Quarterly Journal of Engineering Geology and Hydrogeology-2009-Parry-499-510

Quarterly Journal of Engineering Geology and Hydrogeology

doi: 10.1144/1470-9236/08-010 2009, v.42; p499-510.Quarterly Journal of Engineering Geology and Hydrogeology

S. Parry and J.R. Hart facing the profession in Hong KongEngineering geology and the reduction of geotechnical risk: challenges

serviceEmail alerting to receive free e-mail alerts when new articles cite this article hereclick

requestPermission to seek permission to re-use all or part of this article hereclick

SubscribeHydrogeology or the Lyell Collection

to subscribe to Quarterly Journal of Engineering Geology andhereclick

Notes

© The Geological Society of London 2014

by Rodney Xerri on January 15, 2014http://qjegh.lyellcollection.org/Downloaded from by Rodney Xerri on January 15, 2014http://qjegh.lyellcollection.org/Downloaded from

http://www.lyellcollection.org/cgi/alerts

http://www.lyellcollection.org/cgi/alerts

http://www.geolsoc.org.uk/permissions

http://www.geolsoc.org.uk/permissions

http://www.lyellcollection.org/site/subscriptions

http://www.lyellcollection.org/site/subscriptions

http://qjegh.lyellcollection.org/




Engineering geology and the reduction of geotechnicalrisk: challenges facing the profession in Hong Kong

S. Parry* & J.R. HartGeoRisk Solutions Ltd, Suite 02, 15/F, Hollywood Centre, 233 Hollywood Road, Sheung Wan, Hong Kong,

PR China*Corresponding author (e-mail: [email protected])

Abstract: There has been considerable discussion on the role of engineering geologists in reducinggeotechnical risk both internationally and in Hong Kong. This paper discusses the problems facing theengineering geological profession by examining the current approach to natural landslide assessments inHong Kong. Although the problems discussed are based on a Hong Kong perspective many areapplicable elsewhere in the world. The assessment and mitigation of landslide risk from natural slopes isa rapidly expanding area of geotechnical work in Hong Kong. Under relatively new legislation, the HongKong Government now requires a natural terrain hazard study to be undertaken for all new developmentsclose to steep natural terrain. In addition to the assessment of new developments Hong Kong has recentlystarted a programme to assess the risk from natural terrain landslides to existing developments. However,with respect to engineering geology, the profession faces a number of challenges including skillsshortages and insufficient training opportunities, and, more broadly, challenges related to certain contractconditions and prevailing market forces. The combination of these factors in Hong Kong has resulted inthe reduction of engineering geological input within the geotechnical industry. This may result in theadoption of over-conservative mitigation works with resulting impacts on cost, safety and the environment.In the worst case, the lack of engineering geological input may result in the failure to recognize thepotential for low-frequency, high-magnitude landslides, with potentially severe future consequences.

The essence of engineering geology and, by association,an engineering geologist is not easy to define or codify,because of the wide scope of the subject, the partiallyimplicit nature of the work, and the dependence on thetraining and experience of the practitioner (Knill 2002;Baynes & Rosenbaum 2004).

However, a key role of engineering geology is toinvestigate and interpret the ground conditions relevantto an engineering project so that the ground can bereliably characterized in engineering terms and anypotential geological hazards identified to minimize geo-technical risks (Morgenstern 2000). As a starting pointthis requires appropriate academic knowledge to evalu-ate the ground from both a geological and an engineer-ing context. However, more importantly, it requiresskilled observations, interpretations and analyses, whichcan only be gained with practical training and experi-ence. However, the availability of this training andexperience is often controlled by external factors varyingfrom market forces and legal requirements to technicalspecifications and tendering methods.

This paper considers the above issues as they affectengineering geologists working on natural landslide riskassessments in Hong Kong. Although this paper iswritten from a Hong Kong and landslide assessmentperspective, the authors consider that many of the issuesdiscussed are equally applicable both to other areas ofthe profession and internationally.

EducationEngineering geology is rarely taught as an undergradu-ate subject and most practitioners undertake a first

degree in geological science followed by an engineering-based postgraduate degree. Until 1996 all engineeringgeologists in Hong Kong were educated, and most wererecruited from, overseas as there were no local geologydegree courses. In 1997 the first earth science studentsbegan graduating in Hong Kong. However, as withelsewhere in the world, concerns have been raised asto whether today’s all-encompassing ‘earth science’,modular-based degree courses provide sufficient depthin the core components of geology (Clarkson 2004).Without these core components graduates entering theprofession or undertaking a postgraduate study in engi-neering geology are at a distinct disadvantage that canbe only partially overcome with further training andexperience (with suitable mentoring) over a relativelylong period of time (Ng et al. 2007).

Market forcesUnfortunately, the availability of local earth sciencegraduates coincided with a downturn in the constructionindustry as a result of the completion of the majorconstruction programmes associated with Hong Kong’sinternational airport and associated infrastructure in1997, closely followed by the Asian financial crisis in1998; both leading to a reduction in infrastructureconstruction expenditure.

During this downturn, the mainstay of geotechnicalwork in Hong Kong became the Landslip PreventiveMeasures (LPM) programme, which focused on theinvestigation, assessment and stabilization of man-made

Quarterly Journal of Engineering Geology and Hydrogeology, 42, 499–510 1470-9236/09 $15.00 � 2009 Geological Society of LondonDOI 10.1144/1470-9236/08-010

by Rodney Xerri on January 15, 2014http://qjegh.lyellcollection.org/Downloaded from



(i.e. cut and fill) slopes. However, the knock-on effect ofa large geotechnical industry chasing a decreasing poolof work has been that the fees have decreased consider-ably, with obvious consequences for the technical inputfor such work. Martin (2003) noted that a formulaicapproach is commonly adopted for the LPM to mini-mize cost, and this is associated with a reduction inengineering geological input. In particular, the vastmajority of slope stabilization work in Hong Kongutilizes soil nails and there seems to be a belief within theindustry that this technique can compensate for alladverse engineering geological conditions.

Despite the findings of Martin (2003) the situationhas not improved. A review of landslides between 1997and 2001 noted that ‘the use of an over-simplifiedgeological and/or hydrogeological model which does notadequately cater for safety-critical features in the groundis the most important cause of major failures of engi-neered cut slopes’ (GEO 2003). Similar findings withrespect to the lack of engineering geological input werenoted during a government review of the LPM process(Parry et al. 2004), with the following problems high-lighted: (1) the relative lack of integration, and henceover-compartmentalization, in engineering geologicalassessments, potentially resulting in information appar-ent at an earlier stage not being acted upon later; (2) thegeneral lack of pertinent engineering geological observa-tions and subsequent interpretation during the aerialphotographic interpretation (API), and thus commonfailure to take account of the geological and geomor-phological setting of the site; (3) the common lack of aclearly stated strategy for ground investigations, whichdo not fully address geological uncertainties; (4) thevariable quality of ground investigation data withrespect to material and mass descriptions; (5) the useof over-simplified geological models, and thereforethe potentially unrepresentative geotechnical modelsadopted for design; (6) the limited information presentedto demonstrate that the adopted design approach wasverified by pertinent engineering geological observationsduring construction.

In summary, the use of engineering geology in geo-technical work has been adversely affected by prevailingmarket forces since 1997 within the geotechnical andcivil engineering industry in Hong Kong, and numbersof suitably qualified engineering geologists havedeclined. This is a worrying trend given the results of thelandslide and LPM reviews discussed above. The 2008–2009 worldwide economic recession has not yet had anysignificant impact on the industry in Hong Kong.Indeed, the Hong Kong Government have pledged tobring on-line and fast-track additional civil engineeringprojects in light of the market downturn. However, thereare already worrying signs that increased competitionfor this work will lead to further reductions in fees,which may further exacerbate the challenges facing theengineering geology profession.

TrainingIn addition to fundamental geological knowledge,engineering geology demands skilled observations, inter-pretations and analyses, often based on limited infor-mation. Ng et al. (2007) noted that engineering geologyis akin to an apprenticeship, where such skills areacquired by appropriate training and supervision frommore experienced practitioners. The sustainability ofengineering geological practice therefore depends on theavailability of the more experienced practitioners andthe amount of time that they have to train and supervise.Although a few companies did establish trainingschemes based on that developed by the GeologicalSociety of London (GSL) (GSL 1994), the scheme islimited in that it is focused on training graduates forground investigation rather than in engineering geologyper se. Furthermore, such schemes require mentors forguidance.

As a result of the downturn in the constructionindustry after 1997, many experienced engineeringgeologists have left Hong Kong. Consequently, manylocal earth science graduates work under the supervisionof engineers who often have little or no knowledge ofengineering geology. Furthermore, many of these gradu-ates are being exposed only to the formulaic approach toLPM and as a result they have limited opportunities todevelop engineering geological skills. As a result there isa trend for local earth science graduates to be restrictedto the role of data gatherers. Perhaps one reason for thisis that local engineering companies receive a governmenttraining subsidy of US$650 per month for each graduateemployee on the Hong Kong Institution of Engineers(HKIE) training scheme, effectively subsidizing theirtraining. There is no financial support for geologicalgraduate training. Training opportunities for would-beengineering geologists are limited to a minority ofcompanies, and few experienced engineering geologistsremain as potential mentors.

Professional qualifications andregistration

There is currently no professional body for engineeringgeologists in Hong Kong. Previously the HKIE Geo-technical Discipline accepted applications from engi-neering geologists. However, this route for qualifyingengineering geologists ceased with the HKIE accessionto the Washington Accord, which requires applicants tohave an engineering degree. Consequently, many engi-neering geologists obtained Chartered Engineer statusthrough the Institute of Materials, Minerals and Mining(IMMM). However, the increasing engineering contentof first degree courses, in particular the introduction ofMEng courses, has made this route more difficult inrecent years. Consequently, a significant number ofprofessional geologists now obtain Chartered Geologist

S. PARRY & J.R. HART500




(CGeol) status through the GSL. The local earth sciencedegree course is not accredited with the GSL andconsequently 6 years postgraduate experience is requiredrather than 5 years before a candidate can apply forCGeol. In comparison, engineering graduates can applyfor full membership of HKIE 1 year after completion ofthe 3 year training course (i.e. after 4 years). Further-more, CGeol does not require a formal examinationand until 2008 could be confirmed without an interviewif the candidate has 10 years experience. This, rightlyor wrongly, results in the attainment of CGeol beingperceived as a less rigorous method of qualificationas compared with HKIE membership. Although theGovernment and some companies do accept IMMM orGSL qualifications as equivalent qualifications to mem-bership of HKIE, they are not accepted in Hong Kongas being a suitable professional qualification for signing-off geotechnical works, on the grounds that they areawarded overseas and that they do not confer qualifica-tions in the discipline of engineering geology (Massey2001). As a result, Membership of the GeotechnicalDiscipline of the HKIE is the de facto professionalqualification for geotechnical practitioners, and throughthis the attainment of Registered Professional Engineer(Geotechnical) status or RPE(G). The guidelines fornatural terrain hazard studies (NTHSs) (Ng et al. 2003)state: ‘as an NTHS forms part of a geotechnical submis-sion, it should be carried out under the charge of ageotechnical engineer who is professionally qualified inHong Kong and who has suitable local experience innatural terrain hazard studies. A suitable qualification isRegistered Professional Engineer (Geotechnical).’ As aresult, there is a trend for local earth scientists toattempt to become geotechnical engineers, with therecognition that brings, rather than engineering geolo-gists. For example, the course literature for the onlyMSc in ‘Applied Geoscience’ in Hong Kong states that‘the Engineering Geology with HKIE ApprovedCourses theme provides the additional courses whichgraduates in Earth Sciences or Geology would need tomeet the entry requirements of the Hong Kong Institu-tion of Engineers in the Geotechnical Discipline’. As aresult of the above, there is a trend for potential

engineering geologists to attempt to train as engineersgiven this profession’s perceived higher status.

Geotechnical control and standardsFollowing large landslides with multiple fatalities fromman-made slopes in the 1970s, a system of geotechnicalcontrol within the construction industry was establishedin Hong Kong in 1977. Under this system geotechnicalreports for both private and public projects are sub-mitted for checking to the Government. To assist withthe geotechnical control a whole series of design manu-als and documents have been produced (GEO 2000).Although these were originally intended as guidance andstandards of good practice, they have now become defacto geotechnical standards. Consequently, adherenceto them generally ensures a smooth passage through thechecking system. Although this system is useful for themass processing of slope upgrading works, it does notgenerally lend itself to innovative geotechnical designbased on site-specific conditions. For example, Poulos(2007) noted that ‘whilst such an approach facilitatesacceptance of design and thus is efficient from the view-point of time it inhibits innovation and may endanger anunwarranted degree of confidence’. As a result there islittle incentive for the use of site-specific engineeringgeological input to develop innovative designs even ifsuch an approach could result in cost savings.

The geotechnical control system has now beenextended to the checking of an NTHS. However, thereare considerable differences between the assessment offoundations or man-made cut and fill slopes and theassessment of natural terrain for landslide hazards(Table 1). An NTHS will result in limited ‘facts’ and isinstead dependent upon site observations, assessmentand interpretation. A further key difference is thatalthough ‘hard’ engineering in the form of soil nailingmay overcome limitations inherent in the LPM pro-gramme for man-made slopes, particularly where engi-neering geological input is low, such a ‘safety net’ maybe neither economically nor environmentally suitable forwidespread use on natural slopes.

Table 1. Comparison of the investigations of man-made slopes and natural slopes

Man-made slope assessment Natural slope assessmentSite of limited extent Sites have a large extent, often comprising multiple catchmentsGround investigation (GI) stations are closely spaced Limited scope for GI given large site and difficult access; thus

it is very dependent on being located in critical areasExposures are available either before or during the GI, or

during constructionExposure limited to rock outcrop, landslide scars and drainage

linesConsiderable amount of published data on geotechnical

propertiesRelatively limited data on the behaviour of natural landslides

in Hong KongIt may be appropriate to use simple classification of material

types (e.g. ‘colluvium’)Simple classifications are inappropriate; classifying the

superficial deposits requires an understanding of landscapeevolution and geomorphological processes

Well-developed software for slope stability analysis Software programs are not appropriate for catchment-wideapplications

REDUCTION OF GEOTECHNICAL RISK 501




Although the principle of geotechnical control isgood, in reality it can lead to a formulaic approach toboth the undertaking and ‘checking’ of the work. Inslope engineering this may be appropriate to an extent,but in an NTHS such an approach may result in anincomplete study being acceptable; for example, if adegraded landslide scar is not identified because ofinadequate aerial photograph interpretation or fieldmapping, say, how can this be ‘checked’?

Tenders and specificationsIn the current fiercely competitive market in HongKong, consultancy fees represent only a small fractionof the construction costs. As such, engineering geologi-cal input, which is commonly regarded as ‘specialist’or ‘non-standard’, is often reduced. For example, theGovernment guidelines on improving the reliability ofslope design state ‘the geotechnical professional respon-sible for the design review should consider the need toseek further specialist advice or second opinion from anexperienced engineering geologist where deemed appro-priate (e.g. for slopes with very complex ground andhydrogeological conditions)’ (GEO 2003). (Note that itis the ‘geotechnical practitioner’ (normally an engineer)who decides what is ‘complex’.)

The Hong Kong approach to construction workstypically involves a feasibility consultancy followed by aseparate design consultancy. The design may be under-taken by a different consultant from the one that carriedout the feasibility study or it may be undertaken by acontractor if the project is tendered as a design and build(D&B) contract. Often in a D&B contract there islimited time at the tender stage and the contactor isheavily reliant on the feasibility report, despite thecommonly included clause that the contractor shouldsatisfy themselves that the ground conditions are asstated in the report. Consequently, if the feasibilityreport is incomplete or imprecise, the burden of risk canbe considerable.

However, regardless of the type of contract, ofteninsufficient time is allowed for the tenderer. Forexample, a recent tender for an NTHS for an area of 150ha required the presentation of the detailed naturalterrain hazard study and recommendation of mitigationmeasures within 5 weeks of commencement of the study.A second tender for an area of 220 ha required thesuccessful consultant to review the findings of the feasi-bility report, all relevant documentary records, ‘carryingout any necessary diagnosis, detailed field inspectionsand detailed aerial photograph interpretation’, under-take probabilistic assessment of debris runout usingdynamic mobility modelling techniques, and plan andprovide supervision of any ground investigation works,laboratory tests and groundwater monitoring required,all within 8 weeks of commencement of the work.

In summary, the problems the profession is facing canbe exacerbated by unrealistic targets with respect tospecifications and time scale. These contractual prob-lems have been further discussed by Baynes (2007).

Requirements for natural landslideassessments in Hong Kong

In Hong Kong a natural terrain hazard study (NTHS) isrequired for new developments close to steep naturalterrain and where changes in land use result in anincrease in risk; for example, redevelopment with ahigher population density. Furthermore, the Govern-ment recently started a long-term programme (theLandslip Prevention and Mitigation (LPMit) Pro-gramme) to assess the risk from natural terrain land-slides to existing developments. Consequently, thedemand for high-quality engineering geological input isgrowing rapidly. This has been discussed in more detailby Parry & Ng (2010).

Although guidelines exist for the type of data to bedocumented for a landslide assessment in Hong Kong(Ng et al. 2003; GEO 2007), ensuring that the correctdata are both recorded and correctly interpreted reliesextensively on skill and judgement. Unlike many regionsin the world where landslide assessments are carried outat regional scales and a degree of error is generallyacceptable, the majority of Hong Kong’s landslideassessments are facility specific and may be only a fewhectares in size. As a result, the appropriateness andreliability of hazard and risk assessments is dependenton the skill of the engineering geologist. For example,the NTHS guidelines (Ng et al. 2003) note three possibleapproaches; a ‘factor of safety approach’, which isusually problematic to apply because of limited geotech-nical data; a ‘quantitative risk assessment (QRA)approach’, which is commonly perceived to be involvedand potentially costly (Ho 2004); and the ‘design eventapproach’, which is most commonly adopted. Thedesign event approach is a formulaic approach based onconsideration of slope geometry, facility type and rela-tive landslide activity. Based on these factors, either a‘conservative event’ with a notional return period of theorder of 100 years, or a ‘worst credible event’ with anotional return period of the order of 1000 years, isadopted as a design event. However, these are oftenequated to ‘recent’ and ‘relict’ landsides and the assess-ment simply becomes the adoption of the largest ofthe applicable landslide type within the catchment.Although this approach may be suitable at the initialstages of a study to provide a rapid assessment of thepotential magnitude of hazards and the risks from thosehazards, it should be re-evaluated following the engi-neering geological mapping.





Engineering geological input fornatural landslide assessments

The above discussion has highlighted the challenges ofeducation, supply and training of engineering geologistsand their use in the geotechnical industry in Hong Kong.The following part of the paper discusses various com-ponents of the assessment of landslide hazard and risk inHong Kong to illustrate the how these challenges aremanifesting themselves.

Desk study

Although comprehensive data gathering is normallyundertaken as part of an NTHS, these data arecommonly poorly evaluated (Parry et al. 2006). Thesynthesis of pertinent data is often critical to the identi-fication of key engineering geological factors at a specificlocation. An isolated report may hold key data but thismay be overlooked because of the desk study beingperceived as something that can be left to the leastexperienced staff with little supervision.

For example, a desk study was undertaken for a sitethat had been subject to two previous landslide hazardstudies and is being redeveloped privately for luxuryhousing. The published 1:20 000 scale geological mapshows a large area of ‘debris flow deposits’ below thesite, and a large depression is apparent, from a reviewof the available topographical maps, in the hillsideabove the area of ‘debris flow deposits’. The presence ofthe depression and the associated superficial depositsbelow suggest that the site may be located within a large,albeit degraded, landslide scar. However, previousground investigation data simply refer to the superficialdeposits as undifferentiated ‘colluvium’, although aquick re-examination of these data indicates that the‘colluvium’ can be subdivided into two broad types: (1)cohesive fine-grained colluvium (typically a silty clay),probably originating from slope wash or numeroussmall landslide deposits over considerable time; (2)clastic colluvium comprising cobbles and boulders,possibly representing debris from a large-magnitude,high-energy, relatively rapid landslide event(s). Theclastic colluvium is >9.5 m thick and in parts is inter-bedded with fine material, suggesting that there mayhave been more than one episode of failure. Thisreinterpretation of the original data is illustrated inFigure 1.

On the basis of the published geological map and theexisting ground investigation it is possible that much ofthe site is situated within a very large landslide(s) andthat the thick clastic colluvium represents landslidedebris associated with this. The potential presence of thisfeature was not identified in either of the previoushazard studies and consequently the potential for afurther large-scale failure of the adjacent, as yet unfailedterrain, was not addressed.

As a second example, the reports associated with theGeotechnical Area Study Programme (GASP) carriedout in the late 1980s, particularly the District Studies at1:2500 scale (Styles & Hansen 1989), contain a wealth ofinformation. However, the GASP studies focused ongeneral geotechnical limitations to development and notspecifically on landslide hazard. Consequently, thesedata need to be carefully interpreted for site-specificlandslide assessments. Figure 2 indicates slopes abovesea cliffs adjacent to Cape Collinson in the east of HongKong Island. Under GASP the slopes are classified asClass IV (extreme geotechnical limitations) where land-slides have already occurred, with the intervening landbetween landslides shown as Class III (high geotechnicallimitations). The actual cliffs are shown as Class II(moderate geotechnical limitations) and yet they aresubject to considerable oceanic wave action and continu-ing marine erosion. From a geomorphological perspec-tive this location is a classic process–response systemwhereby the response of the foreshore system to thismarine erosion is undercutting of the cliffs followed bystructurally controlled rock falls. These in turn result indestabilization of the overlying soils forming the slopesabove, which has resulted in debris slides in the past.These same processes are operating along the entiresection of coast and therefore the whole terrain unit islikely to have a similar probability, magnitude andfrequency of landsliding.

Typically, existing information is presented at facevalue without due consideration of its origins andintended applications. This is examined further withrespect to landslide data, below. The engineering geol-ogist should be involved in every stage of the desk studyto ensure that: (1) critical data are not overlooked; (2)data are used appropriately and with caution; (3) dataare synthesized to highlight the key engineering geologi-cal factors at a given site, which should be explored andexamined further beyond the desk study; (4) areas ofuncertainty are identified and recorded.

Existing landslide inventories

The GEO has developed detailed territory-wide land-slide inventories since the mid-1990s. The first was theNatural Terrain Landslide Inventory or NTLI, whichwas derived from an interpretation of all high-level(flight height >10 000 feet) aerial photographs (King1999). However, there are limitations with this dataset,in particular its use of high-level photographs only andits identification of ‘relict’ landslides (vegetated in thefirst year of identification; as opposed to ‘recent’ land-slides, which are devoid of vegetation in the first year ofcoverage) predominantly based on relatively poor-quality, high-level aerial photographs taken in 1945.Consequently, a new dataset was produced using bothhigh- and low-level aerial photographs. This forms theEnhanced Natural Terrain Landslide Inventory or





Fig. 1. Reinterpretation of existing drillhole logs at a luxury development site, suggesting the presence of extensive landslide debris,as opposed to homogeneous ‘colluvium’.

Fig. 2. Coastal landslides in east Hong Kong Island and comparison with the GASP map. Class II, moderate geotechnicallimitations; Class III, high geotechnical limitations; Class IV, extreme geotechnical limitations.





ENTLI (Maunsell Fugro Joint Venture 2007). Theresulting dataset is greatly improved, in particular inthe identification of ‘relict’ landslides, which are nowbased on high-quality territory-wide, low-level 1963aerial photographs.

However, there are still limitations to this improvedinventory. For example, although guidelines were pro-duced to improve consistency with respect to the inter-pretation of relict landslides (Parry et al. 2006), theinterpreters’ classification varied with experience. Fur-thermore, given the project constraints (e.g. over 105 000aerial photographs were reviewed in 16 months) theinterpreters could examine them only for relatively clearevidence of landslides (e.g. obvious scarps) with verylittle time available to consider geomorphological set-tings, often critical with respect to the identification ofolder degraded and possibly landslide-related features.

A further landslide inventory in Hong Kong is theLarge Landslide Database (Scott Wilson 1999), whichwas generated primarily by API and focused on land-slides >20 m in width. However, in a recent regionalnatural terrain hazard study some 11 out of 23 landslidesfrom this database were reassessed, again by API, as notbeing large landslides (Maunsell Fugro Joint Venture2004). It is unclear whether this discrepancy reflects theexperience of the interpreters, as the original largelandslide database was compiled by experienced over-seas interpreters with little local experience whereas theregional study was undertaken by less experienced butlocal interpreters. Field mapping of the features was notundertaken for either study or for the ENTLI and,additionally, there was no attempt to systematicallyassess the large landslide database in the ENTLI study.

There is concern that the existing landslide invento-ries, which have been developed without field verifica-tion, will be used mechanically and without appreciationof their compilation methods and limitations, and thatcritically, degraded landforms, which may represent high-magnitude, low-frequency landslide events and that arenot identified in the datasets, will be overlooked. Figure 3indicates the positions of landslides within the existingdatabases and the additional landslides identified duringthe site-specific API. In the example, a possible largedebris lobe was identified below several possible degradedsource areas, which are not included in the landslideinventories. Based on the field mapping of this feature themorphology and materials indicate that the lobe couldhave been deposited as a single large rock avalanche.Both API and field mapping are discussed below.

Although landslide databases exist, the data areindicative only. The limitations in the data must beunderstood and accounted for when examined. How-ever, most of these limitations can be overcome duringthe detailed studies associated with the NTHS. In par-ticular, verification of landslides identified during deskstudies and API is critical and should be carried out byengineering geologists experienced in landslides.

Aerial photograph interpretations

The limitations of API have been well documented bothoverseas (Fookes et al. 1991; Hart et al. 2009) and inHong Kong (Parry et al. 2006). As Fookes et al. (1991)noted, ‘all aerial photographic interpretation must beviewed with caution. When well done and confirmed byground truthing exercises it can be a very valuable toolon any site. However, it should be borne in mind that itis an interpretation and hence subject to the interpreters’training, experience, skill and judgement as well as otherfactors such as the availability of additional infor-mation, type, scale and quality of the photographs, timeavailable and the extent, if any of ground checkingwork.’ Despite these limitations, in Hong Kong there isoften a misconception that aerial photographs display‘facts’, which can be extracted by staff inexperienced inAPI and geomorphological or geological mapping. As aresult, many APIs typically record recent site develop-ments but do not address the evolution of the landscapeand the geomorphological and geological processesinvolved, all of which are critical to the understanding oflandslide hazards (Parry & Ruse 2000).

Even with the use of more experienced interpreters,there are limitations with respect to the application ofAPI alone to the interpretation of landslide-relatedfeatures (Fookes et al. 1991; Hart et al. 2009). Forexample, the identification of relict landslides requirescareful judgement, as landslide scars progressively losetheir morphological definition, at rates depending onfactors such as natural re-vegetation, landslide size,material and location. Furthermore, evidence of land-slide debris is commonly obscured by erosion or veg-etation growth. The certainty with which a feature canbe interpreted and identified as a landslide will varydepending upon on a number of factors including thefollowing.

(1) Original source volumes. Larger failure scars tendto be preserved longer in the landscape. Sewell &Campbell (2005) reported that the upper bound age forthe relict landslides they examined was 34 000 years .With respect to a lower bound age, Evans et al. (1997)suggested that a landslide source would typically be 90%re-vegetated after 20 years.

(2) The presence of debris below a scarp. This is theclearest evidence that a landslide has occurred. How-ever, unless the landslide is relatively large in size or thedebris is deposited as levees outside the drainage line,fluvial erosion commonly results in the removal orreworking of the debris with time.

(3) The sharpness of the scarp. Landslide scarps tendto degrade with time, although this may be affected bysubsequent minor failures of the oversteepened scarp area.

(4) The position of the depression within the land-scape. Landslides are commonly interpreted as suchwhen a depression is inconsistent with the adjacentlandform.





The identification of relict landslides requires verycareful judgement given that landslide scars progres-sively lose their morphological definition at rates thatare dependent on factors such as vegetation, landslidesize, material characteristics and location, and theevidence for associated landslide debris is rarely evi-

dent because of erosion or vegetation growth. Identifi-cation from API is also problematic for recentlandslides where complete detachment does not occur(refer to the example in Fig. 4b). Also, there is atendency to overestimate recent landslide runout fromAPI, as a result of post-landslide, fluvial erosion of the

Fig. 3. (a) Landslides identified in the various inventories. (b) Site-specific API identified a possible large degraded landslide eventnot included in the landslide inventories shown in (a). (c) Engineering geological mapping indicated that lobate landforms within thepossible landslide feature identified from API, have separate origins. For example, a distinct lobate deposit of angular tosub-angular, slightly to moderately decomposed, clast-supported cobbles and boulders was mapped below a large concavedepression. The field evidence suggests that this deposit may have been formed by a large rock avalanche.





landslide source area and reworking of the landslidedebris.

Alternative remote sensing approaches have beenexamined in Hong Kong but have not been widelyadopted to date because of the excellent quality andcoverage of the available aerial photographs. However,a trial of airborne LIDAR has recently been undertakenand this dataset is proving to be valuable especially inareas where dense vegetation is present in the aerialphotographs. However, the processing of LIDAR datais not straightforward and specialist input is required toensure that techniques such as digital elevation model(DEM) gridding and shaded relief maximize the relevantfeatures, as well as ensuring that the data are notmisinterpreted (Parry & Jonas 2007). That said, LIDARhas its limitations (refer to Fig. 4b), like any remotesensing technique, and at the site-specific scale, fieldmapping is essential.

In summary, although API is an extremely useful andimportant tool in landslide hazard assessments, theresolution of the photographs restricts the study to adetailed overview of a site and provides focus for theexamination and evaluation of key features as part ofthe detailed field assessment. As with the review ofexisting data, the API should be carried out by engineer-ing geologists experienced in landslides.

Engineering geological maps

In the authors’ opinions, engineering geological map-ping, including limited ground investigation, is the keyto the identification of landslide hazard and the subse-quent evaluation of risk at a site-specific scale. Much ofthe focus of previous landslide assessments in HongKong has been on identification of landslide locations,

predominantly from API, and often without field inspec-tions. As a result, these assessments emphasize wherelandslides have occurred in the recent past (last 80 years)and locations of possible relict landslide activity, thefiner and critical details of which are extremely difficultto ascertain from API alone. The emphasis is thus onwhere failures have occurred rather than where failurescould occur. With the reliance on routine API andAPI-generated databases there is the potential for high-magnitude, low-frequency landslide events to be missed.These concerns were also raised by the Slope SafetyTechnical Review Board (SSTRB 2007). Furthermore,when considering the possibility of such landslide events,it is not only the initial landslide volume but also thepotential for both multiple source volumes and/orsignificant material entrainment that also have to beconsidered. Although in Hong Kong the most recentchannelized debris flows have not involved significantentrainment, this could occur with the right combinationof geological and geomorphological conditions. Forexample, the 1990 Tsing Shan debris flow comprised aninitial landslide of 450 m3, which induced a secondlandslide of 2500 m3. The debris subsequently acceler-ated over a cliff, resulting in extensive entrainment of thematerials below, with the final volume estimated at19 000 m3 (King 1999).

Careful field mapping can also identify single land-slide events and therefore allow the selection of amore representative volume for the design of landslidemitigation measures. Ideally, the field work should becarried out by the same personnel that carried outthe API. Based on the field mapping, focused groundinvestigation, particularly trial pitting, is likely to berequired to analyse the superficial deposits to determinetheir origin, relative age, and process of deposition.

Fig. 4. Identification and evaluation of landslide events during field mapping and ground investigation; (a) shows a series oflandslide events interpreted from trial pit investigation through a debris lobe; (b) shows extensive tension crack and minor dis-placement associated with an area of distressed ground that was not identified from API or LIDAR but only during field mapping.





Landslides are extremely complicated events, which canbe associated with a variety of depositional environ-ments (Corominas et al. 1996). Unfortunately, there is atendency in Hong Kong to categorize all deposits associ-ated with mass-wasting processes as ‘colluvium’, whichcan lead to considerable overestimations of the size ofprevious landslide events. This has already been high-lighted from separate studies illustrated in Figures 1and 3. In another landslide hazard study, detailed engi-neering geological field mapping located a debris lobeassociated with a recorded 1966 landslide within densevegetation. However, given the limited exposure in thefield, trial pits were excavated to assess the depth of thedebris. These revealed evidence of previous mass move-ments prior to 1966 (Fig. 4a). The horizon beneath the1966 debris comprised sand and well-sorted, subroundedgravel with a well-developed imbrication fabric, suggest-ing that it was deposited by fluvially dominated pro-cesses, possibly during a debris flood. The lowest layercomprised cobbles and boulders, suggesting a relativelyhigh-energy landslide event. This layer is underlain bydense structureless colluvium. Each layer exhibited atopsoil horizon and the stratigraphy suggests that land-slide events in this drainage line may have been decreas-ing in energy, and changing from debris flow to debrisfloods, with time.

During the same study, an area of distressed groundwas observed during field mapping above the main scarpof another 1966 landslide, which, although recorded inthe NTLI, was not recorded in the ENTLI. Althoughthe volume of the landslide debris mobilized from thesource area was relatively small (20 m3), a larger volume(130 m3) of ground moved but did not detach signifi-cantly. The surface expression of this movement is a25 m long, up to 1.5 m high tension crack (Fig. 4b),which was not recorded in any of the landslide invento-ries nor from the site-specific API given that the groundis covered with dense vegetation. Interestingly, thisfeature was also not identified from LIDAR (Parry &Jonas 2007).

In addition to slope angle, one of the main controlson debris runout and therefore the potential hazardfrom any channelized debris flows is the nature anddimensions of the drainage channel. A useful tool forassessing the potential for debris channelization (andtherefore longer runout potential), is the measurementof the channelization ratio (CR) (average channel width/depth at any given section of channel). Unfortunately,many studies in Hong Kong fail to take into account theinterrelationship between CR and landslide volumes,and CR is commonly estimated from the dimensions ofoften poorly developed drainage channels, resulting inan over-conservative CR and excessive estimates ofrunout along channels that are not realistic.

A further problem is the age of relict landslides.Commonly, many large-scale relict landslide scarps areassumed to be associated with environmental conditions

that differ from those occurring today or are assigned tobe in excess of 1000 years old and therefore beyond the‘worst credible event’ designated in the design eventapproach. Consequently, they are removed from thehazard assessment on this basis. However, only limitedabsolute age dating of landslides has been undertaken inHong Kong, which, combined with the current knowl-edge of climate change, and the lack of scientific rigour,makes such assumptions problematic.

At the site-specific scale of landslide risk assessmentsin Hong Kong, engineering geological mapping carriedout by experienced personnel is recommended to evalu-ate landscape evolution, form, processes and materials,all of which are critical to identify, characterize, predictand mitigate landslide hazards.

Ground investigation

Although engineering geological mapping should be thekey focus of any assessment, ground investigation is alsorequired to confirm critical components of the geologicaland hazard models. The requirement for ground inves-tigation works will normally only be apparent during theengineering geological mapping. In Hong Kong groundinvestigation, including the production of interpretativelogs, is carried out by ground investigation contractors.However, the contractor is rarely provided with theresults of the desk study or the initial hazard model.Furthermore, the majority of logging geologists mainlyundertake work related to deep foundations, typicallyassociated with the location of rockhead. Consequently,the resulting logs are unlikely to be of sufficient detail toallow the identification of pertinent information for alandslide hazard assessment (Fig. 5).

DiscussionBased on the above observations the conditions for thedevelopment of the engineering geological profession inHong Kong can be seen as far from ideal. Limited coregeological knowledge, restricted training opportunitiesand the serious shortage of experienced practitionerscombined with low fees and limited scope of practicehave resulted in only a few local earth science graduatesreceiving the appropriate training to obtain the neces-sary skills and experience to become proficient engineer-ing geologists.

However, this situation has not appeared overnightor without warning. The fact that the industry has failedto respond to these challenges suggests that engineeringgeology may be perceived as having little relevance totoday’s construction industry in Hong Kong.

It is with this background that the relatively newlegislation concerned with the assessment of naturallandslide hazards has to be viewed. Based on theauthors’ experience there are worrying signs that the lowconsideration now given to engineering geological input





to geotechnical works is being extended into the assess-ment of natural terrain hazards and risk. However,unlike other geotechnical works in Hong Kong, theunique nature of natural landslide assessment meansthat the current practice of conservative design andreliance on the geotechnical control system are unlikelyto provide an adequate safety net for these inadequacies.Unless this trend is reversed two scenarios are possible:(1) over-conservative assessments of the hazards posedby natural terrain in Hong Kong, resulting in extensiveand potentially unnecessary mitigation works with con-siderable environmental impacts; (2) the risk that high-magnitude, low-frequency events may not be detected,with potentially more serious consequences.

Acknowledgements. The authors would like to thank colleaguesat GeoRisk Solutions Ltd., in particular A. Moore, for assist-ance with the paper.

ReferencesB, F. 2007. Sources of geotechnical risk. In: Proceedings

of the Conference: Engineering Geology in GeotechnicalRisk Management. Hong Kong Regional Group of theGeological Society of London, 1–31.

B, F. & R, M. 2004. Discussion arising fromthe First Hans Cloos Lecture, by John Knill. Bulletin ofEngineering Geology and the Environment, 63, 89–90.

C, E. 2004. Academic geology in the UK: a matter ofgrave concern. European Geologist, June, 6–7.

C, J., R, J. & F, P. . 1996. Debrisflows. In: D, R., B, D., S, L. & I,M. (eds) Landslide Recognition: Identification, Movementand Causes. Wiley, New York, Chpt 7.3.

E, N.C., H, S.W. & K, J.P. 1997. The NaturalTerrain Landslide Study Phases I & II. Special ProjectReport No. SPR 5/97, Geotechnical Engineering Office,Hong Kong.

F, P.G., D, S.G. & L, J.M. 1991. Some observa-tions on a comparative aerial photography interpretationof a landslipped area. Quarterly Journal of EngineeringGeology, 24, 249–265.

GEO 2000. List of Technical Documents. Geotechnical Engi-neering Office, Hong Kong, GEO Technical GuidanceDocument No. 1 (TGN 1).

GEO 2003. Enhancing the Reliability and Robustness of Engi-neered Cut Soil Slope. Geotechnical Engineering Office,Hong Kong, Technical Guidance Note, TGN 11.

GEO 2007. Engineering Geological Practice in Hong Kong.Geotechnical Engineering Office, Hong Kong, Publica-tion, 1/2007.

GSL 1994. Training Guide for Engineering Geologists in theCivil Engineering and Construction Industry. GeologicalSociety, London.

H, A.B, G, J.S. & M, A.E. 2009. Somelimitations in the interpretation of vertical stereo photo-graphic images for a landscape investigation. QuarterlyJournal of Engineering Geology and Hydrogeology, 42,21–30.

H, K.K.S. 2004. Keynote Paper: Recent advances in geotech-nology for slope stabilisation and landslide mitigation—perspectives from Hong Kong. In: Proceedings of NinthInternational Symposium on Landslides, 28 June–2 July2004, Rio de Janeiro, Brazil, 1507–1560.

K, J.P. 1999. The 1990 Tsing Shan debris flow. In: H,K.K.S. & L, K.S. (eds) Proceedings of the 14th SoutheastAsian Geotechnical Conference. Balkema, Rotterdam,783–788.

K, J. 2002. Core values: The First Hans Cloos Lecture. In:Proceedings of the 9th Congress of the InternationalAssociation for Engineering Geology and the Environment,Durban, South Africa. 1–46.

M, R.P. 2003. Review of geological aspects of slopeengineering. Transactions of the Hong Kong Institution ofEngineers, 10, 26–33.

M, J.B. 2001. Qualification, registration and licensing ofgeotechnical practitioners in Hong Kong. PanellistsReport. In: H, K.K.S. & L, K.S. (eds) Proceedings of the

Fig. 5. Comparison of ground investigation contractor and project engineering geologist’s trial pit logs.





14th Southeast Asian Geotechnical Conference. Balkema,Rotterdam, 623–628.

M F J V 2004. Final Report. NaturalTerrain Hazard Study for Tsing Shan Foothill Area. Geo-technical Engineering Office, Civil Engineering Depart-ment, Hong Kong.

M F J V 2007. Final Report on Com-pilation of the Enhanced Natural Terrain Landslide Inven-tory (ENTLI). Agreement CE15/2005. Natural TerrainLandslide Identification—Feasibility Study. GeotechnicalEngineering Office, Civil Engineering Department, HongKong.

M, N.R. 2000. Common ground. In: Proceedingsof the International Conference on Geotechnical and Geo-logical Engineering, GeoEng 200, Melbourne. TechnicalPublishing, Basel.

N, K.C., P, S., K, J.P., F, C.A.M. & S, R.2003. Guidelines for Natural Terrain Hazard Studies. Geo-technical Engineering Office, Hong Kong, GEO Report,138.

N, K.C., T, J.W. & P, S. 2007. Engineeringgeological practice in Hong Kong. In: GeotechnicalAdvancements in Hong Kong since 1970s. Proceedings ofthe Hong Kong Institution of Engineers Annual Seminar.The Hong Kong Institute of Engineers, Hong Kong,11–22.

P, S. & J, D.A. 2007. The use of LIDAR forlandslide hazard assessments: Hong Kong case studies. In:Proceedings of the Conference: Engineering Geology inGeotechnical Risk Management. Hong Kong RegionalGroup of the Geological Society of London, 155–161.

P, S. & N, K.C. 2010. The assessment of landslide riskfrom natural slopes in Hong Kong: an engineering geo-

logical perspective. Quarterly Journal of EngineeringGeology and Hydrogeology, 43, (in press) .

P, S. & R, M.E. 2000. The importance of geomorphol-ogy for natural terrain hazard studies. In: Proceedings ofthe Conference Natural Terrain—a Constraint on Develop-ment? Institute of Mining and Metallurgy, Hong KongBranch, 89–100.

P, S., L, A.M.H. & C, S.D.G. 2004. TrialGeotechnical Audit (2003) of LPM Slopes. GeotechnicalEngineering Office, Hong Kong, Geological Report, GR2/2004.

P, S., R, M.E. & N, K.C. 2006. Assessment ofnatural terrain landslide risk in Hong Kong: an engineer-ing geological perspective. In: Proceedings of the Inter-national Association of Engineering Geology Conference,Nottingham, 2006, Paper 299.

P, H.G. 2007. Piling design for Hong Kong—an externalviewpoint. In: Geotechnical Advancements in Hong Kongsince 1970s. Proceedings of the Hong Kong Institution ofEngineers Annual Seminar. The Hong Kong Institute ofEngineers, Hong Kong, 1–9.

S W 1999. Specialist API Services for the NaturalTerrain Landslide Study: Large Landslide Study for theGeotechnical Engineering Office, Hong Kong Government.Scott Wilson, Hong Kong.

S, R.J. & C, S.D.G. 2005. Report on the Datingof Natural Terrain Landslides in Hong Kong. GeotechnicalEngineering Office, Hong Kong, GEO Report, 170.

SSTRB 2007. Report to the Geotechnical Engineering Office,Civil Engineering and Development Department, HongKong. Slope Safety Technical Review Board Report, 16.

S, K.A. & H, A. 1989. Geotechnical Area StudiesProgramme: Territory of Hong Kong. Geotechnical Con-trol Office, Hong Kong, GASP Report, XII.

Received 4 March 2008; accepted 29 June 2009.



