Chau K.T., Wong R.H.C., Liu J. and Lee C.F. (2003) “Rockfall hazard analysis for Hong Kong based on rockfall inventory” Rock Mechanics and Rock Engineering, Vol. 36, No. 5, pp

Rock Mech. Rock Engng. (2003) 36 (5), 383–408

DOI 10.1007/s00603-002-0035-zRock Mechanics

and Rock EngineeringPrinted in Austria

Rockfall Hazard Analysis for Hong KongBased on Rockfall Inventory

By

K. T. Chau1, R. H. C. Wong1, J. Liu1 and C. F. Lee2

1 Department of Civil and Structural Engineering, The Hong Kong Polytechnic University,Kowloon, Hong Kong, China

2 Department of Civil Engineering, The University of Hong Kong, Pokfulam,Hong Kong, China

Received August 8, 2001; accepted July 24, 2002;Published online December 19, 2002 # Springer-Verlag 2002

Summary

This paper compiles and analyzes the rockfall data in Hong Kong in the last fifty years. A simplerockfall hazard analysis is presented based on this rockfall inventory. A frequency-magnituderelation, which is analogous to the Gutenberg-Richter relation for earthquake occurrence, is pro-posed for rockfall, and direct correlation between rockfall frequency and the daily rainfall isobserved. Data analysis shows that a threshold daily rainfall of about 150–200 mm is expected inorder to trigger rockfall events in Hong Kong. Among the 368 rockfall events in the 13 year periodfrom 1984 to 1996 in Hong Kong, 35% of the incidents lead to blockage of or damage to roads,22% lead to damages or evacuation of squatter huts, 21% lead to blockage of pedestrian pavementand footpath, and 15% affect buildings, such as housing apartments and schools. Only 15% ofthese rockfalls fell onto open space and caused negligible effects on human activities. Most ofthese rockfall events occurred during heavy rain and when ‘‘landslide warning’’ should has beenissued by the Hong Kong Observatory, thus only 6% of these events led to injury or casualty, cardamages, and damages to public utilities. Rockfall hazard zonation maps in terms of the spatialdistribution of previous rockfalls are proposed for both Kowloon Peninsula and Hong KongIsland.

Keywords: Rockfall, hazard map, rockfall inventory, hazard analysis, Hong Kong.

1. Introduction

Rockfall has been a serious threat to highways (e.g. Bunce et al., 1997), railways

(e.g. Brawner and Wyllie, 1976) and residential areas (e.g. Budetta and Santo,

1994) in many mountainous areas of the world. It was estimated that 10 percents

of the world’s population live in mountainous terrain and 36 percents of the

land area of the world are mountainous (Gerrard, 1990). Due to the continuous

expansion of the world’s population, human activities has been ever-increasing in

the formerly remote mountainous areas (e.g. Porter and Orombelli, 1981). Actu-

ally, rockfall has been occurring regularly in Hong Kong (e.g. Chau et al., 1998)

and other parts of the world, such as Canada, USA, Austria, Italy, Japan and

mainland China. In some mountainous areas, rockfall can be extremely frequent;

for example, Gardner (1970) reported a total of 563 rockfalls during only 842

hours of field observation in Lake Louise area in Canada. Actually most of the

rockfall events were unreported, and only those falling on highways, resident areas

or causing damages or casualties will capture public attention. On February 10,

1996, the huge rockfall, that crushed a tunnel along highway 229 on Hokkaido

Island of Japan and killed 20 people, captured much international attention to the

rockfall problem (Ishijima and Roegiers, 1996; Yamagishi, 2000).

The first engineering treatment on rockfall problem was given by Ritchie

(1963). Based on the observation of a series of full scale tests for the fall of basaltic

boulders over real slopes, Ritchie (1963) proposed simple guidelines for the design

of boulder trap ditch or trench at the toe of rock slopes. The method was sub-

sequently compiled into design charts (Fookes and Sweeey, 1976; Whiteside, 1984;

Wyllie, 1987). The main drawback of Ritchie (1963) approach is that impact

dynamics for bouncing boulders was altogether ignored. Since then, many theo-

retical, laboratory and field studies have been published. Without going into the

details, we refer the readers to the review articles on rockfall and related problems

by Piteau and Peckover (1978), Whalley (1984), Richards (1988), Fornaro et al.

(1990), Giani (1992), Flageollet and Weber (1996), and Chau (1997).

To determine the location of the probable rockfall events, a rating system, such

as the Rockfall Hazard Rating System (RHRS) by the Oregon Department of

Transportation (Pierson et al., 1990; Pierson, 1992), can be used. Another simple

approach for rockfall hazard estimation is based on the past rockfall events (e.g.

Porter and Orombelli, 1981; Hungr and Evans, 1998a; Evans and Hungr, 1993).

Once a location is identified with high risk, the probability of the maximum travel

distance and maximum energy of impact of probable rockfall events at the loca-

tion are normally assessed using computer simulations (e.g. Wu, 1985; Bozzolo

and Pamini, 1986; Spang and Sonser, 1995; Hoek, 1987; Hungr and Evans, 1988a–

b; Paronuzzi, 1989; Kobayashi et al., 1990; Azzoni et al., 1995; Descoeudres and

Zimmermann, 1987; Pfeiffer and Bowen, 1989). Most of the existing computer

programs model boulders as a point mass and the fall trajectories along the slope

are controlled by coefficients of restitution. Another completely different numeri-

cal approach for rockfall simulations involve the use of Discontinuous Deforma-

tion Analysis (DDA) (e.g. Lin et al., 1996). With the help of these computer

simulations, hazard and risk assessments of rockfall can be done.

In this paper, we present a simple and cheap method to assess the rockfall

hazard by using rockfall inventory. Hong Kong will be employed as an example.

Rockfall inventory has been mainly compiled and analyzed in the field of geomor-

phology, such as Rapp (1960), Bjerrum and Jorstad (1968), Gardner (1970, 1977,

1980), Luckman (1976), and Douglas (1980). The main problem with the rockfall

data analysis is the incompleteness of any data set as many events either have been

unreported or left unnoticed.

384 K. T. Chau et al.

The only analyses on the rockfall data of Hong Kong are by GEO (1996a) and

Chau et al. (1998). However, both studies are preliminary and rather incomplete

in terms of the rockfall hazard on various facilities and relation between rainfall

and rockfall. In the present analysis, the frequency-magnitude curve and the diur-

nal rockfall frequency plot proposed by Chau et al. (1998) will be re-examined.

Another new analysis of the present study is on the spatial distribution of rockfall

hazard. Although the landslide hazard zonation has been used (e.g. Kienholz,

1978; Moser, 1978; Malgot and Mahr, 1979; Varnes, 1984; Hansen, 1984; Brabb,

1991; PIARC, 1997), very few zonation maps for rockfall alone are available

(e.g. Simpson, 1973). And this kind of rockfall hazard zonation map has not been

proposed for Hong Kong. Therefore, zonation maps for rockfall hazard will also

be proposed for both Kowloon Peninsula and Hong Kong Island based on rock-

fall inventory.

The results of the present analysis should provide useful information on the

plausible location of rockfall, frequency of rockfall activity, time of year when

activity is the highest, size or quantity of rockfall per event, and correlation

between rockfall and rainfall. All this information is useful in assessing the hazard

of rockfall. Although the focus of the data analysis to be presented is restricted to

Hong Kong, the methodology of data analysis proposed here can be applied to

other parts of the world without much modification. In particular, the same

approach can be used to assess rockfall triggered by frost-thaw cycles, seismic

activity, human constructions. But under such circumstance, correlation between

rockfall and the triggering events should be examined, instead of the rockfall-

rainfall relation.

Despite of the simplicity of the present approach, there are certain limitations.

For example, rockfall inventories are not always available, and not necessarily

sufficient to establish zonation maps. In addition, to make the zonation map more

reliable, geological study like structural analysis of cliffs and photogrammetry

should be included in the database in the future. However, these aspects are out of

the scope of the present study.

2. General Geological and Climatic Conditions for Hong Kong

Before we consider the rockfall problems in Hong Kong in the next section, it is

informative to briefly introduce the geological and climatic conditions of Hong

Kong.

2.1 Geological Conditions of Hong Kong

Hong Kong lies on the southern edge of an ancient land mass. The total area of

Hong Kong is about 1000 square kilometres. The natural terrain of Hong Kong is

dominated by mountains and hills with steep slopes. Figure 1 shows an aerial

photograph of the hilly Hong Kong Island and Kowloon Peninsula. The photo-

graph was taken on July 5, 2001 from about 15,000 feet. The highest point in

Hong Kong is Tai Mo Shan at about 957 metres above sea level. The mountains

Rockfall Hazard Analysis for Hong Kong 385

consist primarily of volcanic and granitic rocks. Low grounds tend to be formed of

granite or sedimentary rocks. In places, hill-slope debris forms a mantle over the

bedrock and alluvium fills some of the valleys. The geological maps of Hong Kong

Island and Kowloon Peninsula are shown in Figs. 2–3 respectively.

The granites are younger than the volcanic rocks, having been intruded in

various stages into the volcanic rock mass about 170 million years ago. These

rocks are from fine to coarse-grained and generally widely jointed with spacings

from 300 mm to 3 m apart. These joints normally display in orthogonal sets.

The volcanic rocks are mainly of Middle and Lower Jurassic age, and of vari-

able type and composition, but consisting mainly of tuffs with some andesite and

rhyolite lava flows, and are frequently metamorphosed. These volcanic rocks are

generally closely jointed, with spacings of 50 mm to 300 mm. These joints fre-

quently predominate on one plane.

Interspersed with the volcanic series is a variety of sedimentary rocks, rang-

ing from water-laid volcanics to conglomerates, sandstones and shales, but their

occurrence is mainly restricted to the north-west New Territories. The only lime-

stone in Hong Kong occurs as boulders in an agglomerate found in a limited

locality in Tolo Harbour.

Dykes, varying in width from 150 mm to 1.5 m, have been intruded into both

the granitic and volcanic rocks. The dykes generally dip very steeply and are

associated with zones of close jointing and metamorphosis.

Fig. 1. An aerial photo taken from the Kowloon Peninsula looking southward at the Hong KongIsland on July 5, 2001 (photo by K. T. Chau). See Fig. 14 for the legends H1, H2, H3, H4, and H6


Fig. 2. The geological map for Hong Kong Island. The superficial geology include granite, tuff, quartz,granodiorite, debris flow deposit, alluvium and reclamation

Fig. 3. The geological map for Kwoloon Peninsula. The legend is the same as for Fig. 2


In general, the outcropping rocks of Hong Kong are highly weathered. The

topography associated with volcanic rocks is typically a rugged terrain, with rocky

outcrops and impressive cliffs. In granites, weathering tends to occur selectively

along the joint planes, often extending to a considerable depth. Because of the

large joint spacing, core blocks of unweathered rock may be left. Near the surface

the rock is completely decomposed but, with depth, the frequency and size of core

boulders increases, until completely unweathered rock is reached. On the coast

such weathering is often well displayed and, where the soil has been eroded away,

a jumble of loose boulders may result. Such isolated boulders are a feature of the

granite-hillsides in Hong Kong, and they are conducive to rockfall. Figure 4 shows

a typical boulder field on the slope of Hong Kong Island.

2.2 Climatic Conditions of Hong Kong

Hong Kong’s climate is sub-tropical. During November and December there are

pleasant breezes, plenty of sunshine and comfortable temperatures with a mean

temperature ranging from 17 to 21 �C. January and February are more cloudy,

with occasional cold fronts followed by dry northerly winds with a mean temper-

ature of about 16 �C. But, it is not uncommon for temperatures to drop below

10 �C in urban areas. Sub-zero temperatures and frost do occur at times on high

ground. March and April can also be very pleasant although there are occasional

spells of high humidity. Fog and drizzle can be particularly troublesome on high

ground. May to August are hot and humid with occasional showers and thunder-

Fig. 4. A photograph of a boulder field on the slope of the Hong Kong Island. The Victoria Harbourcan be seen down slope (photo from GEO)


storms. Afternoon temperatures often exceed 31 �C whereas at night, temperatures

generally remain around 26 �C with high humidity. September is the month during

which Hong Kong is most likely to be affected by tropical cyclones, although

typhoons are not unusual at any time between May and November.

If the center of a typhoon comes close to Hong Kong, wind speed may be up to

118 kilometres per hour or more, and rain can become heavy and widespread.

Heavy rain from tropical cyclones may last for a few days and subsequent land-

slides, rockfalls and flooding may cause considerably economical loss.

The mean annual rainfall ranges from around 1300 millimetres along the coast

to more than 3000 millimetres on the high ground. Typically, about 80 percent

of the rain falls between May and September. The wettest month is August with

an average monthly rainfall of about 400 millimetres. Severe weather phenom-

ena that can affect Hong Kong include tropical cyclones, strong winter monsoon

winds, and thunderstorms with associated squalls that are most frequent from

April to September.

3. Rockfall Inventory in Hong Kong

Although the rockfall inventory can be gathered from direct observation (e.g.

Gardner, 1977), by rockfall traps (e.g. Luckman, 1976), and lichenometry (e.g.

McCarroll et al., 1998), a method to interpret historical rockfall events from the

size of lichens growing on boulders, in the present study the rockfall data are

mainly the events reported in Hong Kong that lead to disturbance of human

activities. In particular, the rockfall data are mainly extracted from the annual

report on ‘‘Hong Kong rainfall and landslides’’ which was published annually

since 1984 (Premchitt, 1985–89; Sui, 1990; Tang, 1991; Evans, 1992; Chen, 1993;

Chan, 1994, 1995; Wong, 1996, 1997). However, some information of these rock-

fall records, such as the time of occurrence, the size of boulder, and even the date

of occurrence, are missing in some of these reports. This makes the compilation

and analysis of the data more difficult. In addition, the geology of the slope, the

slope angle, the vertical and horizontal travel distances of the boulder were not

reported in these reports. For a more complete analysis, all these geological data

are essential (e.g. Culshaw and Bell, 1991). In addition, the structural analysis of

cliffs and photogrammetry should preferably be incorporated into the hazard

analysis. Unfortunately, such information is not readily available for Hong Kong.

The latest reports (1992–96) seem to be much more complete than the reports

before 1992. All of the data analysis reported here is only up to 1996. Before 1984,

only some isolated reports are compiled after some specific heavy rainstorms

(Choot, 1983; Hudson, 1982; Tang, 1982). In addition to these reports, some

additional rockfall data were also found in Atkins Haswell (1995), CED (1995),

and GEO (1996a–b). In addition, rockfall caused by blasting was also included in

the present analysis (e.g. Evans and Irfan, 1991).

Although it is very unlikely that the data compiled here are complete, this data

set is considered to be reliable and should constitute a good basis for hazard

analysis which is considered next.


4. Analysis of Rockfall Data

4.1 Frequency and Magnitude Relation

The scale of most of the geological events is believed to increase with the decrease

of frequency of occurrence (Gardner, 1980). The scale of frequent events is nor-

mally too small to cause any significant impact, while the rare large scale event

may lead to major economic and life losses. The frequency magnitude of rockfall

or boulder fall has been studied by Gardner (1977) for Lake Louise Area, Gardner

(1980) for Highwood Pass Area, Luckman (1976) for Jasper National Park,

Douglas (1980) for Country Antrim in North Ireland, Bjerrum and Jorstad (1968)

for Norway, Rapp (1960) for Karkevagge, Wieczorek et al. (1995) for Yosemite

Valley, and Bunce et al. (1997) for Argillite Cut on BC Highway 99 in Canada. A

very rough frequency-magnitude curve was also proposed by CED (1995) for

Hong Kong; however, we consider that the data compiled in CED (1995) is rather

incomplete. In particular, the curve by CED (1995) is obtained by analyzing 169

rockfall data from the 11 year period of 1984–1994.

In this study, we analyze a total of 201 data from the 5 years of 1992–1996,

since we believe that data before 1992 are incomplete. Note that 11 of the data

within this period are missing, and a volume of 1 m3 is assigned to them, which is

presumably the largest size of rockfalls that may have been left unnoticed (i.e. any

boulder size larger than that size is likely to be recorded). Some of the volumes of

boulders used in our analysis are estimated based on the photographs taken at the

time of occurrence. Figure 5 shows the number of rockfall per year versus the

Fig. 5. The average number of rockfall per year versus volume of boulders for Hong Kong (Curves 1and 2), together with Gardner’s (1977) result for Lake Louise Area and Gardner’s (1980) result forHighwood Pass Area. The results for both Lake Louise and Highwood Pass Areas have been scaled up

to a study area of 1,000 km2 (¼ area of Hong Kong) for the purpose of comparison


volume of the boulder, together with the frequency-magnitude curve (solid lines)

proposed by Gardner (1977) for Lake Louise Area and by Gardner (1980) for

Highwood Pass Area. To make the results comparable to that for Hong Kong and

in contrast to the analysis by Chau et al. (1998), both the frequency-magnitude

curves for Lake Louise and Highwood Pass areas have been scaled up to a study

area of 1,000 km2 (about the area of Hong Kong). Motivated by the study of

Gardner (1980) and Chau et al. (1998), the cell size is selected as logarithmic (i.e.

0.01 to 0.1, etc.). The average of the data within each logarithmic volume cell is

plotted as a open circle, and the error bar indicates ‘‘plus’’ or ‘‘minus’’ one stan-

dard deviation of the data within each cell of volume increments. In general, there

are three different ways to interpret the frequency magnitude curve. For example,

Gardner (1977, 1980) assumed a decreasing magnitude-frequency curve (as shown

in Fig. 1 of Chau et al., 1998), Whalley (1974) proposed a modal magnitude-

frequency curve (see Fig. A1 of Whalley, 1974), and Wieczorek et al. (1995)

assumed a fractal or self-similar distribution. The log-normal distribution with a

model size is believed to approximate magnitude and frequency of forces in geo-

morphic processes (Wolman and Miller, 1960), while the self-similar scaling law

seems to describe many phenomena in nature (e.g. Scholz, 1990). Figure 5 shows

the data fit by using both decreasing magnitude-frequency curve (Curve 1) and

modal magnitude-frequency curve (Curve 2). The modal size of the rockfall

in Hong Kong is about 3 m3. In contrast to the conclusion by Chau et al. (1998),

Fig. 5 illustrates that the frequency of rockfall for a specific boulder size is about

the same for both Hong Kong and for Lake Louise and Highwood Pass Areas in

Canada if the study areas are about the same size.

The following self-similar frequency-magnitude relation, which is analogous

to the Gutenberg-Richter relation for earthquake (Scholz, 1990), is proposed by

Wieczorek et al. (1995):

log10 N ¼ a� b log10 V ; ð1Þ

where V and N are the volume of rockfall and the cumulative number of rockfall

with scale of V or larger for a specified period of time, and a and b are constants to

be determined by data.

Figure 6 plots the cumulative number N for rockfall events with volume V or

larger versus the volume of boulders for Hong Kong, together with the frequency-

magnitude for Yosemite Valley established by Wieczorek et al. (1995) (note that

which is scaled down to cumulative number per year). For the rockfall data from

1992 to 1996, the following power law is established:

log10 N ¼ 1:86� 0:896 log10 V; ð2Þ

where N is the cumulative number of rockfall with volume larger or equal to V per

year. Although the frequency-magnitude relations for Hong Kong and Yosemite

Valley given in Fig. 6 seem to be comparable, it should be noted that the area

of the Yosemite Valley is only about 10 km2 comparing to about 1,000 km2 for

Hong Kong. Thus, it can be concluded that the rockfall frequency is much higher

in Yosemite Valley than that in Hong Kong. This may be due to the fact that the


climatic changes in Yosemite Valley are more extreme than in Hong Kong. Ad-

ditional rockfall events are also likely triggered by frost-thaw cycles and seismic

activities. In addition, Yosemite National Park is one of the most visited parks in

USA, and it is unlikely that any sizeable events in the Yosemite Valley will be left

unnoticed.

This simple relation can be used to estimate the return period of some larger

events. For example, for a rockfall of size 1,000 m3 a return period of 6.7 years

is expected. But, of course, the accuracy of this prediction depends on the complete-

ness of our data set and the validity of the self-similar model proposed in (1).

4.2 Diurnal and Seasonal Rockfall Frequency

Diurnal rockfall frequency was established by Gardner (1967, 1970, 1971) for the

North Faces in the Lake Louise Area, by Luckman (1976) for Surprise Valley of

Jasper National Park in Canada, by Church et al. (1979) for Baffin Island in

Canada, and by Chau et al. (1998) for Hong Kong from 129 data from 1949–

1995. Recently, we have updated the diurnal frequency plot of Chau et al. (1998)

by including 15 more data in 1996. The updated plot is similar to Fig. 2 of Chau

et al. (1998), and thus it will not be reported here. As expected, the main conclu-

sion is that most of the rockfalls occurred from 8:00 to 19:00 because many events

occurred at night time are not likely to have been reported reliably.

The seasonal variations of the rockfall events have been plotted for Norway

by Bjerrum and Jorstad (1968), for the railway line in Fraser Canyon of British

Fig. 6. Self-similar scaling relations for the cumulative number of rockfall per year versus volume ofboulders for Hong Kong (from data in 1992 to 1996) and for Yosemite Valley. The volume-frequencypower law relation for rockfalls in Yosemite Valley from 1900 to 1992 by Wieczorek et al. (1995) is

scaled down to cumulative number of rockfall per year for the purpose of comparison


Columbia by Peckover (1975), for Lake Louise Area by Gardner (1967), for

County Antrim North Ireland by Douglas (1980), and for Hong Kong by Chau

et al. (1998). However, Fig. 3 of Chau et al. (1998) based on rockfall data from

1984 to 1995 while the monthly rainfall average from 1989 to 1995. We have

re-plotted the seasonal change for 1989–1996 with the corresponding monthly

rainfall average. Our new plot is basically the same as Fig. 3 of Chau et al. (1998),

and therefore will not be reported here. In general, very good correlation between

the monthly rainfall and monthly rockfall frequency is observed (Chau et al.,

1998).

4.3 Rockfall Frequency and Daily Rainfall Record

To further examine the relation between the rainfall and rockfall event, Fig. 7

plots the average number of rockfall per each raining day with various daily rain-

fall intensity. The plot is based on the analysis of rockfall data from 1989 to 1996

or a total of 245 data (note the date of occurrence for some of this data is missing).

For example, in the period 1989–1996 the average number of rockfall per day

is 0.06 if the daily rainfall R is between 0 and 50 mm, 0.37 for 50 mm < R <100 mm, 1.17 for 100 mm < R < 150 mm, 4.17 for 150 mm < R < 200 mm,

5.75 for 200 mm < R < 250 mm, 6.00 for 250 mm < R < 300 mm, and 12.5 for

300 mm < R < 350 mm. It is clear from Fig. 7 that a higher daily rainfall yields

to higher probability of rockfall occurrence in Hong Kong. In addition, it can be

concluded that if the daily rainfall is about 100 to 150 mm, there will be a high

chance of having at least one rockfall occurrence. For the cases of extremely high

daily rainfall of over 300 mm, it is expected that 12 rockfalls may occur. But, it

should be emphasized that within the eight year period from 1989 to 1996, there

are only two such days (324.1 mm on May 8, 1992 and 322.8 mm on May 20,

1989). Therefore, this prediction of the number of rockfall under extreme con-

ditions should not be taken too seriously as the data set is too small to have any

statistical significance.

Fig. 7. The average number of rockfall per each raining day versus the daily rainfall


However, rainfall histogram before any serious rockfall or landslide event

normally shows that not only the daily rainfall intensity but also the cumulative

amount of continuous rainfall prior to the event are important to the onset of

rockfall and landslide. Therefore, we also plot the two day cumulative rainfall

prior to the event versus the number of rockfall occurrence in Fig. 8. Compar-

ing to Fig. 7, the range of each cell is increased from 50 mm to 100 mm. The

2-day cumulative rainfall seems again to correlate well with the rockfall incidents,

although a higher cumulative rainfall does not necessarily yield a higher number

of events as shown in Fig. 8. In Fig. 9, the 3-day cumulative rainfall versus the

number of rockfall occurrence is given. The correlation between the number of

rockfall and the cumulative rainfall is not so straightforward comparing to Figs.

7–8. As shown in Fig. 9, the average number of rockfall event is smaller for the

days with 600–700 mm cumulative rainfall than the days with cumulative rain-

fall of 500–600 mm. We have also tried to correlate 4-day and 5-day cumulative

rainfalls to the number of rockfall events, the results are very similar to those

observed in Fig. 9.

Fig. 8. The average number of rockfall per day versus the two-day cumulative rainfall prior to the dayof rockfall

Fig. 9. The average number of rockfall per day versus the three-day cumulative rainfall prior to the dayof rockfall


We conclude that the daily and 2-day cumulative rainfall are the most reliable

indicators on the frequency of rockfall events, and the 3-day or longer cumulative

rainfall seems to have less direct consequence on rockfall occurrence.

4.4 The Impacts of Rockfall in Hong Kong

Size of rockfall can, in general, range from less than 1 m3 to over 1 million m3,

such as the Elm, Frank, and Madison Canyon rockfalls and rockslides (e.g.

Voight, 1978a,b). Large scale rockfalls, rockslides and rock avalanches are some-

times difficult to distinguish from one and other, because these events normally

involve both fall and slide mechanisms. A commonly-adopted classification of

rockfall based on the size of fall was proposed by Whalley (1974, 1984): debris fall

(<10 m3), blockfall (>100 m3, a large block which may fragment during travel),

cliff fall (104–106 m3), and Bergsturz (>106 m3, a fall or slide which may travel a

considerable distance). Some local geological settings may be more conducive to

rockfall phenomenon, such as conglomerate formations having soft matrix mate-

rial and rock mass with preferential and well-developed joint sets. Rockfall can

be induced or triggered by heavy rainfall, great temperature change, freeze-thaw

process (e.g. Prior et al., 1971), local fracturing or creeping of rock joints due to

stress concentration, vibrations from blasting, earthquakes and traffic, animal and

human activities, volcanic activities (e.g. Crandall and Fahnestock, 1965), wedg-

ing effect due to roots of vegetation, or simply erosion process (Wieczorek et al.,

1995). The rockfall data compiled by Peckover (1975) from 1933 to 1970 for the

Fraser Canyon of the Yale subdivision of the Canadian National Railway clearly

indicate a direct correlation between rockfall event and precipitation. In addition,

rockfall has clearly been associated with earthquakes (e.g. Broili, 1980; Kobayashi

et al., 1990; Harp and Noble, 1993); and rockfall events have been correlated to

earthquake intensity (Harp and Wilson, 1995).

From 1949 to 1996, there are a total of 30 fatal rockfall events that killed a

total of 52 people, and a total of 42 rockfalls that led to either injury or death, or

both (GEO, 1996b; Chau et al., 1998). There is also a recent report on boulder falls

from natural terrain for Hong Kong (GEO, 1998). The most recent fatal rockfall

occurred on August 18, 1995, during which a boulder of approximately 26 tonnes

broke away from the outcropping bedrock on top of a 30 m high rock slope along

the Tuen Mun Highway (see Fig. 10). The boulder collided with a van before

landed on the central reserve of Tuen Mun Highway; the driver died instantly and

one passenger was injured and hospitalized. The few weeks after the incident was a

nightmare for all residents in the northwestern New Territories, as the Tuen Mun

Highway was closed for remedial works. The most fatal rockfall incident killed 6

persons; it happened 3 times in the history of Hong Kong, twice in Shaukeiwan in

1959 and 1964 and once in Shek Kip Mei in 1964 (as shown in Table 1). The Shek

Kip Mei rockfall buried a building while the two events in Shaukeiwan killed resi-

dents living in squatters. Table 1 compiles all rockfall events in Hong Kong from

1949 to 1996, which caused at least 2 deaths or 4 injuries. Most of the rockfall

events listed in Table 1 were associated with heavy rainfall.


Figure 11 plots the cumulative number of fatalities and injuries from 1949

till 1996. To examine the relation between casualties and rainfall, the cumulative

rainfall in meter is also plotted. In the 47 year period, we have compiled a total of

52 deaths and 78 injuries. Thus, the average annual life loss is 1.11 death/year. The

average population of Hong Kong from 1949 to 1996 is estimated as 4.27 millions,

and this leads to an annual probability of death of an individual in Hong Kong

as 2:59� 10�7. As discussed by Chau et al. (1998), this risk level is smaller than or

comparable to the acceptable risk level for landslides set by various authors (e.g.

Fell, 1994; Finley and Fell, 1997; Bunce et al., 1997), although it appears that no

acceptable risk for rockfall alone has been proposed. As shown clearly in Fig. 11,

the recent fatality rate per year per person is actually decreasing. For example,

in the last 25 years (1971–1996) there are only 9 deaths and the average annual

fatality rate is 0.36; assuming an average population of 5.32 millions in this 25

year period, we arrive at an annual fatality rate per person of 6:7� 10�8, which is

4 times smaller than the 50 year average. In the 15 year period from 1981 to 1996,

there are only 4 deaths, which corresponds to a fatality rate of 0.27 death/year

in Hong Kong. Assuming an average population of 5.8 millions in this 15 year

period, we have the annual fatality rate per person of 4:66� 10�8. In terms of the

effect of rainfall, from 1949 to 1975 every meter of cumulative rainfall in Hong

Kong roughly leads to 0.84 injuries and 0.71 fatalities. However, these estimations

drop to 0.56 injury and 0.17 death per every meter of cumulative rainfall for the 12

year period from 1975 to 1996. Thus, it is fair to say that the number of fatality

due to rockfall have diminished in recent years in Hong Kong, and this recent

drop is probably due to the result of the ‘‘Landslip Preventive Measures’’ (LPM)

program taken by the Hong Kong Government.

Fig. 10. A photograph of the Tuen Highway rockfall occurred on August 18, 1995 (photo fromMing Pao Daily)


In Hong Kong, over 6 millions people are living within an area of around

1,000 km2, of which over 70% of the land are hilly or mountainous. It is expected

that rockfall events in Hong Kong can lead to serious consequences (see for

example Fig. 4). In the light of hazard analysis, it is of interest to know, at least

statistically, where in Hong Kong or what kind of facilities is at a higher risk than

others. Most of the previous rockfalls occurred in Hong Kong have led to injuries,

fatalities, blockage of roads or pedestrian footpath, damages to apartment build-

ings or squatter huts, or damages to car park or cars. Figure 12 plots the cumu-

lative number of events having influence on road (blockage or damage), squatter

(damage or evacuation), footpath (blockage or damage), building (damage or

evacuation), and open space (no direct influence) from 1984 to 1996. Note that

some of these rockfall events may affect, say, both road, footpath, and squatter

Table 1. All rockfall events in the period 1949–1995 that led to at least 2 fatalities or 4 injuries


huts; therefore, double count of influences is allowed in our data analysis. As

shown in Fig. 12, the roads in Hong Kong are at the highest risk in terms of being

impacted by boulders, following by squatter areas, pedestrian pavements or foot-

paths, and buildings. In recent years, the incidents of rockfall affecting road have

Fig. 11. The cumulative fatalities, injuries and rainfall (in meters) versus the years from 1949 to 1996

Fig. 12. The cumulative number of rockfalls affecting road, squatter, footpath, and building, andlanding on open space versus the years from 1984 to 1996


increased while the risk of squatter areas being affected seems to drop slightly. This

may be due to the rapid expansion of the highway system in Hong Kong in recent years

and the decrease in squatter areas because of public housing. Therefore, when rockfall

risk is high (say under heavy rainfall), it seems that people should avoid to be on road

(say to go to work or to rush home if they are at offices). Data analyses for this 13 year

period also show that about 4% of rockfalls lead to injury or death, 4% lead to damage

or blockage of parking space, 2% affect stairways and other facilities adjacent to

buildings, and 2% lead to car damages. These low rates of direct hit of rockfall on cars

and people suggest that the landslip warning system adopted by the Hong Kong Gov-

ernment during heavy rainfall has been effective in reducing fatality or injury. Figure 13

shows a rockfall event occurred on July 14, 1986 at the Kennedy Town Police Quarter

leading to a car damage. The three boulders are of the size of the car.

5. Rockfall Hazard Zonation Maps for Hong Kong

5.1 Rockfall Hazard Analysis and Zonation

There are various means of assessing rockfall hazard (e.g. Pierson et al., 1990;

Evans and Hungr, 1993; Hungr and Evans, 1988b). For a cut slope next to high-

way, the Rockfall Hazard Rating System (RHRS) has been proposed based

upon nine slope parameters (Pierson, 1992; Pierson et al., 1990). For larger areas,

Cancelli and Crosta (1994) proposed a rockfall hazard analysis using the ‘‘Expert

Semi-Quantitative method’’ (Hudson, 1992). In particular, fifteen parameters that

may influence rockfall occurrence are first identified, interactions among these

parameters are formulated in matrix form, weightings are then assigned to each

Fig. 13. A photograph of the rockfall event occurred on July 14, 1986 at the Kennedy Town PoliceQuarter leading to car damage


parameter, and finally zonations are proposed for various probabilities of rockfall

occurrence.

Another popular approach for assessing the potential hazard of landslides is

the use of hazard zonation map (e.g. Varnes, 1984; Hansen, 1984; Brabb, 1991).

However, there are relatively few applications of zonation maps to rockfall hazard

alone (e.g. Simpson, 1973; Culshaw and Bell, 1991). Therefore, a simple rockfall

hazard map based on rockfall inventory will be proposed next.

5.2 Hazard Maps for Hong Kong Island and Kowloon

For the construction of zonation map, it is advisable to consider not only the

rockfall history at a particular location but also the local geology, slope angle,

existing joint sets in rock mass, and local climate changes. To simplify our analy-

sis, in the section we will only consider the rockfall hazard based solely on the

previous rockfall events at a location. According to the terminology of Hansen

(1984), this approach is called ‘‘historical method’’. Contour maps of rockfall

hazard for Hong Kong Island and Kowloon Peninsula are plotted in Figs. 14–15.

These two areas in Hong Kong are selected for analysis because most of the

people in Hong Kong either live or work in these two areas. The scale of these

maps is in the order of 1 :100,000. The maps for these regions are first divided

Fig. 14. The zonation map of rockfall hazard for Hong Kong Island. Each contour line incrementmeans 1 rockfall per 13 years per 0.25 km2 or 0.31 rockfall per year per square kilometer. The locationlegends H1, H2, H3, H4, H5, H6, and H7 are for Tin Hau, Tai Hang, Shaukeiwan, Magazine Gap,Jardine’s Outlook, Kennedy Town, and Repulse Bay respectively. Some of these locations are also

shown in Fig. 1


into square grids or cells of size of 500 m. And all previous rockfalls in these two

areas from 1984 to 1996 are identified within the grids. The cumulative number of

rockfalls within each cell is summed for all records in the 13 year period. Then

contour lines are proposed based on the cumulative number of rockfall within

each cell; therefore 1 unit of contour line increment on these maps represents 1

rockfall per 13 years per 0.25 km2 (i.e. 0.31 rockfall per year per km2). The total

number of rockfall used for Hong Kong Island and Kowloon Peninsula from 1984

to 1996 are 179 and 92 respectively.

As shown in Fig. 14, there are a few places on the Hong Kong Island of higher

hazard: including Tin Hau (12 rockfalls), Tai Hang (6 rockfalls), Shaukeiwan

(6 rockfalls), Magazine Gap (6 rockfalls), Jardine’s Outlook (5 rockfalls), Kennedy

Town (4 rockfalls), and Repulse Bay (4 rockfalls). These locations are labelled as

H1, H2, H3, H4, H5, H6 and H7 respectively in Fig. 14 and some of these loca-

tions were also identified in Fig. 1. For Kowloon Peninsula, the rockfall hazard

appears to be lower. As shown in Fig. 15, the places with maximum rockfall haz-

ards include Lei Yue Mun (6 rockfalls), Sau Mau Ping (5 rockfalls), Lai Chi kok

(4 rockfalls), Choi Hung (4 rockfalls), and Cha Kwo Ling (4 rockfalls). These

higher hazard locations are labelled as K1, K2, K3, K4, and K5. These rockfall

hazard zonation maps should be useful in land-use planning and remedial

measures.

The recent rockfall of volume of 570 m3, that occurred on December 4, 1997

on Sau Mau Pang Road after a blasting operation nearby, confirmed the predic-

tion by Fig. 14 that Sau Mau Pang is indeed one of the high rockfall hazard

Fig. 15. The zonation map of rockfall hazard for Kowloon Peninsula. Each contour line incrementmeans 1 rockfall per 13 years per 0.25 km2 or 0.31 rockfall per year per square kilometer. The locationlegends K1, K2, K3, K4, and K5 are for Lei Yue Mun, Sau Mau Ping, Lai Chi Kok, Choi Hung, and

Cha Kwo Ling respectively


locations in the Kowloon Peninsula. Note that the rockfall hazard zonation map is

based on data from 1984 to 1996.

5.3 Limitations and Future Development

As mentioned earlier, the hazard analysis based upon rockfall inventory does not

take into consideration many inherent conditions for rockfall occurrence. Accord-

ing to Varnes (1984), these include the geology (such as rock types and rock joint

information), geomorphology (such as slope angles and slope curvature profile),

and hydrologic conditions (such as subsurface water level and precipitation) and

climate (such as temperature and frost action), and vegetation (such as the cushion

effect during impact). Although some, if not all, of these conditions and param-

eters are intrinsically inside the rockfall data, they cannot be distinguished sepa-

rately by simply looking at the rockfall inventory in terms of temporal and spatial

distribution. In addition, the hazard derived in the present study resulted from

records of a restricted time interval of 13 years. There is clearly a need to identify

rockfall hazard with respect to each of these parameters and conditions. This

is particularly important for regions with no, poor or very incomplete rockfall

inventory. Therefore, there is a necessity to integrate in the future the proposed

rockfall zonation with a rating system including the main structural and geo-

mechanical parameters of the rock slope in addition to the triggering factors.

Once rockfall hazard map is available, rockfall risk map can, in general, be

constructed for indicating the potential consequences of rockfall on human life,

facilities, and human activities, in terms of fatalities, monetary loss or land loss.

The current trend of rockfall hazard analysis is the incorporation of GIS with

hazard maps (Duarte and Marquinez, 2002). For discussions on the related topic

of landslide hazard and risk analyses, a number of review articles are available

(e.g. Einstein, 1988, 1997; Leroi, 1997; Fell and Hartford, 1997; Hungr, 1997;

Hansen, 1984; Varnes, 1984; Hartlen and Viberg, 1988; Brand, 1988; PIARC,

1997).

6. Conclusion

We have compiled and analyzed over 400 rockfall data reported in Hong Kong

from 1949 to 1996. A decreasing magnitude-frequency curve, a log-normal modal

magnitude-frequency curve, and a self-similar magnitude-frequency power law are

proposed for rockfall events in Hong Kong. The rockfall frequency in Hong Kong

is about the same as in Lake Louise and Highwood Pass Areas in Canada, but

appears to be lower than that for Yosemite valley from 1900 to 1992. The number

of daily rockfalls have been plotted versus to 1-day, 2-day, and 3-day rainfall data

from 1989 to 1996. It is concluded that a threshold daily rainfall of about 150–

200 mm is expected in order to trigger rockfall events in Hong Kong. Based on

the data analysis from 1984 to 1996, boulders are most likely to land on roads,

followed by squatter, footpath and building. Within the same period, 4% of rock-

fall leads to injury or death, 4% leads to damage or blockage of parking space,


2% affects stairways and other facilities adjacent to buildings, and 2% leads to car

damages. In addition, preliminary hazard zonation maps in terms of the spatial

distribution of previous rockfall events are proposed for both Kowloon Peninsula

and Hong Kong Island.

It is hoped that the method of rockfall hazard analysis presented in this study

will motivate similar analyses for the rockfall hazard assessment for other parts of

the world, and will lead to a more standardized approach for rockfall hazard and

zonation analyses.

Acknowledgements

The work was supported by the Research Grant Council (RGC) of the Hong Kong SpecialAdministrative Region through CERG Poject No. HKPolyU 5079/97E and HKPolyUinternal Project No. 351/446. The first author is grateful to Mr. W. K. Pun of Geotechni-cal Engineering Office of Hong Kong, Mr. Bernhard Hosle of Geobrugg, Switzerland,Mr. Gregory Won of Roads and Traffic Authority, Australia, and Dr. James Hamel ofthe Hamel Geotechnical Consultants, Pennsylvania, USA and his postgraduate studentMr. Leo Chan for providing helpful information on rockfall.

References

Atkins Haswell (1995): Quantitative landslide risk assessment for the squatter villages in LeiYue Mun. CED, GEO Agreement GEO 3/95.

Azzoni, A., La Barbera, G., Zaninetti, A. (1995): Analysis and prediction of rockfalls usinga mathematical model. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 32, 709–724.

Bjerrum, L., Jorstad, F. (1968): Stability of rock slopes in Norway. Norwegian Geotech.Inst. Publ. 79, 1–11.

Bozzolo, D., Pamini, R. (1986): Simulation of rock falls down a valley side. Acta Mechan-ica 63, 113–130.

Brabb, E. E. (1991): The world landslide problem. Episodes 14, 52–61.

Brand, E. W. (1988): Landslide risk assessment in Hong Kong. In: Landslides, 5th Int.Symp. on Landslides, Vol. 2. Balkema, Rotterdam, 1059–1074.

Brawner, C. O., Wyllie, D. (1976): Rock slope stability on railway projects. Am. RailwayEngng. Assoc. Bull. 656, 449–474.

Broili, L. (1980): Considerations on rock slope stability during earthquakes. Rock Mech.Suppl. 10, 47–61 (in German with English abstract).

Budetta, P., Santo, A. (1994): Morphostructural evolution and related kinematics of rock-falls in Campania (southern Italy): A case study. Engng. Geol. 36, 197–210.

Bunce, C. M., Cruden, D. M., Morgenstern, N. R. (1997): Assessment of the hazard fromrock fall on a highway. Can. Geotech. J. 34, 344–356.

Cancelli, A., Crosta, G. (1994): Hazard and risk assessment in rockfall prone areas. In:Skipp, B. O. (ed.), Risk and reliability in ground engineering. Thomas Telford, 177–190.

Chan, W. L. (1994): Hong Kong rainfall and landslides in 1993. GEO Report 43.

Chan, W. L. (1995): Hong Kong rainfall and landslides in 1994. GEO Report 54.


Chau, K. T. (1997): Rockfall, landslides and slope failures. In: Asaoka, A., Adachi, T.,Oka, F. (ed.), Deformation and progressive failure of geomechanics. Pergamon Press,Oxford, 907–921.

Chau, K. T., Wong, R. H. C., Lee, C. F. (1998): Rockfall problems in Hong Kong andsome new experimental results for coefficient of restitution. Int. J. Rock Mech. Min. Sci.35(4–5), 662–663, Paper No. 007.

Chen, P. K. H. (1993): Hong Kong rainfall and landslides in 1992. GEO Report 35.

Choot, E. B. (1983): Landslips caused by the June 1983 rainstorm. GEO Report 27.

Church, M., Stock, R. F., Ryder, J. M. (1979): Contemporary sedimentary environmentson Baffin Island, N.W.T., Canada: Debris slope accumulations. Arctic Alpine Res.11(4), 371–402.

Crandell, D. R., Fahnestock, R. K. (1965): Rockfalls and avalanches from Little TahomaPeak on Mount Rainier, Washington. U.S. Geological Survey Bull. 1221(A), 1–30.

Culshaw, M. G., Bell, F. G. (1991): The rockfalls of James Valley, St. Helena. In: Bell,F. G. (ed.), Landslides. Balkema, Rotterdam.

Civil Engineering Department (CED) (1995): Feasibility Study for QRA of Boulder fallhazard in Hong Kong: Final Report, ERM-Hong Kong Ltd./CED Agreement No. CE58/95, CED.

Descoeudres, F., Zimmermann, T. H. (1987): Three-dimensional dynamic calculation ofrockfalls. In: Proc., 6-th Int. Congress on Rock Mech., 337–342.

Douglas, G. R. (1980): Magnitude frequency study of rockfall in Co. Antrim, N. Ireland.Earth Surface Processes 5, 123–129.

Duarte, R. M., Marquinez, J. (2002): The influence of environmental and lithologic factorson rockfall at a regional scale: an evaluation using GIS. Geomorphology 43, 117–136.

Einstein, H. H. (1988): Landslide risk assessment procedure. In: Landslides, 5th Int. Symp.on Landslides, Vol. 2. Balkema, Rotterdam, 1075–1090.

Einstein, H. H. (1997): Landslide risk-Systematic approaches to assessment. In: Cruden,D. M., Fell, R. (eds.), Landslide risk assessment. Balkema, Rotterdam, 25–50.

Evans, N. C. (1992): Hong Kong rainfall and landslides in 1991. GEO Report 20.

Evans, N. C., Irfan, T. Y. (1991): Landslide studies 1991: Blast-induced rock slide at ShauKei Wan. Special Project Report SPR 6/91, GEO.

Evans, S. G., Hungr, O. (1993): The assessment of rockfall hazard at the base of talusslopes. Can. Geotech. J. 30, 620–636.

Fell, R. (1994): Landslide risk assessment and acceptable risk. Can. Geotech. J. 31,261–272.

Fell, R., Hartford, D. (1997): Landslide risk management. In: Cruden, D. M., Fell, R.(eds.), Landslide risk assessment. Balkema, Rotterdam, 51–109.

Finley, P. J., Fell, R. (1997): Landslides: risk perception and acceptance. Can. Geotech. J.34, 169–188.

Flageollet, J. C., Weber, D. (1996): Fall. In: Dikau, R., Brunsden, D., Schrott, L., Ibsen,M. L. (eds.), Landslide recognition-identification, movement and causes. John Wiley,Chichester, 13–28.

Fookes, P. G., Sweeney, M. (1976): Stabilization and control of local rock falls anddegrading rock slopes. Quart. J. Engng. Geol. 9, 37–55.


Fornaro, M., Peila, D., Nebbia, M. (1990): Block falls on rock slopes – Application of anumerical simulation program to some real cases. In: Price, D. G. (ed.), Proc., 6th Int.Congress IAEG. Balkema, Rotterdam, 2173–2180.

Gardner, J. (1967): Notes on avalanches, icefalls, and rockfalls in the Lake Louise District,July and August, 1966. Can. Alpine J. 50, 90–95.

Gardner, J. (1970): Rockfall: A geomorphic process in high mountain terrain. AlbertanGeographer 6, 15–20.

Gardner, J. (1977): High magnitude rockfall-rockslide: Frequency and geomorphic signifi-cance in the Highwood Pass area, Alberta. Assoc. Am. Geographers J., Great Plains-Rocky Mountain Division 6, 228–238.

Gardner, J. (1980): Frequency, magnitude, and spatial distribution of mountain rockfallsand rockslides in the Highwood Pass area, Alberta, Canada. In: Coates, D. R., Vitek,J. D. (eds.), Thresholds in geomorphology. Allen and Unwin, London, 267–295.

Gerrard, A. J. (1990): Mountain environments: an examination of the physical geographyof mountains. Belhaven Press, London.

Geotechnical Engineering Office (GEO) (1996a): QRA of Boulder fall hazards in HongKong: Phase 2 Study. November 29, 1996, C1453.

Geotechnical Engineering Office (GEO) (1996b): Compilation of a database on landslideconsequence. Submitted by Mitchell, McFarlane, Brentnall & Partners Int. Ltd., Agree-ment No. GEO 8/95, Final Report for Special Project Division, GEO.

Geotechnical Engineering Office (GEO) (1998): Landslides and Boulder falls from naturalterrain: Risk guidelines. Final report, GEO E53-68655.

Giani, G. P. (1992): Rockfalls, topples and buckles. In: Rock slope stability analysis.Chapter 7, Balkema, Rotterdam, 191–207.

Hansen, H. (1984): Landslide hazard analysis. In: Brunsden, D., Prior, D. B. (eds.), Slopeinstability. Chapter 13. Wiley, New York, 523–602.

Harp, E. L., Noble, M. A. (1993): An engineering rock classification to evaluate seismicrock-fall susceptibility and its application to the Wasatch Front. Bull. Assoc. Eng. Geol.30(3), 293–319.

Harp, E. L., Wilson, R. C. (1995): Shaking intensity thresholds for rock falls and slides:evidence from 1987 Whittier Narrows and Superstition Hills earthquake strong-motionrecords. Bull. Seism. Soc. Am. 85(6), 1739–1757.

Hartlen, J., Viberg L. (1988): Evaluation of landslide hazard. In: Landslides, 5th Int. Symp.on Landslides, Vol. 2. Balkema, Rotterdam, 1037–1057.

Hoek, E. (1987): Rockfall – a program in BASIC for the analysis of rockfalls from slopes,Unpublished notes, Golder Associates/University of Toronto.

Hudson, J. A. (1992): Rock engineering systems: theory and practice. Ellis, Horwood.

Hudson, R. R. (1982): Report on the rainstorm of August 1982. GEO Report 26.

Hungr, O. (1997): Some methods of landslide hazard intensity mapping. In: Cruden, D. M.,Fell, R. (eds.), Landslide risk assessment. Balkema, Rotterdam, 215–226.

Hungr, O., Evans, S. G. (1988a): Engineering evaluation of fragmental rockfall hazards.In: Bonnard, C. (ed.), Landslides. Proc., 5th Int. Symp. on Landslides, Vol. 1. Balkema,Rotterdam, 685–690.

Hungr, O., Evans, S. G. (1988b): Engineering aspects of rockfall hazards in Canada. Geo-logical Survey of Canada, Ottawa, Open File 2061.


Ishijima, Y., Roegiers, J. C. (1996): Slope failure kills 20 people. ISRM News J. 3(3),28–29.

Kienholz, H. (1978): Maps of geomorphology and natural hazards of Grindelwald, Switzerland:Scale 1 :10,000. Arctic Alpine Res. 10(2), 169–184.

Kobayashi, Y., Harp, E. L., Kagawa, T. (1990): Simulation of rockfalls triggered byearthquakes. Rock Mech. Rock Engng. 23, 1–20.

Leroi, E. (1997): Landslide risk mapping: Problems, limitation and developments. In: Cruden,D. M., Fell, R. (eds.), Landslide risk assessment. Balkema, Rotterdam, 239–250.

Lin, C. T., Amadei, B., Jung, J., Dwyer, J. (1996): Extensions of discontinuous deformationanalysis for jointed rock masses. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 33(7),671–694.

Luckman, B. H. (1976): Rockfalls and rockfall inventory data: Some observations from Sur-prise Valley, Jasper National Park, Canada. Earth Surface Processes 1, 287–298.

Malgot, J., Mahr, T. (1979): Engineering geological mapping of the west Carpathian landslideareas. Bull. Int. Assoc. Engng. Geol. 19, 116–121.

McCaroll, D., Shakesby, R. A., Matthews, J. A. (1998). Spatial and temporal patterns of lateHolocene rockfall activity on a Norwegian talus slope: a lichenometric and simulationmodeling approach. Arctic Alpine Res. 30(1), 51–60.

Moser, M. (1978): Proposals for geotechnical maps concerning slope stability potential inmountain watersheds. Bull. Int. Assoc. Engng. Geol. 17, 100–108.

Paronuzzi, P. (1989): Probabilistic approach for design optimization of rockfall protectivebarriers. Quart. J. Engng. Geol. 22, 175–183.

Peckover, F. L. (1975): Treatment of rock falls on railway lines. Bull. Am. Railway Engng.Assoc. (Chicago) 653, 471–503.

Peckover, F. L., Kerr, J. W. G. (1977): Treatment and maintenance of rock slopes on trans-portation routes. Can. Geotech. J. 14, 487–507.

Pfeiffer, T. J., Bowen, T. D. (1989): Computer simulation of rockfalls. Bull. Assoc. Eng. Geol.26(1), 135–146.

PIARC-World Road Association (1997): Landslides-techniques for evaluating hazard. PIARCTechnical Committee on Earthworks, Drainage, Subgrade (C12), 1–117.

Pierson, L. A., Davis, S. A., Van Vickle, R. (1990): The rockfall hazard rating systemimplementation manual. Oregon State Highway Division, Salem, OR.

Pierson, L. A. (1992): Rockfall hazard rating system. In: Rockfall prediction and control andlandslides case histories. Transportation Research Record No. 1343, 6–13.

Piteau, D. R., Peckover, F. L. (1978): Engineering of rock slopes. In: Schuster, R. L., Krizek,R. J. (eds.), Landslides: analysis and control. National Academy of Sciences, WashingtonDC, 192–228.

Porter, S. C., Orombelli, G. (1981): Alpine rockfall hazards. Am. Sci. 69, 67–75.

Premchitt, J. (1985): Hong Kong rainfall and landslides in 1984. GEO Report 1.






Prior, D. B., Stephens, N., Douglas, G. R. (1971): Some examples of mudflow and rockfallactivity in north-east Ireland. In: Brunsden, D. (ed.), Slopes: Form and process. Inst. Br.Geogr., Special Publication No. 3, 129–140.

Rapp, A. (1960): Recent developments of mountain slopes in Karkevagge and surroundings.Geogr. Ann. 42, 1–158.

Richards, L. R. (1988): Rockfall protection: A review of current analytical and designmethods. In: Meeting on rockfall dynamics and protective works effectiveness. Bergamo,pp. 11-1–11-13.

Ritchie, A. M. (1963): Evaluation of rockfall and its control. Highway Res. Rec. 17, 13–28.

Scholz, C. (1990): The mechanics of earthquakes and faulting. Cambridge University Press,Cambridge.

Simpson, H. E. (1973): Map showing areas of potential rockfalls in the Golden QuadrangleJefferson County, Colorado. U.S. Geol. Survey Map, I-761-C.

Spang, R. M., Sonser, H. (1995): Optimized rockfall protection by ‘‘ROCKFALL’’. In:Proc. 8th Int. Congr. Rock Mech., Tokyo, Vol. 3. 1233–1242.

Siu, K. L. (1990): Hong Kong rainfall and landslides in 1989. GEO Report 6.

Tang, K. Y. (1991): Hong Kong rainfall and landslides in 1990. GEO Report 14.

Tang, M. C. (1982): Report on the rainstorm of May 1982. GEO Report 26.

Varnes, D. J. (1984): Landslide hazard zonation: a review of principles and practice. UNESCO,France, 1–63.

Voight, B. (ed.) (1978a): Rockslides and avalanches, 1: Natural phenomena. Dev. Geotech.Engng. 14A, Elsevier, Amsterdam.

Voight, B. (ed.) (1978b): Rockslides and avalanches, 2: Engineering sites. Dev. Geotech.Engng. 14B, Elsevier, Amsterdam.

Whalley, W. B. (1974): The mechanics of high magnitude-low frequency rock failure and itsimportance in mountainous areas. Geogr. Papers 27, Reading University, 1–48.

Whalley, W. B. (1984): Rockfalls. In: Brunsden, D., Prior, D. B. (eds.), Slope instability.Chapter 7, John Wiley & Sons, New York, 217–256.

Whiteside, P. (1984): Discussion on ‘Rockfall and its control’. Hong Kong Engineer June 1984,12.

Wieczorek, G. F., Nishenko, S. P., Varnes, D. J. (1995): Analysis of rock falls in theYosemite Valley, California. In: Proc., 35th US Symp. on Rock Mech., Balkema, Rotter-dam, 85–89.

Wolman, M. G., Miller, J. P. (1960): Magnitude and frequency of forces in geomorphic pro-cesses. J. Geology 68, 54–74.

Wong, C. K. L. (1996): Hong Kong rainfall and landslides in 1995. GEO Report 59.

Wong, C. K. L. (1997): Hong Kong rainfall and landslides in 1996. Special Project DivisionReport SPR7/97.

Wu, S.-S. (1985): Rockfall evaluation by computer simulation. Transport. Res. Rec. 1031,1–5.

Wyllie, D. (1987): Stemming rock falls. Railway Track and Structures 83(4), 13–14.

Yamagishi, H. (2000): Recent landslides in western Hokkaido, Japan. Pure Appl. Geophys. 157,1115–1134.


Zvelebil, J. (1988): Rhythmical patterns of rockfall evolution in NNW Bohemia. In: Bonnard,C. (ed.), Landslides, Vol. 1. 803–809.

Author’s address: Prof. Kam Tim Chau, Department of Civil and Structural Engi-neering, The Hong Kong Polytechnic University, Kowloon, Hong Kong, China; e-mail:[email protected].

408 K. T. Chau et al.: Rockfall Hazard Analysis for Hong Kong

Documents

Chau K.T., Wong R.H.C., Liu J. and Lee C.F. (2003) “Rockfall hazard analysis for Hong Kong based on rockfall inventory” Rock Mechanics and Rock Engineering, Vol. 36, No. 5, pp