Engineering Geology 189 (2015) 32–47

Contents lists available at ScienceDirect

Engineering Geology

j ourna l homepage: www.e lsev ie r .com/ locate /enggeo

Soil classification analysis based on piezocone penetration test data —

A case study from a quick-clay landslide site in southwestern Sweden

Abbas Abbaszadeh Shahri a,b,⁎, Alireza Malehmir b, Christopher Juhlin b

a Faculty of Civil Engineering, Islamic Azad University, Rudehen branch, Tehran, Iranb Uppsala University, Department of Earth Sciences, Villavägen 16, SE75236 Uppsala, Sweden

⁎ Corresponding author.E-mail address: abbas.abbaszadeh@geo.uu.se (A. Abba

http://dx.doi.org/10.1016/j.enggeo.2015.01.0220013-7952/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 September 2014Received in revised form 9 December 2014Accepted 20 January 2015Available online 11 February 2015

Keywords:CPTu dataSoil classification chartsSensitive fine-grained soilsLandslide

Cone penetration test (CPT) and piezocone penetration test (CPTu)methods can be effective in site characteriza-tion and are important for soil profiling and classification. Extensive experience exists that relates CPT and CPTuresults to soil type/state and these are often the preferred in-situ tools for subsurface investigations, soil explora-tion and the evaluation of different engineering soil properties, compared with conventional laboratory testing.In this study, eightmethods, consisting of traditional and nontraditional ones, for soil classification and predictionof soil type and state using CPTu data were examined and tested. The CPTu data not only provide valuable infor-mation on soil types, but also can be used for deriving correlations with engineering soil properties for the pur-poses of hazard analyses and design of foundations. We carried out and established a region-specific correlationbetween CPTu data and soil properties for three location test points in southwestern Sweden at a quick-clay land-slide site. First, an analysis of the available CPTu datawas performed and then classification of the soils wasmade.After determination and identification of the soil profiles, the resultswere further evaluatedusing theUnified SoilClassification System (USCS) and converted to CPT–SPT charts. We summarize results from each of thesemethods, including a description of the available charts and their performance for soil classification. FromCPTu data, we provide high-resolution soil profiles from the three test points. Our work allowed the detectionof potential sensitive fine-grained clays, which are responsible for landslides in the study area, however, inmost cases occurring above coarse-grained materials. These materials were also detected in the CPTu data andshow good correspondence with available reflection seismic profiles from the site.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Soil classifications (type and/or state) into groups that have similargeotechnical behavior can provide useful guidance for geotechnicalengineers. The conventional method for determining a soil type is bylaboratory classification of samples retrieved from boreholes (Caiet al., 2011). However, in thefield, and in practice, the Standard Penetra-tion Test (SPT) and the cone penetration test (CPT) are two commonlyused in-situ tests for assessing and classifying soil types. The CPT hasproven to be the more reliable, cost effective and valuable tool in char-acterizing subsurface conditions and in assessing various soil properties,including the estimation of the liquefaction potential at a particularsite. If a CPT device is also capable of simultaneously measuring porewater pressure, then the measurements are referred to as a piezoconepenetration test (PCPT) or CPTu (Baligh et al., 1980; Tumay et al.,1981; Zuidberg et al., 1982; Lunne et al., 1997; Cai et al., 2010).

Advocates of the CPT method argue its capacity to define a continu-ous depth profile and its repeatability (Robertson, 2004). Its ability todetect thin layers makes the CPT method superior to the SPT method.

szadeh Shahri).

Although, there are several classification-based algorithms to predictthe soil type and state using CPT or CPTu data (e.g., Begemann, 1965;Douglas and Olsen, 1981; Jones and Rust, 1982; Senneset and Janbu,1985; Robertson et al., 1986; Olsen and Malone, 1988; Campanellaand Robertson, 1988; Robertson, 1990, 2009; Jefferies and Davies,1991, 1993; Olsen, 1994; Olsen and Koester, 1995; Olsen and Mitchell,1995; Eslami and Fellenius, 1997; Robertson and Wride, 1998;Jefferies and Been, 2006; Jung et al., 2008; Schneider et al., 2008; Cetinand Ozan, 2009), the proposed charts by Robertson et al. (1986) andRobertson (1990) have been popular and widely used (Long, 2008;Cai et al., 2011). The non-normalized charts by Robertson et al. (1986)define 12 Soil Behavior Type (SBT) zones, whereas, the normalizedcharts by Robertson (1990) define 9 zones. This difference has gener-ated some confusion among geotechnical engineers and led Robertson(2010, 2012) to provide an update of the charts.

In this paper we aim to classify soil types in a quick-clay landslide sitein southwestern Sweden on the basis of CPTu data. Our ultimate goal is todetermine and identify sensitive fine-grained clays, which may beresponsible for landslides. The study area (Fig. 1) has been the subjectof several geophysical studies, such as reflection and refraction seismic(Malehmir et al., 2013a, 2013b; Adamczyk et al., 2013, 2014; Lundberget al., 2014) and electromagnetic (Shan et al., 2014) investigations, as

Fig. 1. (A) Locationmapof the study area, (B) landslide riskmap of Sweden showing the location ofmajor landslides and (C) the location of the test points in our study area. CPTu data from7202, 7203 and 7206 are used in this study. CPT point 7202 occurs within amajor quick-clay landslide scar (noticeable in Lidar data). Black lines show the location of existing geophysicalprofiles. Reflection seismic data from lines 5 and 4 are presented and compared with our results in this study. Modified from Malehmir et al. (2013a, 2013b).

33A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

well as geoelectrical and CPTmeasurements (Löfroth et al., 2011).We usethe CPT data presented by Löfroth et al. (2011) in this study. In order toprovide insight on the applicability of the various soil classificationmethods for assessment of soil types and states using CPTu site investi-gations we tested eight CPT based classification methods. Our objectivewas to determine which one(s) provide the highest resolution soilprofile and if detection of sensitive clay is possible. After the analysiswe compared the results of each method with the laboratory UnifiedSoil Classification System (USCS) provided by Löfroth et al. (2011).We show that among the tested methods, the proposed ones byRobertson et al. (1986), Robertson (2010), Brouwer (2007) and Zhang

and Tumay (1999) give good results, but the original method proposedby Robertson et al. (1986) has the highest quality.

2. Soil type classification methods

In this study the proposed methods by Robertson et al. (1986),Robertson (1990), Olsen and Mitchell (1995), Swedish NationalReport (1995), Eslami and Fellenius (1997), Jefferies and Been (2006),Brouwer (2007) and Robertson (2010) have been used and tested. Weprovide here the basic ideas behind each of the mentioned soil classifi-cation methods and the various formulas employed by them. Note that

Table 1Summary of field and laboratory measurements in the study area (data from Löfroth et al., 2011).

Testpoint

Depth(m)

Ground water table(GWT) (m)

Undrained shearstrength (Kpa)

Clay content(%)

PH Mass of bulkmaterial (g)

Mass of centrifugedpore water (g)

Centrifuged porewater (% bulk)

7201 37.98 1 0.06–0.63 63–71 8.2–9.3 516.2–546.7 76.6–116.4 14–237202 19.99 1.7 0.43–2.8 53–69 7.4–7.6 546.5 96.7 187203 22.58 1 0.28–2.9 12–58 7.4–8.17206 33.86 1 0.22–6.82 6–80 7.1–8.47208 37.98 1 0.51–4.7 55–75 8.3–9.4 453–566 76.4–81.8 147209 30.98 1 0.08–1.2 38–64 7–9.2 547.8–548.6 89.9–101.7 16–19

34 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

soil type and state have interchangeably been used in the literature andsometimes in this paper too. With the soil type we refer to grain sizedistribution (e.g., clay, silt, sand) and the soil state to the geotechnicalproperties or behavior of the soil (e.g., dense, contractive, liquefiable).

Robertson et al. (1986): Robertson et al. (1986) and Campanella andRobertson (1988) were among the first to present a chart based on thepiezocone with the cone resistance corrected for pore pressure at theshoulder. By knowing the cone resistance (qc), friction ratio (Rf) andsleeve friction (fs), Robertson et al. (1986) and Campanella andRobertson (1988) proposed Eq. (1), which can be used to estimate soiltype using CPT data:

Rf ¼f sqc

� 100: ð1Þ

The Robertson et al. (1986) profiling chart identifies regions (codedby numbers) that separate soil types into twelve zones. A novel fea-ture in the profiling chart is the delineation of zones 1, 11, and 12,representing the soil response, thus enabling the CPTumethod to deter-mine more than just the grain size. The separation between the zones(e.g., zones 3 through 10) indicates a gradual transition from fine-grained to coarse-grained soil.

Based on Robertson et al. (1986), but also ASTM standard D5778-95(2000) and D5778-12 (2012), the corrected cone resistance for porepressure (qt) can be obtained using Eq. (2), which includes the mea-sured pore pressure (u2) and the ratio between the unaffected conebase and the pore water pressure to total shoulder area (a):

qt ¼ qc þ u2 1−að Þ: ð2Þ

Fig. 2. CPTu plots from the three test points in the stud

The Robertson et al. (1986) profiling chart introduces a pore pres-sure ratio (Bq) that is defined by Eq. (3). By considering the in-situpore pressure (u0) and the total and effective overburden stress (σV

and σV′), Robertson et al. (1986) and Robertson (1990) proposed,although they initially opposed this way of calculating Bq (Wroth,1988), that for the calculation of the Bq, normalized cone resistance(qcnrm or Qt1) and normalized friction factor (Rcnrm or Fr) Eqs. (4) and(5), respectively, can be used:

Bq ¼u2−u0

qt−σv¼ Δu

qt−σvð3Þ

qcrnm ¼ Qt1 ¼ qt−σv

σ 0v

ð4Þ

Rcnrm ¼ Fr ¼f s

qt−σ 0v: ð5Þ

By considering the water pressure at the base of the sleeve (u2), thecross sectional area of the sleeve at the base (Asb), the water pressure atthe top of the sleeve (u3), the cross sectional area of sleeve at the top(As) and the surface area of the sleeve (As), Lunne et al. (1997) sug-gested Eqs. (6) and (7) to correct for the sleeve friction (ft) and theerror due to unequal areas (Δfs), respectively:

f t ¼ f s−u2 � Asb−u3 � Astð Þ

Asð6Þ

y area. See the location of the test points in Fig. 1.

35A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

Δ f s ¼ f t− f s ¼u2 � Asb−u3 � Ast

As: ð7Þ

The Bq-chart shows directly zones where the u2 pore pressuresbecome smaller than the initial pore pressures (u0) in the soil duringthe advancement of the penetrometer, resulting in negative Bq-values.Otherwise, the Bq-chart appears to be an alternative, rather than anauxiliary chart; one can use one or the other depending on preference.However, near the upper envelopes, a CPTu data point falling in aparticular soil-type zone in the friction ratio chart will not alwaysappear in the same soil-type zone in the Bq-chart. Robertson et al.(1986) points out that “occasionally soils will fall within differentzones on each chart” and recommends that the user studies the porepressure rate of dissipation (if measured) to decide which zone appliesto the questioned data.

Robertson (1990): proposed a refined version of the Robertson et al.(1986) profiling chart, plotting the normalized cone resistance (qcnrm)against the normalized friction ratio (Rcnrm) in a cone resistance chart.The numbered regions in the profiling chart separate the soil states(or behavior) into nine zones. The two first and two last soil types arethe same as those used by Robertson et al. (1986) and zones 3 through7 correspond to former zone 3 through 10. The Robertson (1990)

Fig. 3. Variations of corrected tip resistance versus sleeve friction and friction ratio for t

normalized profiling chart has been extensively used in engineeringpractice (as has the Robertson et al., 1986 chart). Schneider et al.(2008) showed that Δu/σ′v is a better form for the normalized porepressure parameter than Bq. However, according to Robertson (2012),the use of Δu/σ′v is not a good approach for estimating the over consol-idation ratio (OCR).

Robertson et al. (1986) and Robertson (1990) stressed that the CPT-based charts are predictive of soil behavior, because the cone respondsto the in-situ mechanical behavior of the soil and not directly to soil-type classification criteria, using geological descriptors that are oftenbased on grain-size distribution and soil plasticity (Robertson, 2012).

Olsen and Mitchell (1995): carried out research on normalization ofthe cone penetration resistance. Their work incorporated over 2 de-cades of field data and an extensive database of chamber test studiesby other researchers to deduce the tip normalization for a number ofdifferent soil types. They produced a strong basis for cone normaliza-tion, and subsequent researchers (e.g., Robertson and Wride, 1998)have referred to this body of work when addressing normalization(Cai et al., 2011). A limitation of normalization based on empiricaldata is that a soil layer must be “uniform” and extend over a sufficientdepth range to be of use in calculating the normalization exponent. Nor-malization data in sands are generally restricted to chamber test resultsbecause of the inherent variability of in-situ deposits due to the typical

he three selected test points; 7202 (A and D), 7203 (B and E) and 7206 (C and F).

36 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

depositional environment associated with this type of soil. For clays,field tests are the only option because of the difficulty of performing ac-curate chamber studies. For soils other than sands and clays, such assilts, high silt content sands, and clayey sands, it is difficult, if not impos-sible, to find “uniform” layers in the field or to produce quality chambertest specimens to generate normalization data.

Swedish National Report (1995): provides a history of investigationof soil strata by penetration testing in Sweden. In the years 1914–1922,the Geotechnical Commission of the Swedish State Railways developedand standardized the Swedish Weight Sounding Method (the term“sounding” is used in the same sense as “penetration testing”).With the aid of this equipment, in combination with sampling, it be-came possible to obtain reliable information about the stratificationof soil layers.

Fig. 4.Variations of excess pore pressure versus corrected cone tip resistance and pore pressuredetailed description of the figure.

This preliminary soil classification based on CPT tests is thus mainlymade on the basis of the relation between the cone resistance, thesleeve friction and the normal in-situ stress conditions. In soft clay, themeasured sleeve friction is very small and relatively unreliable, but inmore consolidated clay, where the cone resistance may be of the samesize as for soft, coarser soil, themeasured values normally become largerandmore reliable. Possible uncertainties in themeasurements of sleevefriction normally have a relatively small influence on this determina-tion. The main exceptions are highly sensitive clays and/or silty-clays.In these soils, the sleeve friction may be very low; at the same time asthe measured stiffness in relation to the overburden pressure placesthe soil in the region for silt. However, very high pore pressures areoften developed in the tests and a check on whether the factor Bq ishigher or lower than 0.6 can be used to judge whether the soil should

with net cone resistance for 7202 (A andD), 7203 (B and E) and 7206 (C and F). See text for

Table 2Correlation level of different soil classification methods and their respective appliedparameters used in this study.

Method Applied parameters Correlation(%)

Robertson et al.(1986)

Corrected cone resistance, friction and porepressure ratio

75

Robertson (1990) Normalized friction ratio and normalized coneresistance

–

Olsen and Mitchell(1995)

Friction ratio and normalized cone resistance 35

Swedish NationalReport (1995)

Pore-water pressure ratio and net coneresistance

55

Eslami and Fellenius(1997)

Sleeve friction and effective cone resistance 50

Jefferies and Been(2006)

Friction ratio and dimensionless penetrationresistance

40

Brouwer (2007) Friction ratio and cone resistance 70Robertson (2010) Both normalized and non-normalized friction

ratios and cone resistance70

Zhang and Tumay(1999)

CPT based soil classification system based onfuzzy logic

70

Fig. 5. Comparison of updated Soil Behavior Type (SBT) charts based on corrected cone resistance

37A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

be classified as silt or sensitive clay. In those cases where the soil hasbeen classified as “clay/organic soil” the classification process passesover to the special classification chart developed for this type of soil.This chart is based on the net cone resistance and Bq.

Eslami and Fellenius (1997): developed a soil profiling methodwhen investigating the use of cone penetrometer data for pile designs.They compiled a database consisting of CPT and CPTu data obtainedfrom boring, sampling, laboratory testing, and routine soil characteriza-tion cases from 18 sources reporting data from 20 sites in 5 countries.They introduced a chart based on non-normalized parameters usingeffective cone resistance (qe) and fs, where qe= (qt − u2). The databasevalues are divided into fivemain soil type categories (sensitive and col-lapsible clay and/or silt, clay and/or silt, silty clay and/or clayey silt,sandy silt and/or silty sand, sand and/or sandy gravel).

The CPT methods (mechanical cones) do not correct for the porepressure on the cone shoulder and the profiling based on CPT datamay not be relevant outside the local area where they are developed.The error due to omitting the pore water pressure correction is largein fine-grained soils and smaller in coarse-grained soils (Fellenius and

(Robertson et al., 1986; A, B and C) and non-normalized CPT (Robertson, 2010; D, E and F).

Fig. 6. Comparison of updated Soil Behavior Type (SBT) chart based on normalized CPT parameters (Robertson, 2010) for (A) 7202, (B) 7203 and (C) 7206, respectively, and (D) soil statemodified from Robertson (1990).

38 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

Eslami, 2000). The effective cone resistance (qe) suffers from the lack ofaccuracy in soft fine-grained soils (Robertson, 2012).

Jefferies and Been (2006): updated an earlier chart by Jefferies andDavies (1993), using the parameter Qt1 (1 − Bq) + 1, to overcome theproblem in soft sensitive soils where Bq N 1 (Eq. (8)). The parameterQt1 (1− Bq) + 1 is simply the effective cone resistance given by:

Qt1 1−Bq

� �þ 1 ¼ qt−u2

σ 0v0

: ð8Þ

Jefferies and Davies (1993) and Robertson and Wride (1998) indi-cated that for ISBT (Soil Behavior Type Index) that Eq. (9) can be applied.

ISBT ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

3:47− logqcpa

� �� �2þ logRf þ 1:22� �2� �� �s

: ð9Þ

Jefferies and Davies (1993) suggested that the SBT index (Ic) couldalso be used to modify empirical correlations that vary with soil type

and proposed a normalization that incorporates the pore pressure di-rectly into a modified normalized cone resistance using Qt1 (1 − Bq)as presented in Eq. (10). Note that the reliability of IC as an index forme-chanical behavior classification of soil has been investigated by Ku et al.(2010). This is a powerful concept and has been usedwhere appropriatein this study.

IC ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

3− log Qt 1−Bq

� �� �� �2 þ 1:5þ 1:3 logFrð Þ2� �� �s

: ð10Þ

To determine the soil type index, Robertson andWride (1998), laterupdated by Zhang et al. (2002), suggested a more generalized normal-ized cone parameter to evaluate soil liquefaction, using normalizationwith a variable stress exponent (n; n ≤ 1).

Qtn ¼ qt−σv

pa� pa

σ 0v

� �nð11Þ

39A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

n ¼ 0:381 ISBTð Þ þ 0:05σ

0

v

pa

!−0:15 ð12Þ

where (qt − σv)/Pa is a dimensionless net cone resistance; Pa/σ′v is astress normalization factor and Pa is an atmospheric pressure with thesame units as for qt and σv.

Olsen and Malone (1988), Robertson (1999), Zhang et al. (2002),Boulanger and Idriss (2004), Moss et al. (2006) and Cetin and Isik(2007) agreed that in the clean sand region, the stress exponent (n) istypically close to 0.5 and in the clay region the stress exponent is closeto 1.0. Only the normalized SBT chart suggested by Jefferies and Been(2006) uses a stress normalization of n = 1.0.

Zhang and Tumay (1999) developed a CPT based soil classificationsystem based on fuzzy logic where the results are presented in theform of percentage probability for clay, silt or sand. Although thisapproach is conceptually attractive and provides some estimate of un-certainty for each SBT zone, results are often misinterpreted as grainsize distribution.

Fortunately, soil classification criteria based on grain-size distribu-tion and plasticity often relate reasonably well to in-situ soil behaviorand hence, there is often good agreement between USCS-based classifi-cation and CPT-based SBT (Molle, 2005). However, several examplescan be given where differences can arise between the two (Robertson,2012).

Brouwer (2007): provided information about cone penetration test-ing, soil sampling techniques and instrumentation on land and rail,near-shore and offshore. More than 25 cone penetrometers, instrumen-tation devices and samplers were described.

Robertson (2010): presented an update to the Robertson et al.(1986) chart to aid in comparison with the normalized chart suggestedby Robertson (1990). One of the major applications of CPT has been thedetermination of soil stratigraphy and the identification of soil type. Thishas typically been accomplished using charts that link cone parametersto soil type. Robertson (2010) proposed a new non-normalized Soil Be-havior Type index (ISBT) that uses the basic non-normalized CPT results.The non-normalized SBT index (ISBT) is essentially the same as the nor-malized SBTn index (Ic), suggested by Robertson andWride (1998), butonly uses the basic CPTmeasurements. In general, the normalized Icwillprovidemore reliable identification of SBT than the non-normalized ISBT,but when the in-situ vertical effective stress is between 50 kPa and150 kPa there is often little difference between the normalized andnon-normalized SBT. The normalized Soil Behavior Type chart andSBTn index (Ic) are recommended for later post-processing of CPT re-sults, but the non-normalized chart and SBT index (ISBT) can be helpfulfor real-time data processing and interpretation.

Fig. 7. Classification of the soil/state type based

3. Study area and available data

Our study area is next to the Göta River valley (Fig. 1) that was likelyformed through erosion of aweak zonewithin the gneissic bedrock. Theregion has experienced several landslides (Millet, 2011; Nadim et al.,2008; Klingberg, 2010). In parts of the valley, the gneissic rocks, withsome elements of diabase and granite, outcrop, but, in general, theyare mostly covered by post- and glacial sediments. The post-glacial sed-iments consistmainly of silt and clay deposits, sometimes verymassive.The glacial clay deposits that have been erodedmay also be over consol-idated (Klingberg, 2010). There is normally a geologic andmorphologicdistinction between the southern and northern parts of the river. In ourstudy area, a thick layer of glacial clay dominates the soil profile down tothe bedrock. This layer is partly underlain by glaciofluvial deposits, till ordifferent combinations of friction soils (Klingberg et al., 2006; Sundborgand Norrman, 1963). Due to the geological history of the valley, wheremuch of the clay was deposited in a marine environment, there is asignificant presence of so-called quick clay in the area. The quick clayexists as layers or planes in clay sediments with normal sensitivity,but in many places the entire clay stratum is comprised of quick clay(Lindskog, 1983).

Our test site (Fig. 1) is about 7 km north of the town of Lilla Edet,where the Göta landslide occurred in 1957. This area has been recentlystudied by the Swedish Geotechnical Institute (SGI) in a nation-wideproject, which included investigations of areas prone to sliding alongthe Göta River (Löfroth et al., 2011). A total of 9 CPTu geotechnical testpoints (Fig. 1) with a maximum depth of 38 m and a groundwatertable between 1 and 1.7 m were made available through this study.The CPTu test points consisted of 7201, 7203 and 7205 in the easternpart of a 30–40 year-old landslide scar, 7202 in the landslide scar and7206, 7207 and 7208 in the western part of it (Fig. 1).

The geotechnical laboratory tests for the classification of the soil andgrain size distribution, determination of density, water content, liquidlimit and undrained shear strength in the undisturbed and remoldedstate by fall cone tests, chemical analysis and pore water extractionwere carried out by SGI (Löfroth et al., 2011). A summary of the lab-oratory analysis is given in Table 1 and CPTu test data for three testpoints are shown in Fig. 2. We considered test points 7202, 7203and 7206 (Figs. 1 and 2) because of their locations in the landslidescar region.

4. Interpretation of CPTu data and soil classifications

CPTu data obtained from the study area show a large variation ofcone resistance below 11 m depth for 7202, 12 m for 7203 and 16 m

on the Swedish National Report (1995).

Fig. 8. Soil/state-type classification based on the proposed chart by Brouwer (2007).

Fig. 10. Distribution of available data for soil-type description according to Jefferies andBeen (2006).

40 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

for 7206 (see ellipses 1 and 2 in Fig. 3A, B, and C). Hence, before analyz-ing the soil classifications, several correlations between corrected coneresistance and measured CPTu parameters were first established. Wecarried out this to investigate which one of the three soil groups (gran-ular, cohesive and cohesionless soils) that our data belong to. On thebasis of Begemann (1965), Sanglerat et al. (1974) and Schmertmann(1975) a variation of the friction ratio in the range of 2 to 5% indicatesa mixture of clay-sand and silts. In the test points of the selected area,the friction ratio varies between 0 and 2.4% (ellipse 1 in Fig. 3D, E, andF). High corrected cone resistance values are normally associated withcohesionless soils such as silt and sand, in which friction ratio tends todecrease with cone resistance (ellipse 2 in Fig. 3D, E, and F).

As presented in Fig. 4, the corrected cone resistance and excess porepressure for the test points show values in the range of 0–5000 KPa forthe corrected tip resistance, the excess pore pressure is predominantlypositive with an increasing trend (see ellipse 1, in Fig. 4A, B, and C),but for values greater than 5000 KPa, the excess pore pressure shows

Fig. 9. Soil-type classification based on the proposed chart by Eslami and Fellenius (1997)and Fellenius and Eslami (2000).

a decreasing trend (see ellipse 2, in Fig. 4A, B, and C). This shows thatthe soils are either dense or over-consolidated, which can produce a de-creasing trend or even negative pore water pressure during cone pene-tration testing. Cai et al. (2011) indicated that high negative porepressures are also possible in the case of very dense fine sands and dila-tive silts.

The variation of pore pressure ratio (Bq), originally formulated byLunne et al. (1997), can be used to identify the soil type and Bq N 0.60indicates clayey, silty and sandy soils (ellipse 2, in Fig. 4D, E and F). Pos-itive Bq (ellipse 1 in Fig. 4D, E and F) indicates the development of excesspositive pore pressure during cone penetration testing and, typically,such soils are classified as being soft to medium consistency clayeysoils (Sharma and Ilamparuthi, 2009).

Fig. 11. Soil/state-type classification based on the proposed chart by Olsen and Mitchell(1995).

41A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

5. Evaluation of CPTu-based soil classification methods in our studyarea

In this paper the proposed classificationmethods by Robertson et al.(1986), Robertson (1990), Robertson (2010), Jefferies and Been (2006),and Eslami and Fellenius (1997), traditional site investigation tech-niques (Swedish National Report, 1995); (Brouwer, 2007); (Olsen andMitchell, 1995) were used and analyzed for the classification of soiltypes and states in the area. Different characteristics of the appliedmethods are summarized in Table 2.

Three typical chart results containing the distribution of CPTu dataand interpretations based on the Robertson et al. (1986), Robertson(1990) and Robertson (2010) methods are presented in Figs. 5 and 6.According to Robertson et al. (1986), the soil type in 7202 varies be-tween zones 1, 6, 7, 8 and 9 and rarely goes to zone3. At 7203 a variationbetween zones 1, 5, 6, 7 and 9 is observedwith only very fewdata pointsfalling into zone8. Based on Robertson (2010), soil types in 7202 changein zones 1, 3, 4, 5 and6, for 7203 vary in zones 1, 3, 4, 5, 6 and 7206 showchanging in zones 1, 3, 4, 5, 6, and 7. Considering Robertson (1990), thesoil states vary between zones 3 to 7, but have predominance in zones 5and 6 (i.e., more sensitivity and nearly normally consolidated).

As indicated in Fig. 7, based on traditional site investigation tech-niques presented by the Swedish National Report (1995), the soiltypes are classified as very silty clay, clays or slightly over consolidatedsilty clays and highly sensitive clays, but most generally map into thefields of normally consolidated clays or slightly over consolidated siltyclays. Based on Brouwer (2007), as presented in Fig. 8, the soils are

Fig. 12. Identified soil profile for test point 7202, (A) Soil Behavior Type Index comparedwith Uprofile indicate sensitive fine grained clays. Color scales are similar to those shown in Fig. 5).

classified as very silty and loose to medium coarse sand and graveland, sometimes, dense gravel. The proposed chart by Eslami andFellenius (1997) and then Fellenius and Eslami (2000) showsmore sen-sitivity in the soils and, according to Fig. 9, a variation between clayeysilt, silty clay, sandy silt and sand can be observed. Considering Fig. 10,the proposed method by Jefferies and Been (2006) for soil classificationshows a wide variation from silty clay to clayey up to gravelly sand. Theproposed chart by Olsen andMitchell (1995) shows variations from siltmixtures to sand and gravel (Fig. 11).

In order to choose and evaluate the different results of the differentmethods, a criterion needed to be developed. We used a success rateconcept for this that is the ratio of the number of correct predictionsto the total number of predictions classifying soils as either coarse-grained or fine-grained. The correct prediction was defined based onthe results from the USCS method (which is based on the laboratorymeasurements and CPTu data) provided by Löfroth et al. (2011) (seeFigs. 12A, 13A and 14A). This analysis showed (for summary seeTable 2) a 75% success rate using Robertson et al. (1986), 70% usingRobertson (2010), 55% using the Swedish National Report (1995), 70%using Brouwer (2007), 50% using Eslami and Fellenius (1997), 40%using Jefferies and Been (2006), 35% using Olsen and Mitchell (1995),and 70% using Zhang and Tumay (1999). Therefore, we considered theRobertson et al. (1986) chart for our study, even though an updatedversion was presented by Robertson (2010).

After selecting the Robertson et al. (1986) chart as our base conceptfor soil classification, we converted all the results, i.e., soil types ob-tained from each method, into depth for determination of soil type

SCS data (dark red dots), (B) Soil Behavior Type zones and (C) soil profile (red zones in soil

42 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

thickness. Results from Robertson et al. (1986) were then comparedwith the USCS data from Löfroth et al. (2011) in order to evaluatetheir match. If there was a match then the solution from Robertsonet al. (1986) would be accepted, otherwise the nearest match fromthe other seven methods was considered to be the correction thatwas required to Robertson et al. (1986).

Because the eight employed methods do not provide comparablegrain-size classification to the USCS method, we defined a criterion inthis study that converts grain sizes obtained from the eight methodsto the USCS soil groups. In this criterion soft clay and clay mixtures,respectively, represent CL and CL–MLgroups (inUSCS). For siltmixturesincluding sandy silt and clayey silt, the ML group in USCS was consid-ered. For sand and sand mixtures consisting of silty sand, gravellysand to sand (medium to dense) and clean sand, and silty-clayey-sand, respectively, the SM and SC–SM groups were considered. There-fore, in our study, zone 3 from Robertson et al. (1986) represents CLand CL–ML groups, zone 4, zone 5 and zone 6 represent SM and SC–SMand zone 7 represents the GM group (Figs. 12B, 13B and 14B).

6. Evaluation of the results

To evaluate the obtained results (Figs. 12C, 13C and 14C), we decid-ed to estimate SPT data from the CPTu data (Fig. 2). The prediction wasdone on the basis of the corrected friction ratio and corrected cone resis-tance (Fig. 15), which uses the chart proposed by Olsen (1988). Thisallowed predicting the SPT-N value (N1) data. From the predicted SPTdata, we then obtained corrected SPT data on the basis of (N1)60 which

Fig. 13. Identified soil profile for test point 7203, (A) Soil Behavior Type Index compared with Usoil profile indicate sensitive fine grained clays. Color scales are similar to those shown in Fig. 5

is 60% of the hammer energy that penetrated 30 cm depth (Olson andStark, 2003). We chose the proposed criterion by Olson and Stark(2003), because it is formulated in terms of (N1)60 and allows for adirect comparisonwith liquefaction susceptibility. Visual inspection be-tween the results from the predicted SPT and soil profiles obtained fromCPT data suggests a good correlation between the two (Fig. 16A, B andC). Moreover, the predicted soil profiles based on the USCS and (N1)60in the selected test points of this study (Fig. 16A, B and C) can be com-pared with the probabilistic method introduced by Zhang and Tumay(1999) (Fig. 17). A relatively good match can also be observed betweenthe probabilistic approach and the CPTu results (e.g., Fig. 17). Note thatwe only show the results for CPTu 7203, but for the other two testpoints similar results were obtained.

7. Discussion

Landslides, as a geological phenomenon are a significant geohazardin Sweden and can occur in special sensitive fine-grained clays, namelyquick clays. Our approach in this study to detect possible sensitive fine-grained clays is promising. Obtained soil profiles, corrected on the basisof the applied methods and available USCS data, show that at test point7202 that some thin layers of sensitive clays at depths between 4m and11 m are present. For test point 7206, the amount of detected sensitivefine-grained clay is even more and the layers are thicker than at testpoint 7202. Depths range from 4 m down to 32 m. For test point 7203,a thick layer of sensitive fine-grained materials between 2 m and13 m is found. Test point 7202 is located in the slide area, perhaps

SCS data (dark green dots), (B) Soil Behavior Type zones and (C) soil profile (red zones in).

Fig. 14. Identified soil profile for test point 7206, (A) Soil Behavior Type Index comparedwith USCS data (dark blue dots), (B) Soil Behavior Type zones and (C) soil profile (red zones in soilprofile indicate sensitive fine grained clays. Color scales are similar to those shown in Fig. 5).

Fig. 15. Prediction of SPT-N values for the selected test points.

43A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

explainingwhy there is lesser amounts of sensitive clays in this locationthan the other two points. This is also reflected in the number of soiltypes identified in this point compared to the two others. Themain rea-son for this difference is interpreted to be that some 15–20m of soil hasbeen removed by the slide (Fig. 1).

On the basis of geophysical data and the occurrences of a coarse-grained layer (detected by reflection seismic methods), Malehmiret al. (2013a, 2013b) suggested the possibility of quick clays formingat about 10 m depth in the area around 7202, about 20 m depth near7203 and 25 m depth near 7206. A comparison between the reflectionseismic results and our data shows some consistency (Fig. 18) suggest-ing sensitive fine-grained materials in similar depth ranges and locatedabove a coarse-grained layer which is generally reflective (Fig. 18).Coarse-grained materials (gravelly sand to sand) (e.g. Fig. 12) werealso detected by the geophysical and seismic methods (Fig. 18). Theover consolidation ratio (OCR) (Fig. 18) shows a sudden increase atthe location of the coarse-grained layer (likely patchy and not continu-ous in large scale) at the selected test points.

The coefficient of permeability using the CPTu data for the coarse-grained layers shows higher values than the other identified layers.This has implications for the proposed scenarios by Malehmir et al.(2013a) concerning the generation and formation of quick clays at thesite. A plausible scenario for the formation of quick clays is that thecoarse-grained materials act as a conduit with direct infiltration (eitherfrom the surface through the clays or from the contact between theclays and the bedrock) towards the Göta River. During the process and

44 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

in the case of excess water, a confined aquifer forms (or becomes overpressured) and leaches salt from the marine clays above. The leachingthen destabilizes the structure of the clay and forms the quick clay(Rankka et al., 2004). The excess water could also, at proper conditions,trigger a landslide at the site by increasing the pore pressure in thepores. This, however, may require other complementary processessuch as by human activities (as seen in a few cases) or sudden (or grad-ual) erosion in the riverbank. More CPTu test points spread out over thestudy area and hydrogeological experiments are required to fully un-derstand this process and verify these scenarios.

Soil classificationmethods using only the cone resistance and sleevefriction give a poor interpretation of the soil types, but by using porepressure and its correction to cone resistances the results for soil classi-fication are significantly improved. If a positive correlation betweencone resistance and sleeve friction (increasing the cone resistancewith sleeve friction) is observed, then this may imply that soil materialsare becoming harder. A careful inspection of Fig. 3 (3D, 3E and 3F) sug-gests an opposite trend (high cone resistances and low friction ratios),

Fig. 16. Computed N1(60) in this study and comparison with the soil profiles for the selected tegroup and light yellow represents silt mixtures group in USCS).

perhaps implying the dominance of cohesionless soils such as silty-sand and sandy soils and low tomedium cone resistances and high fric-tion ratios that indicate the presence of silt, clay and clayey soils.

As shown in ellipse 2 in Fig. 4D, E, and F, corrected cone resistancesless than 5000 KPa imply that the pore pressure ratio varies between0 and 2. This means the soil may be classified as silty sand or clayeysand. Ellipse 1 in Fig. 4D, E and F indicates that for cone resistancesmore than 5000 KPa the pore pressure ratio will go towards zero. Inthis case the soils can be classified as consolidated clay, clay mixturesand also medium dense sand. When Bq is applied the USCS data showbetter resolution and correlation with the employed classificationcharts. If Bq is not considered, then correlation is poor and the classifica-tion of soil based solely on cone resistance and sleeve friction does notcorrelate well in many cases.

Difficulties in taking into account soil mixtures when there are twoor more soil types are one of the main problems causing errors inCPTu charts for soil classifications. This is an inherent limitation becausesoil classification on the basis of cone test results for all soil types is not

st points (red layers corresponds the clay group, green indicates sand and sand mixtures

45A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

practical and perhaps impossible. Therefore, many of these charts arebasedmore on the soil behavioral types, such as the contracting or dilat-ing nature of the materials than the soil composition (Cai et al., 2011).As reported by Robertson et al. (1986), a recognized problem associatedwith soil classification charts on the basis of CPTu data is that a soil cangradually change its apparent classification as cone penetration in-creases with depth. This is because measured cone resistance, sleevefriction, and pore pressure all tend to increase with increasing overbur-den pressure. Since most of the data used to develop the soil classifica-tion charts in this study were obtained from soundings carried out atdepths of about 30 m we expect some errors, especially at greaterdepths. At great depths, however, normalized CPTu measurementswould be able to account for the effects of the overburden pressure.This implies that using CPT for greater depths provides a quick soiltype interpretation. It also implies fewer required samples for laborato-ry investigations and a resulting significant reduction in field operationscan be expected in site investigations. Soil classification in the laborato-ry using collected samples from boreholes is the primary conventionalmethod for determining soil stratigraphy, but the main challenge ofusing the CPT or CPTu data for soil profiling is that samples are not re-trieved for laboratory testing.

8. Conclusions

There are several classification methods on the basis of CPT data,which enable predicting soil type and state. However, the main pointis that their validity still needs to be verified for local cases becausethe original soils used in the soil classification chart development willgenerally be. In this study results of CPTu tests conducted in southwest-ern Swedenwere used to predict the soil depth profile in a region proneto quick-clay landslides. Eight soil classification methods based on CPTdata, including Robertson et al. (1986), Robertson (2010), Jefferies andBeen (2006), Swedish National Report (1995), Brouwer (2007),Eslami and Fellenius (1997), Olsen and Mitchell (1995) and Zhang

Fig. 17. Comparison of the obtained results in this study by USCS classification with Zh

and Tumay (1999), were applied for soil type and state prediction andinterpretation.

The present analysis shows that the applied classification methodsare capable of interpreting the subsurface soil type, but their accuracyshould be checked and verified. By taking into account that appliedCPT charts do not provide accurate predictions of soil types based ongrain size distribution, we employed the USCS laboratory results tohave better resolution for soil type determination. Using USCS data asa standard for soil type recognition in all of the appliedmethods impliesthat we can predict soil type from CPT data. In our case, a best agree-ment and compatibility was obtained using the Robertson et al.(1986) method. By converting the identified soil types to depth for theother methods used in this study, our base concept was then modified.To verify the results, wemade a comparisonwith the predicted SPT datafor the selected area and borehole soil types with the probabilistic basemethod of Zhang and Tumay (1999).

Considering that sensitive fine-grained clays, and especially quickclays, are responsible for most landslides in Sweden, this paper helpsto recognize and identify zones potentially comprising sensitive claysin the study area. The identified soil profiles, which estimate soil staterather than actual soil type based on the USCS, compare acceptablywell with the applied soil classification charts and are generally consis-tent with the results predicted by probabilistic methods (Zhang andTumay, 1999) suggesting some of the quick clays likely form above acoarse-grained layer that has been delineated by surface geophysicalstudies (Malehmir et al., 2013a, 2013b).

Acknowledgments

SGI kindly provided the CPT data for which we are grateful. A.Abbaszadeh Shahri carried out this research while visiting UppsalaUniversity as a post-doctoral fellow. Formas (through GeoInfra-Trust(252-2012-1907), www.trust-geoinfra.se) and SEG (Society ofExploration Geophysicists through its Geoscientists without Borders,www.seg.org/gwb) are gratefully acknowledged for funding A.

ang and Tumay (1999) proposed method. Only results for 7202 are shown here.

Fig. 18. 3D viewshowing two seismic sections (modified fromMalehmir et al., 2013a) in the vicinity of theCPT test points. The data clearly image an interpreted coarse-grained layer abovethe dipping bedrock surface (towards the river) and correlate well with the results (soil type and OCR) obtained in this study. The yellow dotted lines indicate the seismic line positionscorresponding to the topography of the area (see Fig. 1). The correspondence between the soil type classification obtained in this study and the reflector geometry strongly suggests that itis a coarse-grained layer generating the reflection.

46 A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

Malehmir's research contribution in this work. Critical reviews by twoanonymous reviewers and the editor, C.H. Juang, significantly improvedthe quality of our work and its presentation for which we are thankful.

References

Adamczyk, A., Malinowski, M., Malehmir, A., 2013. Application of first arrival tomographyto characterize a quick-clay landslide site in southwest Sweden. Acta Geophys. 61,1057–1073.

Adamczyk, A., Malinowski, M., Malehmir, A., 2014. High-resolution near surface velocitymodel building using full-waveform inversion — a case study from southwestSweden. Geophys. J. Int. 197, 1693–1704.

ASTM D5778-12, 2012. Standard Test Method for Electronic Friction Cone and PiezoconePenetration Testing of Soils. ASTM International, West Conshohocken, PA http://dx.doi.org/10.1520/D5778-12.

ASTM D5778-95, 2000. Standard Test Method for Performing Electronic Friction Cone andPiezocone Penetration Testing of Soils. ASTM International, West Conshohocken, PAhttp://dx.doi.org/10.1520/D5778-95R00.

Baligh, M.M., Vivatrat, V., Ladd, C.C., 1980. Cone penetration in soil profiling. J. Geotech.Eng. 112 (7), 727–745.

Begemann, H.K.S., 1965. The friction jacket cone as an aid in determining the soil profile.Proceedings, 6th International Conference on Soil Mechanics and Foundation Engi-neering, Montreal, Quebec, Canada. vol. I, pp. 17–20.

Boulanger, R.W., Idriss, I.M., 2004. State normalization of penetration resistance and theeffect of overburden stress on liquefaction resistance. Proceedings of 11th interna-tional conference on soil dynamics and earthquake engineering and 3rd internationalconference on earthquake geotechnical engineering. University of California, Berke-ley, CA, pp. 484–491.

Brouwer, H., 2007. In Situ Soil Testing, Chapter 2, CF Cone (Cone Penetration Test).Lankelma, UK 978-1-86081-951-3, p. 144 (JJ M Brouwer).

Cai, G.J., Liu, S.Y., Tong, L.Y., 2010. Field evaluation of deformation characteristics of alacustrine clay deposit using seismic piezocone tests. Eng. Geol. 116 (3), 251–260.

Cai, J., Liu, S., Puppala, A.J., 2011. Comparison of CPT charts for soil classification using PCPTdata: example from clay deposits in Jiangsu Province. China Eng. Geol. 121, 89–96.

Campanella, R.G., Robertson, P.K., 1988. Current status of the piezocone test. Proceedingsof First International Symposium on Penetration Testing, ISOPT-1, 1, (93–116).Orlando, March 22–24.

Cetin, K.O., Isik, N.S., 2007. Probabilistic assessment of stress normalization for CPT data.J. Geotech. Geoenviron. Eng. 133 (7), 887–897.

Cetin, K.O., Ozan, C., 2009. CPT-based probabilistic soil characterization and classification.J. Geotech. Geoenviron. Eng. 135 (1), 84–107.

Douglas, B.J., Olsen, R.S., 1981. Soil classification using electric cone penetrometer. Pro-ceedings of Symposium on Cone Penetration Testing and Experience, GeotechnicalEngineering Division, ASCE. St. Louis, Missouri, October 1981, pp. 209–227.

Eslami, A., Fellenius, B.H., 1997. Pile capacity by direct CPT and CPTu methods applied to102 case histories. Can. Geotech. J. 34 (6), 880–898.

Fellenius, B.H., Eslami, A., 2000. Soil profile interpretation form CPTu data. Proceeding ofYear 2000 Geotechnics Conference, Southeast Asian Geotechnical Society, Asian Insti-tute of Technology, Bangkok, Thailand. vol. 1, pp. 163–171.

Jefferies, M.G., Been, K., 2006. Soil Liquefaction — A Critical State Approach. Taylor &Francis (ISBN 0-419-16170-8 478).

Jefferies, M.G., Davies, M.O., 1991. Soil classification by the cone penetration test: discus-sion. Can. Geotech. J. 28 (1), 173–176.

Jefferies, M.G., Davies, M.P., 1993. Use of CPTU to estimate equivalent SPT N60. Geotech.Test. J. ASTM 16 (4), 458–468.

Jones, G.A., Rust, E., 1982. Piezometer penetration testing CUPT. Proc., 2nd EuropeanSymp. on Penetration Testing, ESOPTII, pp. 607–613.

Jung, B.C., Gardoni, P., Biscontin, G., 2008. Probabilistic soil identification based on conepenetration tests. Geotechnique 58 (7), 591–603.

Klingberg, F., 2010. Bottenförhållanden i Göta Älv: SGU-rapport 2010:7. SverigesGeologiska Undersökning, Göteborg.

Klingberg, F., Påsse, T., Levander, J., 2006. K43: Bottenförhållanden och geologiskutveckling i Göta älv. Sveriges Geologiska Undersökning, Göteborg.

Ku, C.S., Juang, C.H., Ou, C.Y., 2010. Reliability of CPT Ic as an index for mechanical behav-ior classification of soils. Geotechnique 60 (11), 861–875. http://dx.doi.org/10.1680/geot.09.P.097.

Lindskog, G., 1983. Brief Report of the Investigation of the Slope Stability Along the Riverin Göta River Valley. Statens Geotekniska Institut, Linköping.

Löfroth, H., Suer, P., Dahlin, T., Leroux, V., Schälin, D., 2011. Quick clay mapping by resis-tivity-surface resistivity, CPTU-R and chemistry to complement other geotechnicalsounding and sampling. Swedish Geotechnical Institute, report GÄU 30.

Long, M., 2008. Design parameters from in situ tests in soft ground — recent develop-ments. Proceedings of Geotechnical and Geophysical Site Characterization. Taylor &Francis Group, pp. 89–116.

Lundberg, E., Malehmir, A., Juhlin, C., Bastani, M., Andersson, M., 2014. High-resolution 3Dreflection seismic investigation over a quick-clay landslide scar in southwest Sweden.Geophysics 79, B97–B107.

47A. Abbaszadeh Shahri et al. / Engineering Geology 189 (2015) 32–47

Lunne, T., Robertson, P.K., Powell, J.J.M., 1997. Cone Penetration Testing in GeotechnicalPractice. Blackie Academic, EF Spon/Routledge Publ., New York (312 pp.).

Malehmir, A., Saleem, U.M., Bastani, M., 2013a. High-resolution reflection seismic investi-gations of quick-clay and associated formations at a landslide scar in southwestSweden. J. Appl. Geophys. 92, 84–102.

Malehmir, A., Bastani, M., Krawczyk, C.M., Gurk, M., Ismail, N., Polom, U., Persson, L.,2013b. Geophysical assessment and geotechnical investigation of quick-clay land-slides — a Swedish case study. Near Surf. Geophys. 11, 341–350.

Millet, D., 2011. River Erosion, Landslides and Slope Development in Göta River, a studyBased on Bathymetric Data and General Limit Equilibrium Slope Stability Analysis.(M.Sc Thesis). Department of Civil and Environmental Engineering, Division ofGeoEngineering, Geotechnical Engineering Research Group, Chalmers university ofTechnology, Göteborg, Sweden.

Molle, J., 2005. The accuracy of the interpretation of CPT based soil classification methodsin soft soils. (MSc Thesis). Section for Engineering Geology, Department of AppliedEarth Sciences, Delf University of Technology, Report No. 242, Report AES/IG/05-25,December.

Moss, R.E.S., Seed, R.B., Olsen, R.S., 2006. Normalizing the CPT for overburden stress.J. Geotech. Geoenviron. Eng. 132 (3), 378–387.

Nadim, F., Pedersen, S.A.S., Schmidt-Thomé, P., Sigmundsson, F., Engdahl, M., 2008.Natural hazards in Nordic countries. Episodes 31 (1), 176–184.

Olsen, R.S., 1988. Using the CPT for dynamic site response characterization. In: Von Thun,J.L. (Ed.), Earthquake Engineering and Soil Dynamics II— Recent Advances in GroundMotion Evaluation: American Society of Civil Engineers Geotechnical Special Publica-tion. 20, pp. 374–388.

Olsen, R.S., 1994. Normalization and Prediction of Geotechnical Properties Using the ConePenetration Test. (PhD Dissertation). University of California at Berkeley, Berkeley,California.

Olsen, R.S., Koester, J.P., 1995. Prediction of liquefaction resistance using the CPT. Proc.,International Symposium on Cone Penetration Testing, CPT 95, Linkoping, Sweden,pp. 251–256.

Olsen, R.S., Malone, P.G., 1988. Soil classification and site characterization using the conepenetrometer test. In: De Ruiter, J. (Ed.), Penetration Testing 1988, ISOPT-1. Balkema,Rotterdam, The Netherlands, pp. 887–893.

Olsen, R.S., Mitchell, J.K., 1995. CPT stress normalization and prediction of soil classifica-tion. Proceedings of the International Symposium on Cone Penetration Testing (CPT'95), Linkoping, Sweden, pp. 251–256.

Olson, S.M., Stark, T.D., 2003. Yield strength ratio and liquefaction analysis of slopes andembankments. J. Geotech. Eng. Am. Soc. Civ. Eng. 129 (8), 727–737.

Rankka, K., Andersson-Sköld, Y., Hultén, C., Larsson, R., Leroux, V., Dahlin, T., 2004. QuickClay in Sweden. Report 65. Swedish Geotechnical Institute, Linköping.

Robertson, P.K., 1990. Soil classification using the cone penetration test. Can. Geotech. J.27 (1), 151–158.

Robertson, P.K., 1999. Estimation of minimumundrained shear strength for flow liquefac-tion using the CPT. Proc., 2nd Int. Conf. on Earthquake Geotechnical Engineering.Balkema, Rotterdam, The Netherlands.

Robertson, P.K., 2004. Evaluating soil liquefaction and post-earthquake deformationsusing the CPT. Keynote Lecture at International Conference on Geotechnical andGeophysical Site Characterization, ISC-2, Porto, Portugal.

Robertson, P.K., 2009. Discussion of ‘CPT-based probabilistic soil characterization andclassification’ by K. Onder Cetin and Cem Ozan. J. Geotech. Geoenviron. Eng. 135(1), 84–107.

Robertson, P.K., 2010. Soil behaviour type from the CPT: an update. 2nd InternationalSymposium on Cone Penetration Testing, CPT'10, Huntington Beach, CA, USA.

Robertson, P.K., 2012. Interpretation of In-Situ Tests-Some Insights. Mitchell Lecture, ISC'4Brazil, Sep., 2012.

Robertson, P.K., Wride, C.E., 1998. Evaluating cyclic liquefaction potential using the conepenetration test. Can. Geotech. J. 35, 442–459.

Robertson, P.K., Campanella, R.G., Gillespie, D., Greig, J., 1986. Use of piezometer cone data.In-Situ '86 Use of In-situ testing in Geotechnical Engineering, GSP 6, ASCE, Reston, VA,Specialty, Publicationpp. 1263–1280.

Sanglerat, G., Nhim, T.V., Sejourne, M., Andina, R., 1974. Direct soil classification by staticpenetrometer with special friction sleeve. Proc., the First European Symposium onPenetration Testing. ESOPT-1, Vol. 2.2, June 5–7, 1974, Stockholm, pp. 337–344.

Schmertmann, J.H., 1975. Measurement of in situ shear strength. Proc., Specialty Confer-ence on In Situ Measurement of Soil Properties: ASCE. vol. 2. ASCE, USA, pp. 57–138(Raleigh).

Schneider, J.A., Randolph, M.F., Mayne, P.W., Ramsey, N.R., 2008. Analysis of factorsinfluencing soil classification using normalized piezocone tip resistance and porepressure parameters. J. Geotech. Geoenviron. Eng. 134 (11), 1569–1586.

Senneset, K., Janbu, N., 1985. Shear strength parameters obtained from static cone pene-tration tests. Proc., Strength Testing of Marine Sediments; Laboratory and In-SituMeasurement: ASTM Special Technical Publication, STP. 883, pp. 41–54.

Shan, C., Bastani, M., Malehmir, A., Persson, L., Engdahl, M., 2014. Integrated 2D modelingand interpretation of geophysical and geotechnical data to delineate quick clays at alandslide site in southwest Sweden. Geophysics 79, EN61–EN75.

Sharma, M.S.R., Ilamparuthi, K., 2009. Interpretation of electric piezocone data of ChennaiCoast. Ocean Eng. 36 (6), 511–520.

Sundborg, Å., Norrman, J., 1963. Göta Älv hydrologi och morfologi- med särskild hänsyntill erosionsprocesserna. Statens Geologiska Undersökning, Avhandlingar ochuppsatser I 4:0 Nr 43, Stockholm, 1963.

Traditional site investigation techniques, Excerpt from "Swedish National Report", In Pro-ceedings of the International Symposium on Cone Penetration Testing (CPT '95),Linkoping, Sweden, Vol.1.

Tumay, M.T., Boggess, R.L., Acar, Y., 1981. Subsurface investigation with piezoconepenetrometer. ASCE GSP on Cone Penetration Testing and Experience, St Louis,MOpp. 325–342.

Wroth, 1988. Penetration Testing— AMore Rigorous Approach to Interpretation. Penetra-tion Testing, ISOPT-1. 90 6191 8014.

Zhang, Z., Tumay, M.T., 1999. Statistical to fuzzy approach toward CPT soil classification.J. Geotech. Geoenviron. Eng. 25 (3), 179–186.

Zhang, G., Robertson, P.K., Brachman, R.W.I., 2002. Estimating liquefaction inducedground settlements from CPT for level ground. Can. Geotech. J. 39 (5), 1168–1180.

Zuidberg, H.M., Schaap, L.H.J., Beringen, F.L., 1982. A penetrometer for simultaneouslymeasuring of cone resistance, sleeve friction and dynamic pore pressure. Proceedingsof the Second European Symposium on Penetration Testing Vol. 2. CRC Press,Amsterdam, pp. 963–970.