Environmental & Engineering Geoscience

Environmental &Engineering GeoscienceMAY 2016 VOLUME XXII, NUMBER 2

THE JOINT PUBLICATION OF THE

ASSOCIATION OF ENVIRONMENTAL AND ENGINEERING GEOLOGISTS

AND THE GEOLOGICAL SOCIETY OF AMERICA

SERVING PROFESSIONALS IN

ENGINEERING GEOLOGY, ENVIRONMENTAL GEOLOGY, AND HYDROGEOLOGY

Environmental & Engineering Geoscience (ISSN 1078-7275) is pub-lished quarterly by the Association of Environmental & EngineeringGeologists (AEG) and the Geological Society of America (GSA).Periodicals postage paid at AEG, 1100 Brandywine Blvd, Suite H,Zanesville, OH 43701-7303 and additional mailing offices.

EDITORIAL OFFICE: Environmental & EngineeringGeoscience journal, Department of Geology, Kent StateUniversity, Kent, OH 44242, U.S.A. phone: 330-672-2968, fax:330-672-7949, [email protected].

CLAIMS: Claims for damaged or not received issues will behonored for 6 months from date of publication. AEG membersshould contact AEG, 1100 Brandywine Blvd, Suite H, Zanesville, OH43701-7303. Phone: 844-331-7867. GSA members who are notmembers of AEG should contact the GSA Member Service cen-ter. All claims must be submitted in writing.

POSTMASTER: Send address changes to AEG, 1100Brandywine Blvd, Suite H, Zanesville, OH 43701-7303. Phone: 844-331-7867. Include both old and new addresses, with ZIP code.Canada agreement number PM40063731. Return undeliverableCanadian addresses to Station A P.O. Box 54, Windsor, ON N9A6J5 Email: [email protected].

DISCLAIMER NOTICE: Authors alone are responsible forviews expressed in articles. Advertisers and their agencies aresolely responsible for the content of all advertisements printed andalso assume responsibility for any claims arising therefromagainst the publisher. AEG and Environmental & EngineeringGeoscience reserve the right to reject any advertising copy.

SUBSCRIPTIONS:

Member subscriptions: AEG members automatically receivedigital access to the journal as part of their AEG membershipdues. Members may order print subscriptions for $60 per year.GSA members who are not members of AEG may order for $60per year on their annual GSA dues statement or by contactingGSA.

Nonmember subscriptions are $295 and may be ordered fromthe subscription department of either organization. A postagedifferential of $10 may apply to nonmember subscribers outsidethe United States, Canada, and Pan America. Contact AEG at844-331-7867; contact GSA Subscription Services, GeologicalSociety of America, P.O. Box 9140, Boulder, CO 80301.

Single copies are $75.00 each. Requests for single copies shouldbe sent to AEG, 1100 Brandywine Blvd, Suite H, Zanesville, OH43701-7303.

© 2016 by the Association of Environmental and EngineeringGeologists

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, orby any information storage and retrieval system, without permission in writing from AEG.

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER

ABDUL SHAKOORDepartment of GeologyKent State University

Kent, OH 44242330-672-2968

[email protected]

BRIAN G. KATZFlorida Department of Environmental

Protection2600 Blair Stone Rd.Tallahassee, FL 32399

[email protected]

EDITORSCover photo

A view of one of the alleys in Rock City, Mountain Lake, Virginia, formed byextension along discontinuities during lateral spreading of the rock mass. Photocourtesy of Abdul Shakoor. See article on page 93.

SUBMISSION OF MANUSCRIPTS

Environmental & Engineering Geoscience (E&EG), is a quar-terly journal devoted to the publication of original papers thatare of potential interest to hydrogeologists, environmental andengineering geologists, and geological engineers working in siteselection, feasibility studies, investigations, design or construc-tion of civil engineering projects or in waste management,groundwater, and related environmental fields. All papers arepeer reviewed.

The editors invite contributions concerning all aspects of envi-ronmental and engineering geology and related disciplines.Recent abstracts can be viewed under “Archive” at the website, “http://eeg.geoscienceworld.org”. Articles that report onresearch, case histories and new methods, and book reviewsare welcome. Discussion papers, which are critiques of print-ed articles and are technical in nature, may be published withreplies from the original author(s). Discussion papers andreplies should be concise.

To submit a manuscript go to http://eeg.allentrack.net. If youhave not used the system before, follow the link at the bottom ofthe page that says New users should register for an account.Choose your own login and password. Further instructions willbe available upon logging into the system. Please carefully readthe “Instructions for Authors”.

Authors do not pay any charge for color figures that are essen-tial to the manuscript. Manuscripts of fewer than 10 pages maybe published as Technical Notes.

For further information, you may contact Dr. Abdul Shakoor atthe editorial office.

JOHN W. BELL

Nevada Bureau of Mines andGeologyRICHARD E. JACKSON

(Book Reviews Editor)Geofirma Engineering, Ltd.JEFFREY R. KEATON

AMEC AmericasPAUL G. MARINOS

National Technical Universityof Athens, GreeceJUNE E. MIRECKI

U.S. Army Corps of EngineersPETER PEHME

Waterloo Geophysics, IncNICHOLAS PINTER

Southern Illinois University

PAUL M. SANTI

Colorado School of MinesROBERT L. SCHUSTER

U.S. Geological SurveyROY J. SHLEMON

R. J. Shlemon& Associates, Inc.GREG M. STOCK

National Park ServiceRESAT ULUSAY

Hacettepe University, TurkeyCHESTER F. “SKIP” WATTS

Radford UniversityTERRY R. WEST

Purdue University

EDITORIAL BOARD

ASSOCIATE EDITORS

JEROME V. DEGRAFF

USDA Forest ServiceTHOMAS J. BURBEY

Virginia Polytechnic InstituteSYED E. HASAN

University of Missouri, Kansas City

ROBERT H. SYDNOR

ConsulantCHESTER F. WATTS (SKIP)Radford University

Environmental &Engineering Geoscience

Volume 22, Number 2, May 2016

Table of Contents

93 Using Discontinuity Mapping to Investigate the Origins of Rock City and Mountain Lake, Giles County,

Virginia

Nidal W. Atallah, Abdul Shakoor, and Chester F. Watts

113 Discovering and Characterizing Abandoned Waste Disposal Sites Using LIDAR and Aerial Photography

Andrew de Wet

131 Identification of Wall Tension Fractures Caused by Earthquakes, Blasting, and Pile Driving

Jeffrey A. Johnson and Alan “Bob” Mutchnick

141 Geologic and Geotechnical Factors Controlling Incipient Slope Instability at a Gravel Quarry, Livermore

Basin, CaliforniaPhilip L. Johnson, Patrick O. Shires, and Timothy P. Sneddon

157 Factors Affecting Failure by Internal Erosion of Geosynthetic Clay Liners Used in Freshwater Reservoirs

Hakki O.Ozhan and Erol Guler

171 Book Review

Geomodels In Engineering Geology—An Introduction

By Peter Fookes, Geoff Pettifer, and Tony Waltham

Review by: Richard Jackson

Using Discontinuity Mapping to Investigate the Origins ofRock City and Mountain Lake, Giles County, Virginia

NIDAL W. ATALLAH

Department of Geology, Kent State University, Kent, OH 44242, [email protected]

ABDUL SHAKOOR1

Department of Geology, Kent State University, Kent, OH 44242, [email protected]

CHESTER F. WATTS

Department of Geology, Radford University, Radford, VA 24142, [email protected]

Key Terms: Mountain Lake, Rock City, LandslideDam, Colluvial Deposits Lateral Spread, DiscontinuityMapping

ABSTRACT

Mountain Lake’s unusual location, near the summit ofSalt Pond Mountain, VA, in the non-glaciated portion ofthe Appalachian Mountains, has prompted geologists tostudy its origin for decades. The northeastern end of thelake abuts an area of heterogeneous colluvial depositsthat contain large rectangular blocks of hard TuscaroraSandstone. This area is known as “Rock City” becauseof the resemblance of the gaps between the rock blocksto streets and alleys in a city. The purpose of this studywas to investigate the origin of Rock City and whetherthe colluvial deposits within its boundaries are part of alandslide that is possibly responsible for the formation ofMountain Lake. Mapping of Rock City included takingglobal positioning satellite readings at the corners ofrock blocks and along the boundaries of other outcrops,then using ArcMap software to generate maps. Usingstereonet analysis, the mode of rock-block displacementwas investigated by comparing the measuredorientations of principal discontinuity sets forming therock-block boundaries with discontinuity orientations ofundisturbed outcrops. Discontinuity data analysisindicates that Rock City is most likely a landslide thatdammed the valley of Pond Drain, forming the lake.The primary mode of slope movement involves lateralspreading associated with extension occurring alongdiscontinuities. The Tuscarora Sand-stone blockscomprising Rock City were detached from a scarp facealong a northwest-southeast–trending joint set and weredisplaced laterally toward the west. A seis-mic eventmay have triggered slope movement; however, more

extensive analysis must be done to rule out otherinterpretations.

INTRODUCTION

Mountain Lake is located in Giles County, VA,about 18 km east of Pearisburg (Figure 1). It is situ-ated approximately 1,200 m above mean sea level(AMSL) AMSL, between the crest and the northwestlimb of a gently plunging anticline (Figure 2). Theanticline is on the Narrows thrust sheet in the Valleyand Ridge physiographic province. The axial plane ofthe anticline trends N60u E (Mills, 1989), and its axisplunges toward the northeast at 7.5u (Parker et al.,1975). Three geologic units underlie the lake: theSilurian Tuscarora Sandstone (Stu) at the northernend, the Ordovician Juniata Sandstone (Oj) underthe middle portion, and the Ordovician Reedsville-Trenton Formation (Ort) at the southern end (Figure2 and Table 1). Mountain Lake is an elongated bodyof water oriented south to northwest. The lake, fed byprecipitation, surface runoff, and a line of lake-bot-tom springs, covers a total surface area of about 1.96 105 m2 within a small watershed (1.3 km2) (Cawleyet al., 2001). The headwaters of four major streams lieclose to the lake: Sartain Branch to the east, DoeCreek to the southwest, Johns Creek to the south,and Pond Drain to the northwest.Mountain Lake has a history of unusual self-drain-

ing behavior. Recent episodes in 2008, 2011, and2012 almost completely drained the lake. Draining ofthe lake revealed the presence of four sinkhole-likedepressions in the lake sediment that has accumulatedover colluvium, with piping holes at their bottoms andsides, near the northeastern and northwestern marginsof the lake (Figure 3). Aside from water loss throughevapotranspiration and surface runoff through Pond1Corresponding author, email: [email protected]

Environmental & Engineering Geoscience, Vol. XXII, No. 2, May 2016, pp. 93–111 93

Drain (only when the lake is full), many studies (Mar-land, 1967; Parker et al., 1975; Cawley, 1999; Cawleyet al., 2001; Jansons et al., 2004; Roningen, 2011;and Joyce, 2012) have suggested that lake-level fluc-tuations are caused by water seeping out of the lakebasin through subterranean pathways at the northernend of the lake. Rock City is the area at the northernend of the lake (Figure 4a) featuring large rectangularrock blocks, measuring up to approximately 140 m2,and other colluvial deposits of rock fragments of vary-ing sizes. The objective of this study was to investigatewhether Rock City is part of a landslide that

contributed material for damming Pond Drain andto determine the mode of displacement of large blockswithin Rock City.

PREVIOUS HYPOTHESES ON THE ORIGIN OFMOUNTAIN LAKE

Karst-Related Hypothesis

Holden (1938) hypothesized that Mountain Lakeformed as a result of dissolution of a calcareous layerin the upper Reedsville-Trenton Formation. Ferguson

Figure 1. Location of the study area.

Figure 2. Geologic map of the study area; modified from geologic map of Giles County, VA, by Schultz et al. (1986).

Atallah, Shakoor, and Watts

94 Environmental & Engineering Geoscience, Vol. XXII, No. 2, May 2016, pp. 93–111

et al. (1939) supported this hypothesis, referring to theorigin as a “natural solution collapse basin.” The exis-tence and depth of calcareous strata in the Reedsville-Trenton Formation has not been confirmed. Holden(1938), Butts (1940), Marland (1967), and Roningen(2011) suggest its presence within the uppermost por-tion of the formation, while others, such as Eckroade(1962) and Parker et al. (1975), propose its presenceat a considerable depth. Sharp (1933) rejected the pos-sibility of a natural collapse basin origin, asserting thatdissolution could not have reached the surface becausethe depth of the soluble limestone is approximately305 m. He also argued that Mountain Lake did notresemble other limestone sinks in the area. Parker et al.(1975) used calculations of formation thicknesses todismiss the collapse basin hypothesis. He argued that“a true sinkhole in the upper Martinsburg [Reeds-ville-Trenton] would have to extend upward throughat least 15 m of Juniata to reach the bottom of Moun-tain Lake, and then further dissolve through 457 m ofMartinsburg [Reedsville- Trenton] to make a sink-hole.” However, Williams (2003), in his Encyclopediaof Caves and Karst Science, provides an account ofsubjacent karst collapse sinkholes in caverns deeperthan 1,000 m in Canada and Russia. Finally, Ronin-gen (2011) identified the presence of “significant car-bonate content” in the uppermost section of theReedsville-Trenton Formation in Narrows, VA. Basedon this observation, and considering the alkalinity of

Mountain Lake, she argued that karst dissolutionshould not be ruled out as a possible hypothesis forMountain Lake’s formation.

Landslide-Dam–Related Hypothesis

The major studies promoting the landslide damhypothesis are those of Rogers (1884), Hutchinsonand Pickford (1932), Sharp (1933), Eckroade (1962),Marland (1967), and Parker et al. (1975). Most ofthese studies suggest that the headwaters of the north-westerly flowing Pond Drain cut through the resistantsandstone ridges, breaching the northwestern end ofthe anticline and carving out a narrow valley. The nar-row valley was then dammed by the colluvial blocks ofthe Tuscarora Sandstone that make up Rock City.The mode of displacement by which the damming

occurred has been attributed to different types ofmass movement and is the primary focus of this study.Hutchinson and Pickford (1932) and Hutchinson(1957) suggested that the damming occurred as a resultof the “caving in of overhanging ledges of hard rockthat were undermined by the stream.” Sharp (1933)suggested that Tuscarora blocks crept downwardfrom the ridge in the form of talus, but he mentionedthe possibility of a rockslide as well. Eckroade (1962)agreed with Sharp’s explanation but added that frostheaving generated additional Tuscarora blocks thatwere displaced, probably by solifluction, further

Table 1. Simplified stratigraphic column in the study area.

Formation Age Rock Type Thickness (m) (Mills, 1990)

Rose Hill (Srh) Silurian Sandstone 45–60Tuscarora (Stu) Silurian Sandstone/orthoquartzite 15–45Juniata (Oj) Ordovician Sandstone 60–110Reedsville-Trenton (Ort) Ordovician Shale (mostly) and limestone 425–490

Figure 3. Drastic water level drop in December 2012 showing the lake-bottom depressions (left) compared to lake level in April 2012 (right).

Origins of Rock City and Mountain Lake


damming the stream valley. In this case, solifluctioncorresponds to the definition provided by Easterbrook(1999): the “downslope movement induced by alter-nate freezing and thawing of debris slopes” or “geli-fluction.” Marland (1967) used 14C-dating of corebottoms to confirm that the colluvial material was pro-duced by solifluction during climate with repeatedfreeze-thaw cycles about 9,180¡ 330 years before pre-sent (YBP) YBP. He also added that the leaky nature

of the damming material prevented the formation ofa “permanent lake” until about 2,000 YBP. Parkeret al. (1975) synthesized a number of modes of displa-cement suggested by earlier works in explaining thedamming process and consequent lake formation.These authors agreed that the damming occurred viatalus or slide rock (Sharp, 1933; Eckroade, 1962)through mass movement by solifluction (Eckroade,1962; Marland, 1967) and vertical collapse of ledges

Figure 4. (a) Aerial view and bathymetric map of Mountain Lake, Rock City, and Pond Drain. North is at top of image. Yellow line on mapmarks location of the cross section shown in b; (b) southwest (left) to northeast (right) cross section of the north end of Mountain Lake atPond Drain, obtained by seismic refraction (C. F. Watts, Radford University, 2013).



due to undercutting (Rogers, 1884; Hutchinson andPickford, 1932). However, they suggested that verticalcollapse, promoted by undercutting of the resistantTuscarora and Rose Hill sandstone beds by erosionof the less resistant Juniata Sandstone, was the pri-mary mode of displacement. Investigations of colluvialdeposits in the Mountain Lake area, not far from thestudy site, led Mills (1981, 1988, 1989, 1990) to suggesttheir lateral transport and the retreat of Tuscaroraescarpments through the process of “topographicinversion of hollows and noses.”

Recent seismic refraction studies, performed byWatts (2013), confirm the presence of a colluvium-filled valley at the northwestern end of the lake (C.F. Watts, oral communication, 2013). The studyrevealed a narrow gorge in the bedrock, at least 30 mdeep and filled with colluvium, at the north end ofthe lake (Figure 4b).When the lake is full, the wateroverflows into Pond Drain (Freeman et al., 2012).This supports the various landslide hypotheses.

Fracture-Lineation–Related Hypothesis

Cawley (1999) conducted fracture trace analysis thatdocumented a lineament trending from SE to NW. Heinterpreted the lineament as a “fracture and probablefault associated with the regional Appalachian foldand thrust tectonics.” Cawley (1999) studied this fea-ture using direct current resistivity techniques and con-cluded that pronounced resistivity lows along thefeature were indicative of a “water-filled fracturezone.” Cawley (1999) proposed that the regional frac-ture had a role in carving out the valley of Pond Drainat the northwestern end of the lake as well as in itsdamming and consequent lake formation. In a laterpublication, Cawley et al. (2001) described the dam-ming process as “incremental settling and breakup ofan overlying resistant rib of Clinch [Tuscarora Sand-stone] bedrock in physical contact with the fault linea-tion.” Additionally, Cawley (1999) and Cawley et al.(2001) suggested that the basin of Mountain Lakewas formed when fine sediments were eroded awayby water seeping through the fracture. This alsoexplains periodic drops in lake water levels. However,in a study by Roningen (2011), the presence of thefracture identified by Cawley (1999) could not be con-firmed using electrical resistivity tomography, jointsampling, or lineament analysis.

RESEARCH METHODS

Investigations of Rock City included mapping thelocations and orientations of the large rectangularrock blocks, the “alleys” and “streets” separating the

blocks along major discontinuities, and other colluvialdeposits. The study also included an investigation of acomplex of scarps consisting of irregular, discontinu-ous cliff outcrops of Tuscarora Sandstone just upslopeof Rock City (Figure 5). Discontinuity data, includingboth bedding planes and joint sets in the rock blocksand outcrops upslope, were analyzed in order to eval-uate the mode and extent of displacement exhibitedby the rock blocks.

Rock City Mapping

A GIS map (Figure 6) was generated using globalpositioning satellite (GPS) readings taken at the cor-ners of rock blocks, at the outcrops upslope of therock blocks, and at the boundaries of boulder fieldscontaining a substantial number of small to medium-sized colluvial boulders. The GPS unit used for map-ping included a Trimble Pro XRT Backpack and aNOMAD Data Logger with Terrasync V6.x. FieldSoftware. The data points collected were differentiallycorrected using GPS PathfinderH Office Software andentered into ESRI ArcMap 10 Office Software to gen-erate the Rock City map. By connecting the coordi-nate points at the rock-block corners, the boundariesof large rock blocks were drawn on the map as idea-lized rectangular-shaped polygons (Figure 6), match-ing the orientations of two principal and nearlyvertical joint sets representing the sides of each block(Figure 7). Tape measurements of block dimensionswere taken to supplement the GPS readings and tocorrect for block shapes that were irregular as a resultof weathering and/or disintegration of the originaljoint surfaces.The rock blocks were categorized into the following

size-based groups: very large (with footprints.50 m2),large (20–50 m2), medium (2.5–20 m2), and small(,2.5 m2). In this study, all blocks with a base areaof $2.5 m2 were considered “rock blocks,” and theirdimensions were measured in detail. Rock blockswith footprint areas of ,2.5 m2 were considered“boulders” and were mapped only where concentratedinto sizeable boulder fields.

Discontinuity Measurements

The orientations of discontinuities within the scarpoutcrops upslope were measured using the WindowMapping Method (Wyllie and Mah, 2004). All jointsencountered within the selected area were measured.These measurements were used to assess whether theseoutcrops represented the in situ bedrock and to estab-lish a baseline for comparing the orientations of dis-continuities within the detached and transported



blocks in Rock City. Similarly, the orientations of alljoints within the rock blocks were measured, especiallythose clearly making up the sides of the blocks (Figure7). Discontinuity measurements were made usingboth a Brunton compass and an Ipad (applicationGeoID 1.61). DIPS 6.0 software (Rocscience, 2012)

was used to generate stereonet plots of discontinuityorientations that were used to identify bedding andprincipal joint sets for both the scarp outcrops andfor individual rock blocks. The dip and dip directionsfor both bedding planes and joint sets were comparedbetween the rock blocks and the scarp outcrops using

Figure 5. Examples of scarp outcrops: (a, b) small separate scarp outcrops and (c–f) continuous cliff-like walls of scarp of varying heights. Thescale in (a) through (f) is approximately 1 cm 5 0.7 m.



DIPS and RockPack III (Watts et al., 2012) softwareprograms, respectively, in order to determine thefollowing:

1) The direction of tilt exhibited by each rock block;2) The direction, clockwise or counterclockwise, and

angle of lateral rotation experienced by rock blocksduring displacement away from the scarp outcrops.The two principal orthogonal joint sets, identifiedin both the scarp outcrops and in each of theblocks, were used to determine rock-block rota-tion; and

3) The type of mass movement that resulted in theformation of Rock City and its connection toMountain Lake’s formation.

ROCK CITY MAPPING RESULTS

Rock City can be divided into two main compo-nents: (1) a belt of multiple scarps of Tuscarora Sand-stone upslope, referred to as the scarp complex in thefollowing discussion, and (2) the overall debris fieldcontaining the large rectangular rock blocks as wellas other coarse colluvial deposits.

Scarp Complex

The scarp complex bounds the debris field to thenorth and consists of partly exposed, cliff outcropsthat are discontinuous and non-linear in nature butthat have an overall trend of N40u–70uW (Figure 6).The scarps vary in height, width, elevation, and aspect.They range from small separate outcrops to sizable cliffwalls (Figure 5). The scarps crop out at different eleva-tions with variable setbacks along any particular stretch,in places giving the appearance of steps or benches. Tohelp visualize the exposure of outcrops at several levels,Figure 8 shows the GPS readings taken at outcropsalong the scarp complex during various trips.Bedding attitudes in the scarp complex are nearly

horizontal (Figure 9). Bedding dips vary from 0u toFigure 7. An example of the two principal joint sets marking theboundaries of a rock block in Rock City.

Figure 6. Rock City boulder field and large block locations.



24u, with an average dip of 7u and a standard deviationof 4.5u. Dip directions in the scarp complex average331u but vary somewhat, with a circular standard devia-tion of 72u. At least four principal joint sets are presentin the scarp complex (Figure 10). The near-vertical joint

sets strike: (1) N35uW, (2) N15uE, (3) N57uE, and (4)N83uE. Sets 1 and 3 are distinctly orthogonal. TheN35uW set may be stress relief joints that developedwhen Pond Drain carved a sandstone-walled valleyprior to the collapse of a portion of the valley wall.

Figure 8. Locations of GPS readings along scarp outcrops where discontinuity measurements were taken.

Figure 9. Bedding plane attitudes for scarp outcrops plotted as poles. Plot generated by DIPS software (Rocscience, 2012).



Variations in bedding-plane and joint orientationswithin the scarp complex most likely exist becausesome blocks that have started to detach from in situstrata within Salt Pond Mountain have tilted orrotated by a small amount. The irregular shape ofthe scarp line (Figure 6) implies that blocks detachedalong pre-existing fractures and joints. The scarp isgenerally a continuous line of outcrops that extendsthe full width of the area of the large rock blocks andboulder fields comprising Rock City, so it can be con-sidered the head scarp for the entire debris field. Also,notwithstanding some irregularities, the standarddeviation of 4.5u for bedding dip in the scarp outcropsmeans that the average dip of 7u can be used for com-parison with bedding-plane orientations within thedetached blocks in Rock City. Bartholomew et al.(2000), authors of the geologic map for the Radfordquadrangle, indicated local dip to the NE near thenortheastern corner of Mountain Lake. This studyfound dip directions primarily to the NE and to theNW and SW. The NW directions represent the NWlimb of the anticline. The SW dips are contrary tothe NE plunge of the anticline, and they possibly indi-cate cambering out of the local hillside, away from thescarp. The NE dips may be the result of some back-rotation from slope movement.

Debris Field

The Rock City portion of the debris field extendsabout 170 m E-W along the lake shore and about100 m N-S from the lake shore to the scarp complex(Figure 6). Its main components are the rock blocks,“streets” and “alleys” separating the blocks, andboulder fields or “jumbles” containing boulders of allsizes. Mapping of Rock City revealed the presence ofat least 57 rock blocks, of which 10 are very large, 21are large, and 26 are medium-sized (Figure 11). Mostrock blocks are rectangular. The passageways betweenrock blocks, referred to as “streets” and “alleys,” varyin width from a few to several meters (Figure 12).They vary from rectilinear to nearly random. Rockswith footprint areas of less than 2.5 m2, designated assmall boulders, are scattered over the entire area ofRock City, but the map documents only concentratedboulder fields (Figure 13). These boulder fields origi-nated from scarp outcrops, collapse of overhangsbetween adjacent blocks, and breakdown of largerrock blocks (Figure 13).The bedding planes of rock blocks dip in all direc-

tions; however, the majority of the dip directionsexceed 180u, indicating that dips toward the NW, W,and SW are more common, with a slight predomi-nance to the NW (Table 2). Exceptions to the overall

Figure 10. Density concentrations of poles for joint set orientations in the scarp outcrops, with the corresponding great circles of the principaljoint sets labeled in blue. Plot generated using DIPS software (Rocscience, 2012).



westward tilting include two main clusters of rockblocks present at the southeastern edge of Rock Cityand another near the scarp complex, which dip mostlysouth and north, respectively.Measurements of joint orientations in rock blocks

indicated that each of the blocks has two principaljoint sets and in some cases other minor/secondarysets (Table 2). The two principal joint sets were, inthe majority of cases, nearly orthogonal, similar tothe orthogonal sets 1 and 3 identified in the scarpcomplex.

DISCONTINUITY MAPPING AND ANALYSISRESULTS

Analysis of Bedding Orientation Data

Bedding planes within the rock blocks in Rock Cityare more scattered than those in the scarp outcrops(Figure 14), suggesting that block orientations becamerandomized during transport. The rock-block beddingplanes also have steeper dips, ranging from 5u to 82u,compared to the scarp area in which bedding dipsrange from 0u to 24u and average 7u. The predomi-nantly westward orientations of rock-block bedding

planes (Figure 14) indicate primarily outward tilting,away from the head scarp (Table 2). Tilt appears toincrease with distance from the scarp and to be morerandomized with smaller blocks. However, the clusterof blocks at the southeastern edge of Rock City (Fig-ure 6) experienced greater tilting toward the SE andSW, most likely because of steeper slopes in that area(Figure 15). The small number of northward-dippingblocks, located very close to the scarp face, suggeststhat after initial detachment, blocks back-rotated.Blocks farther away from the scarp moved by lateralspreading and by outward tilting from the scarp.

Analysis of Joint Orientation Data

The orthogonal joint sets 1 and 3 in the scarp facewere used as references to compare to joint orienta-tions in rock blocks. The purpose was to determinethe degree of lateral rotation experienced by the blocksduring displacement away from the scarp. As lateralrotation exceeds 40u, distinguishing between the jointfaces and between clockwise and counterclockwiserotation becomes very difficult. The same problemmay occur for rock blocks that do not form a perfect

Figure 11. Examples of rock-block sizes: rock block #16 (top) is ofmedium size, and rock block #28 (bottom) is of very large size.

Figure 12. Rock City streets and alleys of varying widths: (top) asmall alley (3 m wide) and (bottom) a large alley “Main Street”(14 m wide).



90u orthogonal system of joints. For example, rockblock #32 shows that it might have experiencedeither a clockwise or counterclockwise rotation of 3u(Table 3). This study focused in part on major rockblocks relatively close to the head scarp, namely thosein Main Street, because of their limited rotation andtheir low degree of tilting. Orientations of joints in

the rock blocks clustered around Main Street (Figure16) demonstrate that these blocks have rotated from10u to 30u in both clockwise and counterclockwisedirections. Additionally, the rock blocks that are adja-cent to one another seem to exhibit similar rotationalmovement. For example, rock blocks 24A, 24B, and25 (Figure 16) rotated counterclockwise about 15u–

Figure 13. Examples of boulder fields (a, b) and the processes supplying material for boulder fields, such as collapsing roofs between rockblocks (c, d), collapsing overhangs (e), and sliding rock blocks from the scarp (f). Approximate scale: 1 cm 5 0.7 m.



Table 2. Rock-block sizes and discontinuity orientations.

Discontinuity Orientations*

Joint Sets (JS)

Rock Bock (RB) No. Block Size (m2) Size Category Bedding Plane** JS#1 JS#2 JS#3 JS#4

RB#1 21.2 Large 13/213 N70E N20W N15E N60WRB#2A 35.0 Large 21/299 N40E N50W N4W —RB#2B 28.1 Large 13/322 N40E N53W North-South N34ERB#3 9.8 Medium 17/339 N70E N13W N85E —RB#4A 34 Large 11/221 N45E N52W N70E N6WRB#4B 9.3 Medium 21/259 N60E N20W N5E —RB#5A 18.0 Medium 13/300 N48E N42W — —RB#5B 9.0 Medium 10/006 N35E N55W N70WRB#6 19.0 Medium 06/314 N45E N45W N12W —RB#7 26.2 Large 06/267 N30E N60W — —RB#8 10.5 Medium 29/354 N85E N5E — —RB#9 14.2 Medium 28/276 N23E N67W — —RB#10 31.8 Large 9/344 N77E N14W — —RB#11 31.9 Large 11/310 N55E N35W — —RB#12A 14.4 Medium 82/201 N25E N65W — —RB#12B 9.7 Medium 23/21 N25E N65W — —RB#12C 6.3 Medium 30/196 N5E N80W N42E —RB#12D 13.6 Medium 16/255 N85E N5W — —RB#12E 8.8 Medium 36/314 N18E N72W — —RB#12F 5.5 Medium 75/197 N12E N85E N15W —RB#12G 10.6 Medium 50/200 N76E N10W — —RB#13 12.6 Medium 15/280 N35E N65W — —RB#14 48.1 Large 32/192 N40E N50W N75E —RB#15 19.3 Medium 18/232 N10E N80W — —RB#16 7.8 Medium 16/260 N84E N6W — —RB#17 25.3 Large 63/238 N35E N80W — —RB#18 38.5 Large 26/203 N70E N3E N35W N40ERB#19 36.6 Large 20/169 N73E N17W N82WRB#20 34.6 Large 11/150 N30E N60WRB#21A 52.4 Very large 6/158 N54E N36W N15ERB#21B 2.7 Medium 50/084 N5E N85E N25W N20ERB#22 54.6 Very large 71/318 N46E N44W — —RB#23 9.6 Medium 32/358 N35E N40W — —RB#24A 66 Very large 11/315 N40E N50W N63W —RB#24B 47.3 Very large 10/013 N40E N50W N70W —RB#25 93.2 Very large 08/318 N35E N55W N34W N66WRB#26 42.9 Large 8/096 N70E N20W — —RB#27 31.9 Large 14/325 N70E N15W — —RB#28 102.4 Very large 07/305 N65E N25W — —RB#29A 64.9 Very large 09/317 N86E N5W N30E —RB#29B 57.4 Very large 09/025 N78E N15W N8W —RB#30A 104.0 Very large 12/305 N20W N85W — —RB#30B 19.3 Medium 20/311 N5E N70W — —RB#31 11.2 Medium 05/293 N82W N8W — —RB#32 142.5 Very large 05/167 N60E N38W — —RB#33 57.1 Very large 24/324 N60E N30W — —RB#34A 35.1 Large 25/318 N42E N48W — —RB#34B 45.3 Large 27/315 N48E N42W — —RB#35A 6.6 Medium 13/281 N45E N42W N7W —RB#35B 4.3 Medium 12/277 N70E N20W — —RB#35C 6.5 Medium 12/285 N65E N25W — —RB#36 18.5 Medium 19/068 N70E N15W — —RB#37 18.6 Medium 07/342 N80E N30W N55E —RB#38 33.1 Large 09/359 N70E N25W N55W —RB#39 26.7 Large 18/026 N50E N30W N10W —RB#40 41.3 Large 20/012 N20E N70E — —RB#41 26.5 Large 33/310 N45E N35W — —

*Strike directions (degrees); dips are near vertical.**Dips/dip directions (degrees).



20u, while rock blocks 26 and 27 rotated clockwiseabout 13u (Table 3). Rock blocks 29A and 29B, atthe western end of Main Street, show a clockwise rota-tion of 29u and 20u, respectively (Table 3).

Joint set orientations for a number of selected rockblocks from other locations in Rock City were alsocompared to the joint orientations in the scarp out-crops (Figure 17 and Table 3). The direction anddegree of rotation for some of these rock blocks canbe described as follows: rock blocks that exhibit acounterclockwise rotation are #7 (25u), #9 (32u),#14 (15u), #20 (25u), #21A (1u), #22 (9u), #34A(13u), and #34B (7u), whereas those that exhibit aclockwise rotation are #10 (20u), #33 (3u), and #19(16u) (Table 3). The comparison does not support a

relationship between the locations of the rock blocksand their degrees of rotation. However, rock blocksimmediately adjacent to Main Street, both to the eastand west, exhibit minimal rotation of just 1u to 15u.The rock blocks in the cluster at the southeasternedge of Rock City had high rotation angles, rangingfrom 13u to 45u. However, the usefulness of comparingdiscontinuity orientations in the scarp to those in therock blocks is limited.

ORIGINS OF ROCK CITY ANDMOUNTAIN LAKE

Head scarp and rock-block discontinuity orienta-tion data support the interpretation that Rock City

Figure 14. Comparison between poles of bedding plane attitudes for the scarp outcrops (red) and the rock blocks mapped in Rock City(black). Plot generated using DIPS software (Rocscience, 2012).



comprises the remnants of a landslide that dammedPond Drain, forming Mountain Lake. This interpreta-tion agrees with several previous studies (Rogers,1884; Hutchinson and Pickford, 1932; Sharp, 1933;Eckroade, 1962; Marland, 1967; and Parker et al.,1975). In addition, seismic refraction data indicatethe presence of a colluvium-filled, buried gorge nearthe northern end of the lake (Figure 4b). The predomi-nance of blocks tilted outward from the scarp anddown the hillside, with principal lateral rotationaround an axis perpendicular to slope, in both clock-wise and counterclockwise directions, also supportsthe landslide hypothesis.The presence of a scarp at the northern edge of

Rock City suggests that displacement of the Tuscarorarock blocks occurred from slope movement ratherthan from karst-related subsidence. The absence of ascarp feature along the entire perimeter of MountainLake suggests that it is unlikely for such a collapse tohave occurred. However, without further subsurfaceinvestigations, the possibility of karst collapse withinthe Reedsville-Trenton Formation cannot be dis-carded. Thus, a catastrophic slope failure appears tobe the most likely mechanism by which the rock blocksof Rock City were displaced.The blocks of Tuscarora Sandstone, which make up

Rock City, broke loose from the scarp face, generallyalong a northwest-southeast–trending joint set, oneof the two orthogonal joint sets. After the initialdetachment, the rock mass broke into rectangularblocks along the orthogonal joint faces. Althoughmany Tuscarora boulders and blocks may have

detached from the scarp as rock falls and collapsingoverhangs, this was not the primary mode of originof Rock City. Because adjacent blocks can generallybe fitted back together across the “alleys,” especiallythe bigger blocks near the scarp complex, most likelythe rock mass moved to the west as either a transla-tional slide or a lateral spread (Cruden and Varnes,1996). Kinematically, translational sliding seems un-likely for the following reasons: (1) bedding dipswithin and below the Tuscarora Sandstone are rela-tively gentle (7u); (2) beds dip into the hillside; (3) theTuscarora Sandstone is underlain by a relatively resis-tant sandstone of the Juniata Formation; (4) theblocks are separated in Rock City, negating the possi-bility of sliding as a single sheet; and (5) the blocks arenot concentrated in the toe area. Therefore, lateralspreading (Cruden and Varnes, 1996) may have beenthe primary mode of movement. Lateral spreads arecommon on gentle slopes and typically involve exten-sional movement similar to the formation of the“streets” and “alleys” in Rock City (Figure 12). Varia-tions in the orientations of alleys and streets documentlittle to considerable independent rotation of rockblocks. Soeters and Van Westen (1996) describe themorphology of lateral spreads as “irregular arrange-ment of large blocks tilting in various directions; blocksize decreases with distance and morphology becomeschaotic; large cracks and linear depressions separatingblocks….” This description is quite similar to the mor-phological features observed in Rock City.Most lateral spreads involve liquefaction or plastic

flow of a subjacent layer (Cruden and Varnes, 1996).

Figure 15. An east-facing photo of the lake and the southeastern edge of Rock City showing a block from the cluster of blocks at thesoutheastern edge of Rock City experiencing greater tilting toward the SE and SW caused by steeper slopes into Mountain Lake.



At Rock City, Juniata Sandstone underlies TuscaroraSandstone, and shale and limestone of the Reedsville-Trenton Formation underlie the Juniata Sandstone(Table 1). The Reedsville-Trenton Formation is atconsiderable depth and is unlikely to have served as abasal zone of plastic flow. Therefore, the Tuscarorasandstone blocks most probably moved over JuniataSandstone. Another possibility is that the blocksmoved over pre-existing colluvial debris covering theslope. The Mountain Lake area has significant

thicknesses of colluvial deposits containing materialsof all sizes (Mills, 1981, 1988, 1990). We have no sub-surface data below Rock City to document the natureof the failure zone. However, Varnes (1978) describedspreads of rock that experienced lateral extension,similar to Rock City, without a well-defined basalshear surface or a zone of plastic flow. Such lateralspreads usually occur near ridge crests (Varnes, 1978).It is not clear what triggered the slope movement

that resulted in the formation of Rock City.

Table 3. Direction and angle of lateral rotation for rock blocks compared to the outcrop scarp complex.

RockBlockNo.

Rock-Block Joint Sets and Angle of Rotation with Respect to ScarpOrthogonal Joints (u) Rock-Block Rotation Summary (u)

RockBlock

JS#1 (NEStrike)

ScarpOutcropJS#1(N57E)

Rotation ofRB

Relativeto JS#1*

RockBlockJS#2(NWStrike)

ScarpOutcropJS#3

(N35W)

Rotation ofRB

Relativeto JS#2*

RotationRelative toSmallerAngle*

AverageRotation(JS1 andJS2)*

Direction ofRotation

RB#1 N70E N57E 13 N20W N35W 15 13 14 ClockwiseRB#10 N77E N57E 20 N14W N35W 21 20 20.5 ClockwiseRB#18 N70E N57E 13 N3E N35W 38 13 25.5 ClockwiseRB#19 N73E N57E 16 N17W N35W 18 16 17 ClockwiseRB#26 N70E N57E 13 N20W N35W 15 13 14 ClockwiseRB#27 N70E N57E 13 N15W N35W 20 13 16.5 ClockwiseRB#28 N65E N57E 8 N25W N35W 10 8 9 ClockwiseRB#29A N86E N57E 29 N5W N35W 30 29 29.5 ClockwiseRB#29B N78E N57E 21 N15W N35W 20 20 20.5 ClockwiseRB#30A N85E N57E 38 N20W N35W 15 15 26.5 ClockwiseRB#33 N60E N57E 3 N30W N35W 5 3 4 ClockwiseRB#36 N70E N57E 13 N15W N35W 20 13 16.5 ClockwiseRB#37 N80E N57E 23 N30W N35W 5 5 14 ClockwiseRB#38 N70E N57E 13 N25W N35W 10 10 11.5 ClockwiseRB#39 N50E N57E −7 N30W N35W 5 5 −1 ClockwiseRB#2A N40E N57E −17 N50W N35W −15 −15 −16 CounterclockwiseRB#2B N40E N57E −17 N53W N35W −18 −17 −17.5 CounterclockwiseRB#4A N45E N57E −12 N52W N35W −17 −12 −14.5 CounterclockwiseRB#5A N48E N57E −9 N42W N35W −7 −7 −8 CounterclockwiseRB#6 N45E N57E −12 N45W N35W −10 −10 −11 CounterclockwiseRB#7 N30E N57E −27 N60W N35W −25 −25 −26 CounterclockwiseRB#9 N23E N57E −34 N67W N35W −32 −32 −33 CounterclockwiseRB#12A N25E N57E −32 N65W N35W −30 −30 −31 CounterclockwiseRB#14 N40E N57E −17 N50W N35W −15 −15 −16 CounterclockwiseRB#15 N10E N57E −47 N80W N35W −45 −45 −46 CounterclockwiseRB#17 N35E N57E −22 N80W N35W −45 −22 −33.5 CounterclockwiseRB#20 N30E N57E −27 N60W N35W −25 −25 −26 CounterclockwiseRB#21A N54E N57E −3 N36W N35W −1 −1 −2 CounterclockwiseRB#22 N46E N57E −11 N44W N35W −9 −9 −10 CounterclockwiseRB#24A N40E N57E −17 N50W N35W −15 −15 −16 CounterclockwiseRB#24B N40E N57E −17 N50W N35W −15 −15 −16 CounterclockwiseRB#25 N35E N57E −22 N55W N35W −20 −20 −21 CounterclockwiseRB#30B N5E N57E −52 N70W N35W −35 −35 −43.5 CounterclockwiseRB#34A N42E N57E −15 N48W N35W −13 −13 −14 CounterclockwiseRB#34B N48E N57E −9 N42W N35W −7 −7 −8 CounterclockwiseRB#40 N20E N57E −37 N70W N35W −35 −35 −36 CounterclockwiseRB#11 N55E N57E −2 N35W N35W 0 0 −1 No rotation shownRB#32 N60E N57E 3 N38W N35W −3 −3 0 Either clockwise or

counterclockwiseRB#41 N45E N57E −12 N35W N35W 0 0 −6 No rotation shown

*Positive values indicate a clockwise rotation, while negative values indicate a counterclockwise rotation.



Considering that weaker formations, such as theJuniata Sandstone, lie directly beneath the TuscaroraSandstone, that the Reedsville-Trenton Shale [Mar-tinsburg] lies at a significant depth below the Tuscar-ora Sandstone, and that the bedding along which atranslational movement could occur dips very gently,a seismic event probably triggered the movement.According to Andrus and Youd (1987), “spreads arethe most common ground failure during earthquakes.”Lateral spreads require considerably strong groundshaking (Keefer, 1984). Giles County, VA, has a

documented history of seismicity, including a massiveearthquake of 1897. This event was the second-largestrecorded earthquake in the southeastern United States(Bollinger and Wheeler, 1988) and is now the third lar-gest, following the August 23, 2011, earthquake inMineral, VA. The U.S. Geological Survey (USGS)estimates that the 1897 earthquake had a Richter mag-nitude of 5.7 and that the 2011 earthquake had a Rich-ter magnitude of 5.8 (USGS, 2011). Mountain Lake islocated approximately 17 km from Pearisburg, VA,the presumed epicenter of the 1897 earthquake.

Figure 16. Comparisons between the principal joint sets within the rock blocks in Main Street (black) and the principal joint sets in the headscarp (red). Plots generated using RockPack III software (Watts et al., 2012).



However, the Giles County seismic zone, as defined byBollinger and Wheeler (1988), has generated earth-quakes even closer to Mountain Lake in historictime. Based on data from 40 historical world-wideearthquakes, supplemented with intensity data fromseveral hundred U.S. earthquakes, Keefer (1984) esti-mated that a magnitude 5.5 earthquake can trigger lat-eral spreads at distances of 12–15 km. As the lakeclearly formed long before earthquake records werekept, it is impossible to judge what magnitudes anddistances to earthquakes may have existed in the Giles

County Seismic Zone, including Mountain Lake, atthe time. Previous studies (Parker et al., 1975; Cawley,1999; and Cawley et al., 2001) have suggested the pos-sibility of earthquakes playing a role in MountainLake’s formation. However, this role was within thecontext of lake-level fluctuations. Earthquakes mayhave adjusted boulders within the landslide dam caus-ing changes in the amount of seepage leaving the lake.Previous studies have suggested rock-block displace-

ment from freeze-thaw cycles (Eckroade, 1962; Mar-land, 1967; and Parker et al., 1975) and runoff from

Figure 17. Comparison between the principal joint sets of selected rock blocks from all portions of Rock City (black) and the principal jointsets in the head scarp (red). Plots generated using RockPack III software (Watts et al., 2012).



storm events (Mills, 1981). Other types of mass move-ment mechanisms may have been creep (Sharp, 1933),solifluction (Eckroade, 1962; Marland, 1967; and Par-ker et al., 1975), and vertical collapse due to undercut-ting (Rogers, 1884; Hutchinson and Pickford, 1932;and Parker et al., 1975). However, none of thesemechanisms explain the extensional features observedin Rock City. Mills (1981, 1988, 1989, 1990) favoredlateral displacement of colluvial deposits in the Moun-tain Lake area; however, he did not suggest lateralspreading.

LIMITATIONS AND FUTURE RESEARCH

The research presented herein has the followinglimitations:

1. No subsurface data are available to document thenature of material that may have served as thebasal zone of shear or plastic flow for a lateralspread to occur. Our hypothesis that Rock Citycomprises a lateral spread is based on the exten-sional features observed within the Rock Cityarea and based on comparisons of discontinuityorientations within rock blocks to discontinuityorientations within head scarp. Future researchshould look into the possibility of drilling withinthe Rock City area to confirm the nature of thematerial below the displaced mass.

2. Although most lateral spreads are caused by earth-quakes, and although Giles County has a docu-mented history of seismic activity, our hypothesisof a seismic event being the triggering mechanismfor lateral spreading, resulting in the formation ofRock City, requires additional documentation.Future research should apply landslide shakingmodels to document whether a seismic event cantrigger lateral spreading in the stratigraphic andstructural settings of the Rock City area and todetermine the magnitude of such an event.

CONCLUSIONS

The conclusions of this study can be summarized asfollows:

1. Rock City may be the remnant of a catastrophicancient slope movement. The large rectangularblocks of Tuscarora Sandstone and other colluvialdeposits comprising Rock City dammed the valleyof Pond Drain, forming Mountain Lake.

2. The primary mode of slope movement in RockCity is lateral extension from lateral spreading.The landslide material detached from a bedrockscarp along northwest-southeast–trending joints

as well as other pre-existing fractures and joints,with individual rock blocks displacing laterallytoward the west. During detachment, the rockblocks separated from each other by a few tomany meters, forming streets and alleys. Duringmovement, the blocks rotated laterally in bothclockwise and counterclockwise directions.

3. A seismic event may have triggered slope move-ment, resulting in the formation of Rock City.The geomorphic setting resulting from Pond Draincarving a valley through the Tuscarora Sandstoneridges appears to have provided a gap in which col-lapsing rock could settle.

ACKNOWLEDGMENTS

The authors would like to thank the three anon-ymous reviewers whose constructive comments greatlyhelped improve the quality of this article.

REFERENCES

ANDRUS, R. D. AND YOUD, T. L., 1987, Subsurface Investigation ofa Liquefaction Induced Lateral Spread, Thousand Springs Val-ley, Idaho: Miscellaneous Paper GL-87-8, Waterways Experi-ment Station, U.S. Army Corps of Engineers, Vicksburg, MS,106 p.

BOLLINGER, G. A. AND WHEELER, R. L., 1988, The Giles County,Virginia, Seismic Zone- Seismological Results and GeologicalInterpretations: U.S. Geological Survey Professional Paper1355, 85 p.

BUTTS, C., 1940, Geology of the Appalachian Valley in Virginia,Part 1: Virginia Geological Survey Bulletin, No. 52, 568 p.

CAWLEY, J. C., 1999, A Re-Evaluation of Mountain Lake, GilesCounty, Virginia: Lake Origins, History and EnvironmentalSystems: PhD Dissertation, Virginia Polytechnic Instituteand State University, 118 p.

CAWLEY, J. C.; PARKER, B. C.; AND PERREN, L. J., 2001, New obser-vations on the geomorphology and origins of Mountain Lake,Virginia: Earth Surface Processes Landforms, Vol. 26, pp.429–440.

CRUDEN, D. M. AND VARNES, D. J., 1996, Landslide types and pro-cesses. In Turner, A. K. and Schuster, R. L. (Editors), Land-slides—Investigation and Mitigation: Special Report 247,Transportation Research Board, National Research Council:National Academy Press, Washington, DC, pp. 37–75.

EASTERBROOK, D. J., 1999, Surface Processes and Landforms, 2nded.: Prentice-Hall, Upper Saddle River, NJ. 546 p.

ECKROADE, W. M., 1962, Geology of the Butt Mountain Area, GilesCounty, Virginia: Master’s Thesis, Virginia Polytechnic Insti-tute, 64 p.

FERGUSON, F. F.; STIREWALT, M. A.; BROWN, T. D.; AND HAYES, W.J., JR., 1939, Studies on the turbellarian fauna of the Moun-tain Lake Biological Station. I. Ecology and Distribution:Journal Elisha Mitchell Scientific Society, Vol. 55, pp.274–288.

FREEMAN, J.; CROOK, E. C.; AND WATTS, C. F., 2012, Seismic refrac-tion survey of landslide colluvium and lacustrine sedimentsat Mountain Lake, Virginia (abstract): Geological SocietyAmerica Abstracts Programs, Vol. 44, No. 7, p. 390.



HOLDEN, R. J., 1938, Geology of Mountain Lake, Virginia: Virgi-nia Journal Science, 73 p.

HUTCHINSON, G. E., 1957, A Treatise on Limnology, Vol. 1, Geogra-phy, Physics and Chemistry: John Wiley and Sons, New York.1015 p.

HUTCHINSON, G. E. AND PICKFORD, G. E., 1932, Limnologicalobservations of Mountain Lake, Virginia: International Rev-enue GesamtenHydrobiologie Hydrographie, Vol. 27, pp.252–264.

JANSONS, M.; PARKER, B. C.; AND JACOB, E. W., 2004, Subterraneanloss and gain of water in Mountain Lake, Virginia: A hydro-logic model: Virginia Journal Science, Vol. 55, No. 3, pp.107–113.

JOYCE, W. L., 2012, Examining Pathways for Water Loss fromMountain Lake, Giles County, V.A.: Master’s Thesis, VirginiaPolytechnic Institute, 44 p.

KEEFER, D. K., 1984, Landslides caused by earthquakes: GeologicalSociety America Bulletin, Vol. 95, No. 4, pp. 406–431.

MARLAND, F. C., 1967, The History of Mountain Lake, GilesCounty, Virginia: An Interpretation Based on Paleolimnology:Ph.D. Dissertation, Virginia Polytechnic Institute and StateUniversity, 129 p.

MILLS, H. H., 1981, Boulder deposits and the retreat of mountainslopes, or, “Gully Gravure” Revisited: Journal Geology, Vol.89, pp. 649–660.

MILLS, H. H., 1988, Surficial Geology and Geomorphology of theMountain Lake Area, Giles County, Virginia, Including Sedi-mentological Studies of Colluvium and Boulder Streams: U.S.Geological Survey Professional Paper 1469, 57 p.

MILLS, H. H., 1989, Hollow form as a function of boulder size inthe Valley and Ridge province, southwestern Virginia: Geol-ogy, Vol. 17, pp. 595–598.

MILLS, H. H., 1990, Thickness and character of regolith on moun-tain slopes in the vicinity of Mountain Lake, Virginia, as indi-cated by seismic refraction, and implications for hillslopeevolution: Geomorphology, Vol. 3, pp. 143–157.

PARKER, B. C.; WOLFE, H. E.; AND HOWARD, R. V., 1975, On theorigin and history of Mountain Lake, Virginia: SoutheasternGeology, Vol. 16, No. 4, pp. 213–226.

ROCSCIENCE, 2012, Graphical and Statistical Analysis of OrientationData: Electronic document, available at http://www.rocscience.com/products/1/Dips

ROGERS, W. B., 1884, Report on geological reconnaissance of thestate of Virginia, made under the appointment of the Boardof Public Works, 1835. Reprinted in Rogers, W. B., A Reprintof Annual Reports and Other Papers on the Geology of the Vir-ginias: D. Appleton Company, New York, pp. 109–110.

RONINGEN, J. M., 2011, Hydrogeologic Controls on Lake Level atMountain Lake, Virginia: Master’s Thesis, Virginia Polytech-nic Institute, 92 p.

SCHULTZ, A. P.; STANLEY, C. B.; GATHRIGHT, T. M., II; RADER, E.K.; BARTHOLOMEW, M. J.; LEWIS, S. E.; AND EVANS, N. H.,1986, Geologic Map of Giles County, Virginia: Virginia Divi-sion of Mineral Resources, Publication 69.

SHARP, H. S., 1933, The origin of Mountain Lake, Virginia: JournalGeology, Vol. 41, pp. 636–641.

SOETERS, R. AND VAN WESTEN, C. J., 1996, Slope instability recog-nition, analysis, and zonation. In Turner, A. K. and Schuster,R. L. (Editors), Landslides—Investigation and Mitigation: Spe-cial Report 247, Transportation Research Board, NationalResearch Council, National Academy Press, Washington,DC, pp. 129–177.

U.S. Geological Survey (USGS), 2011, Electronic document, avail-able at http://earthquake.usgs.gov/earthquakes/events/2011virginia/overview.php]

VARNES, D. J., 1978, Slope movement—Types and processes. InSchuster, R. L. and Krizek, R. J. (Editors), Landslides—Analysis and Control: Special Report 176, TransportationResearch Board, National Research Council, Washington,DC, pp. 12–33.

WATTS, C. F., 2013, Mountain Lake local geology and bathymetryoverlays: KML files, Radford University https://sites.google.com/a/vt.edu/radford-geology—mountain-lake-files/google-earth-files

WATTS, C. F.; GILLIAM, D.; HROVATIC, M.; AND HONG, H., 2012,ROCKPACK III for Windows: Rock Slope Stability Computer-ized Analysis Package, Part One—Stereonet Analysis: Rock-Ware, Inc., Earth Science and GIS software, Golden, CO 80401.

WILLIAMS, P., 2003, Dolines. In Gunn, J. (Editor), Encyclopedia ofCaves and Karst Science: Fitzroy Dearborn, London, U.K.,pp. 304–310.

WYLLIE, D. C. AND MAH, C. W., 2004, Rock Slope Engineering:Civil and Mining, 4th ed.: Spon Press/Taylor and FrancisGroup, London, U.K. 432 p.



Discovering and Characterizing Abandoned WasteDisposal Sites Using LIDAR and Aerial Photography

ANDREW DE WET1

Department of Earth & Environment, Franklin and Marshall College, Lancaster, PA 17603

Key Terms: LIDAR, Historical Aerial Photographs,Remote Sensing, Waste Disposal Sites, Landfills, GIS,Site Investigations

ABSTRACT

Currently there are 1,908 active landfills across theUnited States; however, in the 1970s, prior to themodern era of sanitary landfills, there may have beenas many as 100,000 landfills and dumps. Most of theselandfills and dumps were unregulated and were aban‐doned, and details about their locations and char‐acteristics are poorly documented. The AbandonedLandfill Inventory produced by the PennsylvaniaDepartment of Environmental Protection lists 2,620waste facilities in the state, of which 1,309 aredescribed as “landfills” or “abandoned landfills.” Thereare 157 reported facilities in Lancaster County, PA.Seventeen are “landfills” or “abandoned landfills”;however, it is clear that most of the landfills anddumps that existed in Lancaster are not listed. A 1971report documented approximately 23 land disposalsites, including eight landfills (one of which was alicensed sanitary landfill), 15 dumps, approximately 44identified informal dumps, and perhaps 200 to 300additional unidentified informal dumps. This study usesLIDAR and historical aerial photography integratedinto a GIS database to investigate three dump sites inLancaster County, PA. The sites had varying degreesof available information about their location andextent. The techniques discussed here can be used toinvestigate other ‘known’ and possible abandoned sitesin order to significantly increase the robustness of theavailable data about these sites, leading to bettermonitoring and even remediation in extreme cases. Asland-use pressures increase with an expanding popu‐lation abandoned landfill sites need to be avoided orused in appropriate ways.

INTRODUCTION

Abandoned Waste Disposal Sites

According to the most recent (2012) data there are1,908 active Municipal Solid Waste landfills in theUnited States (EPA, 2014). The total number of land-fills has been declining for decades; however, the totalcapacity has increased by 65 percent since 1980 and iscurrently around 251 million tons per year (Center forSustainable Systems, 2014; EPA, 2014). Landfills havethe potential to seriously affect environmental andhuman health (Herndon et al., 1990; Suflita et al., 1992),but landfill technology has improved significantly inrecent decades, and most landfills in the United Statesare now well documented and monitored by variousstate and federal agencies (EPA, 2014). Potentiallymore troubling is that prior to the modern era of regu-lated sanitary landfills, municipal and other waste wasdisposed of in thousands of open dumps, haphazard(also referred to as promiscuous) dumps, and unregu-lated landfills across the United States. Precise num-bers are unavailable, but estimates suggest that thereare up to 100,000 dumps and landfills in these cate-gories (Suflita et al., 1992). This number does notinclude the countless small farm and household dumpsthat have existed since colonial times.This study’s focus is on using remote sensing techni-

ques to discover and characterize abandoned wastesites by examining three sites in Lancaster County,PA. Lancaster County has a long history of humanoccupation. Prior to colonial times the area was occu-pied by Native Americans, primarily from the Susque-hannock Group (Kent, 1984; Wallace, 1989). In theearly 1700s the area was settled and largely deforestedby Europeans. Lancaster County was founded in 1729,and the City was founded in 1730. The populationgrew rapidly, particularly from the 1950s on, and iscurrently over 530,000 (US Census Bureau). Thepopulation density is 550 per square mile, double thePennsylvania average (US Census Data for 2010).The county is also an important agricultural area,with nearly 6,000 farms and a growing Amish commu-nity (Lancaster Farmland Trust, 2015). While thecounty is only one of 67 counties in Pennsylvania,according to a Lancaster County Planning Commission1Corresponding author. [email protected]


report in 1971 (Lancaster County Board of Commis-sioners, 1971) there were approximately 23 waste dispo-sal sites, including eight landfills (including one licensedsanitary landfill), 15 dumps, approximately 44 identifiedhaphazard dumps and perhaps 200 to 300 additionalunidentified haphazard dumps in the county at thetime. In most cases the report does not give precise loca-tions for these waste disposal sites. More recently thePennsylvania Department of Environmental Protection(PADEP) has produced an inventory of abandonedlandfills in Pennsylvania, part of the Abandoned Land-fill Inventory (ALI) Project, which collects geospatialand descriptive data for closed and abandoned landfillsthroughout the state of Pennsylvania (www.psda.psu.edu). ALI site locations were determined using historicrecords such as microfiche, index cards, topographicmaps, and staff personal files. These data were thencompiled into site lists and incorporated into a GISdatabase (“Municipal Waste Operations”) availablefrom the Pennsylvania Spatial Data Access (PASDA)on the Pennsylvania Geospatial Data ClearinghouseWeb site (www.pasda.psu.edu). Of the 2,620 facilitieslisted, 1,309 are described as “landfills” or “abandonedlandfills.” There are 157 reported locations in LancasterCounty. Seventeen are “landfills” or “abandoned land-fills”; however, none of the landfills examined in thisstudy are in the ALI database. But many of these formerwaste disposal sites are known to local LancasterCounty officials and residents or are recorded in variousunpublished documents and reports held by local gov-ernment agencies. There is concern that as time passesmuch of this local knowledge will be lost. This alsoraises the question of how many other former dumps/landfills remain to be re-discovered in the future.In 1965 the US Congress passed the Solid Waste

Disposal Act, and in the early 1970s many waste dis-posal sites were closed because of increased publicawareness of the environmental problems associatedwith unregulated waste disposal and because morestrenuous legislative and oversight requirements werebeing implemented (Lancaster County Board of Com-missioners, 1987). Lancaster County is typical of coun-ties across the United States and elsewhere.Unidentified former waste disposal sites are a nationaland international issue that warrants attention forhuman and environmental health and safety reasons.In addition, increased land use pressure in urbanand suburban areas may lead to re-use of these loca-tions without an understanding of their past history,with potential negative health and environmentalconsequences.

Study Locations

This study demonstrates how a combination ofLIDAR and historical aerial photographs provides apowerful method for identifying abandoned waste dis-posal sites and determining detailed information abouteach site’s history and specific characteristics. Threesites in Lancaster County were investigated usingthis combined technique; two sites are fairly well-documented former landfills; the other site was ‘dis-covered’ and characterized in this study (Figure 1).All three locations were field checked. Using this tech-nique numerous other sites in Lancaster County havebeen identified as potential waste disposal sites thatshould be investigated further. The waste disposallocations discussed here were active between the1940s and the late 1960s. Local government reports,unpublished documents, and “anecdotal knowledge”provided varying amounts of information about theirexistence and approximate location.The first site—Baker Woodlands/Spalding Conser-

vancy—has previously been studied using a variety oftechniques including near-surface geophysics (de Wetet al., 1998, 1999). Approximately 100 acres of thisarea was recently designated as a nature conservancyby Franklin & Marshall College (Spalding Conser-vancy). This site includes several landfills that wereactive during the 1950s and 1960s (de Wet et al., 1998,1999) and received Lancaster City municipal waste,debris from an urban renewal project in LancasterCity, and waste from several local manufacturing com-panies. Because the location of former waste depositson the site are already fairly well defined it is an excel-lent place to test the value of using newly availableLIDAR data combined with historical aerial photogra-phy to discover and characterize abandoned landfills.Between 1962 and 1968, after the Baker Woodlands

landfills were closed, Lancaster City municipal anddemolition waste went to another site, now located inthe County Park (Lancaster County Commissioners,1973). This location is partially documented in localgovernment reports but was delineated in detail hereusing LIDAR data and historical aerial imagery andis the second location investigated in this study(referred to as the County Park site).The third site, the South Duke Street landfills, is

substantially older; it was active during the 1940s andis located very close to Lancaster City within thefloodplain of the Conestoga River. Little is knownabout this landfill, but field observations suggest thatit also received primarily municipal waste.After the 1960s, Lancaster City waste was disposed

of in the Manor Township landfill, located close to thecurrently active Frey Farm sanitary landfill (Figure 1).Presently, most Lancaster County household waste

de Wet


is incinerated, and the ash and other municipal wasteare buried in the Frey Farm sanitary landfill.

Topographic Information Using LIDAR

Since landfills typically modify the land surface inways that are distinguishable from natural processes,

high-resolution spatial information can be used toidentify and characterize specific features. In somecases, landfills simply change the topography throughthe addition of material on top of the original naturalsurface. The resulting topographic features are usuallydistinct from naturally produced landforms. In othercases, landfills occupy areas that were previously

Figure 1. Map showing the three sites studied here and the location of the municipal waste operations in Lancaster County included in theAbandoned Landfill Inventory (ALI) produced by the Pennsylvania Department of Environmental Protection. There are 157 reportedfacilities in Lancaster County, PA, of which 17 are “landfills” or “abandoned landfills.” Study site 1 is the Baker Woodlands/SpaldingConservancy site, study site 2 is the Lancaster County Park site, study site 3 is the South Duke Street site, and study site 4 is Frey Farmlandfill, the current sanitary landfill used for most of the disposal of most of the municipal waste generated in Lancaster County (principallyin the form of ash produced by the Resource Recovery Facility located along the Susquehanna River in the northwest part of the county).



disrupted by human activities such as quarrying or sur-face mining, and in these situations landfill detectioncan be challenging if the natural grade of the landscapewas reestablished when the landfill closed. Most land-fills and dumps used prior to 1970, however, were merelyabandoned with little or no effort to return the landscapeto its natural state. Even landfills that were ‘graded’prior to closure likely leave a telltale topographic signa-ture that should be evident in detailed topographic datasets. While digital topographic data such as digital eleva-tion models (DEMs) derived from Shuttle Radar Topo-graphy Mission and Advanced Spaceborne ThermalEmission and Reflection Mission have been availablefor years and cover most of the United States, thesedata sets are not of sufficiently high resolution to behelpful in identifying the relatively small topographicchanges associated with landfills. Recently, however,high-resolution LIDAR data with typical horizontalresolutions of ,1 m and vertical resolutions of ,0.5 mare becoming available for large parts of the UnitedStates. According to an inventory of known high-resolu-tion digital elevation data sources conducted as part ofthe National Enhanced Elevation Assessment (NEEA)in the summer of 2011, LIDAR data have been collectedover 28 percent of the conterminous United Statesand Hawaii (http://nationalmap.gov/3DEP/neea.html).In the United States, elevation data are availablethrough The National Map (nationalmap.gov), andwhile lower resolution data are available across theUnited States, high-resolution (1-m) bare-earth DEMdata will be populated as new data is acquired in2015 and beyond (nationalmap.gov/elevation.html). Inmany cases individual states have partial or completehigh-resolution (,1-m) LIDAR coverage that is publi-cally available. For example, in Pennsylvania, data areavailable from the Pennsylvania Spatial Data Access(PASDA) Web site (www.pasda.psu.edu) and in Mary-land from the GeoSpatial Data Center (http://dnrweb.dnr.state.md.us/gis/data/index.asp). These high-resolution elevation data offer a way to observeand characterize landscape changes, and despitesome limitations, these data provide detailed topo-graphic information even in vegetated areas. Thisis particularly important since many old landfillswere left to re-vegetate and may now be forested orbrush covered.

Historical Aerial Photography

Numerous human activities modify topography, somore than just LIDAR information is needed to re-discover and characterize former waste disposal sites.Current and historical aerial photography providesthis necessary additional information. Current aerialphotography includes very high-resolution normal

color and color infrared (CIR) imagery that providesdetailed information about recent changes in vegeta-tion and the land surface, which in combinationwith topographic data is useful for detecting formerlandfills. Older aerial photography is usually blackand white and varies in spatial resolution. In manyparts of the United States, aerial photography datesback into the 1930s and occasionally even into the1920s. Since the 1940s virtually every part of theU.S. land surface has been photographed, withupdates every several years. Historical aerial photo-graphy to detect and assess former landfills andhazardous waste sites is an established technique(Erb et al., 1981; Lyon 1987; Pope et al., 1996;de Wet et al., 1998, 1999; Biotto et al., 2009; andSilvestri and Omri, 2009), but there are limitationsto it, especially because a wide variety of other humanactivities can disrupt the earth’s surface and be con-fused with landfills. Since landfills operate over thetimescale of years and produce significant surface dis-ruption and modification, combining aerial photogra-phy with co-registered high-resolution LIDAR dataand its derivatives yields enhanced detection andcharacterization of landfills.

METHODS

LIDAR and Aerial Photography

LIDAR data for the three sites investigated herewere obtained from two different sources—the U.S.Geological Survey (USGS) and the PennsylvaniaDepartment of Conservation and Natural Resources(PA DCNR). LIDAR data used for the analysis ofthe Baker Woodland/Spalding Conservancy site wereobtained from the USGS. The data were acquired in2004 by AERO-METRIC, Inc., using an airborneOPTECH ALTM 30/70 sensor. USGS post-processedthe data with Realm software and used TerraScansoftware. AERO-METRIC provided 15 ground truthpoints for quality assurance and quality control checksagainst global positioning system (GPS) Rapid-Staticsurvey methods. LIDAR returns were filtered andclassified to produce a last-return bare-earth data set.The bare-earth returns have a vertical root meansquare error of better than 15 cm relative to NorthAmerican Vertical Datum of 1988 and a nominal hor-izontal spacing of better than 2.0 m based on Univer-sal Transverse Mercator (UTM) coordinate systemrelated to the North American Datum of 1983. Inthis study, the LIDAR bare-earth returns wereimported into an ArcGIS Terrain model and triangu-lated. The triangulated data were rasterized to aground resolution of 1 m for further processing andvisualization. The LIDAR data for sites 2 and 3 were

de Wet


downloaded from the PASDA Web site (www.pasda.psu.edu) and were produced by the PAMAP Program,PA DCNR, Bureau of Topographic and GeologicSurvey. The data were collected in 2006 and are avail-able in a processed bare-earth surface DEM formatusing the Pennsylvania State Plane coordinate systemand have a horizontal ground resolution of ,1 mand a typical vertical resolution of less than 1 m.

The LIDAR data were processed and displayedusing different rendering techniques, including basicgray-scale or color-coded schemes. These imagesform, in essence, a highly detailed and accurate topo-graphic map of the earth’s surface. This type of map

was combined with additional spatial data for inter-pretation and visualization purposes.For this study, hillshading, slope, and aspect pro‐

cessing techniques were employed to highlight topogra-phy. Hillshading, or illuminating a topographic surfacefrom one direction, emphasizes features oriented at rightangles to the illumination direction and de-emphasizesfeatures parallel to the illumination direction. An illumi-nation angle of 45u above horizontal was used in thisstudy. To compensate for any loss of features due toone-direction illumination, composite multiple hillshad-ing was performed using Spatial Analyst using Mul‐tiple Direction Oblique Weighted (MDOW) analysis

Figure 2. WDOW LIDAR map and topographic profiles of study site 1 (Baker Woodlands/Spalding Conservancy) and surrounding area.The properties owned by Franklin & Marshall College are shown. The western property has been designated by the college as a conservancy(Spalding Conservancy). It is presently mostly wooded and includes much of the area affected by the brick-making activities (brick kilns, claypits, etc.) and later landfills. The eastern property comprises mainly sports fields but includes some woods that were also affected by claymining and landfills.



(Mark, 1992) (Figure 2). Using ArcMap, a hillshade grid(shade_flt or shade_int) was placed above the bare-earthgrid in the layers tree. The hillshade layer was thendisplayed using a gray-scale or monochromatic gradi‐ent color scheme, made 50 percent transparent, andstretched to ‘minimum-maximum.’ The bare-earth ele-vation data were displayed using a different color, show-ing where land use/land cover relates to elevation. Thebare-earth elevation was also stretched using ‘mini-mum-maximum.’ The raster shade_flt is kept as a float-ing point raster to allow for certain operations, such ashistogram stretching or lightening in ArcGIS or Photo-shop, which required additional gray-scale bit-depth.A significant advantage of this technique is that it inte-grates complex information into one layer.

Slope and aspect were derived using standard tech-niques in ArcGIS, and the resulting data were com-bined with other spatial data, such as aerialphotographs. The LIDAR data set forms the three-dimensional (3-D) visualization base for the studysites. Detailed 3-D characterization of each site wasobtained by combining the LIDAR-derived DEMwith historical aerial photography and other spatialdata sets.In this study, historical aerial photographs were

downloaded from the Penn Pilot Web site or scannedfrom photographs archived in the Departmentof Earth & Environment at Franklin & MarshallCollege (Table 1). The scanned historical photographstypically had a pixel resolution of ,2 m and were

Table 1. Details of the aerial photographs used for each study site.

Site Number Date Source Scale Digital Resolution (ft) Coord.

1, Spalding Conservancy ahg102193 4/29/1940 Penn Pilot1 1:20,000 0.686 (2.25) Noneahg102194 4/29/1940 Penn Pilot1 0.686 (2.25)AHG-3D-22 6/17/1947 F&M College 0.896 (2.94)GS-OY 1 59 4/21/1951 F&M College Noneahg_6r_18 11/11/1957 Penn Pilot2 1:20,000 0.677 (2.22)ahg_7r_44 6/7/1958 F&M College 0.643 (2.11)ahg_7r_45 06/071958 F&M College 0.643 (2.11)ahg_7r_46 06/071958 F&M College 0.643 (2.11)AHG-6EE-280 5/23/1964 F&M College 0.24 (0.8)ahg_3mm_19 7/5/1971 Penn Pilot3 1:20,000 0.67 (2.19) None173-189 10/27/1974 F&M College42071-178-57 10/29/1978 F&M College4

27002360PAS 2007 PASDA5 — 0.3 (0.98) PA st pl7

2, County Park AHG-3D-22 6/17/1947 F&M College 0.896 (2.94) Noneahg_2r_64 9/27/1957 Penn Pilot2 1:20,000 0.677 (2.22) NoneAHG-6EE-242 5/23/1964 F&M College 1:20,000 0.25 (0.8) Noneahg_3mm_103 7/5/1971 Penn Pilot3 0.67 (2.19) None173-191 10/27/1974 F&M College None42071-178-57 10/29/1978 F&M College4 None26002370PAS 2007 PASDA5 — 0.3 (0.98) PA st pl7

18TUK885295 2012 PASDA6 — 0.3 (0.98) PA st pl7

18TUK885310 2012 PASDA6 — 0.3 (0.98)18TUK900295 2012 PASDA6 — 0.3 (0.98)18TUK900310 2012 PASDA6 — 0.3 (0.98)

3, South Duke Street ahg102101 4/29/1940 Penn Pilot1 1:20,000 0.69 (2.25) Noneahg102102 4/29/1940 Penn Pilot1 1:20,000 0.69 (2.25)ahg102103 4/29/1940 Penn Pilot1 1:20,000 0.69 (2.25)ahg102139 4/29/1940 Penn Pilot1 1:20,000 0.69 (2.25)ahg102140 4/29/1940 Penn Pilot1 1:20,000 0.69 (2.25)AHG-3D-22 4/17/1947 F&M College 0.90 (2.94) NoneAHG-3D-23 4/17/1947 F&M College 0.90 (2.94)GS-OY 1 59 4/21/1951 F&M College Noneahg_2r_64 9/27/1957 Penn Pilot2 1:20,000 0.68 (2.22) None26002370PAS 2007 PASDA5 — 0.3 (0.98) PA st pl7

18TUK900310 2012 PASDA6 — 0.3 (0.98) PA st pl7

1Produced by: U.S. Department of Agriculture (USDA) Agricultural Adjustment Administration Northeast Division.2USDA Commodity Stabilization Service.3USDA Agricultural Stabilization and Conservation Service.4USDA.5PAMAP Program, PA Department of Conservation and Natural Resources, Bureau of Topographic and Geologic Survey.6Lancaster County.7PA st pl: Pennsylvania State Plane.

de Wet


georeferenced using standard methods in ArcGIS.Recent aerial photographs (PAMAP_cycle2; and Lan-caster County, PA RGB and CIR Orthoimages) wereobtained from the PASDA Web site and integratedinto the ArcGIS database. The PAMAP_cycle2orthorectified digital raster images were taken in2007 and have a horizontal ground resolution of,0.5 m (1 ft) and use the Pennsylvania State PlaneSouth coordinate system (Table 1). The LancasterCounty, PA, True Color (RGB) and Color InfraRed(CIR) Orthoimages were taken in 2012 and have aresolution of 0.076 m and are available in eitherUTM or PA State Plane South coordinates (Table 1).

Database and Abandoned Landfill Determination

Combining LIDAR and historical aerial photo-graphs allows for planiform analysis of former land-fills, leading to detailed spatial and temporalinformation about their location and extent. Thisforms the backbone of the GIS database used in thisstudy, but the data set also includes roads, topographiccontours, geology, soils, land parcels, and hydrology(all available from the PASDA Web site). Spatialdata were then analyzed either individually or in com-bination with other data sets in the GIS. For example,initial inspection of the hillshaded LIDAR DEM inthe target areas revealed many raised, relatively flatareas surrounded by slopes or peculiar hummocky ter-rain strongly suggestive of an anthropogenically mod-ified surface. Viewing the slope derived from theLIDAR DEM further emphasized these areas. Inother cases, hummocky terrain suggestive of dumpingwas observed. These data were then overlaid on recentaerial photographs to exclude areas that have residen-tial or commercial development in order to focus onpotential landfill sites. It was assumed that it was un‐likely that new housing or commercial developmentswere established on abandoned landfills.

Georeferenced historical aerial photographs ofpotential landfill sites were studied. Aerial photo-graphs taken before, during, and after the landfillswere active were used to map and determine the timingand extent of each landfill. Where available, stereopairs of historical aerial photographs were examinedmanually and were used in conjunction with the digitaldata for estimating landfill boundaries and to providesemi-quantitative topographic information. Aerialextent and cross-sectional profiles were measuredusing standard techniques in ArcGIS. Landfill depthor thickness was estimated either by extrapolating thepre-landfill surface from the surrounding area, in thecase of landfills in which material was just added tothe surface, or depth estimates from stereo aerialphotos, where material was dumped into excavated

pits. The volumes of the landfills were then calculatedbased on the aerial extent and the estimated averagedepths. Bedrock geology, soil type, and drainagepatterns were examined to determine potential envi‐ronmental impacts of landfill leachate. Historicaldocuments such as newspapers and magazine articlesprovide rich details about a site’s past, although theygenerally only refer to major events or projects andso were used to supplement other data sources.Remote-sensing can potentially provide detailed

information about waste disposal sites, but on-sitefield observations, mapping, near-surface geophysics,and sampling, where possible, will always be neededto compliment remotely derived information.

CASE STUDY 1: BAKER WOODLANDS/SPALDING CONSERVANCY

Background

The Baker Woodlands site was chosen because ithas been well studied, including analysis of aerialphotos, field observations, mapping, and near-surfacegeophysics (Figures 1 and 2) (de Wet et al., 1998,1999; De Wet and Sternberg, 1999). Prior studieshave focused on the history of commercial brick mak-ing (Horning, 1992), farming and other land-use his-tory (de Wet and Sternberg, 1999; de Wet et al.,1999, 2000; and Carlson et al., 2000), and the site’secological recovery (de Wet et al., 1998). These studiesrevealed a complex land-use history dating back toEuropean colonization of the area in the 1700s.A railway defines the northern boundary of the site,

and it is part of the original Philadelphia and Colum-bia Railroad that was completed in 1834 (Wilson,1985). Substantial land impacts occurred after the con-struction of a brick factory in 1920 (Horning, 1992). Alarge portion of the site was graded for the construc-tion of the brick-making facilities, and between 1920and the 1940s clay was excavated, creating pits cover-ing 10 acres that were probably up to 10 m (30 ft) deep(de Wet et al., 1999). Many of these open pits weresubsequently used as landfills in the 1950s and 1960s.After the brick factory closed in the late 1970s,the buildings were demolished, with the debris leftlargely in place. Most of the site has been allowed tore-vegetate, thus obscuring the surface topographyand restricting near-surface geophysical surveys tonarrow transects cut in swathes through the woodsand scrub. The vegetation now consists of native andnumerous non-native invasive species such as Norwaymaple (Acer platanoides), Ailanthus “tree-of-heaven”(Ailanthus altissima), English ivy (Hedera helix), andMultifora rose (Rosa multiflora) (de Wet et al., 1998).



A large part of the original Baker Woodlands wasrecently designated a conservation area for recreationand long-term ecological and environmental research(Spalding Conservancy). Detailed understanding ofthe site’s land-use history is important in shapingthe conservation plan. As a result of the conservancydesignation, clear-cutting transects for geophysicalsurveys are less likely to be permitted, meaning thatnon-destructive techniques such as LIDAR willbecome even more important in the future. Conserva-tion land restoration plans include actively managingwoodland areas, enhancing and expanding an existingwetland, restoring and preserving historical structureson site, construction of a trail system, and adding sev-eral deer exclosures.

Results

The LIDAR maps show elevation variationsacross the site and adjacent areas (Figure 2). Theeffect of extensive and complex alterations to thetopography related to decades, and in some cases,centuries, of land-use change is clearly evident. Thecentral and south portions are generally suburbandevelopment and associated infrastructure. Mostobvious is the topographic patchwork of modifica-tions for houses and streets. The northern area isdominated by major roads, a railway line, sportsfields, and commercial and industrial buildings andparking lots. The flat sports fields and the flat inter-polated area of commercial and industrial buildingsites are clearly evident. Located between these areasis the Spalding Conservancy, with its complex andhighly variable topographic signature consisting ofsmooth regions interspersed with hummocky terrainand areas of parallel ridges and depressions on thescale of 1 to 10 m.Figure 2 shows a bare-ground MDOW layer (50

percent transparent) overlain on the LIDAR topo-graphic layer with railways, streams, property bound-aries of the Baker Woodlands/Spalding Conservancy,and the landfill locations. The combination ofMDOW bare ground on LIDAR topography revealedsubtle variations in topography. Overlaying theMDOW layer (50 percent transparent) on the topo-graphic layer showed multiple shadows and revealednew details, especially in the low-lying areas adjacentto the Little Conestoga Creek and within the residen-tial area south of the Spalding Conservancy.LIDAR data were used to derive a slope grid, in

which the steepness of the slope is represented byincreasing shades of gray. Slope maps were combinedwith color-coded topographic results to emphasizeslope variations. Landfills in particular are often char-acterized by broad flat areas bounded by steep slopes;

thus, slope maps are particularly useful in defining theedges and size of former landfills, as seen in Figure 2.High-resolution cross sections extracted from theLIDAR data reveal topographic variations across thesite, illustrating its complex past (Figures 2 and 3).Historical aerial photographs from 1936 show that

farming had given way to the brick works, and photo-graphs from the 1940s show numerous buildings andexcavations on the site. Photographs from the 1950sand 1960s show major site disruptions when largeareas were de-vegetated or covered with debris asso-ciated with landfill activity.

The Landfills

Historical documents (Horning, 1992), field obser-vations (de Wet et al., 1999), and aerial photographsindicate that the brick work’s clay pits were convertedto landfills during the late 1950s and early 1960s (deWet et al., 1998). The landfill areas are apparent inthe LIDAR maps and photographs (Figures 2 and3). Two landfills are located within the present Spald-ing Conservancy; one landfill is located north of theConservancy (north of the railway line), and thereare several landfills northwest of the Conservancy(not shown on these maps) (Figure 3). All of the for-mer landfills appear as positive topographic features,and they have a distinctive LIDAR texture becausedebris was not graded or smoothed after beingdumped into the former clay pits. This producedsmall-scale, hummocky surfaces evident in theLIDAR results. The contrast between the disruptedlandfill surfaces and the relatively undisturbed farmfields is particularly clear along the boundary of thelandfill north of the railway and outside of the Spald-ing Conservancy (Figure 3). This landfill boundary isparticularly well known because it was excavated in2012 to stabilize the ground for the construction ofa new rail yard (Figure 4). Slight topographic varia-tion in each landfill coincides with different episodesof landfill activity, as evident in aerial photographs(Figure 3) and near-surface geophysical data (deWet et al., 1999).Prior to its excavation, the northern landfill covered

an area of 33,400 m2 (8.25 acres) and had an averagethickness of 4 m, resulting in a volume of 133,600 m3

(174,742 yd3) (Figure 3 and Table 2). The two landfillslocated in the Spalding Conservancy covered areas of10,904 m2 (2.69 acres) and 47,378 m2 (11.7 acres),and assuming an average thickness of 4 m, these land-fills have volumes of approximately 40,000 m3 (52,318yd3) and 190,000 m3 (248,510 yd3), respectively. Lea-chate-contaminated seeps occur on the downslopesides of both landfills and discharge into the adjacentwetlands before flowing into the Little Conestoga

de Wet


Creek (Figure 3). The wetlands appear to act as a sinkfor much of the metal content of the leachate (Wilsonet al., 2006).

Visualization of the landfills is enhanced by usingthe LIDAR results as an elevation base and by

rendering the historical aerial photographs in 3-D ontop of it. Figure 5 shows the 1964 historical aerialphotograph rendered in 3-D using the LIDAR dataas the elevation base. Vertical exaggeration is 106.Multiple layers were combined and rendered in 3-D

Figure 3. (a) Location map; (b) WDOW shaded relief LIDAR map showing parts of the site affected by the landfills and the location of theclay pits, lime kiln, and associated limestone quarry; (c) 1964 aerial photograph showing the main landfill area, location of topographic profileA-B, and the location of currently active seeps; (d) topographic profile A-B showing the location of the landfill and the seeps downslope of thelandfill.

Table 2. Details of the landfills/dumps at each study site.

Site Dates Active Approximate Dimensions (m) Area (m2) Area (acres) Volume (m3)

1, Spalding Conservancy (north) Late 1950s–1962 365 6 90 6 4 33,400 8.25 133,6001, Spalding Conservancy (west) Late 1950s–1962 150 6 80 6 4? 10,904 2.69 43,6161, Spalding Conservancy South (east) Late 1950s–1962 135 6 360 6 4 47,378 11.71 189,5122, County Park 1962–1968 860 6 200 6 5 187,761 46.40 938,8053, South Duke Street (north) Mid-1940s–early 1950-s 162 6 105 6 6 13,063 3.23 78,3783, South Duke Street (south) Mid-1940s–early 1950-s 185 6 45 6 5 7,161 1.77 35,805



to enhance the overall interpretation. Alternatively,multiple layers can be rendered in 3-D and then dis-placed vertically to allow several layers to be visua-lized in 3-D simultaneously. Clay pit topographic

lows and topographic landfill highs are readily charac-terized using this representation method.

Clay Excavations

Clay pit distribution was mapped and characterizedbased on the LIDAR images (Figure 2, cross sectionsA-B and C-D). Excavations extend to the southernedge of the site and follow the orientation of the regio-nal geology toward the east. Some of the excavatedareas were later filled with landfill debris, as discussedabove, but several others were abandoned after theclay was mined out, and the LIDAR data clearlyshow these excavated areas as linear and arcuate pitsand benches (Figure 2). After the 1940s, clay excava-tion moved north of the railway line and west of theLittle Conestoga Creek.

Geological Features—Regional Structure

Cambro-Ordovician limestone with thin graphiticor micaceous beds, the Conestoga Formation, consti-tutes the local bedrock (Meisler and Becher, 1971;de Wet et al., 1999). Bedrock influences the ENE-WSW orientation of numerous small streams, while

Figure 4. Field photograph from study site 1 showing theexcavation of the landfill located north of the railway. Up to 4 mof relatively unconsolidated landfill debris was removed andreplaced by stable material in order to construct a railyard.

Figure 5. Three-dimensional view of the site looking toward the ENE. The 1964 aerial photograph is draped over the LIDAR data (106vertical exaggeration). The landfills are clearly visible and correspond to relative topographic highs. In 2013 the wetland area was expanded(light green) from the original palustrine wetlands (dark green) that existed between the main landfill and the Little Conestoga Creek in theforeground.

de Wet


the major drainage (Little Conestoga Creek) cutsacross the geological structure, flowing south to theConestoga River (Figure 2). As revealed in theLIDAR results, bedrock controls the distribution ofclay pits and, therefore, landfill locations. It alsodirects groundwater flow and hence the distributionof landfill leachate seeps.

A previously unknown structural feature in thesouthern residential area became apparent with theLIDAR results (Figure 2). The feature occurs along aridge that separates two minor drainages and isoriented N70uE. It parallels the general orientation ofthe bedrock but is difficult to detect in the field becauseof the density of trees and houses. Close examinationof the 1940 aerial photographs, which record pre-sub-urban development, reveals a very subtle break inslope that reflects the structural lineament revealedby LIDAR. Field observations of numerous veinquartz float blocks in the area suggest that the linea-ment is related to fractures filled by resistant quartzveins that produce a topographic high. The LIDARresults also delineated a former limestone quarry in awooded part of the adjacent neighborhood.

Other Features

The oldest constructed features recognized on thesite are a lime kiln and several small associated lime-stone quarries. These features are associated with thelimestone ridge that parallels the regional geologicalstructure (Figure 2; cross section A-B). This area wasthe only part of the site not affected by either brickmaking or subsequent activities. Mature trees and rela-tively undisturbed soils occur along the ridge.

Debris from the demolition of the brickworks build-ings is visible in the LIDAR maps (Figure 3), andfoundations and walls are traced out. Beehive kilnsfor baking bricks are also visible, based on the dis‐tribution of their demolition debris. Documenting his-torical structures is important for site characterization,and in this case, the Spalding Conservancy manage-ment plan may include removal of building debris,making this study’s maps particularly important interms of historical knowledge preservation and re‐moval methods.

The site’s original hydrology has been significantlymodified by land use over time. Seeps associated withthe landfills drain leachate with elevated iron contentthat is naturally attenuated by wetlands downslope(de Wet et al., 2000). LIDAR data revealed newdetails of the site’s hydrology and clarified relation-ships between the drainage areas and wetlands (Fig-ures 3 and 5). The exceptional accuracy of theLIDAR made it possible to map surface water flowfrom the seeps near the landfill area, through the

wetlands, to a discharge point into the Little Cones-toga Creek. This information is guiding ongoing mon-itoring of the leachate produced by the landfills.Numerous enigmatic features visible in the 1940 (andlater) aerial photographs of the wetland areas arenow recognized as pits (partially in-filled) and moundsthrough this study’s LIDAR results. These features arepreviously unknown remnants of the 1920–1940 claymining operations but have been evaluated and incor-porated into the Conservancy’s wetland expansionproject (Figure 5).New results from this study have significantly

increased our confidence in determining the formerlandfill dimensions. This is particularly important aswe quantify current environmental conditions andmake recommendations for ongoing site managementat the Conservancy.

CASE STUDY 2: LANCASTER COUNTY PARK

Background

This site is located within a large meander in theConestoga River and is bounded on the west and northby the Conestoga River and to the southeast by MillCreek (Figure 6). Most of this area is currently acounty park, but between 1962 and 1968 much of itwas used as a landfill for Lancaster City and the sur-rounding townships. There are a number of countydocuments and maps for this site, and historical aerialphotographs show that before the early 1960s, flatterparts of the area were farmland and steep slopes werewooded. The underlying bedrock is Cambro-Ordovi-cian limestone (the Conestoga Formation, as at site1), with a regional strike and dip such that ground-water moving through the landfill flows downslopeeast and west along fractures and karst cavities towardthe Conestoga River and Mill Creek. Since the land‐fill is unlined, leachate seeps surrounding the landfillbecame active soon after the landfill closed. In theearly 1970s a leachate report was produced thatincluded several proposals for remediation (LancasterCounty Commissioners, 1973), but it is unclear if anyof the plan was implemented since leachate continuesto discharge into the river and stream and periodicallyraises public concern (Rutter, 2008). Recent analysesnote the presence of iron, nickel, mercury, zinc,arsenic, chloroethane, and benzene, but the landfillhas been removed from a list of possible Superfundcleanup sites, suggesting that solute concentrationsare low (Rutter, 2008).



Results: Location and Extent of the Landfill

The landfill’s outline is defined by a relatively flatarea bounded by regular steep sloping sides that con-trast with natural slopes in the surrounding area

(Figure 6). The slope map derived from the LIDARresults clearly distinguishes the relatively flat uppersurface of the landfill from the surrounding topogra-phy. Its steep side slopes formed as debris accumulatedover the original hillsides. The landfill comprises three

Figure 6. Maps and topographic profiles of study site 2 Lancaster County Park landfill, (a) slope map derived from the LIDAR data with theoutline of the landfill. Topographic profile A-A9 is located north of the landfill, B-B9 shows the northern part of the landfill and the location ofthe seeps at the base of the slope, and C-C9 shows the southern part of the landfill. The thickness of the landfill was determined by acombination of extrapolating the topography from outside of the landfill, changes in the slope, and observations from stereo-pairs ofhistorical aerial photographs. (b) Colorized topography derived from LIDAR data overlain on the 1971 aerial photograph with the locationof the landfill. The photograph, taken shortly after the closure of the landfill, clearly shows the maximum extent of the landfill.

de Wet


sections—a smaller, lower northern section and twolarger upper sections. The overall landfill area is187,761 m3 (,46 acres) and varies up to 10 m deepalong the western edge (Table 2). Assuming an averagedepth of 5 m, the volume of debris is almost1,000,000 m3 (1,307,951 yd3). Historical photographsconfirm that the landfill started after 1957 but before1964. Photographs from 1971 and 1978 show somecontinued disruption of the landfill’s surface; however,stereo-pair photograph examination indicates thatwhile new material was not added after 1971, somereworking of the landfill surface occurred.

Maps of the approximate extent of the landfill areavailable in local documents (Lancaster County Com-missioners, 1973) and confirm these LIDAR andphotography results in terms of landfill area. The localmaps, however, do not include any information aboutthe landfill’s depth or in-fill history. This study showsthat in 1964 the southwestern part of the landfill wasactive while material was being excavated from thenorthern section of the landfill. The excavated north-ern section was later backfilled with debris, resultingin a significantly thicker layer of waste material thanwould have been expected by simply comparing thepre-1962 topography with the current topography.The older southwest part of the landfill comprisesrefuse principally hauled from the city and surroundingareas, whereas the later northern part likely includesappreciable amounts of demolition debris from late-1960s Lancaster City urban renewal projects.

CASE STUDY 3: SOUTH DUKE STREET

Background

The South Duke Street site is the oldest and leastwell-documented location examined (Figure 7). Therewere two dump sites located in the floodplain of theConestoga River, only a few meters from the riverchannel. They were both active during the 1940s. Littleinformation is available about the type of material inthese dumps, but field observations indicate thatmuch of the material was probably household waste,most likely from nearby Lancaster City.

Results

The LIDAR data show several locations in whichmaterial appears to have been added to the floodplainalong the Conestoga River (Figure 7). These areas arerecognized by a relatively flat upper surface, steepbounding slopes, and planiform shapes. The oldestaerial photographs available (1940) indicate someactivity (bare ground) at two of these locations, butit is unclear whether any dumping of material had

begun at that time. The 1947 aerial photographclearly shows debris being added at the northernsite, covering an area of 100 m by 100 m (10,000 m2,2.5 acres), and it extended to within 30 m of theConestoga River. The southern site was also activeat this time, and because the floodplain was narrowerhere, the debris was located very close to the river andextended for 120 m parallel to the river along thefloodplain. By 1950, there is no evidence of activedumping at the southern site, but there is evidencefor continued activity at the northern site. The areaof exposed ground in 1950 is very similar to the final‘footprint’ for both dumps, as determined from theLIDAR results. Ultimately, the northern fan-shapeddump covered approximately 13,000 m2 (3.2 acres),and the southern dump covered 7,200 m2 (1.8 acres)(Table 2). By estimating the original topography ofthis area and comparing it to the LIDAR topographyit is possible to determine the approximate volume ofmaterial (Figure 7). The depth of the northern dumpvaried from a few meters to over 15 m (average, ,6 m),resulting in an approximate volume of 78,000 m3

(102,020 yd3). The southern dump was much smallerand varied in depth from a few meters to 8 m (aver-age, ,5 m), resulting in a volume of around 36,000 m3

(47,086 yd3).Both dumps are adjacent to the Conestoga River

and are located in floodplain sediment; therefore,leachate and even solid waste entered the river in thepast and may still possibly do so. In the past, water-ways and wetlands were considered a convenient wayto dispose of waste, but we now understand their vul-nerabilities. It makes sense to look along waterwaysclose to urban areas for abandoned waste disposalsites. Using the combination of LIDAR and historicalaerial photographs several other sites along the Cones-toga River that exhibit modification from their naturalstate and may be old landfills have been identified.Preliminary field observation confirms significantdebris at these locations, but more detailed work isrequired to determine the nature of the material andwhether it poses any danger to human health or theenvironment.

DISCUSSION

There are a number of techniques to identify formerlandfills and dumps but they all have drawbacks. Forexample, topographic data such as 10-m DEMs or7.5-minute quadrangle topographic maps usually donot offer sufficient vertical or horizontal resolution tobe useful for detailed land-use analysis. Field mappingmay miss subtle changes in terrain or be challengingas a result of vegetation or lack of site access. Near-surface geophysics requires equipment, time on the



ground, and possible vegetation clearing for transectlines. Although aerial photography provides informa-tion about land-use change through time, it is gener-ally limited to the years after 1930, and there aretechnical shortcomings inherent in older photographicsystems. For these reasons, the combined method used

in this study creates a robust model that is widelyapplicable to other waste disposal and land-use pro-blems. LIDAR data are becoming widely availableas a result of private, state, and federal efforts (forexample, Myers, 2009), and historical aerial photo-graphic coverage of the United States is generally good.

Figure 7. Maps and topographic profiles of study site 3—South Duke Street landfills, (a) colorized topographic map derived from theLIDAR data with the outline of the landfills and location of the topographic profiles. The estimated thickness of the landfills was determinedby a combination of extrapolating the topography from outside of the landfills, changes in the slope, and observations from stereo-pairs ofhistorical aerial photographs. (b) Colorized topography derived from LIDAR data overlain on the 1947 aerial photograph with the locationof the landfills. The photograph was taken during active dumping at the landfills. Activity at the site was not evident in the 1950 aerialphotographs.

de Wet


New Techniques, New Information

LIDAR has been used to understand a wide varietyof natural processes where geomorphic features areimportant, such as tectonics, landslides, glaciers,streams, and coastal processes as well as ecologicalstudies, such as forest characterization, wetland map-ping, and biomass determination (Lyon and Greene,1992; Lefsky et al., 2002; Mitasova et al., 2004;Robertson et al., 2004; Drake et al., 2003; Arrowsmithand Zielke, 2009; and Haneberg et al., 2009). LIDARis also being used for risk assessment and understand-ing human-environment interactions, including coastalerosion, volcanic hazards, flood risk mapping, land-use, and infrastructure mapping (Hodgson et al.,2003; Neelz et al., 2006; Rufin-Soler et al., 2008; Bis-son et al., 2009; Tralli et al., 2009; and Zhou andXie, 2009). LIDAR is now recognized as a key proce-dure for documenting natural land-surface change.

In many cases, human impact on the environmentmight be suspected but the exact nature of the activitymight not be obvious. As demonstrated in this study,the combination of LIDAR with aerial photographyoffers an effective method with which to distinguishnatural from anthropogenically altered terrains andto locate and characterize long-abandoned sites, suchas the South Duke Street dumps. By combiningLIDAR data and its derivatives with other data sets,such as historical aerial photography, geophysical sur-veys, and field observations, as described here, detailedtemporally and spatially complete interpretations ofthree former waste disposal locations were produced.According to PADEP there are 17 potential formerlandfills or dumps in Lancaster County, but the reportdid not include the three locations documented here.This strongly suggests that many more dumps remainto be discovered in the area and that this is an issuethat affects counties, cities, and countries globally.Even if some minor dumps are never identified, hope-fully they will be the dumps with the smallest environ-mental impact. Systematic use of a LIDAR and aerialphotography combination, particularly if used withina GIS database, will permit agencies to locate andcharacterize major abandoned waste disposal siteswith significant efficiency.

While former waste disposal sites are one example inwhich the combination of LIDAR and historical aerialphotographs becomes a powerful tool for document-ing and characterizing past land use, there are a myr-iad of other applications; for example, formerabandoned industrial sites in urban areas remain anissue despite the many successes of the BrownfieldsProgram (Greenberg and Hollander, 2006). Many ofthese sites are conveniently located and could bereturned to productive use either as new industrial

sites, as development opportunities, or as public parksand green spaces. Returning these sites to productiveuse requires thorough site characterization to deter-mine the nature of past activity and whether any envir-onmental concerns or hazards exist.The ability to virtually view a site in 3-D by combin-

ing LIDAR with a wide variety of current and historicaldata sets is useful for analytical purposes but also canhelp municipal planners and others to visualize sitechanges through time. For example, by rendering histor-ical aerial photographs in 3-D using LIDAR results as abase, it is now straightforward to see how current topo-graphic high areas correspond to landfills that wereactive in the late 1950s to early 1960s at the BakerWoodlands/Spalding Conservancy site. This effectivelytakes the viewer back in time to see the landscape as itappeared in the past—an especially powerful visualiza-tion tool for students, environmental scientists, urbanplanners, and other constituents (Figure 5).

Significance: Long-Term Implications of AbandonedWaste Disposal Sites

Waste disposal sites may have an impact on humanhealth, on the environment, on site engineering, and/oron local aesthetics. Human health issues may arisefrom leachate contamination of surface or ground-water (Suflita et al., 1992), physical harm fromexposed debris, or airborne contaminates, contribut-ing to asthma and other respiratory problems (EPA,2008). In almost all cases, waste disposal sites priorto the 1970s were unlined, and few were capped withanything more than a thin soil cover. Rainwater perco-lates easily through them, carrying soluble materialinto local surface and/or groundwater systems (Cum-mins, 1968). In limestone karst terrains, such as theConestoga Formation bedrock underlying the threeLancaster County dumps documented here, leachatemay accumulate and travel significant distances under-ground. At the County Park waste disposal site docu-mented in this study, leachate seeps drain along theperimeter of the former landfill and in some placecross public walking trails (Lancaster County Com-missioners, 1973; Rutter, 2008).Exposed debris may cause physical harm, particu-

larly to children and pets, if rusting metal, sharp plas-tics, and large objects (such as old refrigerators) areaccessible on the ground surface. Unstable footingdue to irregular ground settling may lead to injury ifa site is open for recreational use. Asbestos was widelyused in ceiling and flooring materials in the early 20thcentury, and demolition debris from that era may con-tain these materials. As the binding agents decay,asbestos and other particulates may be released intothe air as the waste disposal site erodes (EPA, 2008).



Old waste disposal sites are likely to have received awide variety of materials because there were no restric-tions on what could be disposed of at these sites priorto the stricter regulations introduced after the 1970s.There is no information available about the type ofmaterial deposited in the Duke Street dump (site 3),and it is unclear whether the material disposed of inthe 1940s was more or less harmful than typical mate-rial disposed of in the 1950s and 1960s. It is likely thatthere were many cases between the 1940s and 1960s ofhazardous material disposal in unregulated municipalwaste sites (Colten and Skinner, 1996). The SpaldingConservancy site received a significant quantity of ceil-ing and floor tile material as waste, much of which isnow exposed on the land surface (de Wet et al.,1999). The age of the landfill not only determines thecontent but also relates to change with time. Forexample, over time the quantity and composition ofleachate may change, as different materials decayat different rates. This influences the composition ofleachate draining from a site (Department of theEnvironment, 1990; Marsh and Garnham, 1996; andMetcalfe and Rochelle, 1999). Study of leachate fromthe County Park landfills shows a change in leachatecomposition over time (Rutter, 2008).The environment is affected by the presence of a

landfill or dump because of the physical disruption tothe local area while the site is active (i.e., excavations,road building, and debris accumulation) and, subse-quently, with land subsidence and poor soil develop-ment, leading to scrubby vegetation, often dominatedby invasive species and thus generally poor wildlifehabitat. Over time, habitat may improve somewhatas the ecosystem recovers, but, as noted for humanhealth, significant wildlife hazards remain. Formerwaste disposal sites may shed trash material into theenvironment for decades as they erode. This affectsan area’s aesthetic value, affecting property valuesand quality of life for nearby residents.Compaction and decomposition of material within

waste disposal sites makes them inherently unstableand, therefore, unsuitable for most development projects.Cavities may form where material readily decays, roofedby less soluble or compactable material that could giveway with increased overburden. If a location is chosenfor development but is atop a former waste disposalsite that is difficult to detect from the surface, such asthe three locations documented here, serious construc-tion problems will arise. Making sound land-use deci-sions for safety, environmental concerns, municipalgrowth, and aesthetics requires an understanding of theprevious land use. A database in which former landfillsand dumps are clearly delineated is an important aspectof municipal planning that needs to be updated as newinformation becomes available. The combination of

LIDAR and aerial photography is an efficient methodof obtaining critical land-use data that should be usedto guide future land-use planning.

CONCLUSIONS

LIDAR is a valuable addition to the tools neededfor accurate site assessment. First, used directly, it pro-vides spatially complete, high vertical- and horizontal-resolution information about the topography of a site.This data set can then be used to derive numerousother useful data sets, including slope, aspect, andshaded relief (particularly using MDOW analysis).Creative symbolization of the topographic data canreveal information such as subtle changes in topogra-phy that provides clues to natural and anthropogenicinfluences on the site. Second, when integrated withother historical and current spatial data sets, such ashistorical aerial photographs and geophysical surveys,LIDAR provides a powerful way to interpret andunderstand a site’s land-use history.LIDAR is particularly useful in areas in which high-

resolution data are needed but are difficult, prohibited,or expensive to obtain. Sites with significant obstruc-tions, including buildings and vegetation, may preventor restrict the use of other mapping techniques that useinstruments such as electronic distance meters or GPS.LIDAR is useful for sites that have restricted accessbecause of hazards, or in sensitive areas such as wet-lands or other special ecological areas. In these circum-stances, using remotely sensed data such as LIDARhas significant advantages.Finally, LIDAR provides a high-resolution base for

the 3-D rendering of other data sets, such as historicaland recent aerial photographs. Visualization in 3-Dcan provide unique insight into the history of an areaand provide a powerful way to analyze and visualizenatural and anthropogenic change over time.The ability to seamlessly transition between differ-

ent data sets is one of the powerful attributes of inte-grating information in a GIS environment, and theresultant data reveals new insights into situations thatcould directly affect human health, environmentalstandards, and future development opportunities.

ACKNOWLEDGMENTS

Mike Rahnis provided technical assistance with theprocessing of the LIDAR data. Chris Williams andCarol de Wet provided helpful comments, and SteveSylvester provided local government documents andpersonal knowledge of sites across Lancaster County.

de Wet


REFERENCES

ARROWSMITH, J. R. AND ZIELKE, O., 2009, Tectonic geomorphologyof the San Andreas Fault zone from high resolution topogra-phy: An example from the Cholame Segment Source: Geomor-phology, Vol. 113, Nos. 1–2, pp. 70–81.

BIOTTO, G.; SILVESTRI, S.; GOBBO, L.; FURLAN, E.; VALENTI S.; AND

ROSSELLI, R., 2009, GIS, multi-criteria and multi-factor spa-tial analysis for the probability assessment of the existence ofillegal landfills: International Journal Geographical Informa-tion Science, Vol. 23, No. 10, pp. 1233–1244.

BISSON, M.; FORNACIAI, A.; BEHNCKE, B.; AND NERI, M., 2009,LiDAR-based digital terrain analysis of an area exposed tothe risk of lava flow invasion: The Zafferana Etnea territory,Mt. Etna (Italy): Natural Hazards, Vol. 50, No. 2, pp.321–334.

CARLSON, E.; DEWET,A.; CAMPBELL, S.; ANDWILLIAMS, A., 2000,Geo-physics used to investigate the structure of a landfill: AbstractsPrograms—Geological Society America, Vol. 32, No. 1.

CENTER FOR SUSTAINABLE SYSTEMS, UNIVERSITY OF MICHIGAN, 2014,Municipal SolidWaste Factsheet: PublicationNo.CSS04-15, 2 p.

COLTEN, C. E. AND SKINNER, P. N., 1996, The Road to Love Canal:Managing Industrial waste Before the EPA: University ofTexas Press, Austin, TX. 217 p.

CUMMINS, R. L., 1968, Effects of Land Disposal of Solid Wastes onWater Quality, Cincinnati, Ohio: U.S. Department of Health,Education and Welfare, SW-2ts, 29 p.

DEPARTMENT OF THE ENVIRONMENT, 1990, Landfilling Wastes, Tech-nical Memorandum for the Disposal of Wastes on LandfillSites: Waste Management Paper 26 (HMSO, London).

DE WET, A. P.; RICHARDSON, J.; AND OLYMPIA, C., 1998, Interac-tions of land-use history and current ecology in a recovering“urban wildland”: Urban Ecosystems, Vol. 2, pp. 237–262.

DE WET, A. P. AND STERNBERG, R. S., 1999, Geophysical investiga-tion of an abandoned landfill: Proceedings SAGEEP, March1999, pp. 143–152.

DE WET, A. P.; STERNBERG, R.; AND WINICK, J., 1999, Interpretingland-use history by integrating aerial photographs, near-surfacegeophysics, and field observations into a digital database: Envir-onmental Engineering Geoscience, Vol. 5, No. 2, pp. 235–254.

DE WET, A. P.; WEAVER, S.; AND ACHEAMPONG, S., 2000, Hydrol-ogy, geochemistry and geophysics of the Baker WoodlandsEnvironmental Research Site, Lancaster, PA: Proceedings13th KECK Geology Consortium Symposium, pp. 245–249.

DRAKE, J. B.; DUBAYAH, R. O.; HOFTON, M.; KNOX, R. G.; CLARK,D. B.; CONDIT, R.; AND BLAIR, J. B., 2003, Above-ground bio-mass estimation in closed canopy neotropical forests usingLIDAR remote sensing: Factors affecting the generality ofrelationships: Global Ecology Biogeography Letters, Vol. 12,No. 2, pp. 147–159.

ENVIRONMENTAL PROTECTION AGENCY (EPA), 2008, Framework forInvestigating Asbestos-Contaminated Superfund Sites: OSWERDirective 9200.0-68, September 2008, 71 p.

ERB, T. L.; PHILIPSON, W. R.; TANG, W. T.; AND LIANG, T., 1981,Analysis of landfills with historic air photos: PhotogrammetricEngineering Remote Sensing, Vol. 47, pp. 1363–1369.

GREENBERG, M. R. AND HOLLANDER, J., 2006, The EnvironmentalProtection Agency’s Brownfields Pilot Program: AmericanJournal Public Health, Vol. 96, No. 2, pp. 277–281.

HANEBERG, W. C.; COLE, W. F.; AND KASALI, G., 2009, High-resolution LIDAR-based landslide hazard mapping andmodeling, UCSF Parnassus Campus, San Francisco, USA:Bulletin Engineering Geology Environment, Vol. 68, No. 2,pp. 263–276.

HERNDON, R. C.; MOERLINS, J. E.; LAMBOU, V. W.; AND GEBHARD,R. L., 1990, Proximity of Pennsylvania Sanitary Landfills to

Wetlands and Deepwater Habitats: EPA/600/S4-89/047 May1990, 4 p.

HODGSON, M. E.; JENSEN, J. R.; TULLIS, J. A.; RIORDAN, K. D.; AND

ARCHER, C. M., 2003, Synergistic use of LIDAR and coloraerial photography for mapping urban parcel imperviousness:Photogrammetric Engineering Remote Sensing, Vol. 69, No. 9,pp. 973–980.

HORNING, R. A., II, 1992, The Lancaster Brick Company 1919–1979: Journal Lancaster Historical Society, Vol. 94, pp. 2–32.

KENT, B. C., 1984, Susquehanna’s Indians. Anthropological Seriesno. 6: Pennsylvania Historical and Museum Commission,Harrisburg, PA.

LANCASTER COUNTY COMMISSIONERS, 1973, County Park LeachateStudy: Huth Engineers, Inc., County of Lancaster, PA, 64 p.

LANCASTER COUNTY BOARD OF COMMISSIONERS, 1971, Solid WasteManagement Plan 1971: Lancaster County Planning Commis-sion, County of Lancaster, PA. 132 p.

LANCASTER COUNTY BOARD OF COMMISSIONERS, 1987, LancasterCounty SolidWasteManagement Plan 1986: Lancaster CountyPlanning Commission, County of Lancaster, PA. 206 p.

LANCASTER FARMLAND TRUST, 2015, Lancaster Farmland Trust: Elec-tronic document, available at http://www.lancasterfarmlandtrust.org/heritage/farming-lancaster.html

LEFSKY, M. A.; COHEN, W. B.; HARDING, D. J.; PARKER, G. G.;ACKER, S. A.; AND GOWER, S. T., 2002, Lidar remote sensingof above-ground biomass in three biomes: Global Ecology Bio-geography Letters, Vol. 11. No. 5, pp. 393–399.

LYON, J. G., 1987, Use of maps, aerial photographs, and otherremote sensor data for practical evaluations of hazardouswaste sites: Photogrammetric Engineering Remote Sensing,Vol. 53, pp. 515–519.

LYON, J. G. AND GREENE, R. G., 1992, Use of aerial photographs tomeasure the historical areal extent of Lake Erie coastal wet-lands: Photogrammetric Engineering Remote Sensing, Vol.58, pp. 1355–1360.

MARK, R., 1992,Multidirectional, Oblique-Weighted, Shaded-ReliefImage of the Island of Hawaii: U.S. Geological Survey, OF-92-422, 5 p.

MARSH, A. H. AND GARNHAM, A., 1996, Investigation, hazardassessment and remediation of existing landfills: GeologicalSociety, London, Engineering Geology Special Publications,Vol. 11, pp. 3–7.

MEISLER, H. AND BECHER, A. E., 1971, Hydrogeology of theCarbonate Rocks of the Lancaster 15-Minute Quadrangle,Southeastern Pennsylvania, Groundwater Report W26: Penn-sylvania Geological Survey, Harrisburg, PA.

METCALFE, R. AND ROCHELLE, C. (Editors), 1999, Chemical con-tainment of waste in the geosphere: Geological Society, Lon-don, Special Publications, Vol. 157, pp. 7–16.

MITASOVA, H.; DRAKE, T. G.; BERNSTEIN, D.; AND HARMON, R. S.,2004, Quantifying rapid changes in coastal topography usingmodern mapping techniques and geographic information sys-tem: Environmental Engineering Geoscience, Vol. 10, No. 1,pp. 1–11.

MYERS, M. D., 2009, USGS: Science at the intersection of land andocean: Sea Technology, Vol. 50, No. 1, pp. 18–21.

NEELZ, S.; PENDER, G.; VILLANUEVA, I.; WRIGHT, N. G.; WILSON,M.; BATES, P.; MASON, D.; AND WHITLOW, C., 2006, Usingremotely sensed data to support flood modeling: ProceedingsInstitution Civil Engineers: Water Management, Vol. 159,No. 1, pp. 35–43.

POPE, P.; VAN EECKHOUT, E.; AND ROFER, C., 1996, Waste site char-acterization through digital analysis of historical aerial photo-graphs: Photogrammetric Engineering Remote Sensing, Vol.62, No. 12, pp. 1387–1394.



ROBERTSON, V. W.; WHITMAN, D.; ZHANG, K.; AND LEATHERMAN,S. P., 2004, Mapping shoreline position using airborne laser alti-metry: Journal Coastal Research, Vol. 20, No. 3, pp. 884–892.

RUFIN-SOLER, C.; GARDEL, A.; AND HEQUETTE, A., 2008, Assessingthe vulnerability of coastal lowlands to marine floodingusing LiDAR data, Sangatte coastal dunes, northern France:Zeitschrift Geomorphologie, Vol. 52, No. 3, pp. 195–211.

RUTTER, J., 2008, Gunk fouls trail at park. Lancaster Newspapers,June 15, 2008: Electronic document, available at http://Lan-caster online.com

SILVESTRI, S. AND OMRI, M., 2009, A method for the remote sensingidentification of uncontrolled landfills: Formulation and vali-dation: International Journal Remote Sensing, Vol. 29, No. 4,pp. 975–989.

SUFLITA, J. M.; GERBA, C. P.; HAM, R. K.; PALMISANO, A. C.;RATHJE, W. L.; AND ROBINSON, J. A., 1992, The world’s largestlandfill—A multidisciplinary investigation: EnvironmentalScience Technology, Vol. 26, No. 8, pp. 1486–1495.

TRALLI, D. M.; BLOM, R. G.; ZLOTNICKI, V.; DONNELLAN, A.; AND

EVANS, D. L., 2009, Satellite remote sensing of earthquake,volcano, flood, landslide and coastal inundation hazards:ISPRS Journal Photogrammetry Remote Sensing, Vol. 59,No. 4, pp. 185–198.

WALLACE, P. A. W., 1989, Indians in Pennsylvania, AnthropologicalSeries no. 5: Pennsylvania Historical and Museum Commis-sion, Harrisburg, PA.

WILSON, M. D.; DE WET, A. P.; AND DAVIES, G. M., 2006, Wetlandsedimentary record as an indicator of the land use at a formerindustrial site: Geological Society America Abstracts Pro-grams, Vol. 38, No. 2, p. 80.

WILSON, W. H., 1985, The Columbia-Philadelphia Railroad and itsSuccessor: American Canal and Transportation Center, York,PA. 72 p.

ZHOU, G. AND XIE, M., 2009, Coastal 3-D morphological changeanalysis using LiDAR series data: A case study of AssateagueIsland National Seashore: Journal Coastal Research, Vol. 25,No. 2, pp. 435–447.

de Wet


Identification of Wall Tension Fractures Caused byEarthquakes, Blasting, and Pile Driving

JEFFREY A. JOHNSON1

3830 Valley Centre Drive, No. 705-804, San Diego, CA 92130

ALAN “BOB” MUTCHNICK

GMU Geotechnical, Inc., 23241 Arroyo Vista, Rancho Santa Margarita, CA [email protected]

Key Terms: Structure Survey, Crack Mapping, SiteInvestigations

ABSTRACT

Determining whether an earthquake, blasting, or piledriving caused non-structural “cosmetic” wall cracksor extended pre-existing fractures is often based onjudgment: The structure presumably vibrated; therefore,new cracks formed, and existing fractures wereextended. The proposed two-part, primarily post-event,fracture mapping and analysis of intensity data aredesigned to test the static, as opposed to vibratory,cause of wall fractures, assuming: (1) tension fracturesform perpendicular to the principal tensile stress(σ3) direction; (2) during structure vibrations, the localstress field rotates relative to the static stress field; (3)stress reversals, common during earthquakes, can causetension fractures at all corners of rectangular doors andwindows; (4) progressively younger en echelon tensionfractures, originating at the corner of a rectangular wallopening, trend toward horizontal due to the transientrotation of the principal stresses, whereas fractures fromnon-vibrational causes trend toward vertical; (5) crackextension occurs when two fractures are linked byhook-shaped fractures; and (6) site-specific intensityobservations and local intensity data provide qualitativedata regarding the timing and amplitude of wall strainsand the potential for co-seismic damage. Co-seismictension fractures can therefore be differentiated frompre-existing cracks, suggesting pre-blasting and pile-driving crack mapping, building condition surveys, andco-seismic ground and structure motion recordingsmay not be required: (1) if the building has no priorexposure to potentially damaging vibrations; and (2) areconnaissance inspection indicates there are no pre-existing horizontal or sub-horizontal wall fractures.

INTRODUCTION

The purpose of this study is to present the criteriaand tests that form the basis of a two-part analysisthat can be used to distinguish between wall crackscaused by vibrations from small or distant earth-quakes, pile driving, blasting events, and other con-struction activities such as demolition and dynamicground improvements, from pre-existing fracturescaused by non-seismic processes (Oriard, 1999; Audell,2004). Part one consists of a reconnaissance structurecondition survey and an exterior site review followedby a post-event fracture analysis. Although important,the proposed part one analysis is not dependent onknowledge of the cause(s) of pre-existing fracturesand data from co-seismic ground motion monitoring,provided the structure(s) has not experienced poten-tially damaging vibrations, and sub-horizontal tohorizontal wall fractures are not observed during thepre-event reconnaissance inspection.The post-event part one data collection is limited to

the orientation of tension fractures, angular changesbetween successive en echelon tension fractures, andevidence of fracture linkage. Part two includes collect-ing and documenting site-specific intensity observa-tions and regional intensity data obtained from theU.S. Geological Survey (USGS) or the CaliforniaGeological Survey (CGS). The collection of intensitydata (e.g., the degree of shaking at a specified place[Richter, 1958]) is important because the data supportthe timing and the relative degree of shaking andtherefore the opportunity for co-seismic wall tensionfracture formation and the extension of pre-existingcracks.Emphasis is placed on wall tension fractures to

determine vibratory damage because: (1) they formperpendicular to the principal tensile stress (σ3),thereby giving an indication of the local stress field inthe wall at the time of their formation (Scholz, 2002;Gudmundsson, 2011); (2) during three-dimensional(3D) structure vibrations, σ3 is generally not horizon-tal (Gudmundsson, 2011); (3) tension fractures often1Corresponding author phone: 858-243-4438; email: [email protected]


form when shear fractures do not because wall cover-ings, such as lath and plaster, drywall, and stucco,are weak in tension compared to their shear and com-pression strengths; and (4) it is unlikely cracking offloor slabs, foundations, driveways, sidewalks, patios,and pool decks that are in contact with the groundcan be used to demonstrate vibratory damages becauseco-seismic damages of this type are “inconsequential,except perhaps in the near fault region” (Bolt et al.,2004), where vibratory damage is identifiable.The reader is referred to Griffith (1924), Scholz

(2002), and Gudmundsson (2011) for a general reviewof fracture mechanics. Other methods that are used todistinguish between pre-existing cracks and fracturescaused by mine, quarry, and construction blastingand pile driving have been documented since the1920s and are discussed by Siskind et al. (1980), Stagget al. (1984), the Transportation Research Board(1997), and Oriard (1999). Oriard (1999) and Audell

(2004) discuss non-seismic crack-formation processesin walls covered with lath and plaster, drywall, andstucco.

BACKGROUND

Fractures

For the purposes of this analysis, it was assumed awall includes two of the three mutually perpendicularprincipal compressive stresses, σ1 and σ3, where σ1. σ3.For a vertical wall at rest, σ1 is approximately vertical,and σ3 is approximately horizontal. The assumption isconsistent with Anderson’s (1951) theory of faulting,where the three principal stresses are arranged so thattwo are parallel and one is perpendicular to a relativelylevel ground surface. For purposes of this analysis,σ2 can be ignored, and the type (i.e., tension or shear)and orientation of a fracture(s) are dependent onthe static or seismic orientation of principal stressesσ1 and σ3.A crack or fracture is a discontinuity with no tensile

strength. There are two types of fractures, extension,which is the result of tensile stress, and shear, whichis the result of either shear or compressive stress.Only tension fractures (i.e., an extension fracture thatforms when σ3 is negative; Gudmundsson, 2011) arerelevant to this discussion.Walls are not homogeneous or isotropic structures.

However, in a vertical wall that lacks stress concentra-tions, such as doors and windows, σ1 is approximatelyvertical, and σ3 is approximately horizontal, andtension fractures will form perpendicular to σ3. Non-vertical tension fractures indicate rotation of theprincipal stresses or local stress field in the wall poten-tially due to either permanent or seasonal differentialfoundation movement or transient vibratory groundmotions. Tension fractures are therefore importantdiagnostic discontinuities because they indicate thelocal stress field at the time of their formation.Discontinuities such as wall openings, often referred

to as stress raisers, increase the stress at corners of theopenings, alter the orientation of the stress trajectories orprincipal stress axes, in the adjoining wall, and there-fore change the direction of fracturing if sufficientadditional stresses are added from vibrational andnon-vibrational causes. Static tension fractures, near-est to the corners of doors and windows, are therefore,although not always (Audell, 2004), inclined ratherthan vertical (Figures 1a and 2a). However, as distancefrom a stress raiser increases, successively formed ten-sion fractures will trend toward vertical (Figures 1band 2b).An integral part of conducting post-event crack map-

ping is determining if pre-existing fractures were

Figure 1. Exterior stucco fractures, most likely caused bydifferential settlement, observed during a post-pile-driving siteinspection. (a) Approximately six tension cracks linked by hook-shaped fractures. Tension crack 1 is the oldest. (b) Angularrelationship between progressively younger fractures trendingtoward vertical indicates fractures caused by a non-seismic process.

Johnson and Mutchnick


extended or increased in length. Crack extension isherein defined as the formation of a new fracture,roughly in the same linear alignment but not connected,that eventually links by hook-shaped tension fractures

with the older fracture (Figures 1a and 2a). Hook-shaped tension fractures are characteristic of the linkageprocess and are caused by curved stress trajectoriesbetween the offset and initially unconnected fractures.Once hook-shaped fractures are evident, fracture exten-sion should be assumed because the fractures are mostlikely acting as a single, hard linked structure. Forexample, fractures 3 and 4 (Figure 1a) exhibit hook-shaped fractures without a clear visual connection. Itis unlikely fractures 5 and 6 would have formed withouta hard link between fractures 3 and 4.Stress reversals during an earthquake cause side-

to-side, as well as up-and-down structure vibrationsand can induce “X” wall tension cracks (Figure 3).“X” cracks are important because they are “charac-teristic of co-seismic structural damage” (Stien-brugge, 1970) and demonstrate σ3 is generally nothorizontal while the structure is vibrating. Small ordistant earthquakes, blasting, and pile driving gener-ally will not result in structural damage such as “X”wall cracks, but they can cause cracking at and nearthe corners of rectangular doors and windows (Fig-ure 4). One indication of damaging vibrations andstress reversals is a set of tension fractures that occurat all corners of a rectangular wall opening (Figures3c and 4). In addition, fractures that extend dueto seismic shaking will trend toward the horizontal(Figure 5a and 5b).Overprinting, a type of crack extension, occurs when

en echelon fractures are caused by two or more pro-cesses (Audell, 2004). The residence shown on Figure6a was inspected and photographed after a number ofquarry blasts. The results of the inspection suggestedfractures 1 to 4 were not caused by the blasting, whereasthe fracture number 5 was. Additional fractures that arealso most likely the result of blasting are highlighted inwhite, to improve visibility (Figure 6b).A second example of overprinting was observed on

a wall of a residence in Los Angeles, CA, affected bya landslide and not seismic shaking (Figure 7). Theslope supporting the residence failed and acceleratedduring the winter rains of 1993, most likely causingfractures 3 and 4. Fractures 1 and 2 were approxi-mately vertical and located at only one corner of thewindow (Figure 7b), suggesting they were the resultof a non-seismic process.

Intensity

Intensity is an observational measure of the co-seis-mic damage to a structure and the degree of shaking atthe site (Richter, 1958). The first descriptive intensityscale intended for general use was the Rossi-Forelscale (De Rossi, 1883). In 1902, Mercalli (1902) intro-duced a scale that reduced some of the problems of

Figure 2. Interior drywall fractures, most likely caused bydifferential settlement, observed during a post-blasting site inspec-tion. (a) Approximately four tension cracks linked by hook-shapedfractures. Tension crack 1 is the oldest. (b) Angular relationshipbetween progressively younger fractures trending toward verticalindicates fractures caused by a non-seismic process.

Identification of Wall Tension Fractures


assigning Rossi-Forel intensities. In 1931, Wood andNeumann (1931) proposed the modified Mercalliintensity (MMI) scale and an abridged versiondesigned for use by observers with varied experiencelevels (Neumann, 1954).MMI is a progressive scale ranging from not felt

(MMI I) to total damage (MMI XII). MMI is impor-tant because of its extensive field testing and updatingover the past 80+ years, and it includes the “felt” inten-sities suitable for supporting or eliminating the poten-tial for co-seismic fracture formation and extension.The 1931 MMI scale, with some modifications, is stillin use today (Stover and Coffman, 1993; Deweyet al., 1995).Selected wording from the 1931 unabridged MMI

scale for estimating MMI IV, V, and VI earthquake

intensities, which includes the formation of fractures,is listed in Table 1. For comparison, criteria used toestimate MMI following the 1994 Northridge, CA,earthquake (Dewey et al., 1995) are also listed. Animportant difference between the unabridged 1931wording describing MMI V and that used by Deweyet al. (1995) was the addition of “hairline cracks ininterior walls.” In the 1931 abridged version, “a fewinstances of cracked plaster” was included in MMI V(Wood and Neumann, 1931, pp. 279–280). Dewey et al.(1995) did not specify the type of wall covering, eventhough the results of blasting studies indicate plasteris more prone to cracking than drywall or stucco.In the meizoseismal region, where structural

damage is common, MMI is determined primarily byfield observations. At greater distances, MMI was,until recently, based in part on questionnaires mailedto post offices and other government facilities by theU.S. Coast and Geodetic Survey and more recentlyby the USGS. The USGS appears to have started thepractice following the 1886 Charleston, SC, earth-quake (Talwani, 2014).Stover and Coffman (1993) amended MMI to

account for post-1931 field experience, recent modifi-cations to the USGS mailed questionnaires, and thepoor reliability of the subjective effects on people inassigning MMI. The U.S. Bureau of Mines (Siskindet al., 1980, 1993) and the Transportation ResearchBoard (1997) also concluded that human perception

Figure 3. “X” wall fractures. (a) Five-story structure in the city ofManagua after the M 6.2 1972 Managua, Nicaragua, earthquake.(b) Low-rise building, San Fernando Valley, CA, after M 6.7 1994Northridge, CA, earthquake. (c) Fractures on four corners of wallopening, M 6.7 1994 Northridge, CA, earthquake.

Figure 4. (a and c) Interior drywall fractures at both corners of adoor observed during a post-blasting site inspection. (b) Over-printing suspected based on increase in fracture angles for one setof linked cracks (30u to 56u) and ,28u angle for isolated fractureswith no clear evidence of linkage. (d) Overprinting suspected basedon similar fracture angles (i.e., 25u shown in d and 28u to 30u on b),asymmetric number of fractures, reduction in fracture angles from15u to approximately horizontal, and no clear evidence of linkage.Lack of evidence of linkage between suspected seismic fracturessuggests cracking due to one or possibly two blasts.



of vibrations is not an accurate gauge of the damagepotential of the vibrations.

Dengler and Dewey (1998) conducted a telephonesurvey following the M 6.7 1994 Northridge, CA,

earthquake, obtaining building-specific intensity infor-mation from a relatively large number of locationscompared to responses obtained from the mailers.Results of their survey were compared with the

Figure 5. (a) Interior wall fractures extending from top-right corner of a door. Un-numbered en echelon tension fractures are most likely theresult of differential settlement. Numbered and linked fractures are likely due to vibrations from a blast. (b) Trend toward horizontal ofnumbered fractures shown on Figure 5a supports origin due to blast vibrations.



USGS MMI data, and an algorithm to estimate whatwas called the “community decimal intensity” (CDI)was developed. CDI estimates are similar but are notidentical to MMI (Dengler and Dewey, 1998).The USGS adopted the Dengler and Dewey (1998)

CDI algorithm and a form of their telephone question-naire for use on the Internet (Wald et al, 1999a,1999c). The program that became known as “Didyou feel it?” (DYFI) has three main sections, includingthe two-part earthquake effects section. Part oneincludes observations related to sounds, the perfor-mance of exterior walls and fences, and the movementof objects within the structure. Part two includesobserved building damage from “hairline cracks inwalls” to “building permanently shifted over founda-tion.” According to the USGS web page, as the dataare collected, CDI values are rounded to integers,average values are computed for individual ZIP codes,and then those values are converted to MMI

intensities. CDI data complement site-specific MMIobservations and are of particular importance if site-specific data do not exist or cannot be obtained.ShakeMaps, posted online by the USGS and the

CGS, are a valuable, rapidly available source ofground motion and instrumental intensity data (Waldet al., 1999b, 1999c). Instrumental intensity ShakeMapsare based on correlations of MMI with peak groundmotion parameters (Wald et al., 1999b; Caprio et al.,2015) and can also be used to supplement site-specificMMI data.Siskind et al. (1980) published a correlation between

blast recordings and damage to structures. Blasts gen-erally contain a high concentration of energy that cancause damage to nearby structures. In addition, airblasts (air overpressures), also caused by blasting, canfracture windows and crack walls. Table 1 contains aportion of the Siskind et al. (1980) damage classifica-tion system. Although numerous studies have pointedout blasts are not earthquakes, our preliminary

Figure 6. Fractures 1, 2, 3, and 4 are most likely the result ofdifferential settlement. Fracture 5 (a) and highlighted fractures (b)are relatively low angle and unlinked, suggesting they formed dueto structure vibrations resulting from possibly one blast thatcontained asymmetric stress reversals (evidence of fractures dueto structure vibrations was not observed at the opposite corner ofthe door).

Figure 7. Linked and relatively low-angle fractures 3 and 4 werethe result of accelerated movement of a 1993 landslide and notstructure vibrations. Seismic exposure occurred prior to theconstruction of the residence and a year later during the 1994Northridge, CA, earthquake. Fractures 1 and 2 may have been theresult of differential settlement. It is unclear if the differentialsettlement was related to the slope movement.



analysis suggests “threshold” damage ranges fromMMI IV to V, and “minor” damage ranges fromMMI V to VI.

The Transportation Research Board (1997) con-cluded vibrations due to pile installation could causewall cracking at distances generally limited to thelength of the pile. In extreme cases, differential settle-ment and resultant foundation and building damage,due to either compaction or consolidation of sandysediments, have been recorded at distances of 400 m(Transportation Research Board, 1997).

ANALYSIS

The objective of the proposed two-part analysis is todetermine if a small or distant earthquake, blasting, orpile driving caused or extended non-structural “cos-metic” wall fractures. Part one consists of pre-eventdata collection, which includes documenting the his-tory of exposure to potentially damaging earthquakesand other sources of ground vibrations and, in thecase of blasting and pile driving, conducting a recon-naissance structure(s) inspection to determine if existingwall fractures include horizontal or approximatelyhorizontal tension cracks and/or fractures at all cor-ners of at least one rectangular wall opening. If theanswer is yes to either of these, detailed pre-eventcrack mapping or a structure condition survey shouldbe considered. Post-event part-one analysis consistsof conducting a detailed analysis of wall fractures

focusing on stress raisers to determine if: (1) there aretension fractures at or near all corners of one ormore of the rectangular wall openings; (2) as the dis-tance from a wall opening increases, successive tensionfractures trend toward horizontal; and (3) there isevidence of hook-shaped linkage fractures betweensuccessive horizontal-trending tension fractures. Ifthe answer is no to all of the above, the probabilityis low that an earthquake, pile driving, or blastingcaused tension fractures or extended existing cracks.Part-two site-specific intensity data should be

collected as soon after the event as possible fromoccupants or by inspection. Six questions, based onTable 1, for estimating a site-specific MMI are listedin Table 2. For example, was wall creaking noticed,did wall hangings move or fall, did any items fallfrom shelves, and were any windows cracked?Although wall creaking by itself (MMI IV) is not suf-ficient to determine whether or not co-seismic wallcracks developed, the absence of creaking would likelypreclude fracture formation. Minimal wall movementis also suggested if hanging pictures and mirrors wereundisturbed and items did not fall from shelves or fur-niture. A cracked window (MMI V) supports fractureformation at the corners of the window and potentiallyat other wall openings including doors.A word of caution: Humans can detect vibrations

well below levels that can cause damage (Richter,1958; Siskind et al., 1980, 1993), and the USGS, theTransportation Research Board (1997), and Oriard

Table 1. Comparison of modified Mercalli intensity and the blasting damage classification.

Rating Modified Mercalli Intensity Blasting Damage Classification

Wood and Neumann (1931) Dewey et al. (1995) Siskind et al. (1980)IVCreaking of walls, frame Walls creaked loudly ThresholdRattling of dishes, windows, doors Buildings shaken moderately to strongly Loosening of paint; small plaster cracks at

joints between construction elementsSwinging of hanging objectsVCracked windows (generally not) A few windows crackedOverturned small or unstable objects A few small objects overturned and fallenMoved small objects, furnishingsSwinging of hanging objects, doors Hanging pictures tilted, out of place, or fallenBroke dishes to some extent Lengthening of old cracks

MinorA few instances of cracked plaster Hairline cracks in interior walls Hairline to 3 mm cracksVISome windows broke out Some windows broken outFall of plaster in small amount A few instances of fallen plaster or damaged Loosening and falling of plasterCracked plaster, especially fine cracks

chimneysFall of loose mortar

un-reinforced masonry chimneys Cracks in masonry around openings nearpartitions

Overturned furniture Light furniture overturnedMoved moderately heavy furnishings Moderately heavy furniture displacedFall of knick-knacks, books, pictures Many small objects overturned and fallenBroke dishes in considerable quantity Many glassware items or dishes broken

Large cracks in interior walls



(1999) concluded that human perception of vibrationsis not an accurate gauge of the damage potential of thevibrations. However, the implied or expressed correla-tion between the human perception of vibrations andbuilding damage suggests a potential need for publicoutreach and pre-event structure condition surveysindependent of the technical need (Oriard, 1999).

CONCLUSIONS

The type of data collected during the proposed two-part analysis can be used to test if non-structural wallfractures were caused by seismic shaking, potentiallyeliminating the need for detailed pre-blasting and piledriving crack mapping and instrumental recordingsof ground or structure motions if, during part one, itis determined the subject structure(s) has not beenexposed to seismic shaking and if non-seismic pro-cesses have not caused horizontal or sub-horizontalwall tension fractures.The part-one post-event crack analysis focuses on

evidence of: (1) tension fractures at all corners of rec-tangular wall openings; (2) successive tension fracturestrending toward the horizontal as distance from thewall opening increases; and (3) fracture extension orhook-shaped tension fractures between successive ten-sion fractures. Part-two site-specific intensity observa-tion questions (Table 2) are designed to determine ifpotentially damaging structure vibrations equaled orexceeded MMI IV (Table 1). It is unlikely for vibra-tory tension fractures to have formed or existing frac-tures to have been extended if MMI was #IV.

ACKNOWLEDGMENTS

We would like to acknowledge and thank the threereviewers, who provided valuable insight and con-structive suggestions and comments.

REFERENCES

ANDERSON, E. M., 1951, The Dynamics of Faulting and Dyke For-mation with Applications to Britain, 2nd ed.: Oliver andBoyd, Edinburgh, U.K.

AUDELL, H. S., 2004, Field Guide to Crack Patterns in Buildings, aGuide to Residential Building Cracks Caused by GeologicHazards: Special Publication No. 16, Association of Engineer-ing Geologists, Denver, CO. Southern California Section, 71 p.

BOLT, B., SOMERVILLE, P., ABRAHAMSON, N., ZERVA, A., 2004,Workshop Proceedings: Effects of Earthquake-induced transi-ent ground surface deformation on at-grade improvements,CUREE Publication No. EDA-04, 41 p.

CAPRIO, M.; TARIGAN, B.; WORDEN, C. B.; WIEMER, S.; AND

WALD, D. J., 2015, Ground motion to intensity conversionequations (GMICEs): A global relationship and evaluation ofregional dependency: Bulletin Seismological Society America,Vol. 105, No. 3, pp. 1476–1490.

DENGLER, L. A. AND DEWEY, J. W., 1998, An intensity survey ofhouseholds affected by the Northridge, California, earthquakeof 17 January, 1994: Bulletin Seismological Society America,Vol. 88, No. 2, pp. 441–462.

DE ROSSI, M. S., 1883, Programma dell'osservatorio ed archive cen-trale geodinamico: Bollettino del Vulcanismo Italiano, Vol. 10,pp. 3–124.

DEWEY, J. W.; REAGOR, B. G.; DENGLER, L.; AND MOLEY, K., 1995,Intensity Distribution and Isoseismal Maps for the Northridge,California, Earthquake of January 17, 1994: U.S. GeologicalSurvey Open-File Report 95-92, 35 p.

GRIFFITH, A. A., 1924, Theory of rupture. In Biezeno, C. B. andBurgers, J. M. (Editors), Proceedings of the First InternationalCongress on Applied Mechanics: Delft, Netherlands, Walt-man, pp. 55–63.

GUDMUNDSSON, A., 2011, Rock Fractures in Geological Processes:Cambridge University Press, Cambridge, U.K., 578 p.

MERCALLI, G., 1902, Sulle modificazioni proposte alla scala sismicaDe Rossi-Forel: Bollettino della Società Sismologica Italiana,Vol. 8, pp. 184–191.

NEUMANN, F., 1954, Earthquake Intensity and RelatedGround Motion: University of Washington Press, Seattle,WA, 77 p.

ORIARD, L. L., 1999, The Effects of Vibrations and EnvironmentalForces, a guide for the investigation of structures: InternationalSociety of Explosive Engineers, Solon, OH, 284 p.

POLLARD, D. D. AND FLETCHER, R. C., 2006, Fundamentals ofStructural Geology: Cambridge University Press, Cambridge,U.K., 500 p.

RICHTER, C. F., 1958, Elementary Seismology: W. H. Freeman andCompany, San Francisco, CA, 768 p.

SCHOLZ, C. H., 2002, The Mechanics of Earthquakes and Faulting,2nd ed.: Cambridge University Press, Cambridge, U.K., 404 p.

SISKIND, D. E.; CRUM, S. V.; AND PLIS, M. N., 1993, Blast Vibra-tions and Other Potential Cause of Damage in Homes near aLarge Surface Coal Mine in Indiana: U.S. Bureau of MinesReport of Investigations 9455.

SISKIND, D. E.; STAGG, M. S.; KOPP, J. W.; AND DOWDING, C. H.,1980, Structure Response and Damage Produced by GroundVibration from Surface Mine Blasting: U.S. Bureau of MinesReport of Investigations 8507.

STAGG, M. S.; SISKIND, D. E.; STEVENS, M. G.; AND DOWDING, C. H.,1984, Effects of Repeated Blasting on a Wood-Frame House: U.S. Bureau of Mines Report of Investigations 8896, 82 p.

STIENBRUGGE, K. V., 1970, Earthquake damage structuralperformance in the United States. In Wiegel, R. L. (Editor),Earthquake Engineering: Prentice-Hall, Inc., Ch. 9,Englewood Cliffs, NJ, pp. 167–226.

Table 2. A list of questions for estimating the potential for fractureformation and extension.

MMI

Did the walls creak loudly? IVDid you hear rattling of dishes, windows, or doors or see

hanging objects swing?IV

Were any small objects overturned or did you see doorsswing?

V

Any hanging pictures tilted, out of place, or fallen? VAny cracked windows? VAny light furniture overturned or heavy furniture

displaced?VI



STOVER, C. W. AND COFFMAN, J. L., 1993, Seismicity of the UnitedStates, 1568–1989 (Revised): U.S. Geological Survey Profes-sional Paper 1527, 418 p.

TALWANI, P., 2014, The impact of the early studies following the1886 Charleston earthquake on the nascent science of seismol-ogy: Seismological Research Letters, Vol. 85, No. 6, pp.1366–1372.

TRANSPORTATION RESEARCH BOARD, 1997, Dynamic Effects of PileInstallations on Adjacent Structures: National CooperativeHighway Research Program Synthesis 253.

WALD, D. J.; DENGLER, Q. L.; AND DEWEY, J., 1999a, Utilization ofthe Internet for rapid community intensity maps: Seismologi-cal Research Letters, Vol. 70, pp. 680–697.

WALD, D. J.; QUITORIANO, V., HEATON, T. H.; AND KANAMORI, H.,1999b, Relationships between peak ground acceleration, peakground velocity and modified Mercalli intensity in California:Earthquake Spectra, Vol. 15, No. 3, pp. 557–564.

WALD, D. J.; QUITORIANO, V.; HEATON, T. H.; KANAMORI, H.;SCRIVNER, C. W.; AND WORDEN, C. B., 1999c, TriNet “Shake-Maps”; rapid generation of peak ground motion and intensitymaps for earthquakes in Southern California: EarthquakeSpectra, Vol. 15, No. 3, pp. 537–555.

WOOD, H. O. AND NEUMANN, F., 1931, Modified Mercalli intensityscale of 1931: Bulletin Seismological Society America, Vol. 21,pp. 277–283.



Geologic and Geotechnical Factors ControllingIncipient Slope Instability at a Gravel Quarry,

Livermore Basin, California

PHILIP L. JOHNSON1

PATRICK O. SHIRESTIMOTHY P. SNEDDON

Cotton, Shires and Associates, 330 Village Lane, Los Gatos, CA 95030

Key Terms: Slope Instability, Gravel Quarry, Land-slide, Stratigraphy

ABSTRACT

Mine pit slopes at Arroyo del Valle Quarry in theLivermore Basin of northern California expose lateQuaternary sandy gravel. In the deep subsurface, thegravel unconformably overlies gently folded lacustrinesediments. Within the lacustrine sediments, a bed ofsheared, unoxidized clay overlies a marl bed, forminga distinctive marker bed couplet. Structure contourson this marker bed show an anticline and a synclinein the vicinity of the quarry. These northwest-strikingfolds are aligned parallel to regional Quaternary foldand thrust belt structures. Where the marker bed dipstoward the pit, slope inclinometers consistentlydeflected toward the pit at the depth of the unoxi‐dized clay. Where the marker bed dips away from thequarry pit, no slope inclinometer deflections wererecorded. Thus, slope instability was controlled by thesite stratigraphy, geologic structure, and the locationof quarry slopes relative to that geologic structure.High pore-water pressures within the unoxidized clayalso contributed to slope instability. Shearing of theunoxidized clay occurred prior to excavation of thequarry pit, and the resulting low residual strength ofthis high-plasticity clay made it particularly vulner‐able to incipient landsliding when lateral confinementwas removed during excavation of the quarry pit.Analysis of the critical region between the quarry pitand the anticline axis showed that the static factor ofsafety remained below 1.5. Seismic displacementanalyses indicated that moderate to large displace‐ments would be anticipated. Thus, depressurizationwells and an earth-fill buttress were designed andimplemented to mitigate deep-seated slope instability.

INTRODUCTION

Mine slope stability has been studied extensively atsites at which mine slopes expose fractured rock, andthe application of rock mechanics to the study ofmine slopes has aided in the design of stable slopesfor long-term mine reclamation and short-term slopestability during mining operations (Hoek and Karzu-lovik, 2000; Wyllie and Mah, 2004). In some cases,rock slopes in open pit mines have experienced large,fast-moving failures (Pankow et al., 2014) that presentsignificant challenges to mining operations.By contrast, the stability of gravel quarry slopes

that expose Quaternary sediments has received lessattention. However, as mining of aggregate resourcesfrom basins within large metropolitan areas continuesand urban development encloses the mined lands, thestability of gravel quarry slopes has become a signifi-cant concern (Doughton, 2009). Modern miningmethods have allowed extraction of aggregate andother resources to significant depths within sedimen-tary basins, and dewatering systems have allowedmining of aggregate well below the groundwatertable. Thus, relatively deep quarry excavations maybe found locally within populated regions. Wherethe mine pits expose coarse granular materials, pitslope stability should be a simple function of slopeangle, slope height, and (relatively high) materialstrengths. However, where the geologic conditionsare more complex, slope stability may be more chal-lenging to achieve.Arroyo del Valle Quarry is a gravel quarry located in

the Livermore Basin of northern California (Figure 1).Sand and gravel have been mined extensively in theLivermore Basin since the early 20th century for useas aggregate (Goldman, 1964; Dupras, 1999). Thequarry is the easternmost of a series of gravel pitswithin the southern portion of the basin. Once miningof an individual pit is completed and dewatering sys-tems are shut down, the pits are allowed to floodslowly by groundwater seepage, becoming artificial1Corresponding author email: [email protected].


lakes. Suburban residential development within theLivermore Basin has grown extensively since the1980s and has approached the margins of the formergravel pits. At Arroyo del Valle Quarry, residentialdevelopment extends up to the northeast boundary ofthe quarry property, currently within approximately125 ft of the quarry pit.The Arroyo del Valle Quarry is located within a

geologically complex setting that may appear to berelatively simple at the surface and within the shallowsubsurface. The mine pit slopes expose dense sandygravel that appears to remain stable at slope inclina-tions of 2H:1V (26.6u) or greater. However, slopeinclinometers installed around the mine pit haverecorded deflections at depth, below the gravelexposed in the quarry pit. In this article, the uniquesite stratigraphy, geologic structure, and hydrogeologyfound in the subsurface at Arroyo del Valle Quarryand the impact of this unique geology on the stabilityof the quarry slopes are described. In addition, ana-lyses of static and seismic slope stability and the designof mitigation measures that were implemented toachieve long-term stability of the former quarry pitslopes are described.

GEOLOGIC SETTING

The roughly east-west–trending late QuaternaryLivermore Basin (Figure 2) is filled with non-marineclastic sediments (Barlock, 1989; Helley and Graymer,1997). During Miocene to early Pleistocene time, theancestral Livermore Basin (Unruh et al., 1997)extended beyond the limits of the late Quaternarybasin, and upper Miocene to lower Pleistocene sedi-mentary rocks of the Sycamore, Tassajara, and Liver-more formations are exposed in the hills that surroundthe late Quaternary basin (Andersen et al., 1995). Thesediments that fill the late Quaternary LivermoreBasin consist of Pleistocene to Holocene alluvial fan,

terrace, and floodplain deposits that overlie the Liver-more Formation (Helley and Graymer, 1997). Earlywork by the California Department of WaterResources (CDWR, 1974) included characterizationof the alluvial aquifers of the late Quaternary Liver-more Basin. These aquifers contain a large volume ofcoarse-grained sediment that was derived from theDiablo Range to the south and transported by alluvialprocesses northward into the basin. Fine-grainedlacustrine sediments form aquitards that cap the aqui-fer units. Ehman et al. (2004) interpreted electric logsand cuttings logs from 37 water wells in the centralportion of the basin (north of Arroyo del ValleQuarry) and used sequence stratigraphic methods tocharacterize the aquifer stratigraphy and to identifyseveral depositional sequences that are bounded byunconformities.Crane (1995, 2007) mapped the northern, eastern,

and western boundaries of the late Quaternary basinas faults that bound the hills that surround the basin.He interpreted the northern boundary of the basin asa southwest vergent thrust fault that bounds theMount Diablo region (Figure 2), while complex thrustand strike slip faulting characterize the hills to the eastand west. Thus, the late Quaternary Livermore Basinis flanked by faulted uplifts that expose tilted andfolded Miocene to lower Pleistocene rocks.Sawyer and Unruh (2004) identified a southwest

vergent fold and thrust belt, the Mount DiabloFold and Thrust Belt (MDFTB), that encompassesthe Mount Diablo region and the Livermore Basin.Compressional deformation of the MDFTB hasbeen attributed to a restraining left stepover betweenthe right lateral Greenvillle and Concord faults(Unruh and Lettis, 1998). Unruh et al. (2007) viewthe Livermore Basin as deformed by active foldingand southwest vergent thrust faulting behind theleading edge of the MDFTB, which extends asfar southwest as the Verona and Williams faults(Figure 2). Several northwest striking and activelygrowing folds were identified within the basin andsurrounding hills (Sawyer and Unruh, 2004). Thus,compressional tectonics of the MDFTB appear tobe responsible for local deformation of LivermoreBasin sediments.

GEOLOGY OF THE ARROYO DEL VALLEQUARRY AREA

Site Geomorphology

The geomorphology of the Arroyo del Valle Quarrysite and surrounding area was mapped using stereopairs of historic aerial photographs that pre-date theexcavation of the quarry (Figure 3). The geomorphic

Figure 1. Location map for the Livermore Basin and surroundinguplifts.

Johnson, Shires, and Sneddon


Figure 2. Simplified map of the Livermore Basin and major mapped faults, modified from Wagner et al. (1990). The labeled primary streamchannels within the basin are Arroyo del Valle (AV) and Arroyo Mocho (AM).

Arroyo Del Valle Quarry


setting of the southern Livermore Basin was shaped byfluvial deposition along Arroyo del Valle, a low-sinu-osity (braided) stream. The stream channel is flankedby uplifts to the northeast and southwest. To thenortheast, a subdued intrabasinal uplift is centeredon the Livermore anticline that may be associatedwith a blind reverse or thrust fault (Sawyer and Unruh,2004). To the southwest, tilted beds of the LivermoreFormation flank the southwest margin of the basin.Between these two uplifts, Arroyo del Valle flowstoward the northwest. A series of fluvial terracesascend from the channel of Arroyo del Valle (youngestto oldest) and up the flanking uplifts northeast andsouthwest of the channel (Figure 3). The Arroyo del

Valle Quarry study area is located within the channeland adjacent alluvial terraces.

Site Stratigraphy

The quarry cut slopes expose only the shallowestportion of the basin stratigraphy. As a result of thelack of surface exposure, the primary tool for charac-terization of the deeper stratigraphy was subsurfaceexploration. Thus, 39 continuously cored borings andtwo large-diameter bucket auger borings were drilledfor this study. The core samples were logged in detail,sedimentary characteristics were described, and mar-ker beds were identified. Correlations between borings

Figure 3. Photogeologic map showing the geomorphic setting of the Arroyo del Valle Quarry area, based upon interpretation of stereo pairsof historic aerial photographs. The polygon in the upper left portion of the map shows the area of Figure 7.



were added to detailed geologic cross sections, andstructure contour maps were compiled using markerbed elevations.

Upper Sandy Gravel Deposits

The quarry pit slopes expose the upper sandy graveldeposits. These poorly sorted, coarse sediments consistof rounded to subangular cobbles, pebbles, and verycoarse to fine sand with local boulders and minorlow-plasticity silt and clay. Where exposed in quarrypit walls, local horizontal stratification and pebbleimbrication are visible. Individual beds of sandygravel have sharp, irregular lower contacts with well-developed erosional relief; this is particularly notice-able where the sandy gravel overlies local fine-grainedbeds (Figure 4). The sandy gravel unit varies in thick-ness from approximately 90 to 100 ft (27–30 m). Theyoungest gravel deposits exposed in the area of thehistoric stream channel are mostly gray, while olderbeds are oxidized to yellowish brown (Munsellcolor 10YR).

Interbedded with the sandy gravel are local bedsof yellowish brown (10YR) clayey silt to silty claythat are generally less than 6 ft (2 m) in thickness(Figure 5). Based upon subsurface exploration, localfine-grained beds within the sandy gravel are mostlydiscontinuous, and where correlation of individualbeds can be accomplished between boreholes, theyappear to be horizontal or subhorizontal.

Lower Fine-Grained Deposits

In borings surrounding the quarry pit, fine-graineddeposits were encountered below the upper sandygravel deposits. The fine-grained strata consist of dis-tinct beds of oxidized silty clay, unoxidized clay,marl, silty fine sand, and sandy silt. The oxidized

clay displays variable coloration from light yellowishbrown to strong brown (Munsell colors range from2.5Y to 7.5YR). Though mostly lacking in stratifica-tion, the oxidized clay is locally laminated. The resultsof Atterberg Limits testing of the oxidized clay indi-cate that the average liquid limit and plasticity indexare 37 and 15 (respectively), which correlate withlow-plasticity clay. The color and texture of the oxi-dized clay strongly contrast with those of the underly-ing unoxidized clay.The unoxidized clay is greenish gray to olive gray

(5G to 5Y), with local lamination and significantlylower silt content (,10 percent silt) than the overlyingoxidized clay. The results of laboratory testing of theunoxidized clay indicate that the average liquid limitand plasticity index are 75 and 48 (respectively), whichplaces this clay in the high-plasticity range. In a few ofthe cores drilled northeast of the quarry pit, small pele-cypod and gastropod shells and shell fragments wereidentified within the unoxidized clay. This strati-graphic unit displays evidence of intense shearingthat includes numerous highly polished surfaces anddevelopment of clay gouge (Figure 6). The unoxidizedclay overlies a white to light gray marl that is highlyreactive to dilute hydrochloric acid. Below the marlis fine silty sand to sandy silt with interbedded siltyclay; these strata are more permeable than the overly-ing unoxidized clay and marl.The sheared, unoxidized clay and underlying marl

form a distinctive and laterally persistent stratigraphiccouplet. This couplet was traced across the quarry siteand adjacent area to the northeast and is designated asthe unoxidized clay-marl marker bed couplet.

Interpretation of Sedimentary DepositionalEnvironments

Upper Sandy Gravel Deposits

The sandy gravel to gravelly sand encountered inthe upper 100 ft (30 m) in the study area appears to

Figure 4. Sandy gravel over a fine-grained bed exposed in theshallow portion of the quarry pit wall.

Figure 5. A fine-grained bed that pinches out laterally (exposed inthe quarry pit wall).



have been deposited in a braided stream environmentsimilar to modern Arroyo del Valle. The predomi-nantly coarse texture of this deposit is consistent witha high-velocity flow regime, and the horizontal stratifi-cation and pebble imbrication are consistent withdeposition within a braided stream environment(Bridge and Lunt, 2006; Miall, 2010). Local discontin-uous fine-grained beds within the upper sandy gravelwere deposited at low flow velocity in an interchannelfloodplain setting.

Lower Fine-Grained Deposits

The lower fine-grained deposits are interpreted aslacustrine sediments. The primary constituents arelocally laminated clay with low silt content, marl,and silty clay. These sediments overlie silty fine sandand sandy silt. Though there is no single characteristicthat defines a lacustrine depositional environment,typical indications include abundant fine-graineddeposits, laterally continuous thin beds, lamination,freshwater fauna, organic-rich sediments, and evapor-ite or carbonate mineralogy (Picard and High, 1972,1981; Platt and Wright, 1991; and Carroll and Bohacs,1999). The laterally persistent fine-grained deposits,local lamination, carbonate-rich sediments, and localmolluscan fauna of the lower fine-grained sedimentarystrata are consistent with a lacustrine setting.

Interpretation of Geologic Structure and Stratigraphy

The study of geologic structure of the lower fine-grained strata has been greatly aided by the recogni-tion of the unoxidized clay-marl marker bed couplet.Through detailed logging of core borings and careful

correlation of the marker bed couplet between theseborings, structure contours were drawn on the top ofthe marker bed couplet (Figure 7). The structure con-tour map shows a northwest-plunging anticline in thenorthern portion of the study area and a northwest-plunging syncline in the southern portion. These foldsalign well with the orientation of other folds and thrustfaults in the MDFTB. This implies that folding of thelacustrine sediments is related to northeast-southwestshortening within the MDFTB.Though the lower fine-grained sedimentary deposits

had not been recognized at the time of quarry excava-tion, the pit was excavated above the southwest-dip-ping limb of the anticline (Figure 7). A cross sectionacross the quarry and northeast flank of the quarryshows the site stratigraphy and geologic structure(Figure 8). In the southwest limb of the anticline, themaker bed couplet strikes roughly 303u to 313u(N47W to N57W) and dips approximately 2u to thesouthwest. The dip of the northeast limb is approxi-mately 4u to the northeast. By contrast, the overlyingupper sandy gravel deposits do not appear to befolded. Based upon this angular discordance and theapparent truncation of the lower fine-grained stratain the northeastern portion of the study area, itappears that an unconformity separates the uppersandy gravel from the lower fine-grained deposits.Thus, we designate them as separate stratigraphicsequences. Though the exact ages of these twosequences is poorly constrained, based upon the topo-graphic setting of these deposits and the contrast withthe steeply tilted and folded Livermore Formationbeds that are exposed in the nearby hills south of thebasin, the lower fine-grained strata appear to be Pleis-tocene in age, and the upper sandy gravel strata appearto be late Pleistocene to Holocene in age.

HYDROGEOLOGY OF THE ARROYO DELVALLE QUARRY AREA

Based upon subsurface exploration, three separateaquifers were identified at the Arroyo del Valle Quarrysite. The shallowest aquifer is within the upper portionof the sandy gravel and above the discontinuous fine-grained beds. Below the discontinuous fine-grainedbeds is a semi-confined aquifer within the lower portionof the sandy gravel. The oxidized clay, unoxidized clay,and marl form an aquitard below the sandy gravel. Thebeds of sand and sandy silt below the unoxidized clayand marl constitute a confined aquifer. Within thestudy area, these three aquifer units are designated asthe upper, middle, and lower aquifers (Figure 9).Though deeper aquifers may be present below thedesignated lower aquifer, this study did not includecharacterization of deeper aquifer stratigraphy.

Figure 6. A core sample exposing a polished surface within thesheared unoxidized clay.



Long-term monitoring of vibrating wire piezometersinstalled in the lower aquifer indicates that piezometricsurface elevations typically ranged from 75 to 102 ft(23–31 m) above the top of the lower aquifer. By con-trast, the middle aquifer water levels have remainedlower than those in the lower aquifer. Thus, the piezo-meter data indicate an upward gradient from the loweraquifer to the middle aquifer (Figure 9). Pore pressuremeasurements from piezometers installed in the unox-idized clay are intermediate between those of the mid-dle and lower aquifers. This upward pore pressuregradient has caused elevated pore pressures withinthe intervening unoxidized clay interval.

SLOPE INSTABILITY

The instability of the quarry pit slopes is not readilyapparent at the ground surface. As a result of the rela-tively small magnitude of displacement, roughly 4 to 5in. (10–12 cm), typical landslide-related landforms(scarps, grabens, and hummocky topography) havenot developed. The initial evidence for instability of

the quarry slopes came from linear cracks that formedin roadways located approximately 125 ft (38 m)northeast of the quarry pit (Figure 10). Following theinitial observation of pavement distress in 2001, multi-ple slope inclinometers were installed around thequarry pit from 2002 to 2006.Based upon detailed logging of continuous core

samples from the slope inclinometer borings, discrete(landslide-type) deflections in the slope inclinometerswere consistently found to correspond to the depth ofthe sheared unoxidized clay. This is shown in adata plot from a slope inclinometer located near thenortheast flank of the quarry pit combined with thestratigraphic column from the same slope inclinometerboring (Figure 11). Even where the quarry pit wallswere locally steeper than 2H:1V (26.6u), the slopeinclinometer casings deflected only within the unoxi-dized clay and not within the upper sandy gravelsequence. Thus, the site stratigraphy exerts strong con-trol on the instability of the quarry slopes.Slope instability is also controlled by geologic struc-

ture. As shown in Figure 7, the slope inclinometer

Figure 7. Site map with structure contours drawn on the top of the unoxidized clay-marl marker bed couplet. The cross-section line indicatesthe location of the geologic cross section (Figures 8, 9, and 13).



displacement vectors are consistently oriented in adown-dip and downslope direction. At the northeastmargin of the quarry, in the area between the anticlineaxis (Figure 7) and syncline that underlies the pit floor,slope inclinometer monitoring data show deflections ina down-dip direction, toward the pit. We refer to thiscritical sliding block where the sheared unoxidizedclay dips toward the quarry pit as the Northeast Block.The anticline axis that forms the updip margin of theNortheast Block acts as a natural barrier to retrogres-sive movement farther to the northeast. Across theanticline axis, the sheared unoxidized clay bed dipsaway from the pit slope, and the slope inclinometerson this fold limb have not deflected.The monitoring results from a representative slope

inclinometer located on the upper bench on the north-east side of the quarry provide a useful record of dis-placement of the northeast block over a period ofseveral years (Figure 12), although it does not includethe estimated several inches of movement thatoccurred prior to slope inclinometer installation. Forthe purposes of this study, the rates of displacementwere averaged and annualized to allow meaningfulcomparison of monitoring periods of varied dura-tions. Between January 2004 and April 2005, slope

inclinometer monitoring results showed that the dis-placement rates averaged 0.26 in. (6.6 mm) per year.Between April and September of 2005, the displace-ment rate increased to an average of 1.0 in. (25.4mm) per year. Between September 2005 and Septem-ber 2006, an average displacement rate of 0.14 in.(3.5 mm) per year was recorded. Thus, prior to imple-mentation of mitigation measures designed to addressslope instability at the quarry pit, the rate of move-ment within the Northeast Block had already slowed.The recent displacement that occurred on the north-

east flank of the quarry pit would be insufficient inmagnitude to produce the well-developed polished

Figure 8. Geologic cross section across the quarry site and adjacent area northeast of the quarry.

Figure 9. Hydrogeologic cross section showing the piezometricsurfaces of the middle and lower aquifers within the NortheastBlock. U, upper aquifer; M, middle aquifer; L, lower aquifer.



surfaces and clay gouge observed in the unoxidizedclay. In addition, the unoxidized clay was encounteredin an intensely sheared state in borings throughout thestudy area, including the fold limb that dips away fromthe quarry pit. Therefore, it appears that the unoxi-dized clay was sheared prior to excavation of thequarry pit.

Based upon the slope inclinometer data, it is clearthat the inclinometer deflections were a response tostress release and removal of lateral confinement dueto quarry pit excavation. This stress release triggereddownslope movement in the weak unoxidized claybed where it dips toward the quarry pit excavation.

The high pore-water pressures within the unoxidizedclay also contributed to instability of the NortheastBlock. The movement of the Northeast Block that fol-lowed pit excavation appears to represent incipientlandsliding because it did not develop into a largermagnitude failure.

GEOTECHNICAL ANALYSIS AND DESIGN OFMITIGATION MEASURES

Limit equilibrium and finite element slope stabilityanalyses were performed to assess the instability andfactor of safety (FS) of the Northeast Block. The

Figure 10. Map of ground cracks that were the first indication of incipient slope instability related to the quarry pit slope.



analyses indicated that the shear strength of the sandygravel was incrementally mobilized as displacementsoccurred within the sheared unoxidized clay. Thisexplains the decline in slope inclinometer deflectionover time. Under static conditions, the slopes shouldexperience little displacement once the gravel strengthhas been fully mobilized. However, under seismic con-ditions, the Northeast Block would be expected toreactivate, potentially leading to much larger displace-ments. Consequently, implementation of mitigationmeasures was necessary to provide for the long-termstability of the Northeast Block during future seismicevents.Though several methods of mitigation were consid-

ered, an earth-fill buttress repair was chosen for itssimplicity and reliability. Limit equilibrium slope sta-bility analyses showed that the pressure exerted by

Figure 12. Graph of cumulative displacement over time at theunoxidized clay interval from a representative slope inclinometeron the northeast flank of the quarry pit.

Figure 11. Slope inclinometer data plot and stratigraphic column from a slope inclinometer located near the northeast flank of the quarry pit.



water within the quarry pit acted to resist movement ofthe Northeast Block. If the pit were drained to allowplacement of an earth-fill buttress, displacementswithin the Northeast Block would likely accelerateuntil the gravel strength was fully mobilized. Thus,mitigation measures designed to stabilize the North-east Block had to maintain short-term stability duringplacement of the buttress.

A two-dimensional limit equilibrium slope stabilityanalysis was performed on a representative cross sec-tion using Spencer’s method (Wright, 1975) to evalu-ate the stability of the Northeast Block duringplacement of the buttress and to calculate the potentialseismic displacements after completion of the buttress.The key parameters for this slope stability analysisincluded topographic profile, rupture surface geome-try, material shear strengths, unit weights, and piezo-metric surface elevations. The material shearstrengths were determined by laboratory testing (con-solidated-undrained triaxial compression and tor-sional ring shear testing) and back-calculationanalyses. The shear strengths for the sandy gravels inparticular were difficult to estimate because of difficul-ties in obtaining undisturbed samples of the sandygravel material for triaxial shear strength testing dueto the large clast size. Therefore, in addition to triaxialcompression testing, back-calculation analyses of thesandy gravel material strengths were performed onareas of the quarry where the sandy gravel wasexposed in near-vertical cut faces of approximately35 ft in height. Several small failures had occurred inthese cut faces, and based on these analyses the labora-tory-derived peak shear strength was actually lowerthan the back-calculated shear strengths. Correlationsbetween gravel particle size and friction angle werealso considered in developing the shear strengths.Based on these sources, representative shear strengthsfor the sandy gravel material were selected as shownin Table 1. The material unit weights were determinedby laboratory testing of undisturbed samples. Theshear strengths and unit weights are compiled inTable 1. The topographic profile and subsurface geo-metry were taken from a representative geologic crosssection (Figure 8), and the piezometric surface eleva-tions were derived from monitoring of piezometerswithin the study area.

In addition to the stratigraphic units previouslydescribed, two additional materials were placed as fillfor the earth-fill buttress: K-in.– (13-mm–) diameterpea gravel and compacted pit run fill. The pea gravelwas produced by processing of material excavatedfrom one of the other nearby gravel quarries. The pitrun material consisted of sandy gravel excavatedfrom a borrow area at the southwest margin of the

quarry pit and placed as engineered fill without addi-tional processing.The piezometric surface elevations were determined

by long-term monitoring of staged vibrating wirepiezometers that were installed within the NortheastBlock and elsewhere around the study area. The piezo-meter sensors were divided into three groups: “uppersensors,” “mid-sensors,” or “lower sensors,” based onthe aquifer stratigraphy and depth of sensor placement(Figure 9). The upper aquifer and discontinuous fine-grained beds within the gravel were assigned the“upper sensor” piezometric surface. The middle aqui-fer was assigned the “mid-sensor” piezometric surface.The unoxidized clay and lower aquifer were assignedthe “lower sensor” piezometric surface. For design ofthe buttress, the highest recorded levels for each sensorgroup were used to estimate the piezometric surfacefor each hydro-stratigraphic unit.The initial analyses indicated that the calculated FS

values ranged from 1.22 to 1.70 (depending on the rup-ture surface analyzed) using peak shear strengths formost materials and residual shear strength for rupturesurface gouge, as shown in Table 1. The unexpectedfinding that FS remained above 1.0 while active land-slide-type movement was apparently underway is bestexplained by the high strength of the sandy gravelmaterial. The slope movement response to the quarryexcavation was concentrated on the unoxidized clay,while the relatively strong sandy gravel near the toeof the slope provided shearing resistance to this deepmovement in the unoxidized clay. Because the amountof movement in the unoxidized clay was relativelysmall, displacement did not reach the point at whichthe peak shear strength of the sandy gravel near thetoe of the slope was fully mobilized. Therefore, whenusing peak strength values for the sandy gravel in theslope stability analyses, the overall static FS of theslope (including the impact of the resisting sandygravel) remained above unity, although local discrete

Table 1. Summary of static material properties.

Material

UnitWeight(pcf)

Cohesion(psf)

FrictionAngle (u)

Sandy gravel 139 200 45Clay beds within sandy gravel 127 1,500 27Oxidized clay 130 1,000 24Sheared unoxidized clay 121 0 11Unoxidized clay gouge

(residual shear strength)121 0 Non-linear

(,6)Lower confined aquifer 133 1,300 28Compacted pea gravel fill 134 0 Non-linear

(,40 to 51)Compacted pit run fill 145 100 39



shearing movement in the unoxidized clay occurred inresponse to the excavation. Once sufficient displace-ment within the unoxidized clay had occurred, theshearing resistance of the sandy gravel was partiallymobilized, and displacement within the unoxidizedclay slowed. These analyses expose an oversimplifyingassumption typically used for limit equilibrium slopestability analysis, that the FS is uniform along all por-tions of the rupture surface analyzed whether the rup-ture surface passes through weak clay at residualstrength or strong gravel with no previous history ofshearing.In addition to the static slope stability analyses that

were conducted, seismic displacements were also esti-mated using a Newmark-type sliding block analysis(Jibson and Jibson, 2003). Three active faults (theCalaveras, Greenville, and Hayward faults, located9.3 to 19.1 km from the site) were identified as mostlikely to affect the site, with moment magnitudes ran-ging from 6.6 to 7.1. A design target response spec-trum was developed for the site based upon multipleattenuation relationships with modifications for near-fault effects. Seven strong motion records (PacificEarthquake Engineering Research Center, 2000) wereselected that were representative of the expectedground motion at the site and scaled as necessary toobtain a relatively close fit between the averageresponse spectrum of the seven motions and the targetdesign response spectrum. One-dimensional equiva-lent-linear seismic response analysis (SHAKE2000,2000) was performed on three representative strati-graphic columns to obtain horizontal equivalent accel-eration time histories at the depth of sliding, whichwere then used in conjunction with yield accelerationsfrom pseudo-static slope stability analyses to estimateseismic displacements (Koragappa et al., 2004).Based upon these analyses, the movement of the

Northeast Block resulted from stress relief upon exca-vation of the free face at the northeast margin of thequarry pit. The previously sheared clay that was at ornear residual shear strength prior to quarry pit excava-tion and the elevated pore pressures within the criticalsliding surface were the primary factors resulting in themovement detected initially as ground cracks in

nearby roads and later by the slope inclinometers.Using partially mobilized upper gravel strengths,piezometric influence from the lower aquifer, andpotential movement that extends northeastward tothe anticline axis, our analysis showed that an indus-try-accepted static FS of greater than 1.5 and seismicdisplacements of less than 15 cm (6 in.) could beachieved by partial filling of the quarry pit with engi-neered fill to an elevation of 390 ft (119 m) above sealevel (Figure 13).

IMPLEMENTATION OF MITIGATIONMEASURES

Between September 2006 and May 2007, 40 depres-surization wells were installed on the northeast flankof the quarry to relieve pore-water pressures withinthe lower aquifer. Based upon the site hydrogeology,lowering of the piezometric surface within the loweraquifer was expected to reduce pore pressures withinthe overlying unoxidized clay and, in turn, to improvethe static FS of the Northeast Block. Immediatelyupon installation, the depressurization wells flowedat the surface under artesian conditions. As antici-pated, the artesian flow resulted in lowering of thepiezometric surface of the lower aquifer to roughlythe elevation of the well discharge pipes. The piezo-metric surface at a representative piezometer withinthe Northeast Block (Figure 14) shows this initialdrop. Following installation of the depressurizationwells, the displacement rates in slope inclinometer SI-2 decreased to an average of 0.04 in. (1 mm) per yearbetween September 2006 and May 2008 (Figure 12).During May 2008, submersible pumps with float-acti-vated controllers were installed in 10 of the wells.Within 9 days of pump activation, the elevation ofthe piezometric surface within the lower aquifer (Fig-ure 14) had dropped by 43 ft (13 m). Within 3 monthsafter pump activation, the piezometric surface eleva-tion had dropped an additional 20 ft (6 m).Following activation of the well pumps and prior to

placement of the fill buttress, the quarry pit (artificiallake) was partially drained. Between June 2008 andAugust 2008, the water surface elevation of the lake

Figure 13. The geologic cross section annotated to show the phases of grading during placement of the earth-fill buttress. Thedepressurization wells are screened in the lower aquifer. U, upper aquifer; M, middle aquifer; L, lower aquifer. The dashed horizontal linerepresents the elevation of the lower aquifer piezometric surface after installation of the depressurization wells.



was lowered approximately 13 ft (4 m) by pumping.During this period of time, some slope inclinometerswithin the Northeast Block showed a modest accelera-tion in deflection rate. Between June 2008 and Novem-ber 2008, the average slope inclinometer displacementrate was 0.18 in. (4.5 mm) per year (Figure 12). Thus,even partial draining of the lake did affect slope stabi-lity, as predicted.

In order to temporarily stabilize the NortheastBlock before the quarry pit was fully drained, a tem-porary buttress was placed on the lake bottom. Thetemporary buttress consisted of pea gravel fill thatwas dropped through the water column and onto thelake bottom using a floating conveyor belt system.Figure 13 shows the buttress in cross-section view.Prior to construction, analysis of the loose dumpedpea gravel fill indicated that it could be susceptible toliquefaction associated with strong seismic groundmotions, so it was necessary to compact the pea gravelfill to provide long-term stability under future seismicloading conditions. Once the lake was drained andthe pea gravel fill could be dewatered, it was excavatedin slots oriented perpendicular to the slope and notmore than 100 ft (30 m) wide in order to preserve thebuttressing effect. The pea gravel was placed back

into the slots in lifts and compacted to 95 percent rela-tive compaction. After recompaction was completed,the elevation of the top of the pea gravel fill was364.5 ft (111.1 m).Following placement and recompaction of the pea

gravel fill, drainage of the pit continued, and pit runmaterial was excavated from a shallow borrow areasouthwest of the pit and placed as compacted fillover the pea gravel using conventional grading meth-ods. As placement of the fill buttress progressed, theslope inclinometer displacement rate slowed to 0.045in. (0.11 mm) per year between November 2008 andApril 2009. The completed fill buttress reached thefinal elevation of 390 ft (118.9 m) during April 2009.Subsequent monitoring of site slope inclinometersand piezometers from 2009 through 2014 has shownno evidence of an increased rate of movement (Figure12) even though the piezometric surface of the loweraquifer has risen since pumping of the depressurizationwells ceased (Figure 14).

CONCLUSIONS

The unique stratigraphy, geologic structure, andhydrogeology at the Arroyo del Valle Quarry site

Figure 14. Piezometric surface elevation data from a representative piezometer installed in the confined lower aquifer on the northeast side ofthe quarry pit. (a) Initial drop in piezometric surface during installation of depressurization wells; (b) brief pumping test on a singledepressurization well; (c) well discharge to the quarry pit maintains a constant piezometric surface elevation of approximately 407 ft (theelevation of the well discharge lines); (d) significant lowering of the piezometric surface during pumping of 10 depressurization wells; (e)return to equilibrium conditions following cessation of well pumping; (f) slight rise of the piezometric surface as the lake surface elevationincreased and the well discharge lines were inundated.



and surrounding area were key factors that led to inci-pient failure of the northeast slope of the quarry pit(the Northeast Block). The highly sheared unoxidizedclay found in the subsurface was weakened to residualshear strength by other geologic processes prior toexcavation of the quarry pit, and this very low shearstrength made it vulnerable to instability once thequarry pit was excavated. The slope inclinometersdeflected only within the unoxidized clay interval;thus, the site stratigraphy strongly controlled incipientlandslide movement. Slope inclinometer displacementsoccurred only where the unoxidized clay dips towardthe quarry pit. Therefore, geologic structure also con-trolled incipient landsliding. Elevated pore-water pres-sures within the unoxidized clay appear to be related tothe high pore pressures within the underlying loweraquifer and the upward hydraulic gradient betweenthe lower and middle aquifers. Thus, the site hydro-geology also contributed to the instability of theNortheast Block.Based upon our analyses, we conclude that incipient

landsliding resulted from the site geologic and hydro-geologic conditions described above combined withstress relief related to excavation of the quarry pit.Initial deflection within the weak unoxidized clayeventually led to partial mobilization of the strengthof the overlying gravel and a decline in slope inclin-ometer deflection rate. However, the static FSremained below the industry-accepted minimum valueof 1.5, and unacceptably large displacements wereanticipated under seismic loading conditions. Thus, itwas necessary to design and implement mitigationmeasures to address the static and seismic stability ofthe Northeast Block.Because the pressure exerted by the water that filled

the former quarry pit provided some counterbalanceto the forces driving slope movement, the lake couldnot be drained without triggering significant accelera-tion in slope movement. Initially, the pore pressureswithin the unoxidized clay were lowered using depres-surization wells screened in the lower aquifer com-bined with pumping of those wells. Then atemporary buttress of pea gravel was placed on thefloor of the former quarry pit before the lake was com-pletely drained. Once drained, the pea gravel wasremoved in slots, compacted as it was placed backinto the slots, and capped with additional engineeredfill that was placed using conventional gradingmethods.Limit equilibrium analyses show that the completed

buttress should provide a FS of greater than 1.5 understatic conditions, even as pore pressures within theunoxidized clay rise. Displacements that could resultfrom strong seismic shaking were calculated at lessthan 6 in. (15 cm) with the earth-fill buttress in place.

Five years of post-construction monitoring haveshown no evidence of renewed slope inclinometerdeflection within the Northeast Block or elsewherearound the quarry site.

ACKNOWLEDGMENTS

This project was completed with the review, assis-tance, and guidance of professors Jonathan Bray,Scott Kiefer, and Matthew Mauldon. Geotechnicallaboratory testing was provided by Cooper TestingLabs, Inc. The authors wish to thank Dale Marcum,Joe Durdella, Jason Nichols, Jonathan Sleeper,Jamie Smith, and many others at Cotton Shires andAssociates, Inc., for their assistance during comple-tion of this project. We wish to thank MichaelHart, Douglas M. Yadon, and an anonymousreviewer for their helpful comments that improvedthis manuscript.

REFERENCES

ANDERSEN, D. W.; ISAACSON, K. A.; AND BARLOCK, V. E., 1995,Neogene evolution of the Livermore Basin within the Califor-nia Coast Ranges. In Fritsche, A. E. (Editor), Cenozoic Paleo-geography of the Western United States II: Pacific SectionSEPM Publication 75, pp. 151–161.

BARLOCK, V., 1989, Sedimentology of the Livermore Gravels(Miocene-Pleistocene), Southern Livermore Valley, California:U.S. Geological Survey Open File Report 89-131, 93 p.

BRIDGE, J. S. AND LUNT, I. A., 2006, Depositional models ofbraided rivers. In Sambrook Smith, G. H.; Best, J. L.; Bris-tow, C. S.; and Petts, G. E. (Editors), Braided Rivers: Interna-tional Association of Sedimentologists Special PublicationNo. 36, pp. 11–50.

CALIFORNIA DEPARTMENT OF WATER RESOURCES (CDWR), 1974,Evaluation of Groundwater Resources, Livermore and SunolValleys: CDWR Bulletin 118-2, 153 p.

CARROLL, A. R. AND BOHACS, K. M., 1999, Stratigraphic classifica-tion of ancient lakes: Balancing tectonic and climatic controls:Geology, Vol. 27, No. 2, pp. 99–102.

CRANE, R., 1995, Geology of the Mount Diablo region and EastBay hills. In Sangines, E. M.; Andersen, D. W.; and Buising,A. V. (Editors), Recent Geologic Studies in the San FranciscoBay Area: Pacific Section SEPM Special Publication, Vol.76, pp. 87–114.

CRANE, R., 2007, Geology of Central California: Pacific SectionAmerican Association of Petroleum Geologists, CD-ROMNo. 4, 237 p.

DOUGHTON, S., 2009, State warned gravel pit of slope instability 4years before landslide: Seattle Times, October 16, 2009.

DUPRAS, D., 1999, Representative industrial mineral mines of theSan Francisco Bay region: Sand and gravel, crushed rock,and limestone. In Wagner, D. L. and Graham, S. A. (Editors),Geologic Field Trips in Northern California: California Divi-sion of Mines and Geology Special Publication 119, pp.235–245.

EHMAN, K. D.; FIGUERS, S.; AND BLAKE, R., 2004, Preliminary stra-tigraphic evaluation, west side of the main basin, Livermore-Amador groundwater basin: unpublished consultant’s reportto Zone 7 Water Agency, 25 p.



GOLDMAN, H. B., 1964, Sand and Gravel in California, and Inven-tory of Deposits, Part B Central California: California Divi-sion of Mines and Geology Bulletin 180-B, 58 p.

HELLEY, E. J. AND GRAYMER, R. W., 1997, Quaternary Geology ofAlameda County and Surrounding Areas, California: U.S.Geological Survey Open File Report 97-97, 1:100,000 scale.

HOEK, E. AND KARZULOVIC, A., 2000, Rock mass properties for sur-face mines. In Hurstrulid, W. A.; McCarter, M. K.; and VanZyl, D. J. A. (Editors), Slope Stability in Surface Mining:SME, Littleton, CO, pp. 59–70.

JIBSON, R. W. AND JIBSON, M. W., 2003, Java Programs for UsingNewmark’s Method and Simplified Decoupled Analysis toModel Slope Performance During Earthquakes: U.S. Geologi-cal Survey, Open File Report 2003-005, CD-ROM.

KORAGAPPA, N.; BRAY, J. D.; MEYER, D.; AND TRAVASAROU, T.,2004, Seismic slope stability of a landfill on young bay mud:Proceedings Waste Tech Landfill Technology Conference,Dallas, TX, May 17–19, 2004, 22 p.

MIALL, A., 2010, Alluvial deposits. In James, N. P. and Dalrymple,R. W. (Editors), Facies Models 4: Geological Society ofCanada, GEOText 6, pP. 105137.

PACIFIC EARTHQUAKE ENGINEERING RESEARCH CENTER, 2000,Ground Motion Database: Electronic document, available athttp://peer.berkeley.edu

PANKOW, K. L.; MOORE, J. R.; HALE, J. M.; KOPER, K. D.;KUBACKI, T.; WHIDDEN, K. M.; AND MCCARTER, M. K.,2014, Massive landslide at Utah copper mine generates wealthof geophysical data: GSA Today, Vol. 24, pp. 4–9.

PICARD, M. D. AND HIGH, L. R., 1972, Criteria for recognizinglacustrine rocks. In Rigby, J. K. and Hamblin, W. K. (Edi-tors), Recognition of Ancient Sedimentary Environments:SEPM Special Publication, No. 16, pp. 108–145.

PICARD, M. D. AND HIGH, L. R., 1981, Physical stratigraphy ofancient lacustrine deposits. In Ethridge, F. G. and Flores, R.M. (Editors), Recent and Ancient Nonmarine Depositional

Environments: SEPM Special Publication, No. 31, pp.108–145.

PLATT, N. H. AND WRIGHT, V. P., 1991, Lacustrine carbonates:Facies models, facies distributions and hydrocarbon aspects.In Anadon, P.; Cabrera, L.; and Kelts, K. (Editors), Lacus-trine Facies Analysis: International Association of Sedimentol-ogists Special Publication, No. 13, pp. 57–74.

SAWYER, T. L. AND UNRUH, J. R., 2004, Characterization of LateQuaternary Deformation of the Mt. Diablo Fold-and-ThrustBelt, Eastern San Francisco Bay Area, California: EOS Trans-actions, Vol. 85, Fall Meeting Supplement, Abstract T32C-02.

SHAKE2000, 2000, A Computer Program for the 1-D Analysis ofGeotechnical Earthquake Engineering Problems: Gustavo A.Ordonez, GeoMotions, LLC, Lacy, WA.

UNRUH, J. R.; DUMITRU, T. A.; AND SAWYER, T. L., 2007, Couplingof early Tertiary extension in the Great Valley forearc withblueschist exhumation in the underlying Franciscan accretion-ary wedge at Mount Diablo, California: Geological SocietyAmerica Bulletin, Vol. 119, pp. 1347–1367.

UNRUH, J. R. AND LETTIS, W. R., 1998, Kinematics of compres-sional deformation in the eastern San Francisco Bay region,California: Geology, Vol. 26, pp. 19–22.

UNRUH, J. R.; SAWYER, T. L.; KNUDSEN, K. L.; AND LETTIS, W. R.,1997,Assessment ofBlindSeismogenic Sources, LivermoreValley,Eastern San Francisco Bay Region, Final Technical Report: U.S.Geological Survey NEHRP Award No. 1434-95-G-2611, 88 p.

WAGNER, D. L.; BORTUGNO, E. J.; AND MCJUNKIN, R. D., 1990,Geologic Map of the San Francisco–San Jose Quadrangle:California Division of Mines and Geology, Regional GeologicMap Series, Map No. 5A, 5 sheets, 1:250,000.

WRIGHT, S. G., 1975, Evaluation of Slope Stability Procedures:American Society of Civil Engineers, preprint 2616, NationalConvention, Denver, CO, 28 p.

WYLLIE, D. C. AND MAH, C. W., 2004, Rock Slope Engineering:Civil and Mining: Taylor & Francis, New York, NY, 456 p.



Factors Affecting Failure by Internal Erosion ofGeosynthetic Clay Liners Used in Freshwater Reservoirs

HAKKI O. OZHAN1

Department of Civil Engineering, Istanbul Kemerburgaz University, Mahmutbey DilmenlerCad, No. 26, Bagcilar-Istanbul, Turkey

EROL GULER2

Department of Civil Engineering, Bogazici University, 34342 Bebek-Istanbul, Turkey

Key Terms: Internal Erosion, Geosynthetic Clay Liner,Freshwater Reservoir, Permittivity

ABSTRACT

Geosynthetic clay liners (GCLs) are often used aslining materials for freshwater reservoirs. To irrigateagricultural land without depleting groundwater, surfacewater is stored in these artificial ponds. In this study,hydraulic conductivity tests were performed on GCLsplaced in flexible-wall permeameters under hydraulicheads of up to 50 m in order to investigate the risk ofinternal erosion. In these tests, base pedestals made ofPlexiglas with uniform circular voids were placedbeneath the GCLs instead of a typical gravel subgrade.The voids in the base pedestal represented the voidsbetween uniform rounded gravel particles. Differenttypes of GCLs were tested. GCL-1 was reinforced usingneedle-punching technology, whereas GCL-2, GCL-3,and GCL-4 were un-reinforced GCLs that wereassembled in the laboratory. We investigated the effectson internal erosion of the void size in the subbase; thegeotextile component that was in contact with thesubbase; the bentonite component; and the manufac‐turing process of the GCLs. Test results indicated thatinternal erosion was directly related to the void diameterof the base pedestal. The resistance of the needle-punched GCL to internal erosion was better than that ofthe un-reinforced GCLs. The degree of internal erosionwas also related to the engineering properties of thegeotextile in contact with the base pedestal. Highertensile strength of the GCL reduced the possiblepotential for internal erosion within it. The type ofbentonite did not have a significant effect on internalerosion.

INTRODUCTION

A GCL is a barrier material that consists of a thinlayer of bentonite (5–15 mm) sandwiched betweentwo geotextiles and/or glued to a geomembrane byusing a water-soluble, non-polluting adhesive (Bouazza,2002; Koerner, 2005;). Geotextile components of GCLsare bonded to the bentonite on both sides with needle-punching, stitch-bonding, adhesives, or just by attach-ing the geotextiles to the wet bentonite. This attachmentis maintained by placing the bentonite on top of thelower geotextile, then wetting the bentonite, and finallylaying the upper geotextile on top of the bentonite layer(Ozhan, 2011). According to the manufacturing pro-cess, GCLs are classified into two groups: reinforcedGCLs and un-reinforcedGCLs. Reinforcement is main-tained by either needle-punching or stitch-bonding.Needle-punched GCLs are reinforced by sewing thetop geotextile through the bentonite into the bottomgeotextile (vonMaubeuge and Heerten, 1994). Bondingis maintained by punching the polypropylene fibersbetween the top and bottom geotextiles, which provideshigher internal shear strength (Bouazza, 2002). Alterna-tively, in stitch-bonding, the GCL is reinforced withparallel rows of sewn yarn (Bouazza, 2002). In an un-reinforced GCL, the high cohesion capability of thewetted bentonite particles provides the bond betweenthe bentonite and the upper and lower geotextiles(Ozhan, 2011), or an adhesive is added to the bentoniteto increase its bonding capacity to the geotextiles(Bouazza, 2002). Reinforced GCLs have greater shearstrength and stronger bonding capability between thebentonite and the geotextiles than un-reinforcedGCLs. The manufacturing process of GCLs (reinforcedor un-reinforced), the manufacturing of the geotextilecomponents of GCLs (woven or non-woven), and thecomposition and form of the bentonite used in GCLs(sodium bentonite or calcium bentonite; granular orpowdered) are the main parameters used to comparedifferent GCLs.GCLs are widely preferred for environmental pro-

tection barriers due to their low hydraulic conductivity1Corresponding author email: [email protected]


(,10−10 m/s), low cost, and ease of installation in bothcover systems and composite bottom liners (Bensonand Meer, 2009). GCLs are used either as part of com-posite liners at the bottom of landfills, canals, storagetanks, and surface impoundments (Hornsey et al.,2010; Kang and Shackelford, 2010; and Rowe andAbdelatty, 2012) or as the sole lining material in fresh-water reservoirs. The collection of surface water intothese GCL-lined ponds necessitates only a smallinvestment and reduces the demand on groundwater.GCL can be used to overlie a wide range of soilsfrom clay to coarse-grained gravel (Ozhan and Guler,2013).High water levels have to be taken into considera-

tion when designing any reservoir. As the depth ofwater increases, the hydraulic head and, consequently,potential for the seepage flow increase, which mightcause the bentonite to erode through the geotextile.This process is known as bentonite extrusion. Afterthe threshold hydraulic head is exceeded, bentoniteparticles can start to flow through openings anddamaged zones within the geotextile (McCook,2007). If this occurs, it will cause the hydraulic conduc-tivity of the GCL to significantly increase and impairthe hydraulic barrier capability of the GCL (Ozhanand Guler, 2013). This process is called internal ero-sion (McCook, 2007). According to Li (2008), grainsize distribution, grain shape, and loading conditionsinfluence the degree of internal erosion. The combinedproperties of the geotextile and the bentonite deter-mine the hydraulic head at which internal erosionbegins. However, when internal erosion begins, thecombined structure begins to change, and excessivedeformations begin to occur in the carrier geotextile,and, consequently, more bentonite particles start toflow. At the end of this process, a total failure in termsof hydraulic conductivity is observed (Ozhan andGuler, 2013).Piping is another process that causes hydraulic fail-

ure of earth materials. Piping is induced by erosionof particles that results in a continuous pipe throughthe material (Jacobson, 2013). Piping initiates at zonesof concentrated leakage through the openings of thematerial, whereas internal erosion is caused by flowthrough the openings of the material (Fell et al.,2003; McCook, 2007). Based upon this definition,hydraulic failure of GCLs due to significant amountof bentonite loss through the openings of the geotextileis caused by internal erosion.Internal erosion also occurs in dams, embankments,

and other structures that consist of earth materials.During internal erosion, the repulsive forces betweensoil particles become greater than the attractive forces,leading to deflocculation and dispersion of the soil(Burns and Ghataora, 2007). The seepage forces

caused by high hydraulic gradients initiate the detach-ment of soil particles (Greene et al., 2010; Muresanet al., 2011; Benahmed and Bonelli, 2012; and Baenaand Toledo, 2014). During the initiation phase, fineparticles erode slowly within the matrix of coarser par-ticles (Chang and Zhang, 2013). After the onset ofinternal erosion, the flow increases and causes a sud-den increase in the hydraulic conductivity of the soil,and an increase in the displacement and flow of soilparticles. Finally, the soil skeleton becomes unstableand collapses (Bendahmane et al., 2006; Chang andZhang, 2013; Rodriguez et al., 2014; and Zhang et al.,2015).

BACKGROUND

Fox et al. (2000) conducted hydraulic conductivitytests in flexible-wall permeameters on both needle-punched and adhesive-bonded GCLs to study theeffects of the particle size of soil cover and the rate ofloading. The GCLs tested were placed beneath gravelswith a particle diameter varying from 12.7 to 50.8 mm.Test results indicated that bentonite extrusion in‐creased with increasing cover soil particle size andrate of loading. The confinement provided by the nee-dle-punching process caused the GCLs to have lessbentonite extrusion and subsequent variability inthickness than the adhesive-bonded GCLs. Fox et al.(2000) further stated that although bentonite displace-ment was observed for some of the GCLs, internal ero-sion was not detected under hydraulic heads of up to0.60 m.The effect of the engineering characteristics of

gravel subgrade on the hydraulic performance ofboth needle-punched and adhesive-bonded GCLs wasinvestigated by Shan and Chen (2003). One of the sub-grade materials was uniformly graded, angularcrushed gravel with diameters ranging from 25.4 to50.8 mm, while the other subgrade material was uni-formly graded, rounded gravel and cobble with dia-meters ranging from 50.8 to 76.2 mm. According tothe test results, a hydraulic head of 0.69 m on theGCLs did not cause internal erosion. However, largerparticle sizes and increased angularity of the subgradecaused more bentonite loss from the GCL than smallerparticle sizes and more rounded shapes of the subgrade(Shan and Chen, 2003).Results of laboratory tests performed by Fox et al.

(2000) and Shan and Chen (2003) indicated thatmore bentonite displacement took place in adhesive-bonded GCLs than in needle-punched GCLs. How-ever, internal erosion causing hydraulic failure of theGCLs was not observed in these studies, and this wasmost likely due to the very low hydraulic heads.

Ozhan and Guler


Rowe and Orsini (2003) conducted hydraulic con-ductivity tests on GCLs using a rigid-wall permea-meter in order to examine the effect of subgrade typeon internal erosion. The subgrades placed beneaththe GCLs were sand, 6 mm gravel, and geonet. Ageonet is an open grid-like geosynthetic material con-sisting of two sets of parallel, polymeric ribs, whichconveys liquids or gases. The primary function of ageonet is drainage. Most of the GCLs placed overgravel or geonet experienced internal erosion underhydraulic heads ranging from 8 to 90 m. However,when sand subgrade was used, there was no internalerosion under hydraulic heads of up to 90 m.

Rowe et al. (2014a) investigated the effects of differ-ent factors such as flow rate and slope angle on down-slope erosion of bentonite from needle-punchedGCLs. In this case, erosion occurred along the down-ward gradient within the liner. According to theresults, the flow rate and slope did not significantlyaffect the time needed to initiate erosion. However,after the erosion holes were formed, higher flow ratesand steeper slopes caused more bentonite particles tobe eroded.

Apart from the laboratory studies considering inter-nal erosion of GCLs, the literature provides few fieldstudies where internal erosion is associated withGCLs. Stam (2000) reported a case study where a sig-nificant amount of leakage was observed through theGCL lining of a reservoir. Internal erosion occurredthrough the non-woven geotextile of the GCL intothe coarse sand subgrade. Orsini and Rowe (2001)indicated that internal erosion took place through theGCLs that were in contact with a coarse gravel sub-grade under hydraulic heads greater than 10 m.

Rowe et al. (2014b) and Brachman et al. (2014) alsoperformed field tests to investigate the effects of GCLtype on down-slope erosion. According to the testresults, erosion did not occur for the GCLs with apolypropylene coating facing up or the GCLs withpolyacrylamide-based polymer-enhanced bentoniteeven after 15 months of exposure. However, erosionbegan after just 6 months of exposure when theGCLs were not coated or polymer enhanced (Brach-man et al., 2014). According to the results, additionalneedle-punching of the GCLs did not reduce the riskof down-slope erosion. Furthermore, granularity ofthe bentonite had almost no effect on the developmentof down-slope erosion (Rowe et al., 2014b).

The laboratory tests and case studies where internalerosion of GCLs was investigated in the field suggestthe factors that affect the amount of bentonite extru-sion, and therefore internal erosion, include coarsenessof the subgrade, height of the hydraulic head, type ofgeotextile and bentonite used in the GCLs, andwhether the GCLs were needle-punched.

The objective of our research was to investigate theperformance of different GCLs against internal ero-sion by placing the GCLs over coarse subgrade mate-rials under high hydraulic heads, which couldsimulate possible field conditions for GCL usage inlining reservoirs. For this purpose, the effects of themanufacturing process, the engineering properties ofthe geotextiles, and the type of bentonite on internalerosion were investigated. In the tests, a base pedestalmade of Plexiglas with uniform circular voids wasused to simulate natural gravel as a subgrade beneaththe GCL. In this way, it was possible to avoid effectsof randomness of the void size of natural earth materi-als. The diameters of the voids were selected as 20, 15,10, and 5 mm. It was shown by Ozhan and Guler(2013) that a perforated base pedestal with uniformcircular voids successfully simulated rounded coarse-grained gravel in terms of internal erosion. BothGCLs reinforced by needle-punching and un-rein-forced GCLs placed over base pedestals with circularvoids of different sizes were tested under hydraulicheads up to 50 m.

MATERIALS

Four different GCLs (one reinforced and three un-reinforced) were tested in this study. The first GCL,Bentomat SS100 (CETCO, 2007), designated asGCL-1, was a reinforced GCL consisting of a layerof granular sodium bentonite between a woven slit-film polypropylene geotextile (108 g/m2) and a non-woven needle-punched polypropylene geotextile (203g/m2). To provide reinforcement, polypropylene fibersfrom the non-woven geotextile were needle-punchedthrough the bentonite to the woven geotextile. Theengineering properties of GCL-1 are listed in Table 1(CETCO, 2007).GCL-2 was assembled in the laboratory using the

same sodium bentonite as GCL-1, sandwiched betweenthe same woven and non-woven geotextiles used inGCL-1 without needle-punching the components.GCL-3 was assembled in the laboratory using calciumbentonite sandwiched between the same woven andnon-woven geotextiles of which GCL-1 was composed.Again, the components were not needle-punched.GCL-4 was assembled using sodium bentonite sand-wiched between the same non-woven geotextile usedin GCL-1 and a woven geotextile with a relativelyhigher tensile strength than the one used in GCL-1,without needle-punching the components.The non-woven geotextile designated as N1 that was

used in all of the GCLs was a polypropylene, staple-fiber, needle-punched, non-woven geotextile. Thewoven geotextile that was used in GCL-1, GCL-2,and GCL-3 specimens was designated as W1, while

Barriers in Freshwater Reservoirs


the woven geotextile used in GCL-4 was designated asW2. Both of the woven geotextiles were polypropy-lene, slit-film geotextiles. The non-woven geotextilethat was used as a filter beneath the perforated basepedestal was designated as NF. The engineering prop-erties of the geotextiles used in this study are listed inTable 2.The sodium bentonite in GCL-1, GCL-2, and GCL-

4 was a naturally occurring, granular, high-swellingWyoming sodium bentonite, whereas the calcium ben-tonite in GCL-3 was a naturally occurring, powderedcalcium bentonite from Ankara, Turkey. The swellindex, liquid limit, and plastic limit were 28 mL/2 g(ASTM D5890, 2011), 344 percent, and 36 percent(ASTM D4318, 2010) for the sodium bentonites, and20 mL/2 g (ASTM D5890, 2011), 141 percent, and41 percent (ASTM D4318, 2010) for the calcium ben-tonite, respectively.

SPECIMEN PREPARATION

A 100-mm-diameter template was placed on speci-men GCL-1 with the woven geotextile side facing up.Then, the outer perimeter was wetted with de-ionizedwater to prevent bentonite loss during cutting. Afterthe bentonite was saturated, the specimen was cutwith a utility knife (Fox et al., 2000; Rowe and Orsini,2003).Preparation of specimens GCL-2, GCL-3, and

GCL-4 was different from that of GCL-1. Bothnon-woven and woven geotextiles with a diameter

of 100 mm were cut separately with scissors, andthe bentonite was placed between the geotextiles.Then, the granular sodium bentonite was wettedwith de-ionized water before placing it between thegeotextiles. The wetting of the bentonite enabled itto bond with the upper and lower geotextiles. Forspecimen GCL-3, the calcium bentonite used wasoriginally in powder form and was pre-moistened tocause expansion before being placed between thegeotextiles (Ozhan, 2011).

METHODOLOGY

Void Size of the Base Pedestal Used Beneath the GCL

In order to standardize the tests and eliminate thevariability of the natural gravel subbase, a perforatedbase pedestal with uniform circular voids was usedinstead of a natural subgrade. The base pedestal wasmade of Plexiglas that had uniform circular voids.The diameter of the circular voids was varied to simu-late different types of subgrade soils. Using the trigo-nometric assumption presented by Ozhan and Guler(2013), the maximum void diameter was calculatedas approximately 20, 15, 10, and 5 mm for a uniformgrain size of 50, 37.5, 25, and 12.5 mm, respectively.The pedestal with hole diameters of 20, 15, 10, and 5mm had 4, 6, 12, and 44 holes, respectively (Figure1a, b, c, and d). The ratio of the area of voids to thetotal area was between 11 and 16 percent. GCL speci-mens were placed over these base pedestals duringthe tests. Ozhan and Guler (2013) showed that the

Table 2. Engineering properties of the geotextile components of the GCLs.

Property Test MethodNon-Woven

Geotextile (N1)Woven

Geotextile (W1)Woven

Geotextile (W2)Non-Woven Geotextile

as a Filter (Nf)

Mass per area (g/m2) ASTM D 5261 203 108 210 120Thickness (mm) ASTM D 5199 2.0 0.4 0.77 1.1Wide-width tensile strength (kN/m) ASTM D 4595 15.4 12.2 44 9Wide-width elongation (%) ASTM D 4595 45 10 15 50Apparent opening size (mm) ASTM D 4751 0.212 0.425 0.380 0.08

Table 1. Engineering properties of GCL-1.

Property Test Method Value

Bentonite swell index ASTM D5890 (2011) min. 24 mL/2 gBentonite mass per unit area ASTM D5993 (2009) min. 4.8 kg/m2

GCL tensile strength EN ISO 10319 (2008) 8 kN/mGCL tensile elongation EN ISO 10319 (2008) 15%GCL index flux ASTM D5887 (2009) 2 6 10−9–2 6 10−10 m3/m2/sGCL permeability ASTM D5887 (2009) 1 6 10−11–1 6 10−12 m/s

Ozhan and Guler


resistance against internal erosion of GCLs placedover granular soils could be successfully simulated byperforated base pedestals with 10-mm- and 15-mm-diameter voids.

Test Procedure

After specimen preparation, hydraulic conductivitytests were performed using a standard flexible-wallpermeameter and the constant-head method (ASTMD5887, 2009). The thickness of a GCL specimen variesthroughout the surface area, which makes it hard toconstruct a constant thickness (Fox et al., 2000). Dueto thickness variation, it is possible to measure the per-meation through the GCL specimen by not taking thethickness parameter into account. For this reason, per-mittivity (Ψ) values were measured in these tests. Theuse of Ψ is more accurate than the use of permeability(k) for thin, compressible geosynthetic products suchas GCLs because there is no need to measure the thick-ness of the GCL specimen (Koerner, 2005).

Permittivity (Ψ) is the ratio of the coefficient of per-meability (k) to the average thickness (L) of the soil.Permittivity (Ψ) is expressed in Eq. 1 as follows:

W ¼ DQA� Dh� Dt

; ð1Þ

where Ψ (1/T) is the permittivity, ΔQ (L3) is the aver-age of inflow and outflow amounts for a given timeinterval, A (L2) is the cross-sectional area of the GCLspecimen, Δh (L) is the hydraulic head difference,and Δt (T) is the interval of time over which the flowΔQ occurs.The configuration of the test setup is shown in Fig-

ure 2. From top to bottom, the setup consisted of arigid top cap, porous stone, filter paper, GCL speci-men, perforated base pedestal, non-woven geotextilefilter (NF), and a rigid bottom cap. A latex membranewas used to prevent side leakage. More details of thetest setup and the specimen preparation in the permea-meter are reported in Ozhan (2011). The GCL-1,GCL-2, and GCL-3 specimens were tested with boththe woven and the non-woven geotextile facing theperforated base pedestal. However, only the wovengeotextile component of the specimen GCL-4 wastested.After the GCL specimen was fully saturated and

consolidated, permeation was initiated by raising thepressure at the top of the specimen to 530 kPa whilethe effluent pressure was kept constant at 515 kPa.As a result, downward flow was produced through theGCL specimen with a pressure difference of 15 kPa(ASTM D5887, 2009). The pressure measured at thelocation where the water entered at top of the GCL spe-cimen was called the influent pressure, whereas the

Figure 1. Perforated base pedestals with: (a) 20 mm, (b) 15 mm, (c) 10 mm, and (d) 5 mm hole diameters.



pressure measured at the location where the water leftthe bottom of theGCL specimen was called the effluentpressure. When the flow became steady or the increasein permittivity was less than one order of magnitude,the head was increased to 5 m by 1 m increments at10 minute intervals. To increase the hydraulic head,both the cell pressure and the influent pressure werekept constant, while the effluent pressure was decreasedto the desired value. At each 5 m increment, the headwas kept constant for 12 days. A duration of 12 dayswas chosen because a significant increase in permittiv-ity, up to three orders of magnitude, was obtainedwithin approximately 12 days for all of the GCLs thatexperienced internal erosion (Ozhan, 2011; Ozhanand Guler, 2013). Therefore, it was concluded that thetest duration was satisfactory for comparison reasons.When internal erosion occurred, the hydraulic conduc-tivity test was terminated; otherwise, the hydraulichead was increased to 50 m in 5 m increments (Ozhan,2011; Ozhan and Guler, 2013).In this study, the maximum applied hydraulic head

was chosen as 50 m because water depths can be ashigh as 40–50 m in freshwater reservoirs (Zohary andOstrovsky, 2010). Consequently, a hydraulic head of50 m was considered sufficient to simulate the worst-case scenario. Moreover, a 50 m hydraulic head wasused to investigate internal erosion of the GCLs inmost of the hydraulic conductivity tests performed byDickinson and Brachman (2010) and Rowe and Orsini(2003).The GCL specimens tested in this study were desig-

nated according to the geotextile type in contact withthe base pedestal (W for woven or NW for non-woven), and the void diameter of the base pedestal(D2 for 2 cm, D1.5 for 1.5 cm, D1 for 1 cm, or D0.5for 0.5 cm). For example, a GCL-1 specimen testedwith a non-woven geotextile over the base pedestal

with a void diameter of 1.5 cm is designated as GCL-1-NW-D1.5. Tests were performed on the woven andnon-woven geotextiles of GCL-1, GCL-2, and GCL-3 using void diameters of 0.5, 1, 1.5, and 2 cm, andon the higher-tensile-strength woven geotextile ofGCL-4 using void diameters of 1, 1.5, and 2 cm.

TEST RESULTS

The results of the hydraulic conductivity tests andthe measured permittivities of the GCLs are listed inTable 3. The hydraulic heads that caused internal ero-sion and the confining, influent, and effluent pressuresare also listed in Table 3.When exposed to a high hydraulic head, at first, the

measured permittivity of the GCL decreased veryslowly. After a while, some bentonite particles beganto erode from the stretched lower geotextile, usuallycausing an increase of up to one order of magnitudein permittivity at the beginning of internal erosion.Higher deformation of the lower geotextile componentof the GCL was detected when compared with thedeformation of the upper geotextile. As time passed,more bentonite particles were extruded through thegeotextile. Finally, an increase up to three orders ofmagnitude in permittivity was measured for all of theGCLs that experienced hydraulic failure. This beha-vior is similar to the test results reported by Roweand Orsini (2003), where hydraulic conductivity ofthe GCL decreased almost one order of magnitudeduring the first 175 hours of permeation under a 10m hydraulic head. Then, the hydraulic conductivitybegan to increase very slowly during the next 155hours of permeation. Finally, hydraulic conductivityincreased drastically within a couple of hours. Themeasured increase in hydraulic conductivity that

Figure 2. Test setup configuration.

Ozhan and Guler


caused the failure of the GCL was more than twoorders of magnitude (Rowe and Orsini, 2003).

Based upon the hydraulic conductivity test results ofspecimens GCL-1, three of the eight tested GCL-1 spe-cimens experienced internal erosion as shown in Fig-ure 3: GCL-1-W-D2 failed under 30 m of hydraulichead, GCL-1-NW-D2 failed under 35 m of hydraulichead, and GCL-1-W-D1.5 failed under 45 m ofhydraulic head. The measured permittivity of thesethree specimens was more than four orders of magni-tude higher at the end of the tests. However, internalerosion did not occur for GCL-1-NW-D1.5, GCL-1-W-D1, GCL-1-NW-D1, GCL-1-W-D0.5, or GCL-1-NW-D0.5, even under 50 m of hydraulic head.

In the hydraulic conductivity test results of GCL-1-NW-D2, the permittivity began to increase afteralmost 175 hours, as shown in Figure 3. The rate ofincrease in permittivity remained constant for approxi-mately 15 hours. Then, the rate of increase increasedand remained almost constant for another 30 hours.Subsequently, a sudden and much higher increase inpermittivity was measured within 1 to 2 hours. Thissudden increase was higher than two and a half ordersof magnitude. The permittivity (Ψ) was 2.30 6 10−9 1/

s at the beginning of the test, increasing to 2.296 10−5

1/s at the end of the test.Based upon the hydraulic conductivity test results of

specimens GCL-2, six of the eight GCL-2 specimensexperienced internal erosion, as shown in Figure 4.The measured permittivity of these specimens wasfour to five orders of magnitude higher at the end ofthe tests. Internal erosion did not occur for GCL-2-W-D0.5 and GCL-2-NW-D0.5, even under 50 m ofhydraulic head.Based upon the hydraulic conductivity test results of

specimens GCL-3, six of the eight GCL-3 specimensexperienced internal erosion, and the permittivityvalues increased by several orders of magnitude, asshown in Figure 5. Again, internal erosion did notoccur in the case of the smallest void diameter (0.5cm) when testing the woven and non-woven geotextilesup to a hydraulic head of 50 m.None of the three GCL-4 specimens experienced

internal erosion, even under 50 m of hydraulic head.The measured permittivity of these specimens wasapproximately half an order of magnitude lower afteralmost 250 hours, as shown in Figure 6. The decreasein permittivity was attributed to the increase in

Table 3. Permittivity test results.

Specimen NameHydraulic Headat Failure (m)

ConfiningPressure(KPa)

InfluentPressure(KPa)

EffluentPressure(KPa)

Permittivity at 50 mHydraulic Head (1/s)

Permittivity BeforeInternal Erosion (1/s)

GCL-1-NW-D2 35 550 530 186.7 — 5.46610−10

GCL-1-W-D2 30 550 530 235.7 — 8.39 6 10−10

GCL-1-NW-D1.5 No failure 550 530 39.5 3.04 6 10−10 —GCL-1-W-D1.5 45 550 530 88.6 — 4.92 6 10−10

GCL-1-NW-D1 No failure 550 530 39.5 9.95 6 10−11 —GCL-1-W-D1 No failure 550 530 39.5 1.32 6 10−10 —GCL-1-NW-D0.5 No failure 550 530 39.5 4.29 6 10−11 —GCL-1-W-D0.5 No failure 550 530 39.5 5.12 6 10−11 —GCL-2-NW-D2 20 550 530 333.8 — 1.25 6 10−9

GCL-2-W-D2 10 550 530 431.9 — 1.67 6 10−9

GCL-2-NW-D1.5 35 550 530 186.7 — 3.85 6 10−10

GCL-2-W-D1.5 15 550 530 382.9 — 3.50 6 10−9

GCL-2-NW-D1 50 550 530 39.5 — 1.32 6 10−10

GCL-2-W-D1 30 550 530 235.7 — 5.59 6 10−10

GCL-2-NW-D0.5 No failure 550 530 39.5 3.97 6 10−11 —GCL-2-W-D0.5 No failure 550 530 39.5 1.08 6 10−10 —GCL-3-NW-D2 20 550 530 333.8 — 1.11 6 10−8

GCL-3-W-D2 5 550 530 481 — 9.72 6 10−9

GCL-3-NW-D1.5 35 550 530 186.7 — 8.61 6 10−9

GCL-3-W-D1.5 15 550 530 382.9 — 1.13 6 10−8

GCL-3-NW-D1 50 550 530 39.5 — 1.18 6 10−9

GCL-3-W-D1 25 550 530 284.8 — 5.54 6 10−9

GCL-3-NW-D0.5 No failure 550 530 39.5 6.77 6 10−10 —GCL-3-W-D0.5 No failure 550 530 39.5 8.69 6 10−10 —GCL-4-W-D2 No failure 550 530 39.5 2.05 6 10−10 —GCL-4-W-D1.5 No failure 550 530 39.5 8.92 6 10−11 —GCL-4-W-D1 No failure 550 530 39.5 6.53 6 10−11 —



seepage stresses and decrease in void ratio of the ben-tonite when the hydraulic head was increased.The zones of the GCL facing the voids on the perfo-

rated base pedestal exhibited a convex geometry and,therefore, were called convex zones, as shown in Figure7a and b. The zones where the GCLwas in contact withthe solid parts of the perforated base pedestal werecalled flat zones. As shown in Figure 7a, a largeamount of bentonite was eroded through the openingsof the woven geotextile. Most of the high increase in

measured permittivity was due to the internal erosionthat occurred through the convex zones of the GCL.The slit-film tapes of the woven geotextile in the convexzones of GCL-1-W-D2 were also damaged and torn, asshown in Figure 7a. Similar behavior was observed inthe non-woven geotextiles (Figure 7b). This showsthat the GCL failed in the convex zones, where internalerosion was the most severe.Specimens GCL-4-W-D1.5 and GCL-2-NW-D0.5

did not experience internal erosion. The appearance

Figure 3. Permittivity (Ψ) vs. elapsed time (t) graph for GCL-1 (the values given in parentheses are the hydraulic heads).


Ozhan and Guler


of these specimens after the termination of the hydrau-lic conductivity tests is shown in Figure 8a and b, andaccording to Figure 8a and b, the depth of the convexzones was smaller, and the slit-film tapes on the wovengeotextile were not damaged in either.

PRACTICAL CONSIDERATIONS

Ozhan and Guler (2013) have shown that perforatedbase pedestals could simulate rounded uniform gravelsubgrades almost perfectly for testing internal erosion.In their study, only two different perforated base ped-estals with uniform 15 mm and 10 mm holes wereused. In our study, the effect of the void size of the sub-base material on internal erosion was modelled by

using four uniform hole sizes. Also, the effects of themanufacturing process of the GCLs, engineering prop-erties of the geotextile components of the GCLs incontact with the subbase, and the type of bentoniteused in the GCLs were investigated. Based upon thetest results, the effects of these different parameterson internal erosion of the GCLs are discussed below.

Effect of the Void Size of the Perforated Base Pedestal

A question that must be addressed is the loss of ben-tonite in the convex zones. Test results show that thefailure of the carrier geotextile did not occur immedi-ately after the application of the hydraulic head. Thecarrier geotextile was the geotextile component of the





GCL placed at the bottom during the manufacturingprocess. Bentonite was sandwiched between the lower,carrier geotextile and the upper, cover geotextile. Thepermittivity of the GCL started to increase slowlyand after 25 to 30 hours, increasing approximatelyone order of magnitude. Our interpretation is thatthis increase in permittivity occurred because the ben-tonite particles began to erode. As a consequence ofthis loss of bentonite particles, the configuration ofthe GCL changed, and excessive deformation occurredin the convex zones of the geotextile. This caused anincrease in the rate of permittivity and, eventually,complete failure. Our opinion is that the failure wasinitiated by internal erosion, which caused the initialloss of bentonite particles, and the final failureoccurred due to the failure of the carrier geotextile.There is a direct relation between internal erosion

and the void size of the subbase material underlying

a GCL. When the void size increased, the depth ofthe convex zones also increased. As a result, a greateramount of bentonite could erode through the openingsand damaged regions in the convex zones, and internalerosion could occur under lower hydraulic heads.GCLs tested over the perforated base pedestal with auniform void diameter of 5 mm did not experienceinternal erosion, even under a hydraulic head of50 m. This indicates that internal erosion becomesless critical as the grain size of the natural soil, andits void size, decreases.In this study, it was shown that placing GCLs over

gravel particles with void diameters of 5 mm or smallerwould not cause internal erosion even under hydraulicheads of 50 m. Thus, it can be concluded that if thesubbase upon which the GCL is placed consists ofcoarser particles, internal erosion can be prevented

Figure 7. (a) Woven geotextile in contact with a perforated basepedestal of 2-cm void size. (b) Non-woven geotextile in contactwith a perforated base pedestal of 1.5 cm void size. Figure 8. (a) Woven geotextile in contact with a perforated base

pedestal of 1.5 cm void size. (b) Non-woven geotextile in contactwith a perforated base pedestal of 0.5 cm void size.

Ozhan and Guler


by placing a sand layer having a nominal thickness of5 cm above the gravel subgrade. In order to preventthe loss of sand particles into the voids in the gravel,either an intermediate geotextile or a sand-mat (sandsandwiched between geotextiles) can be placed overthe gravel subbase.

Effect of the Manufacturing Process of GCLs

The performance of GCL-1 against internal erosionwas better than that of specimens GCL-2 and GCL-3.Specimens GCL-1-W-D2, GCL-1-NW-D2, and GCL-1-W-D1.5 were the only GCL-1 specimens that experi-enced internal erosion, whereas all of the GCL-2 andGCL-3 specimens tested over the perforated base ped-estal with uniform void diameters of 20, 15, and 10mm experienced internal erosion. Hydraulic headsthat caused internal erosion on GCL-1 specimenswere also higher than those on GCL-2 and GCL-3 spe-cimens when the perforated base pedestal beneath theGCLs had the same void size. The results of the per-mittivity tests in which GCL-1, GCL-2, and GCL-3specimens experienced internal erosion are summa‐rized in Figure 9. As shown in Figure 9, the needle-punched specimens (GCL-1) performed better, indi-cating that it contributed to the prevention of internalerosion under high hydraulic heads.

Effect of the Geotextile Component of GCLs

The geotextile type played a significant role in inter-nal erosion. Among the tests conducted with thewoven geotextile side facing the perforated base pedes-tal, GCL-4 performed the best. The main reason forthis was the higher tensile strength of the woven

geotextile component (W2), which prevented the for-mation of defects on the slit-film tapes. Bulging ofthe carrier geotextile into the voids, which caused theformation of the convex zones, might have contributedto the failure of the GCL. Sample W2, with a highertensile strength, seemed to perform the best andappeared to have the least amount of bulging intothe voids simulated by the base pedestal.As shown in Figure 9, GCL-1, GCL-2, and GCL-3

specimens tested with their woven geotextile compo-nent (W1) over the perforated base pedestal experi-enced internal erosion under lower hydraulic headsthan the GCL specimens tested with non-wovengeotextiles.It can be speculated that one of the reasons for the

better performance of the non-woven geotextile (N1)than the woven geotextile (W1) was the capability ofN1 to deform extensively without losing its mechanicalproperties. Probably, the rigid slit-film tapes of thewoven geotextile (W1) got damaged and caused a sig-nificant increase in the opening size in the convexzones more easily than those of the non-wovengeotextile.The apparent opening size of the geotextile in con-

tact with the perforated base pedestal can also influ-ence internal erosion. The woven geotextile (W1) hadan opening size of 0.425 mm, and the non-woven geo-textile (N1) had an opening size of 0.212 mm (ASTMD4751, 2012). This indicates that using geotextileswith smaller opening sizes can be beneficial.The comparison between the behavior of the GCLs

with N1 and W2 facing the perforated base pedestalmight be considered to contradict the above statement,because W2 performed much better. However, ouropinion is that the higher tensile strength of W2 pre-vented the openings of the geotextile from enlarging.

Effect of the Bentonite Component of GCLs

Although calcium bentonite has a higher hydraulicconductivity than sodium bentonite (Gleason et al.,1997), both GCL-2 and GCL-3 specimens performedsimilarly against internal erosion. This indicates thatthe bentonite type does not influence internal erosionsignificantly for the un-reinforced GCLs tested in thisstudy. Furthermore, at the beginning of the tests, adecrease in permittivity was measured for all of theGCL specimens composed of either sodium bentoniteor calcium bentonite that experienced internal erosion.This decrease was attributed to the increase in seepagestresses caused by the increase in hydraulic head, whichresulted in a decrease in the void ratio of the bentonite(Rowe and Orsini, 2003). However, to come to a deci-sive conclusion, further studies with permeation peri-ods of several years should be performed.

Figure 9. Hydraulic head at failure and void diameter of perforatedbase pedestal.



CONCLUSIONS

The diameter of the voids of the perforated basepedestal plays a significant role in internal erosion.As the void size increased, internal erosion occurredunder lower hydraulic heads. For the test configura-tions chosen and the GCLs used in this study, no inter-nal erosion occurred for the base pedestal with a voiddiameter of 5 mm.The manufacturing process (reinforced with needle-

punching versus un-reinforced) plays a significant rolein internal erosion. The performance of the GCLs thatwere manufactured by needle-punching was betterthan that of the GCLs that were assembled in thelaboratory without needle-punching, indicating that aproper bonding of the two geotextile componentsthrough needle-punching was beneficial.The engineering properties of the geotextile compo-

nents also play a significant role in internal erosion.The GCL with the woven geotextile (W2) facing theperforated base pedestal performed the best. This wasdue to the high tensile strength of W2, which helpedto prevent excessive deformation. However, the per-formance of the non-woven geotextile (N1) was betterthan that of the woven geotextile (W1) against internalerosion. This result was interpreted as being due to thecapacity of the non-woven geotextile to deform exten-sively without significantly changing its mechanicalproperties, e.g., the apparent opening size. The smallerinitial opening size of the non-woven geotextile wasalso instrumental in preventing internal erosion.The type of bentonite (sodium versus calcium bento-

nite) used in the un-reinforced GCLs did not signifi-cantly affect internal erosion.In conclusion, GCLs have been shown to be used

successfully as a barrier if the above conditions are fol-lowed and the same materials are used.

ACKNOWLEDGMENTS

The authors are appreciative of the financial sup-port provided by the Scientific Research Project Foun-dation of Turkey (Project No. 07HA401). The authorsalso sincerely thank Dr. Nigel Webb and Dr. SusanEnglish of the Colloid Environmental TechnologiesCompany (CETCO) for providing GCL-1 specimensand the geotextile and bentonite components ofGCL-1 specimens in order to use them for assemblingspecimens GCL-2, GCL-3, and GCL-4.

REFERENCES

ASTM STANDARD D4318, 2010, Standard Test Methods for LiquidLimit, Plastic Limit, and Plasticity Index of Soils: ASTMInternational, West Conshohocken, PA.

ASTM STANDARD D4595, 2011, Standard Test Method for TensileProperties of Geotextiles by the Wide-Width Strip Method:ASTM International, West Conshohocken, PA.

ASTM STANDARD D4751, 2012, Standard Test Method for Deter-mining Apparent Opening Size of a Geotextile: ASTM Interna-tional, West Conshohocken, PA.

ASTM STANDARD D5199, 2012, Standard Test Method for Measur-ing the Nominal Thickness of Geosynthetics: ASTM Interna-tional, West Conshohocken, PA.

ASTM STANDARD D5261, 2010, Standard Test Method for Measur-ing Mass per Unit Area of Geotextiles: ASTM International,West Conshohocken, PA.

ASTM STANDARD D5887, 2009, Standard Test Method for Mea-surement of Index Flux through Saturated Geosynthetic ClayLiner Specimens Using a Flexible-Wall Permeameter: ASTMInternational, West Conshohocken, PA.

ASTM STANDARD D5890, 2011, Standard Test Method for SwellIndex of Clay Mineral Component of Geosynthetic Clay Liners:West Conshohocken, PA.

ASTM STANDARD D5993, 2009, Standard Test Method for Measur-ing Mass per Unit of Geosynthetic Clay Liners: ASTM Inter-national, West Conshohocken, PA.

BAENA, C. M. AND TOLEDO, M. A., 2014, An experimentally veri-fied model of the seepage progress due to dissolution of solu-ble particles in foundations subject to intergranular flow:Environmental Earth Sciences, Vol. 72, pp. 3369–3382.

BENAHMED, N. AND BONELLI, S., 2012, Investigating concentratedleak erosion behaviour of cohesive soils by performing holeerosion tests: European Journal of Environmental and CivilEngineering, Vol. 16, No. 1, pp. 43–58.

BENDAHMANE, F.; MAROT, D.; ROSQUOET, F.; AND ALEXIS, A., 2006,Characterization of internal erosion in sand kaolin soils:Revue Européenne de Génie Civil, Vol. 10, No. 4, pp. 505–520.

BENSON, C. H. AND MEER, S. R., 2009, Relative abundance ofmonovalent and divalent cations and the impact of desicca-tion on geosynthetic clay liners: Journal of Geotechnical andGeoenvironmental Engineering, Vol. 135, No. 3, pp. 349–358.

BOUAZZA, A., 2002, Geosynthetic clay liners: Geotextiles and Geo-membranes, Vol. 20, pp. 3–17.

BRACHMAN, R. W. I.; RENTZ, A.; ROWE, R. K.; AND TAKE, W. A.,2014, Classification and quantification of downslope erosionfrom a GCL when covered only by a black geomembrane:Canadian Geotechnical Journal, Vol. 52, No. 4, pp. 395–412,doi: 10.1139/cgj-2014-0241.

BURNS, B. AND GHATAORA, G. S., 2007, Internal erosion of kaolin.In Puppala, A. J.; Hudyma, N.; and Likos, W. J. (Editors),GSP 162, Problematic Soils and Rocks and In Situ Character-ization, GeoDenver 2007: American Society of Civil Engineers,Denver, CO, pp. 1–8.

CETCO, 2007, BENTOMAT SS100 Certified Properties TechnicalData Sheet: CETCO, Hoffman Estates, IL.

CHANG, D. S. AND ZHANG, L. M., 2013, Critical hydraulic gradientsof internal erosion under complex stress states: Journal of Geo-technical and Geoenvironmental Engineering, Vol. 139, No. 9,pp. 1454–1467.

DICKINSON, S. AND BRACHMAN, R. W. I., 2010, Permeability andinternal erosion of a GCL beneath coarse gravel: Geosyn-thetics International, Vol. 17, No. 3, pp. 112–123.

EN ISO 10319, 2008, Geosynthetics—Wide-Width Tensile Test, 2nded.: International Organization for Standardization, Geneva,Switzerland.

FELL, R.; WAN, C. F.; CYGANIEWICZ, J.; AND FOSTER, M., 2003,Time for development of internal erosion and piping inembankment dams: Journal of Geotechnical and Geoenviron-mental Engineering, Vol. 129, No. 4, pp. 307–314.

Ozhan and Guler


FOX, P. J.; DE BATTISTA, D. J.; AND MAST, D. G., 2000, Hydraulicperformance of geosynthetic clay liners under gravelcover soils: Geotextiles and Geomembranes, Vol. 18, pp.179–201.

GLEASON, M. H.; DANIEL, D. H.; AND EYKHOLT, G. R., 1997, Cal-cium and sodium bentonite for hydraulic containment appli-cations: Journal of Geotechnical and GeoenvironmentalEngineering, Vol. 123, No. 5, pp. 438–445.

GREENE, B. H.; CROCK, J.; MOSKOVITZ, L.; AND PREMOZIC, J. W.,2010, Evaluation of seepage, internal erosion, and remedialalternatives for East Branch Dam, Elk County, Pennsylvania:Environmental & Engineering Geoscience, Vol. 16, August2010, pp. 229–243.

HORNSEY, W. P.; SCHEIRS, J.; GATES, W. P.; AND BOUAZZA, A., 2010,The impact of mining solutions/liquors on geosynthetics: Geo-textiles and Geomembranes, Vol. 28, No. 2, pp. 191–198.

JACOBSON, T., 2013, An Analysis on Soil Properties on PredictingCritical Hydraulic Gradients for Piping Progression in SandySoils: M.Sc. Thesis, Utah State University, Logan, UT, 38 p.

KANG, J. B. AND SHACKELFORD, C. D., 2010, Consolidation of ageosynthetic clay liner under isotropic states of stress: Journalof Geotechnical and Geoenvironmental Engineering, Vol. 136,No. 1, pp. 253–259.

KOERNER, R. M., 2005, Designing with Geosynthetics, 5th ed.:Prentice Hall, Upper Saddle River, NJ, 816 p.

LI, M., 2008, Seepage Induced Instability in Widely Graded Soils:Ph.D. Thesis, University of British Colombia, Vancouver, B.C., Canada, 300 p.

MCCOOK, D. K., 2007, Filter Diaphragms: Part 628 Dams—National Engineering Handbook: U.S. Department of Agricul-ture, National Resources Conservation Service, Washington,D.C. 58 p.

MURESAN, B.; SAIYOURI, N.; GUEFRECH, A.; AND HICHER, P. Y.,2011, Internal erosion of chemically reinforced granular mate-rials: A granulometric approach: Journal of Hydrology, Vol.411, pp. 178–184.

ORSINI, C. AND ROWE, R. K., 2001, Testing procedure and results forthe study of internal erosion of geosynthetic clay liners: In Pro-ceedings of the Geosynthetics 2001: Industrial Fabrics Associa-tion International, Portland, OR, pp. 189–201.

OZHAN, H. O., 2011, Internal Erosion of Geosynthetic Clay Linersunder High Hydraulic Heads: Ph.D. Thesis, Bogazici Univer-sity, Istanbul, Turkey, 222 p.

OZHAN, H. O. AND GULER, E., 2013, Use of perforated base pedes-tal to simulate the gravel subbase in evaluating the internal

erosion of geosynthetic clay liners: ASTM Geotechnical Test-ing Journal, Vol. 36, No. 3, pp. 418–428.

RODRIGUEZ, V.; GUTIERREZ, F.; GREEN, A. G.; CARBONEL, D.;HORSTMEYER, H.; AND SCHMELZBACH, C., 2014, Characterizingsagging and collapse sinkholes in a mantled karst by means ofground penetrating radar (GPR): Environmental & Engineer-ing Geoscience, Vol. 20, May 2014, pp. 109–132.

ROWE, R. K., AND ABDELATTY, K., 2012, Effect of a calcium-richsoil on the performance of an overlying GCL: Journal of Geo-technical and Geoenvironmental Engineering, Vol. 138, No. 4,pp. 423–431.

ROWE, R. K.; ASHE, L. E.; TAKE, W. A.; AND BRACHMAN, R. W. I.,2014a, Factors affecting the down-slope erosion of bentonitein a GCL: Geotextiles and Geomembranes, Vol. 42, No. 5,pp. 445–456, doi:10.1016/j.geotexmem.2014.07.002.

ROWE, R. K.; BRACHMAN, R. W. I.; TAKE, W. A.; ASHE, L. E.; AND

RENTZ, A., 2014b, Field and laboratory observations of down-slope bentonite migration in exposed composite liners. InBouazza, A.; Brown, B.; and Yuen. S (Editors), 7th Interna-tional Conference on Environmental Geotechnics: CurranAssociates, Inc. Melbourne, Australia, pp. 1–18.

ROWE, R. K. AND ORSINI, C., 2003, Effect of GCL and subgradetype on internal erosion in GCLs under high gradients: Geo-textiles and Geomembranes, Vol. 21, No. 1, pp. 1–24.

SHAN, H. Y. AND CHEN, R. H., 2003, Effect of gravel subgrade onhydraulic performance of geosynthetic clay liner: Geotextilesand Geomembranes, Vol. 21, No. 6, pp. 339–354.

STAM, T. G., 2000, Geosynthetic clay liner field performance. InProceedings of the 14th GRI Conference (Hot Topics inGeosynthetics): Geosynthetic Institute, Las Vegas, NV, pp.242–254.

VON MAUBEUGE, K. P. AND HEERTEN, G., 1994, Needle-punchedgeosynthetic clay liners. In Proceedings of the 8th GRI Confer-ence: Geosynthetic Research Institute, Philadelphia, PA, pp.129–207.

ZHANG, X.; DONG, J.; HUANG, Z.; NIE, X.; WONG, H.; WANG, J.;SUN, W.; LEO, C.; AND WANG, J., 2015, Study of soil structuresstrength and stiffness loss based on thermodynamics and con-tinuum mechanics: Environmental Earth Sciences, Vol. 73,No. 8, pp. 4143–4149.

ZOHARY, T. AND OSTROVSKY, I., 2010, Ecological impacts of exces-sive water level fluctuations in stratified freshwater lakes:Inland Waters, Vol. 1, pp. 47–59.



Book Review

Geomodels in Engineering Geology—An Introduction

(Peter Fookes, Geoff Pettifer, and Tony Waltham)

Review by: Richard Jackson

11 Venus Crescent, Geofirma Engineering Ltd., Heidelberg, Ontario N0B 2M1, Canada

This is an unusual and welcome contribution to theengineering geology literature. The preamble outlinesthe concept of a “geomodel” and its depiction in simpli-fied block diagrams is useful for the engineering geologistand geotechnical engineer. The geomodel approach wasdeveloped in theUKbecauseof thepopularity ofFookes’Glossop Lecture before the Engineering Geology Groupof the Geological Society of London, which was subse-quently published as Fookes (1997). It is written in theBritish tradition of engineering geomorphology.

The purpose of the book is “to help engineers visua-lize the three-dimensional geology and to act as aquick introduction to new or unfamiliar ground orenvironments for geologists and engineers.” The blockdiagrams are to be employed as “springboards” in thedesign and construction of geotechnical models, which,perhaps, indicates the authors’ desire to get the engi-neering geologists on site well before geoengineerslike me develop too many conclusions!

My colleague, Rob Sengebush of INTERA, Inc., inAlbuquerque, NM, developed one such geomodel of aPleistocene alluvial-fan aquifer cutting through brack-ish Eocene sediments in the heart of San Diego, CA,which we published in this journal recently (Sengebushet al., 2015, Figure 5).

However, this volume is unusual in several ways.The book itself is laid out in landscape, not portrait,style to accommodate the numerous illustrations. It isdivided into five parts:

1) underlying factors: climate and geology;2) near-surface ground changes;3) basic geological environments influencing engi‐

neering;4) ground investigations; and5) case histories and some basic ground characteris-

tics and properties.

An appendix, which defines the various geotechnicalproblems associated with different types of soils, abibliography of textbooks, and a list of the locationsof the numerous photographs complete the text.

Each of the parts (chapters really) comprises severalsections, each of which contains a block diagramdrawn by one of the authors (Pettifer), text, and sum-mary tables, followed by numerous color photographsfrom across the world. Thus, Part 3 begins with a sec-tion led by a block diagram of glacial environments,which is supplemented by three pages of text and tableson “glacial landforms” and “engineering in glacialenvironments.” The section is completed by a dozenphotographs of glaciers, moraines, drumlins, andtills—adequate for Brits perhaps but not enough, Ithink, for most Canadians, Scandinavians, and manyAmericans who live their lives on these materials.The format is instructive, although I would prefer

having fewer and larger photographs so that thedetails are more evident. This same format isfollowed in the other sections of Part 3, i.e., periglacialenvironments; temperate environments such as theMediterranean; relict periglacial environments insouthern Britain; hot desert environments; savannaenvironments; hot, wet tropical environments; andmountain environments. The photographs are excel-lent, and many of the block diagrams are superb.One might have hoped that the authors would have

digested some modern hydrogeological thinkingbecause the section on groundwater and “permeability”is poor. Nearly 40 years after Freeze and Cherry (1979)laid out the basic principles of modern hydrogeology,the authors show little appreciation of shallow ground-water flow systems and their relation to the topographicfeatures. Reading about the “coefficient of permeabil-ity,” a term dead here since the 1970s, rather than“hydraulic conductivity” is disappointing.This is a very British book written for a British

audience and might not be appreciated by many inNorth America. Although probably one quarter ofthe photographs are from the Americas, the authorshave not cited the North American scientific literatureto any significant degree—just 3 of ,70 referencesare by North American authors or institutions. This isdisappointing when they have cast their photographic


net so wide and it undermines the adoption hereof this useful book.Nevertheless, for those of us with a working interest

in geomodels, this is a welcome monograph illustratedwith some truly splendid block diagrams. Hopefully,this book will serve to keep alive this pictorial tradition.

REFERENCES

FOOKES, P., 1997, First Glossop Lecture: Geology for engineers:The geological model, prediction and performance: QuarterlyJournal of Engineering Geology, Vol. 30, pp. 293–424.

FOOKES, P.; PETTIFER, G.; AND WALTHAM, T., 2015, Geomodelsin Engineering Geology—An Introduction: Whittles Pub-lishing, Taylor & Francis, Caithness, Scotland, UK. Avail-able in North America through CRC Press, Boca Raton,Florida, 176 p. ISBN 9781498740043, Paperback US$70.https://www.crcpress.com/Geomodels-in-Engineering-Geology-An-Introduction/Fookes-Pettifer-Waltham/9781498740043

FREEZE, R. A. AND CHERRY, J. A., 1979, Groundwater: Prentice-Hall Inc., Englewood Cliffs, NJ.

SENGEBUSH, R. M.; HEAGLE, D. J.; AND JACKSON, R. E., 2015, Thelate Quaternary history and groundwater quality of a coastalaquifer, San Diego, California: Environmental & EngineeringGeoscience, Vol. XVIII, No. 4, pp. 249–275.

Book Review


Environmental & Engineering Geoscience

Documents

Transcript of Environmental & Engineering Geoscience