ORI GIN AL PA PER

Systematic documentation of landslide events in Limbearea (Mt Cameroon Volcano, SW Cameroon): geometry,controlling, and triggering factors

V. B. Che • M. Kervyn • G. G. J. Ernst • P. Trefois • S. Ayonghe •

P. Jacobs • E. Van Ranst • C. E. Suh

Received: 6 October 2009 / Accepted: 26 January 2011 / Published online: 23 March 2011� Springer Science+Business Media B.V. 2011

Abstract Limbe town and surrounding areas, on the SE foot slopes of the active Mt

Cameroon Volcano, have experienced numerous small-scale shallow landslides within the

last 20 years. These resulted in the loss of *30 lives and significant damage to farmland

and properties. Landslides and their scars are identified in the field, and their geometry

systematically measured to construct a landslide inventory map for the study area. Specific

landslides are investigated in detail to identify site-specific controlling and triggering

factors. This is to constrain key input parameters and their variability for subsequent

susceptibility and risk modeling, for immediate local and regional applications in land-use

planning. It will also enable a rapid exploration of remediation strategies that are currently

lacking in the SW and NW regions of Cameroon. Typical slides within the study area are

small-scale, shallow, translational earth, and debris slides though some rotational earth

slides were also documented. The depletion zones have mean widths of 22 m ± 16.7 m

and lengths of 25 ± 23 standard deviation. Estimated aerial extents of landslide scars and

volume of generated debris range from 101 to 104 m2 and 2 to 5 9 104 m3, respectively. A

key finding is that most slope instabilities within the study area are associated with and

appear to be exacerbated by man-made factors such as excavation, anarchical construction,

and deforestation of steep slopes. High intensity rainfall notably during localized storms is

the principal triggering factor identified so far. The findings from this case study have

relevance to understanding some key aspects of locally devastating slope instabilities that

commonly occur on intensely weathered steep terrains across subtropical Africa and in the

subtropics worldwide and affecting an ever denser and most vulnerable population.

V. B. Che (&) � S. Ayonghe � C. E. SuhDepartment of Geology and Environmental Science, University of Buea, Buea, Cameroone-mail: chevivianbih@yahoo.com

M. KervynDepartment of Geography, Vrije Universiteit Brussels, Free University of Brussels, Brussels, Belgium

V. B. Che � M. Kervyn � G. G. J. Ernst � P. Jacobs � E. Van RanstDepartment of Geology and Soil Science, Ghent University, Ghent, Belgium

P. TrefoisRoyal Museum for Central Africa, Tervuren, Belgium

123

Nat Hazards (2011) 59:47–74DOI 10.1007/s11069-011-9738-3

Keywords Cameroon � Landslide � Scar geometry � Sliding mechanism � Causal and

triggering factors

1 Introduction

Landslides are recognized and well-studied geomorphic hazards affecting numerous areas

in the world especially across the subtropics, where intense and prolonged rainfalls are

dominant. They represent a major process through which hill slope developed (Ahmad

and McCalpin 1999; Ayalew and Yamagishi 2004; Knapen et al. 2006) and are one of

the most common and damaging natural hazards threatening life, property, and the

livelihood of millions of persons (Ahmad et al. 1994). Milestone studies of landslides are

provided by Varnes (1978), Cruden and Varnes (1996), Carrara et al. (1999), Guzzetti

et al. (2003) and Davies et al. (2005). Some of these studies aim specifically at assessing

landslide susceptibility through the use of statistic or heuristic approaches (Dai and Lee

2002; Suzen and Doyuran 2004; Knapen et al. 2006; Buh 2009). Others have focused on

characterizing the geometric properties of landslide (Hovius et al. 1997; Malamud et al.

2004a, b) and others in evaluating the run-out behavior of slides defining what has been

termed the angle of reach or angle of apparent cohesion defined by the ratio of height

drop to horizontal distance for a given landslide (Corominas 1996; Dahl et al. 2010).

Most of these studies are based on slide descriptions from papers, reports and on the

interpretation of aerial photographs (Wen et al. 2004), rather than on first-hand field

studies.

In Cameroon, landslides have been reported mainly in the Bamboutos (Zogning et al.

2007; Ayonghe and Ntasin 2008), Limbe, Bafaka, and in the northwest region of the

country (Lambi and Ngwana 1991; Ayonghe et al. 1999, 2002, 2004; Ngole et al. 2007;

Thierry et al. 2008). These authors provide descriptive analyses of the factors conditioning

and triggering these slides, evaluate the amount of damage caused, while on the other hand,

little has been published with regard to field-measured geometric characteristics of land-

slides or landslide scars. From archived landslide records in Cameroon, a total death toll of

146 persons have been recorded in the last three decade giving an average of 4.4 landslide

related deaths per year assuming a poison distribution.

At Mt Cameroon (MC), though volcanic and low-intensity earth tremors, lava flows and

toxic ash falls all pose threats as eruptions historically recurred every 10–20 years on

average (Suh et al. 2003), recurrent small-scale landslides have been the main cause of

fatalities and destruction of local community livelihood. Hence, small landslides can be

rated as the most significant and recurrent geohazard after floods. Despite their frequent

occurrence and dramatic impact for local communities, no systematic data on this geo-

morphic process has yet been collected for the Limbe region. As a result, no relevant

geohazard or risk modeling has been developed or remediation strategies explored par-

ticularly in the Limbe area. Hence, there is a need for the construction of a national

database for these phenomena.

In the broader NW, W, and SW regions of Cameroon, crater lake out gassing has caused

two major disasters in the 1980s and remains a concern that is being tackled at Lake Nyos

and Monoun so that small landslides are arguably the most common and recurrent geo-

hazard in these regions where *6 million people live. Landslides in this region have

considerable social and economic consequences (Ayonghe et al. 2004, Ayanji 2004;

Ayonghe and Ntasin 2008). They result in the destruction of farm land, subsistent and

48 Nat Hazards (2011) 59:47–74

123

industrial cash and food crops, disrupt transportation means, cause immediate damage to

infrastructure and probably contribute to soil and biodiversity loss even though this has not

been monitored in the affected areas. Although mass movements are a prime hazard within

this region, information about landslides in Cameroon is generally restricted.

This paper systematically document and quantify for the first time the overall geometric

parameters of small-volume slides in Limbe based on field measurements, and establishes a

frequency-size statistic of the observed slides in order to quantify the occurrence of slides.

It also aims at identifying in qualitative terms the causal and triggering factors for a few

landslides for which sufficient observations are available. This is to help constraints for

subsequent modeling, monitoring, and remediation efforts that might be applied in Limbe

or other areas affected by such small-scale failures. Geometric and site-specific charac-

teristics of landslides will assist in understanding the sliding mechanism and in the iden-

tification of the causal and triggering factors. This effort constitutes the first essential step

toward deriving a landslide susceptibility map, to help implement future practice that aim

at reducing landslide occurrence or impact via remediation, land-use, and urban planning

efforts. The findings from the present study also have some generic value to help document

small-scale slope instability problem elsewhere in the subtropics besides providing insights

into small-scale devastating geohazards affecting vulnerable people already affected by

absolute poverty.

2 Description of study area

Limbe is a coastal town [*85,000 inhabitants in 2005 (Bureau Central des Recensements

et des Etudes de Population 2010)] that lies on the SSE foot slopes of MC, a 4,100-m-high,

active lava-dominated volcano (Fig. 1). Most of the inhabitants of this area are low income

earners. The most important economic activities in this area include farming, with indi-

viduals engaged in subsistence and cash crop production (Cassava, banana, cocoa, maize,

palm and peanut), fishing, and petty trading. Recently, there has been an explosion in the

construction of houses with reinforce concrete, and cement block though traditional

wooded houses constructed with wood locally called ‘‘carabot’’ are also numerous. These

construction works hardly respect standard building codes or regulations put in place by

the authorities of the Ministry of Town Planning and Housing.

Geomorphologically, for its main part, the study area is made up of ridges and

deeply incised ravines with a general W–E orientation (Fig. 1), at high angle to the

general NE–SW orientation of MC and gently sloping foot slopes of MC composed of

multiple porphyritic basaltic lava flows, punctuated by several strombolian pyroclastic

cones to the W and NW and lahar deposit to the E of the study area (Fig. 2). These

ridges form part of the Limbe-Mabeta volcanic massif, made up of degraded and deeply

weathered Tertiary basaltic lava flows (Hasselo 1961). Individual ridges are separated

by asymmetric V-shaped valleys occupied by perennial and/or ephemeral streams.

These streams either empty themselves directly into the ocean or into the delta around

Mabeta (Fig. 1).

The main rock types within this area include basalts, basanites, lahar deposits, and

pyroclastic materials. Theses rocks either lie exposed at the surface or are covered by

extremely fertile dark brown, reddish brown, yellowish and/or pale yellow sticky, clay, silt

and silty clay soils derived from in situ intense weathering. Soil thicknesses range from a

few centimetres to more than 10 m in some areas. Soil formation and vegetation recovery

Nat Hazards (2011) 59:47–74 49

123

Fig. 1 Location of study area, Limbe and its surrounding indicating the main geomorphologiccharacteristics, trend of the hydrographic network and morphology of some pyroclastic cones. Includedare the maps of Cameroon indicating the location of Mt Cameroon in the SW region, and location of studyarea on the SE foot slope of MC

50 Nat Hazards (2011) 59:47–74

123

rates are so far unquantified but are clearly extremely rapid so that areas destroyed by

landslides experience full vegetation recovery in less than 5 years. The vegetation is

mainly secondary tropical forest made up of deep rooted trees with heights that range from

3 m to over 10 m, characterized by interrupted canopy, wild palms, banana, rubber and

palm plantations, indigenous subsistence farms and fruit tree (mango, plum, avocado).

Only a small portion (\5%) of the study area is covered by mangrove forest. The sec-

ondary forest developed after intense deforestation of the primary tropical forest

*50 years ago for the installation of agro-industrial complexes and progressive

urbanization.

Fig. 2 Geologic map of the Limbe and its environs modified by Endeley et al. (2001) and Thierry et al.(2008) and validated with field observations

Nat Hazards (2011) 59:47–74 51

123

Elevations in the study area range from 0 m at sea level to about 1,200 m a.s.l. Slope

gradient derived from a 20-m Digital Elevation Model (DEM) obtained by digitalizing

contours from a 1:50,000 topographic map Buea, NB-32-IV range from 0� to 43�. Slopes

with gradient greater than 20�, where slides are most likely to occur, make up *7% of the

study area. Field measurements indicate that DEM-derived slopes are systematically lower

than the maximum slope at any location. The climate in the study area is subequatorial with

two distinct seasons—a 4-month dry season from November to mid-March and an 8-month

rainy season that runs from mid-March to November with mean annual rainfall amount of

*3,100 mm ± 1,100 standard deviation.

Cognizance of the fact that the study area lies on the slopes of active MC it is affected

by low intensity seismic activity that is monitored by the Unit for Volcanological and

Geophysical Research Centre (ARGV) located at Ekona (located out of the study area).

Seismicity here is characterized by low-intensity earth quake that might not even be felt at

the epicenter. An average of 15 events per day was recorded prior to the most recent

eruption on the mountain in 1999. This average increased to over 200 seismic events on the

27th and 28th of March when the eruption actually began (Aka 2001 in Buh 2010). On the

other hand, the month of June 2001 was characterized by 0–8 low-magnitude seismic

events per/day, and there was no observed increase prior to or after the 27th of June 2001,

suggesting that seismicity was not a significance triggering factor of the slide occurrence

(Buh 2009). Thus, the importance of seismic activities as a trigger is not evaluated in this

study given most of the recorded slides occurred in 2001.

3 Method of investigation

Several landslide activities have been reported within the Limbe municipality. The earliest

recorded landslide event within the study area occurred at Cassava farm (Fig. 1) in Sep-

tember 1989 following heavy rains (Lambi and Ngwana 1991). The most severe of these

mass movement events accompanied by floods occurred on the afternoon of June 27, 2001

affecting *3,000 inhabitants in the neighborhoods of Mabeta New layout, Towe, Unity

quarters, Livanda Congo, Bonjo, and Mukuka (Ayonghe et al. 2004; Ayanji 2004; Thierry

et al. 2008). Approximately 23 persons were killed, 50 injured, *120 houses completely

destroyed, community infrastructure (roads, water and electricity supply line, communi-

cation systems, schools, and churches) and farmland destroyed. Total losses were esti-

mated at 1.5 billion FRS CFA (*3 million US dollars) (Ayanji 2004). Since 2001,

repeated landslide events have been recorded with 6 more casualties and at least one slide

reported every 2 years within the study area. Single events that generate multiple slides are

however rare. Frequent flooding in Limbe result from the non-existence or presence of

poor drainage systems, poor land-use patterns, rapid urbanization, and non-application of

standard building codes/measures proposed by the Ministry of Town Planning and

Housing. For this reason, Limbe and its neighborhood were chosen as the pilot area for the

present study to constrain the geometric characteristics of landslide scars and to determine

the factors responsible for slope failure.

This study involved detailed field observations and mapping of the geometric config-

uration of landslide scars formed during June 27, 2001, September 26, 2005, July 14, 2006,

July to August 2008, June 29–30, 2009, and August 6, 2009 sliding events at Mabeta,

Bonjo, Makuka, Towe, Mandoli, Unity quarters, Livanda South, and Kie Village,

respectively (Fig. 1). Due to the lack of recent and regular aerial photography surveying of

the region, only landslide scars observed in the field are accounted for in this study. Field

52 Nat Hazards (2011) 59:47–74

123

mapping and interviews with the local inhabitants that witness the landslide event or its

impact provided insights into the factors responsible for landslide occurrence within the

study area. Landslides were recognized from sharp changes in vegetation type, presence of

bare crescent (acute downslope) shaped scarps, sharp depression on the landscape as well

as the presence of displaced material at the foot of the scar (jumbled-up mixture of subsoil

and topsoil).

UTM geographic coordinates (WGS84 datum) for the scars were obtained at the left

margin, center and right margins of the crown and at the toe with a Garmin Etrex GPS

receiver to map out the outline of the scar. Measurements of the width of rupture (depletion

zone) (Wr), length of rupture (Lr), and scarp’s height (h) (Fig. 3) were obtained with a

graduated surveyor tape. From these parameters, the area of the rupture zone (A) and the

volume (Vl) of displaced material were estimated using the follow standard formulae:

A ¼ Lr�Wr

(Guthrie and Evans 2004)

Vl ¼ 1

6�P� Lr�Wr� h

(Cruden and Varnes 1996)

The latter equation assumes that each slide has an elliptical shape, which appears to be a

reasonable first-order approximation based on the detailed field observations. It is worth

noting that Lr is difficult to measure because the outline of the rupture surface around the

foot is usually buried under displaced material. Hence, Lr was extrapolated from the

outline of the main scarp and slide flanks with the elevation along the extrapolation line

taken as the elevation at the foot. Slide run-out distance was measured for all fresh slides

and ignored in cases where the slide trail had either been completely re-colonized by

vegetation or converted into farmland. Slope gradient and orientation before failure were

estimated by measuring the gradient and orientation of the slide-adjacent slopes with a

Silva compass/clinometer.

Cruden and Varnes (1996)’s classification was used for reference to a classic description

framework in this study. According to their definitions, an earth slide refers to failure in

Fig. 3 Sketch of 2009 debris slide at Moliwe and a view of the geometric parameters measured in the field.A represents the width of the rupture surface measured at its widest end, B length of surface of rupture(depletion zone) measured from the head scarp to the foot of the slide and C is total run-out distance (totallength) from scarp to the toe of the slide. Nomenclature after Dikau et al. (1996) and Knapen et al. (2006)

Nat Hazards (2011) 59:47–74 53

123

which more than 80% of the debris is less than 2 mm across, whereas a debris slide refers

to downslope mass movement in which 20–80% of the debris is coarser than 2 mm. This

distinction was made on the basis of granulometry characterization of field samples,

accounting for the proportion of large boulders observed in the field and not represented in

the samples. These slides are further classified as translational and rotational slides,

respectively. Rotational slides have a curve (spoon-shaped) slip surface, show backward

rotation of trees within the debris, and generally result in slope reversal, while translational

slides are characterized by a planar slip surface and show forward rotation with no slope

reversal (Dikau et al. 1996).

Rainfall data obtained from 10 manually operated rain gauges located within the study

area (Fig. 4) and managed by the Cameroon Development Cooperation (CDC) provided at

least 20 and at most 36 years (1975–2009) of monthly rainfall, 4 of which had 10 years

(2000–2009) of daily rainfall, whereas 6 automated self-emptying Oregon rain gauges

Chop farm Zone

Makuka ZoneBonjo

Zone

Mabeta Zone

Kie Zone

Zone

Fig. 4 Locations of landslide scars, zones, and rain station that provided rainfall data for this analysis. Twosets of rain stations are observed—manually operated rain stations managed by the CDC from whichmonthly data from 1973 to 2009 and automatic self-emptying Oregon rain station planted in 2009 for thisresearch. Red dots with white rims represent scar described in detail in this study

54 Nat Hazards (2011) 59:47–74

123

capable of measuring total daily rainfall provide daily rainfall from April (2009 to August

2010). Mean annual rainfall ranges from 2,100 to 4,600 mm with a total of 104–212 rainy

days a year and mean annual temperature of *26�C. Maximum monthly rainfall occurs in

June, July, and August, which coincide with the occurrence or recorded landslides. Mean

monthly rainfall amount is *400 mm for June and 680 mm for July and August. Pre-

cipitation occurs in the form of light to heavy rains typically associated with spatially

localized rainstorms that vary in duration from a few minutes to over 4 days in a role.

Daily records indicated that local rainstorms produce huge gradients in rainfall even on a

5-km spatial scale around MC. Hence, very different amounts of rainfall are recorded at the

different stations, due to the highly localized nature of rainstorms. It is therefore likely that

rainfall at a particular site could be higher or lower than reported. In this study, 5-km buffer

zones were constructed around each slide and we mention the highest daily rainfall

recorded for stations located within the 5-km buffer zone or data for the closest rain gauge,

if located less than 1 km from the observed landslide.

4 Slide description

A total of 62 slides, 52 recent (i.e. slides with well-defined margins, head scarp, with no or

partially developed drainage channels) and 10 older landslides (i.e. slides where margins,

head scarp have been degraded) scars were observed. These slides together produced a

total volume of *105 m3 of debris from a total area of *3.3 km2, which is *0.46% of the

study area. Mean width of the depletion zone (Wr) is 22 m ± 16.7 m standard deviation

(sd), and length (Lr) 25 ± 23 m sd. Individual aerial extent and volume range from a few

m2 to 9.3 9 103 m2 and 2.5 m3 to 5 9 104 m3, respectively. Most of these failures initiate

at mid slope rather than at the shoulder or top of the slope. Slope length below the scarp

(drop in height of the depletion zone) range from 1 to 81 m with a mean of 18 m ± 16sd.

The accumulation zone of 87% of all observe slides had either been washed away by

stream water, reworked, covered by vegetation, or converted into farmland; hence, actual

run-out distance could not be measured. Hence, these values in general underestimate the

total mass or volume of material that moved downslope or the total area affected.

The observed slides can be grouped into 6 landslide zones: Mabeta zone which covers

Mabeta, Cassava farm, Towe, and Unity quarter, the Bonjo zone made of Bonjo, and

Mandoli, the Chopfarm, Makuka, Moliwe, and Kie landslide zones, respectively, with a

few isolated cases that do not fall within any of these zones (Fig. 4). Geometric charac-

teristics of the individual slides observed within the study area are presented in Table 1.

One slide within each of the zone is described in detail hereafter as type example to

illustrate the geological setting, the material involved and the potential causal factors.

Human, material, economic, and environmental losses associated with slides can vary

greatly depending on their location. Documented impacts of slides within the study area

are summarized on Table 2. From Tables 1 and 2, it can be seen that no correlation exists

between the size of slides and their associated impacts. Indeed, slides with whatever size

will have the greatest impact when they occur close or within inhabited areas.

Another key observation is that during field visits, most of the slides that occurred in

2009 were associated with slopes that had been newly excavated for construction purposes.

In addition, areas that were affected by the 2001 slide event had been rebuilt without the

application of formal stabilization measures. It is therefore obvious that despite the threats

posed by nature on the inhabitants of these hazardous terrains, poverty will force indi-

viduals to resettle in the same area because they really have no other choice or alternatives.

Nat Hazards (2011) 59:47–74 55

123

Ta

ble

1G

eom

etri

cch

arac

teri

stic

so

fal

lo

bse

rved

scar

sm

easu

red

inth

eL

imbe

area

(so

uth

ern

foo

tsl

op

eso

fM

tC

amer

oo

n)

Sli

de

IDL

on

git

ud

ein

UT

ML

atit

ud

ein

UT

ME

levat

ion

(m)

Wid

tho

fsc

arp

(m)

Hei

gh

to

fsc

arp

(m)

Len

gth

(m)

Init

ial

slo

pe

Vo

lum

e(m

3)

Are

a(m

2)

Ty

pe

15

23

,90

04

41

,61

35

83

1.6

2.8

23

.72

21

,098

74

9R

ota

tio

nal

25

23

,72

04

40

,61

61

14

85

.01

0.0

11

0.0

30

48

,97

69

,350

Tra

nsl

atio

nal

35

27

,48

54

38

,38

41

57

30

.05

.04

7.9

N/A

3,7

64

1,4

37

Tra

nsl

atio

nal

45

26

,51

14

38

,69

71

94

39

.03

.02

7.6

46

1,6

91

1,0

76

Tra

nsl

atio

nal

55

24

,51

74

41

,94

21

24

24

.53

.01

08

.02

64

,158

2,6

46

Tra

nsl

atio

nal

65

24

,98

04

41

,95

71

10

17

.93

.52

0.0

32

65

63

58

Tra

nsl

atio

nal

75

24

,96

74

41

,94

21

05

40

.04

.04

0.0

27

3,3

52

1,6

00

Tra

nsl

atio

nal

85

24

,96

24

42

,03

18

13

7.0

5.0

29

.02

82

,810

1,0

73

Tra

nsl

atio

nal

95

23

,73

74

41

,71

63

61

2.5

5.0

13

.03

24

26

16

3T

ran

slat

ion

al

10

52

3,7

10

44

1,7

47

33

21

.22

.02

7.0

28

60

05

72

Tra

nsl

atio

nal

11

52

4,7

41

44

2,1

01

43

9.5

2.5

22

.63

22

81

21

5T

ran

slat

ion

al

12

52

4,7

24

44

3,9

04

78

11

.53

.51

4.5

28

30

61

67

Tra

nsl

atio

nal

13

52

4,7

12

44

3,9

52

10

42

2.5

6.0

57

.63

24

,073

1,2

96

Tra

nsl

atio

nal

14

52

4,7

06

44

3,9

02

76

16

.52

.01

0.0

28

17

31

65

Tra

nsl

atio

nal

15

52

4,6

37

44

3,8

81

86

17

.02

.56

.03

21

34

10

2T

ran

slat

ion

al

16

52

7,9

67

45

0,0

28

18

68

.41

.16

.32

83

05

3T

ran

slat

ion

al

17

52

8,1

41

45

0,1

64

21

32

0.3

1.5

6.2

28

99

12

6R

ota

tio

nal

18

52

5,6

26

44

3,9

12

11

21

2.7

8.3

5.0

45

27

66

4T

ran

slat

ion

al

19

52

5,6

83

44

3,9

32

12

99

.02

.51

5.0

28

17

71

35

Tra

nsl

atio

nal

20

52

3,7

65

44

1,4

18

10

31

8.2

5.3

34

.32

21

,733

62

4T

ran

slat

ion

al

21

52

3,9

03

44

1,5

01

96

20

.92

.03

9.8

36

87

18

32

Tra

nsl

atio

nal

22

52

3,9

07

44

1,4

91

84

30

.01

.51

5.1

22

35

64

53

Tra

nsl

atio

nal

23

52

9,9

36

44

9,2

42

18

43

.03

.00

.57

02

2R

ock

fall

24

52

9,9

51

44

9,2

53

18

23

9.4

2.0

0.5

70

21

20

Tra

nsl

atio

nal

25

53

0,0

86

44

9,3

39

18

47

9.0

5.6

12

.03

02

,781

94

8T

ran

slat

ion

al

56 Nat Hazards (2011) 59:47–74

123

Ta

ble

1co

nti

nu

ed

Sli

de

IDL

on

git

ud

ein

UT

ML

atit

ud

ein

UT

ME

levat

ion

(m)

Wid

tho

fsc

arp

(m)

Hei

gh

to

fsc

arp

(m)

Len

gth

(m)

Init

ial

slo

pe

Vo

lum

e(m

3)

Are

a(m

2)

Ty

pe

26

52

8,8

70

44

9,1

64

17

91

5.0

1.2

3.1

46

29

47

Tra

nsl

atio

nal

27

52

9,8

01

44

8,7

35

19

63

0.0

8.0

––

––

Tra

nsl

atio

nal

28

52

5,2

00

44

1,8

95

12

45

5.1

5.0

52

.12

67

,519

2,8

71

Tra

nsl

atio

nal

29

52

4,5

07

44

1,9

68

12

11

5.0

3.5

3.0

60

83

45

Tra

nsl

atio

nal

30

52

3,9

68

44

1,8

88

14

15

.01

0.0

1.0

70

79

15

Tra

nsl

atio

nal

31

53

4,6

75

45

3,2

20

28

98

.01

.01

2.0

05

09

6T

ran

slat

ion

al

32

53

4,7

40

45

3,4

28

32

22

5.0

3.0

13

.00

51

13

25

Tra

nsl

atio

nal

33

53

4,7

53

45

3,4

27

32

01

2.0

2.0

12

.00

15

11

44

Tra

nsl

atio

nal

34

53

7,7

88

45

2,2

14

11

03

4.0

1.0

10

.04

01

78

34

0T

ran

slat

ion

al

35

52

3,6

19

44

4,2

80

33

8.0

0.9

––

––

Tra

nsl

atio

nal

36

52

3,8

04

44

4,2

80

34

7.0

5.0

––

––

Tra

nsl

atio

nal

37

53

1,2

82

45

0,7

18

17

4–

––

––

–T

ran

slat

ion

al

38

51

9,7

28

44

9,2

74

62

8–

––

––

–T

ran

slat

ion

al

39

52

5,0

17

44

8,1

09

24

60

.00

.00

.0–

––

Tra

nsl

atio

nal

41

52

3,4

66

44

0,9

66

14

01

9.2

2.8

32

.02

9.8

90

16

14

Tra

nsl

atio

nal

40

52

4,9

28

44

3,9

80

10

38

.01

.00

.0–

––

Tra

nsl

atio

nal

42

52

7,5

51

45

0,4

55

22

18

.01

.50

.0–

––

Tra

nsl

atio

nal

43

52

7,5

79

45

0,4

03

24

21

3.6

5.2

0.0

––

–T

ran

slat

ion

al

44

52

5,7

24

44

3,9

41

13

33

0.0

7.0

0.0

––

Tra

nsl

atio

nal

45

52

5,9

76

44

4,1

47

22

34

3.1

2.0

42

.13

21

,901

1,8

15

Tra

nsl

atio

nal

47

52

4,7

54

44

4,6

22

16

43

1.1

4.0

45

.03

02

,932

1,4

00

Tra

nsl

atio

nal

48

52

4,9

34

44

4,7

48

16

21

5.0

2.5

30

.04

05

89

45

0T

ran

slat

ion

al

49

52

7,3

19

45

0,4

54

22

81

6.0

2.5

31

.72

06

64

50

7T

ran

slat

ion

al

50

52

4,7

25

44

4,0

33

14

78

.82

.01

3.2

58

12

21

16

Tra

nsl

atio

nal

51

52

5,2

28

44

1,9

75

97

8.0

1.0

10

.0–

42

80

Tra

nsl

atio

nal

Nat Hazards (2011) 59:47–74 57

123

Ta

ble

1co

nti

nu

ed

Sli

de

IDL

on

git

ud

ein

UT

ML

atit

ud

ein

UT

ME

levat

ion

(m)

Wid

tho

fsc

arp

(m)

Hei

gh

to

fsc

arp

(m)

Len

gth

(m)

Init

ial

slo

pe

Vo

lum

e(m

3)

Are

a(m

2)

Ty

pe

52

52

9,3

49

45

9,7

34

71

69

.01

.02

0.0

30

94

18

0T

ran

slat

ion

al

53

52

8,0

52

45

3,4

05

46

11

0.0

1.0

50

.03

62

62

50

0T

ran

slat

ion

al

54

52

2,9

67

44

4,2

29

65

4.0

1.0

10

.06

52

14

0T

ran

slat

ion

al

55

52

2,9

52

44

4,2

82

83

9.5

1.0

9.0

64

45

86

Tra

nsl

atio

nal

56

52

2,9

47

44

4,3

09

86

10

.01

.07

.06

43

77

0T

ran

slat

ion

al

57

52

3,7

85

44

5,1

74

10

61

7.7

3.0

6.0

40

16

71

06

Tra

nsl

atio

nal

58

52

7,4

58

44

8,8

62

18

76

.01

.06

.03

01

93

6T

ran

slat

ion

al

59

52

7,9

74

44

9,0

37

20

42

0.8

2.3

61

.62

61

,539

1,2

77

Tra

nsl

atio

nal

61

52

0,0

80

44

4,2

12

13

85

2.5

63

6.0

34

5,9

40

1,8

90

Rota

tio

nal

62

51

9,8

27

44

4,2

35

80

46

.26

45

.03

66

,534

2,0

79

Tra

nsl

atio

nal

Sli

de

typ

eaf

ter

clas

sifi

cati

on

sch

eme

by

Cru

den

and

Var

nes

(19

96)

and

Kn

apen

etal

.(2

00

6)

58 Nat Hazards (2011) 59:47–74

123

Table 2 Date of landslide occurrence and associated damage in the Limbe area

SlideID

Locationname

Year ofoccurrence

Farmland

Culvert Buildingsdestroyed

Roadblock

Casualties/injury

Otherdamage

1 Bonjo 26-Sep-05 x x – –

2 Mondoli 27-Jun-01 x – –

3 Chop farm 27-Jun-01 x x – –

4 Chop farm 27-Jun-01 – – x – –

5 Makuka 27-Jun-01 x x – – –

6 Makuka 27-Jun-01 – – –

7 Makuka 27-Jun-01 x – – – – –

8 Makuka Unknown – – –

9 Bonjo 2004 x – – –

10 Bonjo 1999 x – – –

11 Makuka 08-Aug – – –

12 Mabeta New layout 27-Jun-01 x x – – –

13 Mabeta New layout 27-Jun-01 x x – – –

14 Mabeta New layout 27-Jun-01 – – – – –

15 Mabeta New layout 27-Jun-01 x – – – – –

16 Mt Mbinde Unknown x – – – – –

17 Mt Mbinde Unknown x – – – – –

18 Mabeta New layout Unknown – – – – –

19 Mabeta New layout Unknown – – – – –

20 Bonjo Unknown x – – – – –

21 Bonjo Unknown x – – – – –

22 Bonjo Unknown – – – – –

23 Ombe1 Unknown x – – – – –

24 Ombe 2 Unknown x – – – – –

25 Ombe 3 Unknown x – – – – –

26 Tomatel 1 Unknown x – – – – –

27 Tomatel 2 Unknown x – – – – –

28 Makuka 27-Jun-01 x – – – – –

29 Makuka Unknown – – – – – –

30 Bonjo Unknown – – – – – –

31 Mutengene 2007 – – – – – –

32 Mutengene 2007 – – – – – –

33 Mutengene 2007 – – – – – –

34 Likomba Unknown – – – – – –

35 Cassava farm 27-Jun-01 – – – – – –

36 Cassava farm 27-Jun-01 – – – – – –

37 Mevio Unknown – – – – –

38 Engel Mount Unknown – x – – – –

39 Mile 4 Unknown – – – – – –

40 Mabeta New layout 27-Jun-01 – – – – – –

41 Mondoli Unknown x – – – – –

42 Mt Mbinde Unknown x – – – – –

Nat Hazards (2011) 59:47–74 59

123

4.1 Mabeta slides

Five first time (slopes that have never been affected by landslide) translational earth slides

Fig. 5 occurred on degraded pyroclastic cones with slope gradient of 30� to 40� on the

afternoon of the 27th of June 2001 after severe rains during which 180 mm of rain in 24 h

was recorded at Krater some 2 km away. People were buried by the floods and landslides

which destroyed their houses. Together, all the slides at Mabeta destroyed four houses and

killed *23 people (Ayanji 2004; Ayonghe et al. 2004; Zogning et al. 2007). Eyewitnesses

reported observing water oozing out of the ground below the foundations of some houses

located downslope and anomalous muddy surface runoff a few hours before sliding. This

crucial observation could be used in risk awareness raising for the biggest of future events

and to understand the process that led to sliding.

The widths of rupture zone (Wr) of individual slides range from 12 to 43 m, Lr from 17

to 58 m and h from 2 to 7 m. The approximate run-out distance of the largest of these scars

is 350 m (slide 1 on Fig. 5). A longitudinal section through slide 3 in Fig. 5 is presented in

Fig. 6. The depletion zone has a well-defined scarp and margins made up of a mixture of

dark red and reddish brown soil and weathered vesicular porphyritic basaltic pyroclastic

Table 2 continued

SlideID

Locationname

Year ofoccurrence

Farmland

Culvert Buildingsdestroyed

Roadblock

Casualties/injury

Otherdamage

43 Mt Mbinde Unknown x – – – – –

44 Mabeta New layout 27-Jun-01 – – – – –

45 Mabeta New layout 27-Jun-01 x – 6 – 14 x

46 Mabeta New layout Unknown – – – – –

47 Towe slide 1 27-Jun-01 x – x – – –

48 Towe slide 2 27-Jun-01 x – x – –

49 Mt Mbinde Unknown x – – –

50 Mabeta New layout Unknown – – –

51 Makuka Unknown – – –

52 Bonduma 14-Jul-06 x – 1 – 4 –

53 Balondo hill Unknown x – – – –

54 Unity quarter 29-Jun-09 – 1 – – –

55 Unity quarter 29-Jun-09 – 1 – 2 –

56 Unity quarter 29-Jun-09 – 1 – – –

57 Lifanda South 29-Jun-09 – 1 – – –

58 Moliwe 29-Jun-09 x – – – – –

59 Moliwe 29-Jun-09 – – – – –

60 Mile one Limbe 29-Jun-09 x – – – – –

61 Kie Village 06-Aug-09 x x x 1 1 Electricitysupplyinterruptedand watertankpartiallyburied

62 Kie Village 2001/06/27 – –

60 Nat Hazards (2011) 59:47–74

123

rocks. The slide trail at the time of observation had well-developed drainage corridors

separated by longitudinal ridges 30–90 cm high and about a 1 or 2 m wide and parallel to

the slide margin.

The local population in Limbe has increased significantly (from a population of 44,561

in 1987 to 84,223 inhabitants in 2005 implying a growth rate of 3.4% for the Limbe

municipality (Bureau Centrale des Recensements et des Etudes de Population 2010). In the

Mabeta New Layout and Unity Quarter areas, characterized by steep slope ([30) and

pyroclastic material, construction works are usually not associated with any formal

Fig. 5 Four translational landslides on a degraded pyroclastic cone at Mabeta. Photograph was acquired in2005, 4 years after the landslide took place. The largest of the scar is already partially revegetated (courtesyTytgat Nele 2008). Note the presence of building on the foot of this slope and changing vegetation cover onthe slope

Fig. 6 Sketch of slide 4 at Mabeta showing characteristics at the time (in year 2009) of observation

Nat Hazards (2011) 59:47–74 61

123

stabilization measures. Individual simply cut little terraces in a haphazard manner to

provide room for construction (Fig. 7) and also cut down trees for crop cultivation. In the

last 20 years, individuals have moved further up the hills around the Mabeta New Layout

to elevations of *120 m a.s.l. to cultivate yams, cassava, and maize. This has increased

the slope length void of vegetation and as such large areas initially covered by primary

forest have been striped of its vegetation and converted into farmland and secondary forest

made of fruit trees, and wild palms. With these observations, it is possible that anarchical

construction, excavation of foot slopes, steep slopes, and deforestation were the main

conditioning factors, whereas intense prolonged rainfall for a day or so and associated soil

saturation were the main triggers of the 2001 slides at Mabeta.

4.2 Makuka debris slide

On the same day in June 2001, several slides occurred at on slopes in Makuka, a quarter in

Limbe, one of these is described in detail. The debris slide occurred on a 26� natural slope

developed on weathered basaltic flow within a secondary forest characterized by the

presence of fruit trees located away from built up areas. It has a width of 55 m, length of

52 m, and a height of 3 m giving an affected area of 2.8 9 103 m2 and a volume of

7.6 9 102 m3. Sliding material is composed of dark porphyritic basaltic rock fragments

and clays. These fragments are a mixture of rounded, sub-angular and angular, partially or

completely weathered, porphyritic basaltic blocks that range in size from a few mm to over

1 m across. The scarp and margins are sharp, and the east margin is characterized by the

presence of 3 curved successive concentric tension cracks. The slide terminates along a

stream channel flowing on jointed, dense, porphyritic basalts. These have diverse orien-

tation with a large fraction between N30� and N60� and N120�–N150�E. At the time of

observation, debris was still visible within the stream channel though some had been

washed away by the stream. Evidence of slope undercutting and material erosion was

observed, indicating that stream undercutting played a role in the occurrence of the

landslide.

Fig. 7 Close view of anarchical construction at Mabeta. Terrace cut into slope to provide space forconstruction without any stabilization measure put in place. Excavated material pile downslope and is apotential tread to downslope located buildings

62 Nat Hazards (2011) 59:47–74

123

4.3 Bonjo earth slide

On the July 21, 2005, a rotational earth landslide occurred along a road cut in Bonjo, a

small locality within the Limbe Municipality. It blocked the lone unpaved road linking the

Military camp of Man ‘‘O’’ war Bay, the Limbe 3 Council area and the town of Limbe

(Fig. 8a) for 2 days before the debris was excavated and the road reopened. At Krater

located *4.5 km away from Bonjo, 120 mm of rain in 24 h was recorded . This slide

occurred on a 22� slope and has well-defined scarp and margins. The crown is charac-

terized by a crescent shaped tension crack, 0.3–1 m wide and *1 m deep. The scar is 25 m

wide, 24 m long with a scarp height (h) of 2.8 m, resulting in an estimated volume of

103 m3 and a total run-out distance of 35 m. Rock clasts range from a few mm to over 1 m

and make up less than 10% of the debris material. Prominent phenocrysts observed in hand

specimens are plagioclase and pyroxenes. The phenocryst makes up about 20% of the rock

mass.

Debris moved from the road itself was piled up on the right side of the road and later

remobilized by intense rainfall that followed a few days after the debris was clear off the

road, causing significant damage to buildings located tens of meters downslope such that

*80 inhabitants were rendered homeless for *6 months. A key additional finding of

crucial relevance for future interventions when clearing a landslide site and for local risk

awareness raising is just how hazardous it can be, to leave much of a former slide heap by

the road side in the vicinity of downslope-located buildings, as illustrated here.

Between June and August 2008, the north-eastern edge of the slide was reactivated

(Fig. 8b) moving the scarp backward by 5 m, giving it a retrogressive character. H

increased to 3.2 m, and the crown still shows the presence of tension cracks and soil

pinnacles. Renewed sliding present rotational and flow type failure characteristics with a

total run-out distance of 24 m. With the paving of the Limbe-Man ‘O’ War bay road at the

end of 2009, a retaining wall was constructed across the slide. Construction works exposed

a beautiful section of the slide (Fig. 9). From the section, it is observed that rocks that

make up the slope are intensely fractured, weathered and show heterogeneous weathering

patterns. Thickness of the soil layer around the Bonjo area measures over 10 m with pale

yellow clays underlain by purplish saprolite (weathered material that still maintain

characteristics of the parent rock while soil refers to weathered material in which all

characteristics of the initial parent have been lost).

4.4 Debris slide at Unity quarters

On June 29–30, 2009, three debris slides were recorded at Unity quarters, a small locality

within the Limbe municipality. All three slides are shallow and translational in nature.

Slide width range from 4 to 9 m, length from 7 to 10 m. Cumulatively, they generated

*90 m3 of debris. One of these slides, though small (9 9 1 9 6 m), caused the collapse

of a house wall, which killed two children in their sleep and injured their mother. The cost

of repairs was evaluated at *4 million FCFA (*6,000 Euros; i.e. equivalent to the yearly

incomes of about 5 subsistence agric farmers in the area). This illustrates the high human

and economic impact, on already poor and vulnerable persons despite their small size and

volume.

These slides occurred on a 63–64� artificial slopes that range from 5 to over 10 m in

height created by the excavation of slope material for construction purpose. These slopes

are made up of loose partially weathered pyroclastic materials with no stabilization

measures implemented. The debris slides were associated with torrential rains.

Nat Hazards (2011) 59:47–74 63

123

Eyewitnesses reported that the rains were so heavy that the distress cries of the victims

were not heard by neighbors only 10 m away. 400 mm of rain in 1 day was recorded at

Krater located *3 km away. From field observations made 2 weeks after the event and

from eyewitness reports, anarchical construction on steep slopes and slope excavation are

Fig. 8 a Field view of the 26th September 2005 scar at Bonjo. Debris from the slide completely sealed offthe unpaved road. The road did not move implying the rupture surface lies above the road. Photographs weretaken after the debris was excavated to open up the road. Excavated debris in the foreground. (CourtesyTytgat Nele 2008). b 2008 reactivation of the 2005 slide at Bonjo. Scarp height increased from 2.8 to 3.2 m.Tension cracks are still visible at the crown area. Scarp is characterized by jointed basaltic blockssandwiched between clays

64 Nat Hazards (2011) 59:47–74

123

possibly the main conditioning factors. Figure 10 represents a sketch of land-use types

observed at Unity quarter.

4.5 Earth slide at Moliwe

Two shallow translational slides on weathered basaltic rocks were reported at Moliwe,

*4.4 miles (7 km) NE of Unity Quarter on the night of June 29–30, 2009 (Fig. 11). This

site was initially occupied by oil palm tree plantations that were cut down in April 2009 to

prepare the land parcel for replanting of new palms. At the time of observation, the area

was covered with climbers (cover crop) (Fig. 11) with absolutely no trees. These failures

according to Cruden and Varnes (1996)’s classification are earth slides involving colluvial

soil developed from the weathering of basaltic lava flows. It is 21 m wide and 61 m long,

with total run-out of 83 m. Sliding was initiated on a 26� slope and has a scarp height of

2.3 m, and the failure plane lies along the soil/saprolite boundary. The estimated volume of

material that slid is 1.5 9 103 m3. Both earth slides (the larger of which is described

above) located *100 m from each other terminated along the Moliwe Stream. Significant

evidence of flooding was observed (i.e. bent vegetation and deposited debris, an indication

that the stream occupied more space than its normal flow channel). Rainfall data from

Moliwe show that 64 mm was recorded on the day the landslide occurred and a total of

308 mm of rain fell in the vicinity in the 7 days preceding the event. These slides did not

cause any significant damage but illustrate that deforestation and stream undercutting are

important predisposing factors for the occurrence of this slide.

4.6 Earth slide at Kie Village

At *3 pm on August 6, 2009, an earth slide occurred at Kie Village (Fig. 12). The slide is

a rotational earth slide that initiated on a 34� slope. It involves the movement of reddish

brown colluvial soils developed from the weathering of thinly bedded pyroclastic material.

The thickness of the soil layer observed at the head scar is over 7 m.

Fig. 9 Profile through Bonjo slide exposed by road construction in 2009. Note the presence of fractures,weathering heterogeneity, and thickness of soil column

Nat Hazards (2011) 59:47–74 65

123

The slide geometry is as follows: 53, 36, and 6 m for the depletion zone’s width, length,

and height, respectively. Total run-out was 138 m. Approximately 6 9 103 m3 of soil was

moved by the sliding process. The left margin of the scarp shows the presence of acute

tension cracks with a 0.5-m displacement of the downthrown block. This slide blocked the

road linking Bota and Idenau, partially buried a 300-m3 water tank that supplies potable

water to local residents, injured one person and partially buried his car, and interrupted

electricity supply for about 2 days. Eyewitnesses reported an increase in the amount of soil

washed away by runoff and anomalous muddy runoff at the base of the slope a few hours

Fig. 10 Sketch of characteristic land-use pattern observed at Unity quarter

Fig. 11 Field view of 2009 slide at Moliwe with indication that stream undercutting and deforestation areresponsible for some of the slide occurrence in the Limbe area

66 Nat Hazards (2011) 59:47–74

123

before sliding indicating incipient failure prior to the major collapse. They also reported

that movement was rapid and came in two batches, the second one occurring *10 min

after the first one.

During field visits to the site prior to the collapse, the slope was observed to have been

cut to construct a dirt track to the water tank. In addition, pouzzolane excavation was

initiated along the dirt track cut into the pyroclastic cone in April 2009, producing an 8–12-

m-high near-vertical wall. From August 5–6, 2009, *145 and 65 mm, respectively, of

rain, were recorded at the Krater rain gauge station, located only 400 m away from the

slide location. A total of 568 mm of rain was recorded in the 7 days preceding the event.

Weathering, slope undercutting to construct a road to the water tank, and the excavation of

pouzzalane could be the conditioning factors here.

5 Discussion and perspectives

5.1 Analysis of the quantitative geometry of landslide

The geometry of the slide depletion zones vary widely throughout the study area. This

result is similar to that observed by Knapen et al. (2006) on the foot slopes of Mt Elgon in

Uganda. According to Fell (1994) classification based on volume, the landslides in this

study can be classified as extremely small, very small, and small slides with volumes that

range from a few m3 to over 5 9 104 m3. The largest of the slides (Mondoli landslide, 27

June 2001) accounts for 45% the total volume of displaced material within the study area.

Approximately 96% of the observed slides were \50 m wide (Fig. 13). The geometry of

slide is not directly proportional to the amount of mage caused, but is closely linked to the

location of the slide with respect to human infrastructure. For example, one of the slides at

Unity quarter with only about 90 m3 of debris killed 2 persons and partially damaged a

house, while larger slides with larger debris volumes located at Moliwe or Makuka only

Fig. 12 View of a rotational landslide at Kie Village developed on weathered pyroclastic conecharacterized by reddish to reddish brown soils. Green structure at the center left edge of the picture is a300-m3 water tank partially buried by slide debris

Nat Hazards (2011) 59:47–74 67

123

resulted in environmental damage and economic damage because these areas area sparsely

populated. Notwithstanding, there are no standard measuring tools to evaluate the mag-

nitude or impact of individual landslides.

A systematic trend in the frequency of landslide size and volume is highlighted when

plotting the volume or area of each slide on a logarithmic scale, after ranking the slides

from the smallest to the largest. As shown on Fig. 14, such graph displays a straight line,

with only the smallest and largest slides offset from this general trend. The ranked slides

can be interpreted as a proxy for the cumulated frequency of slide sizes. It, for example,

shows that 75% of the slides have a surface area of\1,000 m2 and volume of\1,700 m3.

Only 10% of the slides have a surface area [1,800 m2 and a volume [4,000 m3. The

breakdown of the linear relationship between the log of the volume and the cumulative

frequency is to be attributed to the non-exhaustive documentation of the smallest and

largest slides, respectively. The largest Mondoli slide is significantly offset from the fre-

quency–size relationship, due to its exceptional depth. Such slides might result from a

different sliding mechanism or must be quite infrequent, resulting in an under represen-

tation in the datasets, as landslides scar get rapidly masked by vegetation in the study area.

This semi-logarithmic relationship between the cumulative frequency and the volume of

the slide is of great interest to assess the probability of occurrence of slide of different

volume in the future. Such relationship is similar to that obtained from geometrical datasets

for other landslides (Hovius et al. 1997; Malamud 2004a, b; Brunetti et al. 2009; Guzzetti

et al. 2009).

Slope gradient at which slides developed on pyroclastic material is generally steeper

ranging from 32 to 45� on natural slope and can go well beyond 63� for slopes with human

intervention. Slides on basaltic flow initiate at lower slope angles (22–36�) except in area

Fig. 13 Graphical presentation of the distribution of the width of observed slides. Most of the slides are lessthan 50 m wide

68 Nat Hazards (2011) 59:47–74

123

where steep slope are generated as a result of road construction and/or building con-

struction where slide can occur on slope greater than 36� (Table 1). The width and depth of

failure does not vary with parent rock type.

Two tail t-test for the significance of the correlation coefficients between the different

geometric parameters (Table 3) shows that there exists a strong positive correlation

between slides’ volume and area (r (Pearson’s correlation coefficient) value of 0.96),

between length and area (r = 0.81), and between slides’ width and area (r = 0.73). All

these correlations are significant at 0.01 confidence interval. These relations suggest that

variation in aerial and volumetric dimension of the depletion zone is generally controlled

by the width and length of the failure and to a lesser extent by its depth. This is opposed to

what exist in the foot slope of Mt Elgon in Uganda, where Knapen et al. (2006) showed

that variation in volume is mostly controlled by the depth of the shear plane. Secondly, the

volume–area correlation suggests that the height of the scarp (depth of failure) does not

vary greatly within the region. The run-out distance shows a moderate but significant

positive correlation with the volume (r = 0.67), implying that the distance over which the

material travel is dependent on the volume of generated debris. There is also a moderate

correlation between the length of the depletion zone and the run-out distance of the slides,

suggesting that slides with long depletion length will also have a longer run on distance.

There is a weak but significant negative correlation between the initial slope angle and the

length of the landslide scar with longer scars observed on gentler slope and shorter length

Fig. 14 Frequency/size distribution of individual landslides in the Limbe area. In the graphs, x-axes showsrank order, from largest to smallest landslide, a y-axes show landslide area, in square meters. b landslidevolume in cubic meters

Nat Hazards (2011) 59:47–74 69

123

on steep slopes. This relationship can be explained by the idea that as the slope becomes

steeper, the distance over which material will move is shorten. A weak significant negative

correlation exists between the width/length ratios and slope gradient (at a 0.05 confidence

level). A moderate but significant positive correlation exists between the scar height and

the run-out distance as well as the length of the depletion zone.

The slip surface in most of the slides in this area is not controlled by a saprolite-fresh

rock boundary as noted in other studies (Wen et al. 2007), but occurs within the saprolite or

at the soil-saprolite boundary probably as a result of heterogeneity imposed by weathering

as reported by Ngole et al. (2007) for the Mabeta area. Deep soil and underlying saprolite

result from intense and prolonged weathering governed by voluminous rainfall and high

temperatures of approximately 26�C.

According to Cruden and Varnes (1996)’s landslide classification based on the type of

material involved in sliding and the sliding mechanism/mode of failure, most of the slide

are shallow translational (96%) and rotational (4%) earth and debris slides. Two of these

slides initiated as shallow translational slides and then transformed into mud flows with

debris thickness above 2.5 m probably as a result of excess water supplied by intense rain.

Measured widths are approximately equal to the length of the depletion zone for rotational

slides observed in this study, while the length is greater than the width for the translational

slides, reason why the slide was approximated to be elliptical. From Terzaghi et al.

(1996)’s classification scheme based on where the slip surface intersects the slope, all the

slides were classified as slope failure because the slip surface intersects the slope above its

base (foot).

5.2 Causal factors

Observed slides can be grouped into six landslide zones characterized based on rock type,

landuse pattern, slope gradient, and soil type. The Mabeta and Kie landslide zones (Fig. 4)

are characterized by the presence of pyroclastic materials, reddish to reddish brown soils

with low permeabilities, sparse vegetation, and slopes with gradient above 32�, with

intense human activity such as construction of houses, foot parts, and roads. The pyro-

clastic slopes are rendered less stable by human intervention especially in cases where

these activities generate artificial slopes exceeding the angle of repose for the loose

granular pyroclastic material which is *30 to 40� depending on grain size and sorting of

Table 3 Correlation matrix of geometric properties of landslides observed in Limbe

Elevation(m)

Width ofscarp (m)

Height ofscarp (m)

Length(m)

Initialslope

Run-outdistance (m)

Volume(m3)

Area(m2)

Elevation (m) 1

Width of scarp (m) -.16 1

Height of scarp (m) -.311* .54** 1

Length (m) -.04 .55** .34** 1

Initial slope -.10 -.26 -.02 -.44** 1

Run-out distance (m) -.15 .49** .47** .66** -.25 1

Volume (m3) -.08 .65** .55** .66** -.16 .70** 1

Area (m2) -.08 .73** .54** .81** -.27 .75** .96** 1

* Correlation is significant at the 0.05 level (2-tailed)

** Correlation is significant at the 0.01 level (2-tailed)

70 Nat Hazards (2011) 59:47–74

123

the material (Riedel et al. 2003). These pyroclastic cones, typically exploited as aggregates

for roads construction and repair or to make concrete blocks, are made up of loose

mechanically weak particles that are very porous, highly permeable and stable in the dry

and natural state. Stability comes from the fact that they were formed from the deposition

of volcanic ejecta such that the particles were laid down at their angle of repose controlled

by the degree of grain-grain chaining between adjacent highly irregular-shaped particles

(Riedel et al. 2003). Human intervention on these steep slopes by excavation caused the

slopes to be steeper than the angle of repose, making them susceptible to instabilities

particularly after prolonged rains.

The Chopfarm, Bonjo, and Makuka Landslide zones (Fig. 3) are located on weathered

lava flows and characterized by secondary forest and/or farm land, gentler slopes and

sparsely inhabited. Excavation activities and other human influences on slopes are less

common when compared to the Mabeta zone, to the notable exception of slides in Bonjo,

which were directly associated with the road cut and construction works, whereas those at

Makuka occur in a more or less natural environment. The Makuka zone is characterized by

high stream density, and most of the slides here terminate along stream channels. Stream

undercutting is likely a more important factor here.

The Moliwe landslide zone lies within industrial palm plantations grown on pale yellow

soils developed from the weathering of basaltic rocks and lahar deposits. It is also sparsely

populated and characterized by gentler slopes. Slides in this zone occur close to streams,

paved and unpaved roads used by CDC tractors and particularly in areas recently

deforested.

In the study area, the construction of houses, roads and foot paths particularly at Unity

quarter, Bonjo, and Mabeta involves the excavation of soil and the creation of small

terraces. This removal changes the slope line, the angle of repose of particles located

upslope, causing water stagnation on the flattened areas and increased infiltration. The

excavated soil in most cases is piled on the downslope end of individual land parcels such

that with intense rain, the loosen soil rapidly absorbs water, becomes saturated, and slides

causing damage to downslope located structures.

Furthermore, high mean annual precipitation that range from *2,100 to 4,600 mm

(CDC meteorological center) concentrated in 3 of an 8-month rainy season results in high

soil moisture contents and soil saturation for long periods of the year, and this can be

considered as one of the causal factors.

5.3 Triggering factors

Analyses of 20–34 years of monthly rainfalls suggest that landslide activities in Limbe are

associated with extreme rainfall events. A threshold for the initiation of slide is difficult to

define because of high spatio-temporal variation in intensity and duration of rainfall within

the area, and of the absence of records of the exact date and time for most of the landslides.

Hence, the link between landslides and specific rainfall events cannot be precisely defined.

However, most of the slides with known dates suggest that they are associated with rainfall

[110 mm in 24 h preceded by 2–4 days with no or limited rainfall (0–4 mm). Other heavy

rainfall events, with daily rain amount[110 and as great as 250 mm particularly preceded

by day with more intense rains, are however not associated with landslide events. Lumb

(1975) suggested a relationship between the preceding 15 days and 24 h rainfall total and

landslide occurrence. However, Brand et al. (1984) argues that in most tropical residual

soils, antecedent rainfall is not a controlling factor in slide initiation. Instead, it seems

reasonable to assume that the likelihood of landslide occurrence is enhanced subsequently

Nat Hazards (2011) 59:47–74 71

123

to periods of sustained rainfall. These observations are true for most of the landslide with

known dates in the Limbe area.

Nevertheless, a successful monitoring of small landslide hazard in the region requires a

dense network of rainfall stations recording rainfall intensity and duration continuously.

Such information needs to be available in real time for scientists to enable them to define

absolute rainfall event susceptible to triggering landslides.

Buh (2009) analyzes seismic activities as possible trigger for the June 2001 landslides in

Limbe and concludes that June 2001 was characterized by 0–8 event low-magnitude events

per/day. Thus, because of the lack of increased seismicity before and/or on the 27th of June

2001, seismicity did not seem to have been a major triggering mechanism. It can thus be

concluded that landslide occurrence within the study area is not linked to a single factor but

occurs in response to an interplay of several preparatory factors with rainfall as probable

trigger. These results are similar to those reported by Kitutu et al. (2009) for the Bududa

district, Eastern Uganda. In addition, the severity/magnitude/nature of the impact of a slide

depends on its location relative to human infrastructural works rather than its type and size.

For example, a small 9 9 1 9 6 m slide at Unity quarter resulted in the loss of two lives

and property damage estimated at *4 million FCFA (*6,000 Euros), whereas a

21 9 2.3 9 62 m slide at Moliwe resulted in severe local environmental damage but did

not lead to any casualty or property damage.

6 Conclusion

From field observations, eyewitness accounts, and data interpretation, the following con-

clusions can be drawn from this field study:

1. Landslides in Limbe are small to very small translational and/or rotational landslides

with mean width of 24 m. Large landslides ([104 m3) are rare and might be triggered

by other processes than intense rainfall. Importantly, our observations highlight the

lack of correlation between landslide size and their impact, which is mostly controlled

by the proximity to vulnerable infrastructures and populations.

2. Landslide occurrence in the Limbe region results from a combination of factors such

as the presence of steep slopes, of pyroclastic material, of thick soil cover, or the

proximity to stream channels. Landslides are especially frequent on old pyroclastic

cones that have undergone significant weathering.

3. Intense and prolonged rainfall ([110 mm in 24 h preceded by 2–4 days with no or

limited rainfall (0–4 mm)) act as the major trigger that initiates failure.

4. Slide occurrence is exacerbated by human interferences in the form of urban

expansion, anarchical construction, slope excavation, and deforestation.

When deciding on the localization of new development projects, it is essential to pay

attention to slope stability issues in order to mitigate potential losses due to landslides. In

addition, construction works on steep slopes especially on the loose mechanically weak

pyroclastic cones should be discouraged. If unavoidable, adequate retention walls, drainage

paths, and slope reinforcement measures should be implemented and maintained to limit

damage resulting from sliding.

This study thus provides important new insights and quantitative constraints to be used

in deterministic modeling of volume-limited slides. It will also serve as a basis to con-

straint a landslide susceptibility assessment based on the identified causative factors. It also

provides data that can be subsequently used in the development and evaluation of slope

72 Nat Hazards (2011) 59:47–74

123

instability mechanisms for particular sites considered at risk in the Limbe region of

Cameroon, or other areas in a similar context where steep weathered volcanic terrains

receive intense and prolonged rainfall in the subtropics worldwide.

Acknowledgments This work was compiled as part of CVB’s PhD thesis sponsored by a grant from theVLaamse Iinter-Universitaire Raad (Flanders, Belgium) in the framework of the project entitled ‘Geo hazardmonitoring within volcanically active S. W Cameroon’. GGJE is supported through postdoctoral fellowshipsfrom the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen. We gratefully acknowledge the CameroonDevelopment Corporation, Tiko, for providing rainfall data for the study area and thank all local people whoprovided crucial eyewitness reports as well as the local authorities who are encouraging this effort andclosely cooperating with us especially Mr. Matute, Major of Limbe I Council. Contribution from J. Nyssenwho reviewed a preliminary draft of this manuscript is heartily acknowledged. The reviews and commentsof two anonymous reviewers have greatly improved the manuscript.

References

Ahmad R, McCalpin JP (1999) Landslide susceptibility maps for the Kingston metropolitan area, Jamaicawith notes on their use. USD publication no. 5

Ayalew L, Yamagishi H (2004) Slope failure in the Blue Nile Basin, as seen from landscape evolutionperspective. Geomorphology 57:95–116

Ayanji EN (2004) A critical assessment of the natural disaster risk management framework in Cameroon.Unpublished end of course training report CEFAM, Buea, 23 p

Ayonghe SN, Ntasin EB (2008) The geological control and triggering mechanism of landslides of the 20thJuly 2003 within the Bamboutos Caldera, Cameroon. J Cameroon Acad Sci 7(3):191–203

Ayonghe SN, Mafany GT, Ntasin EB, Samalang P (1999) Seismically activated swarms of landslides,tension cracks and rockfall after heavy rainfall in Bafaka, Cameroon. Nat Hazards 19:13–27

Ayonghe SN, Suh CE, Ntasin EB, Samalang P, Fantong W (2002) Hydrologically, seismically and tec-tonically triggered landslides along the Cameroon Volcanic Line. Cameroon Geosci Rev19(4):325–335

Ayonghe SN, Ntasin EB, Samalang P, Suh CE (2004) The June 27, 2001 landslide on volcanic cones inLimbe, Mount Cameroon, West Africa. J Afr Earth Sci 39:435–439

Brand EW, Premchitt J, Phillipson HB (1984) Relationship between rainfall and landslides. In: Proceedingsof the 4th international symposium on landslides, vol 1, Toronto, ON, 16–21 Sept 1984. BiTechPublishers, Vancouver, BC, pp 377–384

Brunetti MT, Guzzetti F, Rossi M (2009) Probability distribution of landslide volume. Non linear Process16:179–188

Buea, Carte provisoire de L’Afrique centrale au 1/50000 Republique Federale du Cameroun feuille NB-32-IV-IB dessine et publie par L’institut Geographique National—Paris (Annexe de Yaounde) en 1963

Buh WG (2009) Geographic information systems based demarcation of risk zones: the case of the LimbeSub-Division-Cameroon. J Disaster Risk Stud 2(1):54–70

Bureau Central des Recensements et des Etudes de Population (2010) Rapport de presentation des resultatsdefinitifs, 68 p

Carrara A, Guzzetti F, Cardinali M, Reichenbach P (1999) Use of GIS technique in the prediction andmonitoring of landslide hazard. Nat Hazards 20:117–135

Corominas J (1996) The angle of reach as mobility index for small and large landslides. Can Geotech J33:206–271

Cruden DM, Varnes DJ (1996) Landslide types and processes, Chap. 3 in landslides: investigations andmitigation. In: Turner AK, Schuster RL (eds) Transportation Research Board, Special Report 247,pp 36–75

Dahl MPJ, Mortensen LE, Veihe A, Jensen NH (2010) A simple qualitative approach for mapping regionallandslide susceptibility in the Faroe Islands. Nat Hazards Earth Syst Sci 10:159–170

Dai FC, Lee CF (2002) Landslides characteristics and slope instability modelling using GIS, Lantau Island,Hong Kong. Geomorphology 42:213–228

Davies MR., Paulen RC, and Hickin AS (2005) Inventory of Holocene landslides, Peace River Area, Alberta(NTS 84C): Alberta Energy and Utilities Board, EUB/AGS Geo-Note 2003-43

Dikau R, Brunsden D, Schrott L, Ibsen M-L (ed) (1996) Landslide recognition: identification, movementand courses. Wiley, 251 p

Nat Hazards (2011) 59:47–74 73

123

Endeley RE, Ayonghe SN, Tchuenteu F (2001) A preliminary hydrogeochemical baseline study of watersources around Mount Cameroon. J Cameroon Acad Sci 1(3):161–168

Fell R (1994) Landslide risk assessment and acceptable risk. Can Geotech J 31:261–272Guthrie RH, Evans SG (2004) Analysis of landslide frequencies and characteristics in a natural system,

coastal British Columbia. Earth Surf Process Landf 29:1321–1339Guzzetti F, Carrara A, Cardinalli M, Reichenbach P (2003) Landslide hazard evaluation a review of current

techniques and their application in a multi-scale study. Geomorphology 31:181–216Guzzetti F, Ardizzone F, Cardinalli M, Rossi M, Valigi D (2009) Landslide volume and landslide mobi-

lization rates in Umbria, central Italy. Earth Planet Sci Lett 279:222–229Hasselo HN (1961) The soils of the lower eastern slopes of the Cameroon Mountain and their suitability for

various perennial crops, Wageningen, 69 pHovius N, Stark CP, Allen PA (1997) Sediment flux from a Mountain belt derived by landslide. Geology

25(3):231–234Kitutu MG, Muwanga A, Poesen J, Deckers JA (2009) Influence of soil properties on landslide occurrences

in Bududa district, Eastern Uganda. Afr Agric Res 4(7):611–620Knapen A, Kitutu MG, Poesen J, Breugelmans W, Deckers J, Muwanga A (2006) Landslides in a densely

populated county at the foot slopes of Mount Elgon (Uganda): characteristic and causal factors.Geomorphology 73:149–165

Lambi CM, Ngwana BS (1991) Human interference and environmental instability. The case of the Limbelandslide. Cameroon Geogr Rev 44–52

Lumb P (1975) Slope failures in Hong Kong. Q J Eng Geol 8:31–65Malamud BD, Turcotte DL, Guzzetti F, Reichenbach P (2004a) Landslide, earthquake and erosion. Earth

Planet Sci Lett 229:45–59Malamud BD, Turcotte DL, Guzzetti F, Reichenbach P (2004b) Landslide inventory and their statistical

properties. Earth Surf Process Landf 29:687–711Ngole VM, Ekosse GE, Ayonghe SN (2007) Physico-chemical, mineralogical and chemical consideration in

understanding the 2001 Mabeta New Layout landslide, Cameroon. J Appl Sci Environ Manage11(2):201–208

Riedel C, Ernst GGJ, Riley M (2003) Control on the growth and geometry of pyroclastic constructs.J Volcanol Geotherm Res 127:121–152

Suh CE, Sparks RSJ, Fitton JG, Ayonghe SN, Annen C, Nana R, Luckman A (2003) The 1999 and 2000eruptions of Mount Cameroon: eruption behaviour and petrochemistry of lava. Bull Volcanol65:267–281

Suzen ML, Doyuran V (2004) Data driven bivariate landslide susceptibility assessment using geographicalinformation systems: method and application to the Asarasuyu catchment, Turkey. Eng Geol71:303–321

Terzaghi K, Peck RB, Mesri G (1996) Soil mechanics in engineering practice, 3rd edn. Wiley, ChichesterThierry P, Stieltjes L, Kouokam E, Ngueya P, Salley P (2008) Multi-hazard risk mapping and assessment on

an active volcano: the GRINP project at Mount Cameroon. Nat Hazards 45(3):429–456Tytgat Nele JL (2008) Susceptibility analysis of the risk of small landslides on the southern slope of Mount

Cameroon Volcano. Unpublished MSc thesis, Ghent University, 132 pVarnes DJ (1978) Slope movement types and processes. In: Schuter RT, Krizek RJ (eds) Special report 176,

Landslide Analysis and Control National Research Council, Washington DC, pp 11–33Wen BP, Ayin A, Duzgoren-Aydin A (2004) Geochemical characteristics of the slip zones of a landslide in

granitic saprolite, Hong Kong: implications for their development and microenvironments. EnvironGeol 47:140–154

Wen BP, Ayin A, Duzgoren-Aydin NS, Li YR, Chen HY, Xiao SD (2007) Residual strength of slip zones oflarge landslides in the three gorge area, China. Eng Geol 93:82–98

Zogning A, Ngouanet C, Tiafack O (2007) The catastrophic geomorphological processes in humid tropicalAfrica: a case study of the recent landslide disasters in Cameroon. Sed Geol 199(1–2):13–17

74 Nat Hazards (2011) 59:47–74

123