flat surfaces requiring minor or no elevation correction.

Mode of data collection and station spacing while GPR data can either be

collected by dragging the antenna over the surface (continuous mode) or by

progressively moving the antenna along the survey line (step mode); the step

mode of data collection is less prone to data degradation and more suitable for

stratigraphic studies although it is more time-consuming. GPR antennas couple to


35/295

distance between survey points must be small enough (usually between 0.25 to

0.5 metres) to be able to resolve steeply dipping and small features as well as to

minimize spatial aliasing of radar reflections. Due to less survey time required,

many GPR surveys are done in continuous mode. Most shielded GPR systems are

designed for continuous data collection mode; therefore, for surveys where

shielded antennas are being considered, care should be taken to ensure optimal

coupling with the ground for efficient transmission of radar signals.

Step size - Spatial resolution places a constraint on the survey design and

selection of antenna centre frequency. Step size (the distance between each data

collection point, also known as station spacing) is extremely important and should

be included in the survey design process. In order to ensure that the ground

response is not spatially aliased in sedimentary studies, a maximum step size of

one metre (less than 1m where antenna frequencies higher than 200MHz are being

employed) should be used to provide detailed horizontal resolution of sedimentary

structures, yet allow profiles to be completed in a timely manner. A typical survey

with 100MHz antenna theoretically should have a step size of 0.25m; however,

larger step sizes can be used where the subsurface stratigraphy is composed of

continuous horizontal layers. If the step size is too large, the data will not

adequately define steeply dipping reflectors or diffraction tails. For 3D GPR

surveys, using a GPR grid spacing larger than a quarter-wavelength in the


36/295

geological studies; hence, very low step sizes are required for thorough definition

of dipping beds and surfaces. The decision to increase or decrease the step size

should be based on a variety of factors including: size of the sedimentary feature

being investigated, dip angle, and areal extent of the survey. From a practical

viewpoint, increasing the station interval reduces data volume and survey time,

yet from a data interpretation standpoint, adhering to the Nyquist close sampling

interval (which requires that measurements be spaced in all horizontal directions

near a quarter-wavelength of the highest signal and noise frequency content to

avoid spatial aliasing) is very important.

Antenna Orientation - Antennas used for most GPR surveys are dipolar and

radiate with a preferred polarity. The antennas are normally oriented so that the

electric field is polarized parallel to the long axis or strike direction of the target;

there is no optimal orientation for an equidimensional target.


37/295

Antenna orientation does affect the quality of collected data (Lutz et al., 2003) and it

should be taken into consideration when planning GPR surveys. The various

arrangements of GPR antenna deployment are illustrated in Figure 5; the most

common orientation that provides the widest angular coverage of a subsurface

reflector is the perpendicular broadside to direction of survey approach (PR-BD).

This orientation is also the easiest for surveying in sedimentary environments. In

some instances, it may be advisable to collect two data sets with orthogonal antenna

orientations in order to extract target information based on coupling angle. If the

antenna system is one which attempts to use a circularly- polarized signal, the antenna

orientation becomes irrelevant; however, as most commercial systems employ

linearly polarized antennas, orientation can be important. Maintaining a consistent

antenna configuration and electronic connections throughout a survey is important

because this will allow changes in polarity, due to increases or decreases in

permittivity (impedance) with depth, to be determined. Most shielded GPR units have

fixed antenna orientation, hence no need of deciding optimal antenna orientation.

GPR units with separate antennas however require determination of optimal antenna

orientation.

2.4 GPR Survey Design

Radar surveys for stratigraphic purpose are carried out either in common offset or


38/295

spacing between the units at each measurement location. The transmitting and receiving

antennas have a specific polarization character for the electromagnetic field generated

and detected. The antennas are placed in a fixed geometry and measurements made at

regular fixed station intervals, as depicted in Figure 6. Data on regular grids at fixed

spacing are normally needed if advanced data processing and visualization techniques are

to be applied. The parameters defining a common offset survey are GPR center

frequency, the recording time window, the time-sampling interval, the station spacing,

the antenna spacing, the line separation spacing, and the antenna orientation. During

surveying, antennas are either dragged along the ground and horizontal distances

recorded on a time-base, which can be converted to a distance-base through manualmarking or measuring wheel, or they are moved in a stepwise manner at fixed horizontal

intervals (step size). Step-mode operation generates more coherent and higher

amplitude reflections, as antennas are stationary during data acquisition. This ensures

more consistent coupling between the antennas and the ground as well as better trace

stacking (Annan and Cosway, 1992).


39/295

The MAL antenna used for this research (250 MHz MAL Geoscience shielded

antenna-Appendix 2) is adapted for single-fold reflection profiling; this GPR unit has

fixed antenna distance and orientation, which makes it impossible to conduct common

midpoint surveys.

Common midpoint (CMP) soundings are primarily used to estimate the radar signal

velocity versus depth in the ground, by varying the antenna spacing (commonly referred

to as offset) and measuring the change of the two-way travel time. At all distances

between source and receiver with a fixed midpoint, most of the energy that is reflected

by, not too steeply dipping, subsurface reflectors comes from the same points straight

below that given midpoint (Figure 7A). From the differences in arrival time versus offset

very accurate velocity estimates can be obtained from all layers. When a CMP

measurement is carried out, different events can be distinguished in the received signal,

when plotted on a travel time versus offset plot as depicted in Figure 7C. These are the

results of the air wave (1), the ground wave (2), the direct reflection from an interface (3),

and the critically refracted airwave (4). Since the velocity in air is always greater than the

largest velocity in the ground, a critical angle of incidence exists where the transmitted

wave travels in the air along the interface. These angles only exist when the wave travels

from a slower medium to a faster medium, such as a reflection from below the ground

that travels toward the earth surface. In a common-offset survey, the above mentioned


40/295


41/295


42/295

2.5 GPR Data Acquisition and Recording

GPR data have the advantage of being viewed directly as they are being recorded in the

field and many GPR equipment control units have basic editing and data processing

capability. Despite this advantage over seismic data acquisition, radar data acquisition

parameters such as antenna frequency, sampling frequency, number of stacks and time

window can not be changed during data processing. To ensure that data acquisition time

is effectively used, laptop computers with installed data processing applications should be

taken to the field and GPR data edited, checked and test-processed in the field before the

completion of field surveys as it is often difficult to fix problems in GPR data that are due

to poor acquisition parameters without having to return to the field.

Detailed record of GPR lines, outcrop photos and their locations should be accurately

documented in the field as the effectiveness of data processing and geological

interpretation hinge on proper documentation of GPR line locations and other outcrop

observations. GPR data sheet such as Table 1 should be completed as each GPR line is

being recorded.

Table 1: Example data sheet that can be used for collecting data from daily GPR surveys


43/295

2.6 GPR Data Processing

GPR signals are usually processed like seismic data although the two are slightly

different. GPR data are most often treated as scalar although as electromagnetic fields,

they are vector quantities. Hence, GPR signals may behave differently due to frequency-

dependent absorption and phase changes at reflections.

In the early days of GPR data processing, much of the data processing was done withseismic data processing software. However, GPR-specific processing software is now

commercially available from packages offered by manufacturers of GPR or as add-ons

for seismic processing software. For this study, REFLEX and GPR Slice processing and

display software were used for data processing and perspective display.

Like seismic reflection data, GPR data require processing aimed at sharpening the signal

waveform by improving the signal to noise ratio of the radar profiles (Reynolds 1997).

The amount of processing depends on the quality of field data obtained, and this can

range from basic processing steps to the more complex application of data processing

algorithms. Data obtained with shielded antennas are often less cluttered with extraneous

signals and usually require less rigorous data processing. In most instances, basic

processing steps can be applied in real-time during data acquisition; however, these might

not be enough to remove extraneous noise from the acquired data. A detailed discussion

of GPR processing steps is important, as the final output of GPR surveys is dependent on

processing steps; radar facies must therefore be compared only between data processed


44/295

criteria can be significantly influenced by the processing steps. Continuous reflection can

appear discontinuous by applying inappropriate gains during data processing and

reflection dips can result from inappropriate processing algorithms; hence, the

importance of proper signal processing steps and outcrop data as ground truth to verify

the fidelity of radar data. This also underscores the importance of documenting data

acquisition and processing steps in GPR publications to allow comparison between

studies.

Commonly applied processing steps on GPR data include:

2. 6.1 Topographic Correction

The collection of topographic data is an essential component of GPR survey as radar data

collected in the field does not take into account topographic variation along a survey line.

This can be corrected by moving traces up and down by an appropriate two-way-time

relative to a common datum, based on knowledge of the velocity, and, therefore, depth

profile of the uppermost part of the radar profile. In order to do this, survey line

topography must be adequately characterized. Topographic surveys are typically

performed using a variety of instruments including optical levels, total station, laser

levels, and differential GPS as these have the required vertical resolution and allowrelatively rapid data collection. Spatial sampling in such surveys should ensure that all

significant breaks in slope are accounted for. While lower frequency antenna can tolerate


45/295

equipment was not available for this study; hence outcrops with flat surfaces were chosen

for this study to avoid the need for elevation correction of the recorded GPR profiles.

Figure 8: Artificial structure on downstream accretion element created by elevation

variation along the profile line. (Profile recorded on Castlegate sandstone outcrop, TusherCanyon, Utah).

2.6.2 Dewow

Wow noise is peculiar to ground penetrating radar and it is as a result of the close

proximity of receiver to transmitter. Dewow is one of GPR processing basic temporal

filtering steps aimed at removing very low frequency components from the data (Figure

9). Very low frequency components of GPR data are associated with either inductive

phenomena or possible instrumentation dynamic range limitations. This process has

historically been done using analog filters in hardware but with the advent of true digital

data acquisition, it has also become a data processing step (Gerlitz et al, 1993). Dewow

filter acts on each trace independently by calculating a mean value of each trace and the


46/295

Figure 9 (A)Display of a single data trace (left) and data section (right) with the lowfrequency wow component masking the real data. (B)Display of a single data trace (left)and data section (right) where the Wow seen in (A) above has been removed with thedewow high pass filter. (Image taken from Annan, 2004).

2. 6.3 Backgound Removal

One of the most common operations specifically applied to GPR data is the use of

B

A


47/295

Figure 10: (A)GPR profile before background removal (B)GPR profile afterbackground removal. (Image taken from http://www.malags.com/Downloads/Product-Brochures.aspx; accessed, August 6, 2010).

In situations where antenna ringing, transmitter reverberation and time synchronous

system artifacts appear, it is very effective in allowing subtle weaker signals, which are

l t t b i ibl i d ti B k d filt li i t t ll

A

B


48/295

(Figure 10 A and B). It is quite effective in relatively lossy materials (conductive

lithologies such as wet or shaly rocks).

2. 6.4 Deconvolution

Deconvolution is aimed at removing effects of a previous filtering operation (Yilmaz,

1987, 2001; Kearey and Brooks, 1991). In both seismics and radar signals, deconvolution

attempts to remove filtering effects resulting from propagation of a source wavelet

through a layered earth, and the recording system response. The intended effect of the

deconvolution process is to shorten pulse length and, therefore, improve vertical

resolution (Figure 11).

It is often tempting to apply seismic processing steps like deconvolution to GPR because

of their kinematic similarity; it is however important to note that deconvolution of GPR

data is not straightforward and rarely yields impressive results (Fowler and Still, 1977;

Payan and Kunt, 1982; LaFleche et al., 1991; Maijala, 1992; Todoeschuck et al., 1992;

Fisher et al., 1996; Arcone et al., 1998). The primary reason for this is that the radar pulse

is often as short and compressed as can be achieved for the given bandwidth and signal-

to-noise conditions. Another important factor is that some of the more standard

deconvolution procedures have underlying assumptions required for wavelet estimation

such as minimum phase and stationarity, which oftentimes are not appropriate for GPR

data (Annan, 1999). The rapid attenuation of GPR signal amplitude means that


49/295

Figure 11: (A)GPR profile before deconvolution (B)GPR profile after deconvolution.

(Image taken from http://www.malags.com/Downloads/Product-Brochures.aspx-accessed on August 6, 2010).

Few instances where deconvolution has proven beneficial occurred when extraneous

A

B


50/295

2. 6.5 Time Gain

Radar signals are prone to rapid signal attenuation as they propagate into the ground.

Signals from greater depths are often very weak, such that simultaneous display of this

information with signals from a shallower depth requires preconditioning for visual

analysis and display. When the amplitude of display is optimal for shallow depth signals,

events from greater depths may be invisible or indiscernible. Gains are required because

the amplitude of a reflected signal decreases with time and depth due to attenuation,

geometrical spreading, partial reflection and scattering (Davis and Annan 1989; Reynolds

1997). "Time gain" applied to radar data attempts to equalize amplitudes by applying

some sort of time-dependent gain function, which compensates for the rapid fall off inradar signals from deeper reflectors. A variety of gain functions can be applied to radar

signals (e.g. constant, linear and exponential gains); it is however important to understand

that the choice of gain adopted might alter the amplitude fidelity of the signals in the

data.

Geological radar surveys are especially susceptible to signal attenuation in mud-rich

environments; hence, for stratigraphic horizon continuity, displaying all the information

irrespective of amplitude fidelity might be important. In this case, manual gain or a

continuously adaptive gain such as AGC (automatic gain control) is often used. With

AGC gain, each data trace is processed such that the average signal is computed over a

time window and then the data point at the centre of the window is amplified (or

i f ti i h it h ld b l t d b d th ifi l f th d


51/295

gain function is chosen, it should be selected based on the specific goal of the survey and

data processing requirement; the objective should be to modify the data while retaining

fidelity without introducing artifacts.

Figure 12: (A) Un-gained Whirlpool Sandstone GPR Profile. (B)Manually-gainedprofile at the Whirlpool Sandstone outcrop at Artpark, USA.

It is important to document the type of gain function applied to radar data while

discussing radar facies as radar reflector character is dependent on the signal processing

steps applied to the data (relative amplitudes and/or phase relationships are changed); for

Queenston Shale

Whirlpool channel sand

A

B

2 6 6 Mi ti


52/295

2.6.6 Migration

Migration relocates reflections to their true spatial position based on the velocity

spectrum of radar signals through the rock layers to produce a real structure map of

subsurface features. Migration algorithms do this by removing diffractions, distortions,

dip displacements and out-of-line reflections resulting from the fact that radar antenna

radiate and receive electromagnetic energy in a complex 3- D cone.

The goal of migration is to make the reflection profile look like the geological structure

in the plane of the survey. It attempts to correctly position subsurface reflection events

(Hatton et al., 1986). However, due to various uncertainties, a significantly improved but

still imperfect image is usually achieved (Figure 13). Such improvements are quitebeneficial in sedimentological studies, where the nature and form of stratigraphic units

and primary sedimentary structure is of utmost importance.

A

B

Migration is generally used for improving section resolution and developing more


53/295

Migration is generally used for improving section resolution and developing more

spatially realistic images of the subsurface and is, perhaps, the most controversial of the

GPR processing techniques. Migration techniques, like deconvolution, were originally

developed for the seismic industry where they are considered as vital for even basic

interpretations. Unfortunately, migration tends to be less successful with GPR, and

although it can be used in relatively homogeneous environments, it is not so good with

complex, heterogeneous environments (lithologies) with variable radar velocity typical of

clastic many rocks. Despite this, migration is still a vital georadar processing step and

classical techniques have been applied successfully to a range of different applications.

Examples include reverse time migration (Sun and Young, 1995; Meats, 1996), FK

migration (Fisher et al., 1994; Pettinelli et al., 1994; Pipan et al., 1996; Yu et al., 1996;

Hayakawa and Kawanaka, 1998) and Kirchhoff migration (in Moran et al., 1998).

Specific GPR-based methods have been developed to overcome some of the limitations

in the seismic data migration algorithms. Examples include matched filter migration(Leuschen and Plumb, 2000); Kirchhoff migration modified for radiation patterns and

interface reflection polarisation (Moran et al., 1998; Van Gestel and Stoffa, 2000);

eccentricity migration for pipe hyperbola collapsing (Christian and Klaus-Peter, 1994),

3D-based vector and topographic migration (Lehman and Green, 2000; Heincke et al.,

2006; Streich et al., 2007) and frequency domain migration for lossy soils (Di and Wang,

2004; Sena et al., 2006; Oden et al., 2007). These new methods are yet to be incorporated


54/295


55/295

pulse is propagated through the host medium, Vm, and the depth to the bounding surface


56/295

p p p g g , m, p g

d, are related by the Equation

t= 2d/Vm

Velocity for depth conversion of the radar profiles recorded in this study was estimated

using this relationship. Although rarely, velocity required for time-depth conversion is

also estimated by direct laboratory measurements of dielectric permittivity on field

samples; measuring travel time between two wells using borehole radar; and

transillumination surveys between two parallel exposures (Annan and Davis, 1976; Topp

et al., 1980; Fisher et al., 1992a; Greaves et al., 1996; Reynolds, 1997; Binley et al.,

2001; Hammon et al., 2002; Tronicke et al., 2002a).

2. 7 Visualization and Display

Ground-penetrating radar has been used for geological imaging since the 1980s and has

since developed into a valuable tool for stratigraphic studies. Most studies utilize

common-offset, 2-D radar reflection profiles to display stratigraphic information obtained

in modern environments and outcrop. Where there is significant lateral variability in

internal structure, pseudo-3D or true 3D surveys may be desirable. Pseudo-3D surveys

involve collecting data on regular or irregular survey grids, usually in two mutually

perpendicular directions, and often displaying results as fence diagrams (for example,

B i t 1995 Ai t l 1996 B i t t l 1996 1999 2000b R b t t l

(2.4)

time in three dimensional seismic imaging revolutionized the way GPR results are


57/295

g g y

displayed and interpreted (Goodman et al., 1995), and recent GPR three-dimensional

imaging produced exciting results for visualization and enhanced interpretation (Conyers

et al., 1997; Jol et al, 2003; Leckebusch, 2003).

While full resolution 3D GPR surveys are generally more informative than single

profiles, it requires significantly increased data acquisition time. Hence, they are seldom

used in geological studies. Data processing for three dimensional GPR surveys is more

time consuming partly because of greater data volumes and complexity, but also due to

lack of efficient commercially available software for three dimensional GPR data

processing; these explain why there are few 3D GPR outcrop studies in the GPR

literature. Not all studies require 3D surveys; staggered GPR lines parallel and orthogonal

to the outcrop face displayed as a fence diagram often yield sufficient architectural details

for stratigraphic interpretation.

2. 8 Causes of GPR Reflections

Although GPR images appear similar to seismic reflections generated across surfaces due

to changes in acoustic impedance in the subsurface, GPR reflections are generated by

changes in electromagnetic impedance due to variation in the electrical and magnetic

properties of the lithologies that an electric pulse transmitted from the antenna into the

ground passes through as it makes its way through lithologies with different electrical and

have been loosely defined. Few studies that have investigated the causes of GPR


58/295

reflection in sediments (Van Dam and Schlager, 2000; Van Dam, 2001) discovered that

propagation of electromagnetic signals is primarily dependent on the electrical

conductivity and magnetic permeability of the medium that the signals pass through.

Electrical conductivity is the ability of the material to transmit electric charges under the

influence of an applied field. In rocks and sediment, this often depends on the dielectric

permittivity associated with variations in water content and ionic concentrations. Water

in sediment pore space normally contains ions, and the electrical conductivity associated

with ion mobility is the dominant factor in determining bulk-material electrical

conductivity. Since water is invariably present in the pore space of sediments and rocks,

it has a dominant effect on electrical properties. Magnetic permeability is the degree of

magnetization of a material that responds linearly to an applied magnetic field; this is

significant in rocks or sediments rich in ferromagnetic minerals. However, in most

sedimentary rocks, the amount of ferromagnetic minerals is minimal and hardly

contributes significantly to generation of radar reflections. Hence, in most GPR surveys

on sediments and rocks, the dominant control on radar transmission and reflection is

electrical conductivity governed by water content and the sediment static conductivity.

The variation in sediment water content often relates to both changes in sediment

porosity and permeability, which in turn is dependent on changes in grain size and fabric

often associated with lamination, cross-bedding and bounding surfaces. Rock property

bed laminae and the associated bottom-set layer (Pryor, 1973; Beard and Weyl, 1973,


59/295

Jordan and Pryor, 1987). Emmet et al (1971) observed that porosity and permeability

parallel to the cross-bed laminae are higher than perpendicular to the laminae. Such

porosity and permeability changes have direct control on water content, which determines

electrical conductivity changes that generate radar reflections (Figure 14).

A

High porosity cross-bedB

5 m

Sedimentary bedding is a product of changes in sediment composition and changes in the


60/295

size, shape, orientation and packing of grains (Collinson and Thompson, 1989) which

results in corresponding changes in porosity (Figure 15).

Figure 15: Bedding in sediments and sedimentary rocks and associated porosity resultingfrom changes in composition, size, shape, orientation and packing of sediment grains.(Image modified from Neal, 1994).

Van Dam et al., (2002a) reported that goethite iron-oxide precipitates, occurring either in

bands or irregular layers, were responsible for significant reflections in the aeolian sandsthey studied. This was due to the higher water retention capacity of goethite with respect

to the host quartz sand, which resulted in higher dielectric permittivity.

Higher porosity and permeability

Lower porosity and

permeabilitty

Textural differences between laminae of differing origins may also control cementation


61/295

patterns and, hence, post depositional diagenesis, which may influence radar reflection

patterns in consolidated sediments.

2. 9 GPR Interpretation Methodology and Radar Stratigraphy

Because seismic reflection and GPR data are analogous in terms of wave propagation

kinematics (Ursin, 1983; Carcione and Cavallini, 1995) as well as reflection and

refraction responses to subsurface discontinuities (McCann et al., 1988; Fisher et al.,

1992a) many of the broad assumptions that underpin processing and interpretation of

seismic reflection data (Sangree and Widmier, 1979; Yilmaz, 1987, 2001) are often

routinely applied to GPR data. However, it must be borne in mind that radar images are

different from seismic images in the following respect:

1. Signal penetration and reflection amplitude are controlled by water saturation,

clay content and magnetic property of the sediment.

2. Radar signals are more prone to scattering making diffraction noise more

prominent in radar profiles.

3. The investigation depth is much less.

4. Resolution is in sub-metre scale.

5. Radar surveys require direct contact between the pulse-transmitting antenna and

the medium which the signals are transmitted through (rocks or water/moist

From the interpretation standpoint, the basic assumption in both techniques is that, at the


62/295

resolution of the survey and after appropriate data processing, reflection profiles will

contain accurate information regarding the external geometry and internal structure of a

sediment body. This implies that the form and orientation of bedding and sedimentary

structures in the plane of the survey will be adequately represented by recorded

reflections, and non-geological reflections can be readily identified and removed by data

processing, or by simply discounting them from the interpretation. While this assumption

holds in many instances, accurate interpretation of GPR data is dependent on the nature

and appropriateness of data processing steps undertaken, the interpretation techniques

employed, and the overall understanding and experience of the interpreter with respect to

GPR; hence, care must be taken to ensure that GPR profiles are not treated as geological

cross sections but interpreted in the context of the local geology and ground-truthed with

outcrop data.

Based on the similarities between GPR and seismic reflection, the concept of seismic

stratigraphy was adapted for interpretation of GPR profiles soon after the realization that

GPR could provide useful data for stratigraphic and sedimentological studies (Baker,

1991; Beres and Haeni, 1991); Jol and Smith (1991) coined the term radar stratigraphy

for this new interpretation technique. Gawthorpe et al (1993) defined radar facies basedon the terminology used to describe seismic facies (Mitchum et al., 1977; Brown and

Fisher 1980) as three-dimensional packages that represent particular combinations of

erosional hiatuses despite acknowledging that while seismic sequences can be mapped


63/295

on at least a basin scale and reflect units that form over hundreds of thousands to millions

of years, radar sequences may be limited to particular depositional environments (e.g. a

point bar in a fluvial channel belt) and may form over a much shorter time scale of tens to

thousands of years. Although, this interpretation methodology has been widely

embraced over the years, it is important to underscore the fact that GPR resolution is an

order of magnitude higher than conventional seismic resolution and coverage of GPR

surveys rarely has the capability to image stratigraphic sequences and sequence

boundaries; it is more appropriate for imaging of meso-scale depositional features or

outcrop scale architectural elements. Although decimeter-scale resolution has been

reported in high resolution seismic surveys, the technology is rarely used for outcrop

studies; hence, ground penetrating radar remains the choice technology for high

resolution outcrop imaging.

To avoid interpretive connotations in describing radar reflections, it is recommended that

unambiguous descriptive criteria such as geometry; dip of reflections, relationship

between reflections be used to describe radar reflections rather than terms with

interpretive connotations like, sequence boundaries, systems tract, etc used in seismic

stratigraphy. Interpretations based solely on processing-dependent attributes such asreflection continuity and amplitude should also be avoided as these are dependent on the

data processing algorithms or gain magnitude applied to the data and are hardly

Reflection packages a, b and c (Figure 16A) could be interpreted as different architectural


64/295

elements based on reflection continuity but the reflectors are seen as continuous in the

better gained data (Figure 16B).

Figure 16:(A) Poorly-gained radar profile giving the impression of facies with lowcontinuity (B) Same data in (A) with gains applied to display all reflectors. GPR dataacquired at Whirlpool Sandstone.

Also, while in most cases, radar reflections parallel bedding and bounding surfaces, care

must be taken to ensure that reflections generated by the water table and diagenetic

horizons, diffractions generated by isolated reflector points, out-of-line reflections,

acb

A

B

0

6

0

6

Depth(m)

Depth(m)

2.10 Radar Facies


65/295

Analysis of 150 published studies in the last thirty years on stratigraphic imaging with

ground penetrating radar reveals an obvious bias towards studies of modern and recent

deposits (Figure 17 and Table A3, Appendix 3). Much of the GPR effort on rocks has

been concentrated on carbonates as the low electrical conductivity of carbonate rocks

make them highly conducive to georadar imaging.

Numberofpublishe

dGPR

studies

rofpublishe

dGPR

studies

A

B

Observations from analyses of published studies also show that highest radar signal


66/295

penetration is observed in carbonate rocks due to their high resistivity. This study

however focuses on the application of GPR imaging technology to the studies of ancient

clastic depositional units.

There is currently no formal methodology for documenting radar profiles and radar

facies; this in addition to different processing and display formats adopted by various

authors make comparison of radar facies rather challenging; but an attempt is made

below to summarize observations and describe radar facies commonly observed in a

variety of depositional environments. Radar facies are defined by: external reflection,

geometry, dip of reflections and relationship between reflections. Due to the paucity of

GPR studies on ancient sediments, most of the radar facies discussed in this section are

from modern sedimentary environments and Quaternary deposits.

Since the inception of its use as a tool for stratigraphic and sedimentological

interpretation, much of the emphasis of GPR studies has been on characterization and

interpretation of radar facies, with radar surfaces and radar packages typically not being

defined. This overemphasis on radar facies analysis, rather than true radar stratigraphy,

led to the common misconception that any radar reflection pattern constituted radar

facies. As a result, use of radar facies to describe reflections not related to primary

sedimentary structure and stratigraphy became rife in the GPR literature. Terms such as

water-table radar facies and hyperbolic or diffraction radar facies can be found

resolution of the radar profile). These examples are in violation of the correct use of the


67/295

term radar facies and, therefore, the true principles of radar stratigraphy. Further

confusion to the interpretation of radar profiles has also been caused by use of

interpretive facies names, such as channel-fill facies and overbank facies. If the

context is not made clear, use of interpretive facies names has the danger of suggesting

that particular depositional environments or sub-environments are characterized only by

certain radar facies which is not true based on the findings from published studies and

GPR outcrop data obtained in this study. Radar facies should therefore be described

without interpretive connotations and an uninterpreted version of the radar facies

provided to provide readers an unprejudiced data and basis for interpretation.

The value of radar facies interpretation is demonstrated in all the case studies for this

research especially at the Shinarump Conglomerate (Figure 21 A), Navajo Sandstone

(Figure 21 B) and Whirlpool Sandstone (Figure 22A). Common radar reflection patterns

(facies) typically observed from a variety of sedimentary environments are discussed

below:

2.10.1 Concave upward/trough-shaped reflectors

Concave-upward ground radar reflectors are observed as valley fill (Ilya, 2006), cut and

fill (Heinz et al, 2003) channel (Figure 18), chute, scour, bar top hollows (Best et al,

2006) and trough cross-beds (Bristow et al, 1999). Documented concave upward GPR

Channels are often several metres to hundreds of metres in width and may have simple to

l fill


68/295

complex fill.

Figure 18: GPR Image of fluvial channels from Dunvegan Formation outcrop (PinkMountain, British Columbia).

Valleys are typically hundreds of metres to kilometre-scale in width, they contain

channels and channel belts and are rarely defined in GPR profiles.

While the various concave-upward radar facies are sometimes geometrically-identical, it

might be difficult to correctly interpret them; they can however be discriminated by their

scale, internal geometry and facies associations. This underlines the importance of

documenting the horizontal and vertical scale of GPR profiles. The geometry of

reflectors often depends on the orientation of the GPR line in relation to the channel axis

(for bedload deposits). GPR profiles along the axis of a channel deposit usually reveal

different reflector geometry compared to profiles perpendicular to the channel axis; hence

it is important to document the orientation of GPR profile from which the radar facies are

obtained.

CH CH

A


69/295

B

2.10.2 Inclined Reflectors

li d d fl b d i i f i h f


70/295

Inclined radar reflectors are observed in a variety of environments; they range from

planar cross-beds, lateral accretion and downstream accretion in fluvial deposit to

Inclined Heterolithic Stratification (IHS) in tidal bars. Inclined radar facies are also

observed in eolian deposits as giant cross-beds and also in deltaic clinoforms (Figure 20).

They are often vital paleocurrent indicators; hence the orientation of the GPR profiles

where they are observed as well as the orientation with respect to channel trend or

regional paleocurrent direction should be carefully documented. They can be

discriminated by the vertical and horizontal scale of the reflectors as well as the

association with contiguous facies, ground-truthed with outcrop observations or sediment

trenches. Examples from modern and ancient sediments are given below:

2.10.3 Reflection-free facies

Reflection-free configuration often indicates one of the following:

massive homogenous lithological units with no contrasting dielectric properties

the presence of highly conductive dissolved minerals in groundwater

magnetic sediment

or the presence of sediments containing high clay content that rapidly attenuates

transmitted electromagnetic signal.

A


71/295

Figure 20: (A) Inclined Radar facies from Fraser and Squamish River (Image takenfrom Wooldridge and Hickins, 2005) (B) Downstream accretion macroform fromCastlegate sandstone, Utah (C) GPR profile at the aeolian Navajo Sandstone from ZionNational Park, Utah revealing a set of cross stratification. (Image taken from Jol et al.,2003).

B

C

deposits can be clearly interpreted in fluvial deposits and interdune deposit discriminated

from dry phase eolian dunes based on reflection geometry and amplitude (Figure 21)


72/295

from dry phase eolian dunes based on reflection geometry and amplitude (Figure 21)

Figure 21: (A)Example of reflection-free radar facies- Shinarump Sandstone (Hurricane

Mesa, Utah) (B)Radar facies from Navajo Sandstone (Moab, Utah)

2.10.4 Horizontal Reflectors

Reflection-free facies

Reflection-free facies

B

Shinarump Conglomerate

Moenkopi Shale

Muddy interdune deposits

Amplitude Scale

Low

High

A

floodplain deposits and lacustrine sediments. They are often recognized by their

reflection amplitudes and facies association; for instance, muddy horizontal overbank


73/295

reflection amplitudes and facies association; for instance, muddy horizontal overbank

deposits are usually high to medium amplitude reflectors with low amplitude at the

base as observed at the Whirlpool Sandstone Figure 22A) while sand-rich intra channel

horizontal reflectors are usually characterized by high amplitude sometimes underlain

by a thin layer of low amplitude reflectors indicating shaly channel base breccia or mud

chips (Figure 22 A and B). Mud beds typically have low amplitude thin horizontal

reflectors with indistinguishable reflectors at the base.

Figure 22: (A)Horizontal radar facies overbank deposit from Whirlpool Sandstone,Artpark, New York (B) Horizontal reflectors from Shinarump sheet conglomerate, Utahinterpreted as upper plane bed flood deposits.

Queenston Shale

Overbank deposit

Channel margin

A

B

Sandy Bedform


74/295

Table 2:Radar facies elements of different sedimentary depositional environments.(Table obtained from Van Overmeeren, 1998)


75/295

( )

Based on this surmise, Van Overmeeren (1998) suggested a compilation of radar facies

atlas from a variety of depositional environments as guide for facies interpretation and


76/295

Figure 24: Radar facies chart of characteristic reflection patterns from varioussedimentary environments. (Image taken from Van Overmeeren, 1998).

Analysis of radar facies observed in this study and from other published GPR studies


77/295

Huggenberger, 1993; Bristow, 1995; Beres et al., 1999; Neal and Roberts, 2000, 2001;

Corbeanu et al., 2001; Heinz, 2001; Neal et al., 2001, Russell et al., 2001; Szerbiak et al.,


78/295

2001; Hornung and Aigner, 2002; ONeal and McGeary, 2002; Heinz and Aigner, 2003;

Skelly et al., 2003). While these studies confirm the usefulness of GPR imaging as a

powerful technique for imaging clastic sedimentary rocks, the idea that radar facies

elements used in interpretation of radar reflection patterns are characteristic of certain

sedimentary depositional environments should be applied with caution while interpreting

outcrop data as

Oluwatosin Akinpelu C 201009 PhD Thesisghgh

Documents

Transcript of Oluwatosin Akinpelu C 201009 PhD Thesisghgh